linux_dsm_epyc7002/drivers/gpu/drm/drm_dp_mst_topology.c

4408 lines
121 KiB
C
Raw Normal View History

drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
/*
* Copyright © 2014 Red Hat
*
* Permission to use, copy, modify, distribute, and sell this software and its
* documentation for any purpose is hereby granted without fee, provided that
* the above copyright notice appear in all copies and that both that copyright
* notice and this permission notice appear in supporting documentation, and
* that the name of the copyright holders not be used in advertising or
* publicity pertaining to distribution of the software without specific,
* written prior permission. The copyright holders make no representations
* about the suitability of this software for any purpose. It is provided "as
* is" without express or implied warranty.
*
* THE COPYRIGHT HOLDERS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS SOFTWARE,
* INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO
* EVENT SHALL THE COPYRIGHT HOLDERS BE LIABLE FOR ANY SPECIAL, INDIRECT OR
* CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE,
* DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER
* TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE
* OF THIS SOFTWARE.
*/
#include <linux/delay.h>
#include <linux/errno.h>
#include <linux/i2c.h>
#include <linux/init.h>
#include <linux/kernel.h>
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
#include <linux/sched.h>
#include <linux/seq_file.h>
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
#include <drm/drm_atomic.h>
#include <drm/drm_atomic_helper.h>
#include <drm/drm_dp_mst_helper.h>
#include <drm/drm_drv.h>
#include <drm/drm_print.h>
drm: Split out drm_probe_helper.h Having the probe helper stuff (which pretty much everyone needs) in the drm_crtc_helper.h file (which atomic drivers should never need) is confusing. Split them out. To make sure I actually achieved the goal here I went through all drivers. And indeed, all atomic drivers are now free of drm_crtc_helper.h includes. v2: Make it compile. There was so much compile fail on arm drivers that I figured I'll better not include any of the acks on v1. v3: Massive rebase because i915 has lost a lot of drmP.h includes, but not all: Through drm_crtc_helper.h > drm_modeset_helper.h -> drmP.h there was still one, which this patch largely removes. Which means rolling out lots more includes all over. This will also conflict with ongoing drmP.h cleanup by others I expect. v3: Rebase on top of atomic bochs. v4: Review from Laurent for bridge/rcar/omap/shmob/core bits: - (re)move some of the added includes, use the better include files in other places (all suggested from Laurent adopted unchanged). - sort alphabetically v5: Actually try to sort them, and while at it, sort all the ones I touch. v6: Rebase onto i915 changes. v7: Rebase once more. Acked-by: Harry Wentland <harry.wentland@amd.com> Acked-by: Sam Ravnborg <sam@ravnborg.org> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jani Nikula <jani.nikula@linux.intel.com> Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Acked-by: Rodrigo Vivi <rodrigo.vivi@intel.com> Acked-by: Benjamin Gaignard <benjamin.gaignard@linaro.org> Acked-by: Jani Nikula <jani.nikula@intel.com> Acked-by: Neil Armstrong <narmstrong@baylibre.com> Acked-by: Oleksandr Andrushchenko <oleksandr_andrushchenko@epam.com> Acked-by: CK Hu <ck.hu@mediatek.com> Acked-by: Alex Deucher <alexander.deucher@amd.com> Acked-by: Sam Ravnborg <sam@ravnborg.org> Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Acked-by: Liviu Dudau <liviu.dudau@arm.com> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Cc: linux-arm-kernel@lists.infradead.org Cc: virtualization@lists.linux-foundation.org Cc: etnaviv@lists.freedesktop.org Cc: linux-samsung-soc@vger.kernel.org Cc: intel-gfx@lists.freedesktop.org Cc: linux-mediatek@lists.infradead.org Cc: linux-amlogic@lists.infradead.org Cc: linux-arm-msm@vger.kernel.org Cc: freedreno@lists.freedesktop.org Cc: nouveau@lists.freedesktop.org Cc: spice-devel@lists.freedesktop.org Cc: amd-gfx@lists.freedesktop.org Cc: linux-renesas-soc@vger.kernel.org Cc: linux-rockchip@lists.infradead.org Cc: linux-stm32@st-md-mailman.stormreply.com Cc: linux-tegra@vger.kernel.org Cc: xen-devel@lists.xen.org Link: https://patchwork.freedesktop.org/patch/msgid/20190117210334.13234-1-daniel.vetter@ffwll.ch
2019-01-18 04:03:34 +07:00
#include <drm/drm_probe_helper.h>
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
drm/dp_mst: Enable registration of AUX devices for MST ports All available downstream ports - physical and logical - are exposed for each MST device. They are listed in /dev/, following the same naming scheme as SST devices by appending an incremental ID. Although all downstream ports are exposed, only some will work as expected. Consider the following topology: +---------+ | ASIC | +---------+ Conn-0| | +----v----+ +----| MST HUB |----+ | +---------+ | | | |Port-1 Port-2| +-----v-----+ +-----v-----+ | MST | | SST | | Display | | Display | +-----------+ +-----------+ |Port-1 x MST Path | MST Device ----------+---------------------------------- sst:0 | MST Hub mst:0-1 | MST Display mst:0-1-1 | MST Display's disconnected DP out mst:0-1-8 | MST Display's internal sink mst:0-2 | SST Display On certain MST displays, the upstream physical port will ACK DPCD reads. However, reads on the local logical port to the internal sink will *NAK*. i.e. reading mst:0-1 ACKs, but mst:0-1-8 NAKs. There may also be duplicates. Some displays will return the same GUID when reading DPCD from both mst:0-1 and mst:0-1-8. There are some device-dependent behavior as well. The MST hub used during testing will actually *ACK* read requests on a disconnected physical port, whereas the MST displays will NAK. In light of these discrepancies, it's simpler to expose all downstream ports - both physical and logical - and let the user decide what to use. v3 changes: * Change WARN_ON_ONCE -> DRM_ERROR on dpcd read errors * Docstring and cosmetic fixes v2 changes: Moved remote aux device (un)registration to new mst connector late register and early unregister helpers. Drivers should call these from their own mst connector function hooks. This is to solve an issue during driver unload, where mst connector devices are unregistered before the remote aux devices are. In a setup where aux devices are created as children of connector devices, the aux device would be removed too early, and uncleanly. Doing so in early_unregister solves this issue, as that is called before connector unregistration. Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Signed-off-by: Leo Li <sunpeng.li@amd.com> Reviewed-by: Lyude Paul <lyude@redhat.com> Signed-off-by: Harry Wentland <harry.wentland@amd.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190723232808.28128-3-sunpeng.li@amd.com
2019-07-24 06:28:01 +07:00
#include "drm_crtc_helper_internal.h"
#include "drm_dp_mst_topology_internal.h"
drm/dp_mst: Enable registration of AUX devices for MST ports All available downstream ports - physical and logical - are exposed for each MST device. They are listed in /dev/, following the same naming scheme as SST devices by appending an incremental ID. Although all downstream ports are exposed, only some will work as expected. Consider the following topology: +---------+ | ASIC | +---------+ Conn-0| | +----v----+ +----| MST HUB |----+ | +---------+ | | | |Port-1 Port-2| +-----v-----+ +-----v-----+ | MST | | SST | | Display | | Display | +-----------+ +-----------+ |Port-1 x MST Path | MST Device ----------+---------------------------------- sst:0 | MST Hub mst:0-1 | MST Display mst:0-1-1 | MST Display's disconnected DP out mst:0-1-8 | MST Display's internal sink mst:0-2 | SST Display On certain MST displays, the upstream physical port will ACK DPCD reads. However, reads on the local logical port to the internal sink will *NAK*. i.e. reading mst:0-1 ACKs, but mst:0-1-8 NAKs. There may also be duplicates. Some displays will return the same GUID when reading DPCD from both mst:0-1 and mst:0-1-8. There are some device-dependent behavior as well. The MST hub used during testing will actually *ACK* read requests on a disconnected physical port, whereas the MST displays will NAK. In light of these discrepancies, it's simpler to expose all downstream ports - both physical and logical - and let the user decide what to use. v3 changes: * Change WARN_ON_ONCE -> DRM_ERROR on dpcd read errors * Docstring and cosmetic fixes v2 changes: Moved remote aux device (un)registration to new mst connector late register and early unregister helpers. Drivers should call these from their own mst connector function hooks. This is to solve an issue during driver unload, where mst connector devices are unregistered before the remote aux devices are. In a setup where aux devices are created as children of connector devices, the aux device would be removed too early, and uncleanly. Doing so in early_unregister solves this issue, as that is called before connector unregistration. Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Signed-off-by: Leo Li <sunpeng.li@amd.com> Reviewed-by: Lyude Paul <lyude@redhat.com> Signed-off-by: Harry Wentland <harry.wentland@amd.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190723232808.28128-3-sunpeng.li@amd.com
2019-07-24 06:28:01 +07:00
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
/**
* DOC: dp mst helper
*
* These functions contain parts of the DisplayPort 1.2a MultiStream Transport
* protocol. The helpers contain a topology manager and bandwidth manager.
* The helpers encapsulate the sending and received of sideband msgs.
*/
static bool dump_dp_payload_table(struct drm_dp_mst_topology_mgr *mgr,
char *buf);
static void drm_dp_mst_topology_put_port(struct drm_dp_mst_port *port);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
static int drm_dp_dpcd_write_payload(struct drm_dp_mst_topology_mgr *mgr,
int id,
struct drm_dp_payload *payload);
drm/dp_mst: Enable registration of AUX devices for MST ports All available downstream ports - physical and logical - are exposed for each MST device. They are listed in /dev/, following the same naming scheme as SST devices by appending an incremental ID. Although all downstream ports are exposed, only some will work as expected. Consider the following topology: +---------+ | ASIC | +---------+ Conn-0| | +----v----+ +----| MST HUB |----+ | +---------+ | | | |Port-1 Port-2| +-----v-----+ +-----v-----+ | MST | | SST | | Display | | Display | +-----------+ +-----------+ |Port-1 x MST Path | MST Device ----------+---------------------------------- sst:0 | MST Hub mst:0-1 | MST Display mst:0-1-1 | MST Display's disconnected DP out mst:0-1-8 | MST Display's internal sink mst:0-2 | SST Display On certain MST displays, the upstream physical port will ACK DPCD reads. However, reads on the local logical port to the internal sink will *NAK*. i.e. reading mst:0-1 ACKs, but mst:0-1-8 NAKs. There may also be duplicates. Some displays will return the same GUID when reading DPCD from both mst:0-1 and mst:0-1-8. There are some device-dependent behavior as well. The MST hub used during testing will actually *ACK* read requests on a disconnected physical port, whereas the MST displays will NAK. In light of these discrepancies, it's simpler to expose all downstream ports - both physical and logical - and let the user decide what to use. v3 changes: * Change WARN_ON_ONCE -> DRM_ERROR on dpcd read errors * Docstring and cosmetic fixes v2 changes: Moved remote aux device (un)registration to new mst connector late register and early unregister helpers. Drivers should call these from their own mst connector function hooks. This is to solve an issue during driver unload, where mst connector devices are unregistered before the remote aux devices are. In a setup where aux devices are created as children of connector devices, the aux device would be removed too early, and uncleanly. Doing so in early_unregister solves this issue, as that is called before connector unregistration. Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Signed-off-by: Leo Li <sunpeng.li@amd.com> Reviewed-by: Lyude Paul <lyude@redhat.com> Signed-off-by: Harry Wentland <harry.wentland@amd.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190723232808.28128-3-sunpeng.li@amd.com
2019-07-24 06:28:01 +07:00
static int drm_dp_send_dpcd_read(struct drm_dp_mst_topology_mgr *mgr,
struct drm_dp_mst_port *port,
int offset, int size, u8 *bytes);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
static int drm_dp_send_dpcd_write(struct drm_dp_mst_topology_mgr *mgr,
struct drm_dp_mst_port *port,
int offset, int size, u8 *bytes);
static void drm_dp_send_link_address(struct drm_dp_mst_topology_mgr *mgr,
struct drm_dp_mst_branch *mstb);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
static int drm_dp_send_enum_path_resources(struct drm_dp_mst_topology_mgr *mgr,
struct drm_dp_mst_branch *mstb,
struct drm_dp_mst_port *port);
static bool drm_dp_validate_guid(struct drm_dp_mst_topology_mgr *mgr,
u8 *guid);
static int drm_dp_mst_register_i2c_bus(struct drm_dp_aux *aux);
static void drm_dp_mst_unregister_i2c_bus(struct drm_dp_aux *aux);
static void drm_dp_mst_kick_tx(struct drm_dp_mst_topology_mgr *mgr);
#define DBG_PREFIX "[dp_mst]"
#define DP_STR(x) [DP_ ## x] = #x
static const char *drm_dp_mst_req_type_str(u8 req_type)
{
static const char * const req_type_str[] = {
DP_STR(GET_MSG_TRANSACTION_VERSION),
DP_STR(LINK_ADDRESS),
DP_STR(CONNECTION_STATUS_NOTIFY),
DP_STR(ENUM_PATH_RESOURCES),
DP_STR(ALLOCATE_PAYLOAD),
DP_STR(QUERY_PAYLOAD),
DP_STR(RESOURCE_STATUS_NOTIFY),
DP_STR(CLEAR_PAYLOAD_ID_TABLE),
DP_STR(REMOTE_DPCD_READ),
DP_STR(REMOTE_DPCD_WRITE),
DP_STR(REMOTE_I2C_READ),
DP_STR(REMOTE_I2C_WRITE),
DP_STR(POWER_UP_PHY),
DP_STR(POWER_DOWN_PHY),
DP_STR(SINK_EVENT_NOTIFY),
DP_STR(QUERY_STREAM_ENC_STATUS),
};
if (req_type >= ARRAY_SIZE(req_type_str) ||
!req_type_str[req_type])
return "unknown";
return req_type_str[req_type];
}
#undef DP_STR
#define DP_STR(x) [DP_NAK_ ## x] = #x
static const char *drm_dp_mst_nak_reason_str(u8 nak_reason)
{
static const char * const nak_reason_str[] = {
DP_STR(WRITE_FAILURE),
DP_STR(INVALID_READ),
DP_STR(CRC_FAILURE),
DP_STR(BAD_PARAM),
DP_STR(DEFER),
DP_STR(LINK_FAILURE),
DP_STR(NO_RESOURCES),
DP_STR(DPCD_FAIL),
DP_STR(I2C_NAK),
DP_STR(ALLOCATE_FAIL),
};
if (nak_reason >= ARRAY_SIZE(nak_reason_str) ||
!nak_reason_str[nak_reason])
return "unknown";
return nak_reason_str[nak_reason];
}
#undef DP_STR
#define DP_STR(x) [DRM_DP_SIDEBAND_TX_ ## x] = #x
static const char *drm_dp_mst_sideband_tx_state_str(int state)
{
static const char * const sideband_reason_str[] = {
DP_STR(QUEUED),
DP_STR(START_SEND),
DP_STR(SENT),
DP_STR(RX),
DP_STR(TIMEOUT),
};
if (state >= ARRAY_SIZE(sideband_reason_str) ||
!sideband_reason_str[state])
return "unknown";
return sideband_reason_str[state];
}
static int
drm_dp_mst_rad_to_str(const u8 rad[8], u8 lct, char *out, size_t len)
{
int i;
u8 unpacked_rad[16];
for (i = 0; i < lct; i++) {
if (i % 2)
unpacked_rad[i] = rad[i / 2] >> 4;
else
unpacked_rad[i] = rad[i / 2] & BIT_MASK(4);
}
/* TODO: Eventually add something to printk so we can format the rad
* like this: 1.2.3
*/
return snprintf(out, len, "%*phC", lct, unpacked_rad);
}
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
/* sideband msg handling */
static u8 drm_dp_msg_header_crc4(const uint8_t *data, size_t num_nibbles)
{
u8 bitmask = 0x80;
u8 bitshift = 7;
u8 array_index = 0;
int number_of_bits = num_nibbles * 4;
u8 remainder = 0;
while (number_of_bits != 0) {
number_of_bits--;
remainder <<= 1;
remainder |= (data[array_index] & bitmask) >> bitshift;
bitmask >>= 1;
bitshift--;
if (bitmask == 0) {
bitmask = 0x80;
bitshift = 7;
array_index++;
}
if ((remainder & 0x10) == 0x10)
remainder ^= 0x13;
}
number_of_bits = 4;
while (number_of_bits != 0) {
number_of_bits--;
remainder <<= 1;
if ((remainder & 0x10) != 0)
remainder ^= 0x13;
}
return remainder;
}
static u8 drm_dp_msg_data_crc4(const uint8_t *data, u8 number_of_bytes)
{
u8 bitmask = 0x80;
u8 bitshift = 7;
u8 array_index = 0;
int number_of_bits = number_of_bytes * 8;
u16 remainder = 0;
while (number_of_bits != 0) {
number_of_bits--;
remainder <<= 1;
remainder |= (data[array_index] & bitmask) >> bitshift;
bitmask >>= 1;
bitshift--;
if (bitmask == 0) {
bitmask = 0x80;
bitshift = 7;
array_index++;
}
if ((remainder & 0x100) == 0x100)
remainder ^= 0xd5;
}
number_of_bits = 8;
while (number_of_bits != 0) {
number_of_bits--;
remainder <<= 1;
if ((remainder & 0x100) != 0)
remainder ^= 0xd5;
}
return remainder & 0xff;
}
static inline u8 drm_dp_calc_sb_hdr_size(struct drm_dp_sideband_msg_hdr *hdr)
{
u8 size = 3;
size += (hdr->lct / 2);
return size;
}
static void drm_dp_encode_sideband_msg_hdr(struct drm_dp_sideband_msg_hdr *hdr,
u8 *buf, int *len)
{
int idx = 0;
int i;
u8 crc4;
buf[idx++] = ((hdr->lct & 0xf) << 4) | (hdr->lcr & 0xf);
for (i = 0; i < (hdr->lct / 2); i++)
buf[idx++] = hdr->rad[i];
buf[idx++] = (hdr->broadcast << 7) | (hdr->path_msg << 6) |
(hdr->msg_len & 0x3f);
buf[idx++] = (hdr->somt << 7) | (hdr->eomt << 6) | (hdr->seqno << 4);
crc4 = drm_dp_msg_header_crc4(buf, (idx * 2) - 1);
buf[idx - 1] |= (crc4 & 0xf);
*len = idx;
}
static bool drm_dp_decode_sideband_msg_hdr(struct drm_dp_sideband_msg_hdr *hdr,
u8 *buf, int buflen, u8 *hdrlen)
{
u8 crc4;
u8 len;
int i;
u8 idx;
if (buf[0] == 0)
return false;
len = 3;
len += ((buf[0] & 0xf0) >> 4) / 2;
if (len > buflen)
return false;
crc4 = drm_dp_msg_header_crc4(buf, (len * 2) - 1);
if ((crc4 & 0xf) != (buf[len - 1] & 0xf)) {
DRM_DEBUG_KMS("crc4 mismatch 0x%x 0x%x\n", crc4, buf[len - 1]);
return false;
}
hdr->lct = (buf[0] & 0xf0) >> 4;
hdr->lcr = (buf[0] & 0xf);
idx = 1;
for (i = 0; i < (hdr->lct / 2); i++)
hdr->rad[i] = buf[idx++];
hdr->broadcast = (buf[idx] >> 7) & 0x1;
hdr->path_msg = (buf[idx] >> 6) & 0x1;
hdr->msg_len = buf[idx] & 0x3f;
idx++;
hdr->somt = (buf[idx] >> 7) & 0x1;
hdr->eomt = (buf[idx] >> 6) & 0x1;
hdr->seqno = (buf[idx] >> 4) & 0x1;
idx++;
*hdrlen = idx;
return true;
}
void
drm_dp_encode_sideband_req(const struct drm_dp_sideband_msg_req_body *req,
struct drm_dp_sideband_msg_tx *raw)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
{
int idx = 0;
int i;
u8 *buf = raw->msg;
buf[idx++] = req->req_type & 0x7f;
switch (req->req_type) {
case DP_ENUM_PATH_RESOURCES:
case DP_POWER_DOWN_PHY:
case DP_POWER_UP_PHY:
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
buf[idx] = (req->u.port_num.port_number & 0xf) << 4;
idx++;
break;
case DP_ALLOCATE_PAYLOAD:
buf[idx] = (req->u.allocate_payload.port_number & 0xf) << 4 |
(req->u.allocate_payload.number_sdp_streams & 0xf);
idx++;
buf[idx] = (req->u.allocate_payload.vcpi & 0x7f);
idx++;
buf[idx] = (req->u.allocate_payload.pbn >> 8);
idx++;
buf[idx] = (req->u.allocate_payload.pbn & 0xff);
idx++;
for (i = 0; i < req->u.allocate_payload.number_sdp_streams / 2; i++) {
buf[idx] = ((req->u.allocate_payload.sdp_stream_sink[i * 2] & 0xf) << 4) |
(req->u.allocate_payload.sdp_stream_sink[i * 2 + 1] & 0xf);
idx++;
}
if (req->u.allocate_payload.number_sdp_streams & 1) {
i = req->u.allocate_payload.number_sdp_streams - 1;
buf[idx] = (req->u.allocate_payload.sdp_stream_sink[i] & 0xf) << 4;
idx++;
}
break;
case DP_QUERY_PAYLOAD:
buf[idx] = (req->u.query_payload.port_number & 0xf) << 4;
idx++;
buf[idx] = (req->u.query_payload.vcpi & 0x7f);
idx++;
break;
case DP_REMOTE_DPCD_READ:
buf[idx] = (req->u.dpcd_read.port_number & 0xf) << 4;
buf[idx] |= ((req->u.dpcd_read.dpcd_address & 0xf0000) >> 16) & 0xf;
idx++;
buf[idx] = (req->u.dpcd_read.dpcd_address & 0xff00) >> 8;
idx++;
buf[idx] = (req->u.dpcd_read.dpcd_address & 0xff);
idx++;
buf[idx] = (req->u.dpcd_read.num_bytes);
idx++;
break;
case DP_REMOTE_DPCD_WRITE:
buf[idx] = (req->u.dpcd_write.port_number & 0xf) << 4;
buf[idx] |= ((req->u.dpcd_write.dpcd_address & 0xf0000) >> 16) & 0xf;
idx++;
buf[idx] = (req->u.dpcd_write.dpcd_address & 0xff00) >> 8;
idx++;
buf[idx] = (req->u.dpcd_write.dpcd_address & 0xff);
idx++;
buf[idx] = (req->u.dpcd_write.num_bytes);
idx++;
memcpy(&buf[idx], req->u.dpcd_write.bytes, req->u.dpcd_write.num_bytes);
idx += req->u.dpcd_write.num_bytes;
break;
case DP_REMOTE_I2C_READ:
buf[idx] = (req->u.i2c_read.port_number & 0xf) << 4;
buf[idx] |= (req->u.i2c_read.num_transactions & 0x3);
idx++;
for (i = 0; i < (req->u.i2c_read.num_transactions & 0x3); i++) {
buf[idx] = req->u.i2c_read.transactions[i].i2c_dev_id & 0x7f;
idx++;
buf[idx] = req->u.i2c_read.transactions[i].num_bytes;
idx++;
memcpy(&buf[idx], req->u.i2c_read.transactions[i].bytes, req->u.i2c_read.transactions[i].num_bytes);
idx += req->u.i2c_read.transactions[i].num_bytes;
buf[idx] = (req->u.i2c_read.transactions[i].no_stop_bit & 0x1) << 5;
buf[idx] |= (req->u.i2c_read.transactions[i].i2c_transaction_delay & 0xf);
idx++;
}
buf[idx] = (req->u.i2c_read.read_i2c_device_id) & 0x7f;
idx++;
buf[idx] = (req->u.i2c_read.num_bytes_read);
idx++;
break;
case DP_REMOTE_I2C_WRITE:
buf[idx] = (req->u.i2c_write.port_number & 0xf) << 4;
idx++;
buf[idx] = (req->u.i2c_write.write_i2c_device_id) & 0x7f;
idx++;
buf[idx] = (req->u.i2c_write.num_bytes);
idx++;
memcpy(&buf[idx], req->u.i2c_write.bytes, req->u.i2c_write.num_bytes);
idx += req->u.i2c_write.num_bytes;
break;
}
raw->cur_len = idx;
}
EXPORT_SYMBOL_FOR_TESTS_ONLY(drm_dp_encode_sideband_req);
/* Decode a sideband request we've encoded, mainly used for debugging */
int
drm_dp_decode_sideband_req(const struct drm_dp_sideband_msg_tx *raw,
struct drm_dp_sideband_msg_req_body *req)
{
const u8 *buf = raw->msg;
int i, idx = 0;
req->req_type = buf[idx++] & 0x7f;
switch (req->req_type) {
case DP_ENUM_PATH_RESOURCES:
case DP_POWER_DOWN_PHY:
case DP_POWER_UP_PHY:
req->u.port_num.port_number = (buf[idx] >> 4) & 0xf;
break;
case DP_ALLOCATE_PAYLOAD:
{
struct drm_dp_allocate_payload *a =
&req->u.allocate_payload;
a->number_sdp_streams = buf[idx] & 0xf;
a->port_number = (buf[idx] >> 4) & 0xf;
WARN_ON(buf[++idx] & 0x80);
a->vcpi = buf[idx] & 0x7f;
a->pbn = buf[++idx] << 8;
a->pbn |= buf[++idx];
idx++;
for (i = 0; i < a->number_sdp_streams; i++) {
a->sdp_stream_sink[i] =
(buf[idx + (i / 2)] >> ((i % 2) ? 0 : 4)) & 0xf;
}
}
break;
case DP_QUERY_PAYLOAD:
req->u.query_payload.port_number = (buf[idx] >> 4) & 0xf;
WARN_ON(buf[++idx] & 0x80);
req->u.query_payload.vcpi = buf[idx] & 0x7f;
break;
case DP_REMOTE_DPCD_READ:
{
struct drm_dp_remote_dpcd_read *r = &req->u.dpcd_read;
r->port_number = (buf[idx] >> 4) & 0xf;
r->dpcd_address = (buf[idx] << 16) & 0xf0000;
r->dpcd_address |= (buf[++idx] << 8) & 0xff00;
r->dpcd_address |= buf[++idx] & 0xff;
r->num_bytes = buf[++idx];
}
break;
case DP_REMOTE_DPCD_WRITE:
{
struct drm_dp_remote_dpcd_write *w =
&req->u.dpcd_write;
w->port_number = (buf[idx] >> 4) & 0xf;
w->dpcd_address = (buf[idx] << 16) & 0xf0000;
w->dpcd_address |= (buf[++idx] << 8) & 0xff00;
w->dpcd_address |= buf[++idx] & 0xff;
w->num_bytes = buf[++idx];
w->bytes = kmemdup(&buf[++idx], w->num_bytes,
GFP_KERNEL);
if (!w->bytes)
return -ENOMEM;
}
break;
case DP_REMOTE_I2C_READ:
{
struct drm_dp_remote_i2c_read *r = &req->u.i2c_read;
struct drm_dp_remote_i2c_read_tx *tx;
bool failed = false;
r->num_transactions = buf[idx] & 0x3;
r->port_number = (buf[idx] >> 4) & 0xf;
for (i = 0; i < r->num_transactions; i++) {
tx = &r->transactions[i];
tx->i2c_dev_id = buf[++idx] & 0x7f;
tx->num_bytes = buf[++idx];
tx->bytes = kmemdup(&buf[++idx],
tx->num_bytes,
GFP_KERNEL);
if (!tx->bytes) {
failed = true;
break;
}
idx += tx->num_bytes;
tx->no_stop_bit = (buf[idx] >> 5) & 0x1;
tx->i2c_transaction_delay = buf[idx] & 0xf;
}
if (failed) {
for (i = 0; i < r->num_transactions; i++)
kfree(tx->bytes);
return -ENOMEM;
}
r->read_i2c_device_id = buf[++idx] & 0x7f;
r->num_bytes_read = buf[++idx];
}
break;
case DP_REMOTE_I2C_WRITE:
{
struct drm_dp_remote_i2c_write *w = &req->u.i2c_write;
w->port_number = (buf[idx] >> 4) & 0xf;
w->write_i2c_device_id = buf[++idx] & 0x7f;
w->num_bytes = buf[++idx];
w->bytes = kmemdup(&buf[++idx], w->num_bytes,
GFP_KERNEL);
if (!w->bytes)
return -ENOMEM;
}
break;
}
return 0;
}
EXPORT_SYMBOL_FOR_TESTS_ONLY(drm_dp_decode_sideband_req);
void
drm_dp_dump_sideband_msg_req_body(const struct drm_dp_sideband_msg_req_body *req,
int indent, struct drm_printer *printer)
{
int i;
#define P(f, ...) drm_printf_indent(printer, indent, f, ##__VA_ARGS__)
if (req->req_type == DP_LINK_ADDRESS) {
/* No contents to print */
P("type=%s\n", drm_dp_mst_req_type_str(req->req_type));
return;
}
P("type=%s contents:\n", drm_dp_mst_req_type_str(req->req_type));
indent++;
switch (req->req_type) {
case DP_ENUM_PATH_RESOURCES:
case DP_POWER_DOWN_PHY:
case DP_POWER_UP_PHY:
P("port=%d\n", req->u.port_num.port_number);
break;
case DP_ALLOCATE_PAYLOAD:
P("port=%d vcpi=%d pbn=%d sdp_streams=%d %*ph\n",
req->u.allocate_payload.port_number,
req->u.allocate_payload.vcpi, req->u.allocate_payload.pbn,
req->u.allocate_payload.number_sdp_streams,
req->u.allocate_payload.number_sdp_streams,
req->u.allocate_payload.sdp_stream_sink);
break;
case DP_QUERY_PAYLOAD:
P("port=%d vcpi=%d\n",
req->u.query_payload.port_number,
req->u.query_payload.vcpi);
break;
case DP_REMOTE_DPCD_READ:
P("port=%d dpcd_addr=%05x len=%d\n",
req->u.dpcd_read.port_number, req->u.dpcd_read.dpcd_address,
req->u.dpcd_read.num_bytes);
break;
case DP_REMOTE_DPCD_WRITE:
P("port=%d addr=%05x len=%d: %*ph\n",
req->u.dpcd_write.port_number,
req->u.dpcd_write.dpcd_address,
req->u.dpcd_write.num_bytes, req->u.dpcd_write.num_bytes,
req->u.dpcd_write.bytes);
break;
case DP_REMOTE_I2C_READ:
P("port=%d num_tx=%d id=%d size=%d:\n",
req->u.i2c_read.port_number,
req->u.i2c_read.num_transactions,
req->u.i2c_read.read_i2c_device_id,
req->u.i2c_read.num_bytes_read);
indent++;
for (i = 0; i < req->u.i2c_read.num_transactions; i++) {
const struct drm_dp_remote_i2c_read_tx *rtx =
&req->u.i2c_read.transactions[i];
P("%d: id=%03d size=%03d no_stop_bit=%d tx_delay=%03d: %*ph\n",
i, rtx->i2c_dev_id, rtx->num_bytes,
rtx->no_stop_bit, rtx->i2c_transaction_delay,
rtx->num_bytes, rtx->bytes);
}
break;
case DP_REMOTE_I2C_WRITE:
P("port=%d id=%d size=%d: %*ph\n",
req->u.i2c_write.port_number,
req->u.i2c_write.write_i2c_device_id,
req->u.i2c_write.num_bytes, req->u.i2c_write.num_bytes,
req->u.i2c_write.bytes);
break;
default:
P("???\n");
break;
}
#undef P
}
EXPORT_SYMBOL_FOR_TESTS_ONLY(drm_dp_dump_sideband_msg_req_body);
static inline void
drm_dp_mst_dump_sideband_msg_tx(struct drm_printer *p,
const struct drm_dp_sideband_msg_tx *txmsg)
{
struct drm_dp_sideband_msg_req_body req;
char buf[64];
int ret;
int i;
drm_dp_mst_rad_to_str(txmsg->dst->rad, txmsg->dst->lct, buf,
sizeof(buf));
drm_printf(p, "txmsg cur_offset=%x cur_len=%x seqno=%x state=%s path_msg=%d dst=%s\n",
txmsg->cur_offset, txmsg->cur_len, txmsg->seqno,
drm_dp_mst_sideband_tx_state_str(txmsg->state),
txmsg->path_msg, buf);
ret = drm_dp_decode_sideband_req(txmsg, &req);
if (ret) {
drm_printf(p, "<failed to decode sideband req: %d>\n", ret);
return;
}
drm_dp_dump_sideband_msg_req_body(&req, 1, p);
switch (req.req_type) {
case DP_REMOTE_DPCD_WRITE:
kfree(req.u.dpcd_write.bytes);
break;
case DP_REMOTE_I2C_READ:
for (i = 0; i < req.u.i2c_read.num_transactions; i++)
kfree(req.u.i2c_read.transactions[i].bytes);
break;
case DP_REMOTE_I2C_WRITE:
kfree(req.u.i2c_write.bytes);
break;
}
}
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
static void drm_dp_crc_sideband_chunk_req(u8 *msg, u8 len)
{
u8 crc4;
crc4 = drm_dp_msg_data_crc4(msg, len);
msg[len] = crc4;
}
static void drm_dp_encode_sideband_reply(struct drm_dp_sideband_msg_reply_body *rep,
struct drm_dp_sideband_msg_tx *raw)
{
int idx = 0;
u8 *buf = raw->msg;
buf[idx++] = (rep->reply_type & 0x1) << 7 | (rep->req_type & 0x7f);
raw->cur_len = idx;
}
/* this adds a chunk of msg to the builder to get the final msg */
static bool drm_dp_sideband_msg_build(struct drm_dp_sideband_msg_rx *msg,
u8 *replybuf, u8 replybuflen, bool hdr)
{
int ret;
u8 crc4;
if (hdr) {
u8 hdrlen;
struct drm_dp_sideband_msg_hdr recv_hdr;
ret = drm_dp_decode_sideband_msg_hdr(&recv_hdr, replybuf, replybuflen, &hdrlen);
if (ret == false) {
print_hex_dump(KERN_DEBUG, "failed hdr", DUMP_PREFIX_NONE, 16, 1, replybuf, replybuflen, false);
return false;
}
/*
* ignore out-of-order messages or messages that are part of a
* failed transaction
*/
if (!recv_hdr.somt && !msg->have_somt)
return false;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
/* get length contained in this portion */
msg->curchunk_len = recv_hdr.msg_len;
msg->curchunk_hdrlen = hdrlen;
/* we have already gotten an somt - don't bother parsing */
if (recv_hdr.somt && msg->have_somt)
return false;
if (recv_hdr.somt) {
memcpy(&msg->initial_hdr, &recv_hdr, sizeof(struct drm_dp_sideband_msg_hdr));
msg->have_somt = true;
}
if (recv_hdr.eomt)
msg->have_eomt = true;
/* copy the bytes for the remainder of this header chunk */
msg->curchunk_idx = min(msg->curchunk_len, (u8)(replybuflen - hdrlen));
memcpy(&msg->chunk[0], replybuf + hdrlen, msg->curchunk_idx);
} else {
memcpy(&msg->chunk[msg->curchunk_idx], replybuf, replybuflen);
msg->curchunk_idx += replybuflen;
}
if (msg->curchunk_idx >= msg->curchunk_len) {
/* do CRC */
crc4 = drm_dp_msg_data_crc4(msg->chunk, msg->curchunk_len - 1);
/* copy chunk into bigger msg */
memcpy(&msg->msg[msg->curlen], msg->chunk, msg->curchunk_len - 1);
msg->curlen += msg->curchunk_len - 1;
}
return true;
}
static bool drm_dp_sideband_parse_link_address(struct drm_dp_sideband_msg_rx *raw,
struct drm_dp_sideband_msg_reply_body *repmsg)
{
int idx = 1;
int i;
memcpy(repmsg->u.link_addr.guid, &raw->msg[idx], 16);
idx += 16;
repmsg->u.link_addr.nports = raw->msg[idx] & 0xf;
idx++;
if (idx > raw->curlen)
goto fail_len;
for (i = 0; i < repmsg->u.link_addr.nports; i++) {
if (raw->msg[idx] & 0x80)
repmsg->u.link_addr.ports[i].input_port = 1;
repmsg->u.link_addr.ports[i].peer_device_type = (raw->msg[idx] >> 4) & 0x7;
repmsg->u.link_addr.ports[i].port_number = (raw->msg[idx] & 0xf);
idx++;
if (idx > raw->curlen)
goto fail_len;
repmsg->u.link_addr.ports[i].mcs = (raw->msg[idx] >> 7) & 0x1;
repmsg->u.link_addr.ports[i].ddps = (raw->msg[idx] >> 6) & 0x1;
if (repmsg->u.link_addr.ports[i].input_port == 0)
repmsg->u.link_addr.ports[i].legacy_device_plug_status = (raw->msg[idx] >> 5) & 0x1;
idx++;
if (idx > raw->curlen)
goto fail_len;
if (repmsg->u.link_addr.ports[i].input_port == 0) {
repmsg->u.link_addr.ports[i].dpcd_revision = (raw->msg[idx]);
idx++;
if (idx > raw->curlen)
goto fail_len;
memcpy(repmsg->u.link_addr.ports[i].peer_guid, &raw->msg[idx], 16);
idx += 16;
if (idx > raw->curlen)
goto fail_len;
repmsg->u.link_addr.ports[i].num_sdp_streams = (raw->msg[idx] >> 4) & 0xf;
repmsg->u.link_addr.ports[i].num_sdp_stream_sinks = (raw->msg[idx] & 0xf);
idx++;
}
if (idx > raw->curlen)
goto fail_len;
}
return true;
fail_len:
DRM_DEBUG_KMS("link address reply parse length fail %d %d\n", idx, raw->curlen);
return false;
}
static bool drm_dp_sideband_parse_remote_dpcd_read(struct drm_dp_sideband_msg_rx *raw,
struct drm_dp_sideband_msg_reply_body *repmsg)
{
int idx = 1;
repmsg->u.remote_dpcd_read_ack.port_number = raw->msg[idx] & 0xf;
idx++;
if (idx > raw->curlen)
goto fail_len;
repmsg->u.remote_dpcd_read_ack.num_bytes = raw->msg[idx];
idx++;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
if (idx > raw->curlen)
goto fail_len;
memcpy(repmsg->u.remote_dpcd_read_ack.bytes, &raw->msg[idx], repmsg->u.remote_dpcd_read_ack.num_bytes);
return true;
fail_len:
DRM_DEBUG_KMS("link address reply parse length fail %d %d\n", idx, raw->curlen);
return false;
}
static bool drm_dp_sideband_parse_remote_dpcd_write(struct drm_dp_sideband_msg_rx *raw,
struct drm_dp_sideband_msg_reply_body *repmsg)
{
int idx = 1;
repmsg->u.remote_dpcd_write_ack.port_number = raw->msg[idx] & 0xf;
idx++;
if (idx > raw->curlen)
goto fail_len;
return true;
fail_len:
DRM_DEBUG_KMS("parse length fail %d %d\n", idx, raw->curlen);
return false;
}
static bool drm_dp_sideband_parse_remote_i2c_read_ack(struct drm_dp_sideband_msg_rx *raw,
struct drm_dp_sideband_msg_reply_body *repmsg)
{
int idx = 1;
repmsg->u.remote_i2c_read_ack.port_number = (raw->msg[idx] & 0xf);
idx++;
if (idx > raw->curlen)
goto fail_len;
repmsg->u.remote_i2c_read_ack.num_bytes = raw->msg[idx];
idx++;
/* TODO check */
memcpy(repmsg->u.remote_i2c_read_ack.bytes, &raw->msg[idx], repmsg->u.remote_i2c_read_ack.num_bytes);
return true;
fail_len:
DRM_DEBUG_KMS("remote i2c reply parse length fail %d %d\n", idx, raw->curlen);
return false;
}
static bool drm_dp_sideband_parse_enum_path_resources_ack(struct drm_dp_sideband_msg_rx *raw,
struct drm_dp_sideband_msg_reply_body *repmsg)
{
int idx = 1;
repmsg->u.path_resources.port_number = (raw->msg[idx] >> 4) & 0xf;
idx++;
if (idx > raw->curlen)
goto fail_len;
repmsg->u.path_resources.full_payload_bw_number = (raw->msg[idx] << 8) | (raw->msg[idx+1]);
idx += 2;
if (idx > raw->curlen)
goto fail_len;
repmsg->u.path_resources.avail_payload_bw_number = (raw->msg[idx] << 8) | (raw->msg[idx+1]);
idx += 2;
if (idx > raw->curlen)
goto fail_len;
return true;
fail_len:
DRM_DEBUG_KMS("enum resource parse length fail %d %d\n", idx, raw->curlen);
return false;
}
static bool drm_dp_sideband_parse_allocate_payload_ack(struct drm_dp_sideband_msg_rx *raw,
struct drm_dp_sideband_msg_reply_body *repmsg)
{
int idx = 1;
repmsg->u.allocate_payload.port_number = (raw->msg[idx] >> 4) & 0xf;
idx++;
if (idx > raw->curlen)
goto fail_len;
repmsg->u.allocate_payload.vcpi = raw->msg[idx];
idx++;
if (idx > raw->curlen)
goto fail_len;
repmsg->u.allocate_payload.allocated_pbn = (raw->msg[idx] << 8) | (raw->msg[idx+1]);
idx += 2;
if (idx > raw->curlen)
goto fail_len;
return true;
fail_len:
DRM_DEBUG_KMS("allocate payload parse length fail %d %d\n", idx, raw->curlen);
return false;
}
static bool drm_dp_sideband_parse_query_payload_ack(struct drm_dp_sideband_msg_rx *raw,
struct drm_dp_sideband_msg_reply_body *repmsg)
{
int idx = 1;
repmsg->u.query_payload.port_number = (raw->msg[idx] >> 4) & 0xf;
idx++;
if (idx > raw->curlen)
goto fail_len;
repmsg->u.query_payload.allocated_pbn = (raw->msg[idx] << 8) | (raw->msg[idx + 1]);
idx += 2;
if (idx > raw->curlen)
goto fail_len;
return true;
fail_len:
DRM_DEBUG_KMS("query payload parse length fail %d %d\n", idx, raw->curlen);
return false;
}
static bool drm_dp_sideband_parse_power_updown_phy_ack(struct drm_dp_sideband_msg_rx *raw,
struct drm_dp_sideband_msg_reply_body *repmsg)
{
int idx = 1;
repmsg->u.port_number.port_number = (raw->msg[idx] >> 4) & 0xf;
idx++;
if (idx > raw->curlen) {
DRM_DEBUG_KMS("power up/down phy parse length fail %d %d\n",
idx, raw->curlen);
return false;
}
return true;
}
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
static bool drm_dp_sideband_parse_reply(struct drm_dp_sideband_msg_rx *raw,
struct drm_dp_sideband_msg_reply_body *msg)
{
memset(msg, 0, sizeof(*msg));
msg->reply_type = (raw->msg[0] & 0x80) >> 7;
msg->req_type = (raw->msg[0] & 0x7f);
if (msg->reply_type == DP_SIDEBAND_REPLY_NAK) {
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
memcpy(msg->u.nak.guid, &raw->msg[1], 16);
msg->u.nak.reason = raw->msg[17];
msg->u.nak.nak_data = raw->msg[18];
return false;
}
switch (msg->req_type) {
case DP_LINK_ADDRESS:
return drm_dp_sideband_parse_link_address(raw, msg);
case DP_QUERY_PAYLOAD:
return drm_dp_sideband_parse_query_payload_ack(raw, msg);
case DP_REMOTE_DPCD_READ:
return drm_dp_sideband_parse_remote_dpcd_read(raw, msg);
case DP_REMOTE_DPCD_WRITE:
return drm_dp_sideband_parse_remote_dpcd_write(raw, msg);
case DP_REMOTE_I2C_READ:
return drm_dp_sideband_parse_remote_i2c_read_ack(raw, msg);
case DP_ENUM_PATH_RESOURCES:
return drm_dp_sideband_parse_enum_path_resources_ack(raw, msg);
case DP_ALLOCATE_PAYLOAD:
return drm_dp_sideband_parse_allocate_payload_ack(raw, msg);
case DP_POWER_DOWN_PHY:
case DP_POWER_UP_PHY:
return drm_dp_sideband_parse_power_updown_phy_ack(raw, msg);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
default:
DRM_ERROR("Got unknown reply 0x%02x (%s)\n", msg->req_type,
drm_dp_mst_req_type_str(msg->req_type));
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
return false;
}
}
static bool drm_dp_sideband_parse_connection_status_notify(struct drm_dp_sideband_msg_rx *raw,
struct drm_dp_sideband_msg_req_body *msg)
{
int idx = 1;
msg->u.conn_stat.port_number = (raw->msg[idx] & 0xf0) >> 4;
idx++;
if (idx > raw->curlen)
goto fail_len;
memcpy(msg->u.conn_stat.guid, &raw->msg[idx], 16);
idx += 16;
if (idx > raw->curlen)
goto fail_len;
msg->u.conn_stat.legacy_device_plug_status = (raw->msg[idx] >> 6) & 0x1;
msg->u.conn_stat.displayport_device_plug_status = (raw->msg[idx] >> 5) & 0x1;
msg->u.conn_stat.message_capability_status = (raw->msg[idx] >> 4) & 0x1;
msg->u.conn_stat.input_port = (raw->msg[idx] >> 3) & 0x1;
msg->u.conn_stat.peer_device_type = (raw->msg[idx] & 0x7);
idx++;
return true;
fail_len:
DRM_DEBUG_KMS("connection status reply parse length fail %d %d\n", idx, raw->curlen);
return false;
}
static bool drm_dp_sideband_parse_resource_status_notify(struct drm_dp_sideband_msg_rx *raw,
struct drm_dp_sideband_msg_req_body *msg)
{
int idx = 1;
msg->u.resource_stat.port_number = (raw->msg[idx] & 0xf0) >> 4;
idx++;
if (idx > raw->curlen)
goto fail_len;
memcpy(msg->u.resource_stat.guid, &raw->msg[idx], 16);
idx += 16;
if (idx > raw->curlen)
goto fail_len;
msg->u.resource_stat.available_pbn = (raw->msg[idx] << 8) | (raw->msg[idx + 1]);
idx++;
return true;
fail_len:
DRM_DEBUG_KMS("resource status reply parse length fail %d %d\n", idx, raw->curlen);
return false;
}
static bool drm_dp_sideband_parse_req(struct drm_dp_sideband_msg_rx *raw,
struct drm_dp_sideband_msg_req_body *msg)
{
memset(msg, 0, sizeof(*msg));
msg->req_type = (raw->msg[0] & 0x7f);
switch (msg->req_type) {
case DP_CONNECTION_STATUS_NOTIFY:
return drm_dp_sideband_parse_connection_status_notify(raw, msg);
case DP_RESOURCE_STATUS_NOTIFY:
return drm_dp_sideband_parse_resource_status_notify(raw, msg);
default:
DRM_ERROR("Got unknown request 0x%02x (%s)\n", msg->req_type,
drm_dp_mst_req_type_str(msg->req_type));
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
return false;
}
}
static int build_dpcd_write(struct drm_dp_sideband_msg_tx *msg, u8 port_num, u32 offset, u8 num_bytes, u8 *bytes)
{
struct drm_dp_sideband_msg_req_body req;
req.req_type = DP_REMOTE_DPCD_WRITE;
req.u.dpcd_write.port_number = port_num;
req.u.dpcd_write.dpcd_address = offset;
req.u.dpcd_write.num_bytes = num_bytes;
req.u.dpcd_write.bytes = bytes;
drm_dp_encode_sideband_req(&req, msg);
return 0;
}
static int build_link_address(struct drm_dp_sideband_msg_tx *msg)
{
struct drm_dp_sideband_msg_req_body req;
req.req_type = DP_LINK_ADDRESS;
drm_dp_encode_sideband_req(&req, msg);
return 0;
}
static int build_enum_path_resources(struct drm_dp_sideband_msg_tx *msg, int port_num)
{
struct drm_dp_sideband_msg_req_body req;
req.req_type = DP_ENUM_PATH_RESOURCES;
req.u.port_num.port_number = port_num;
drm_dp_encode_sideband_req(&req, msg);
msg->path_msg = true;
return 0;
}
static int build_allocate_payload(struct drm_dp_sideband_msg_tx *msg, int port_num,
u8 vcpi, uint16_t pbn,
u8 number_sdp_streams,
u8 *sdp_stream_sink)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
{
struct drm_dp_sideband_msg_req_body req;
memset(&req, 0, sizeof(req));
req.req_type = DP_ALLOCATE_PAYLOAD;
req.u.allocate_payload.port_number = port_num;
req.u.allocate_payload.vcpi = vcpi;
req.u.allocate_payload.pbn = pbn;
req.u.allocate_payload.number_sdp_streams = number_sdp_streams;
memcpy(req.u.allocate_payload.sdp_stream_sink, sdp_stream_sink,
number_sdp_streams);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
drm_dp_encode_sideband_req(&req, msg);
msg->path_msg = true;
return 0;
}
static int build_power_updown_phy(struct drm_dp_sideband_msg_tx *msg,
int port_num, bool power_up)
{
struct drm_dp_sideband_msg_req_body req;
if (power_up)
req.req_type = DP_POWER_UP_PHY;
else
req.req_type = DP_POWER_DOWN_PHY;
req.u.port_num.port_number = port_num;
drm_dp_encode_sideband_req(&req, msg);
msg->path_msg = true;
return 0;
}
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
static int drm_dp_mst_assign_payload_id(struct drm_dp_mst_topology_mgr *mgr,
struct drm_dp_vcpi *vcpi)
{
int ret, vcpi_ret;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
mutex_lock(&mgr->payload_lock);
ret = find_first_zero_bit(&mgr->payload_mask, mgr->max_payloads + 1);
if (ret > mgr->max_payloads) {
ret = -EINVAL;
DRM_DEBUG_KMS("out of payload ids %d\n", ret);
goto out_unlock;
}
vcpi_ret = find_first_zero_bit(&mgr->vcpi_mask, mgr->max_payloads + 1);
if (vcpi_ret > mgr->max_payloads) {
ret = -EINVAL;
DRM_DEBUG_KMS("out of vcpi ids %d\n", ret);
goto out_unlock;
}
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
set_bit(ret, &mgr->payload_mask);
set_bit(vcpi_ret, &mgr->vcpi_mask);
vcpi->vcpi = vcpi_ret + 1;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
mgr->proposed_vcpis[ret - 1] = vcpi;
out_unlock:
mutex_unlock(&mgr->payload_lock);
return ret;
}
static void drm_dp_mst_put_payload_id(struct drm_dp_mst_topology_mgr *mgr,
int vcpi)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
{
int i;
if (vcpi == 0)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
return;
mutex_lock(&mgr->payload_lock);
DRM_DEBUG_KMS("putting payload %d\n", vcpi);
clear_bit(vcpi - 1, &mgr->vcpi_mask);
for (i = 0; i < mgr->max_payloads; i++) {
if (mgr->proposed_vcpis[i] &&
mgr->proposed_vcpis[i]->vcpi == vcpi) {
mgr->proposed_vcpis[i] = NULL;
clear_bit(i + 1, &mgr->payload_mask);
}
}
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
mutex_unlock(&mgr->payload_lock);
}
static bool check_txmsg_state(struct drm_dp_mst_topology_mgr *mgr,
struct drm_dp_sideband_msg_tx *txmsg)
{
unsigned int state;
/*
* All updates to txmsg->state are protected by mgr->qlock, and the two
* cases we check here are terminal states. For those the barriers
* provided by the wake_up/wait_event pair are enough.
*/
state = READ_ONCE(txmsg->state);
return (state == DRM_DP_SIDEBAND_TX_RX ||
state == DRM_DP_SIDEBAND_TX_TIMEOUT);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
static int drm_dp_mst_wait_tx_reply(struct drm_dp_mst_branch *mstb,
struct drm_dp_sideband_msg_tx *txmsg)
{
struct drm_dp_mst_topology_mgr *mgr = mstb->mgr;
int ret;
ret = wait_event_timeout(mgr->tx_waitq,
check_txmsg_state(mgr, txmsg),
(4 * HZ));
mutex_lock(&mstb->mgr->qlock);
if (ret > 0) {
if (txmsg->state == DRM_DP_SIDEBAND_TX_TIMEOUT) {
ret = -EIO;
goto out;
}
} else {
DRM_DEBUG_KMS("timedout msg send %p %d %d\n", txmsg, txmsg->state, txmsg->seqno);
/* dump some state */
ret = -EIO;
/* remove from q */
if (txmsg->state == DRM_DP_SIDEBAND_TX_QUEUED ||
txmsg->state == DRM_DP_SIDEBAND_TX_START_SEND) {
list_del(&txmsg->next);
}
if (txmsg->state == DRM_DP_SIDEBAND_TX_START_SEND ||
txmsg->state == DRM_DP_SIDEBAND_TX_SENT) {
mstb->tx_slots[txmsg->seqno] = NULL;
}
}
out:
if (unlikely(ret == -EIO) && drm_debug_enabled(DRM_UT_DP)) {
struct drm_printer p = drm_debug_printer(DBG_PREFIX);
drm_dp_mst_dump_sideband_msg_tx(&p, txmsg);
}
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
mutex_unlock(&mgr->qlock);
return ret;
}
static struct drm_dp_mst_branch *drm_dp_add_mst_branch_device(u8 lct, u8 *rad)
{
struct drm_dp_mst_branch *mstb;
mstb = kzalloc(sizeof(*mstb), GFP_KERNEL);
if (!mstb)
return NULL;
mstb->lct = lct;
if (lct > 1)
memcpy(mstb->rad, rad, lct / 2);
INIT_LIST_HEAD(&mstb->ports);
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
kref_init(&mstb->topology_kref);
kref_init(&mstb->malloc_kref);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
return mstb;
}
static void drm_dp_free_mst_branch_device(struct kref *kref)
{
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
struct drm_dp_mst_branch *mstb =
container_of(kref, struct drm_dp_mst_branch, malloc_kref);
if (mstb->port_parent)
drm_dp_mst_put_port_malloc(mstb->port_parent);
kfree(mstb);
}
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
/**
* DOC: Branch device and port refcounting
*
* Topology refcount overview
* ~~~~~~~~~~~~~~~~~~~~~~~~~~
*
* The refcounting schemes for &struct drm_dp_mst_branch and &struct
* drm_dp_mst_port are somewhat unusual. Both ports and branch devices have
* two different kinds of refcounts: topology refcounts, and malloc refcounts.
*
* Topology refcounts are not exposed to drivers, and are handled internally
* by the DP MST helpers. The helpers use them in order to prevent the
* in-memory topology state from being changed in the middle of critical
* operations like changing the internal state of payload allocations. This
* means each branch and port will be considered to be connected to the rest
* of the topology until its topology refcount reaches zero. Additionally,
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
* for ports this means that their associated &struct drm_connector will stay
* registered with userspace until the port's refcount reaches 0.
*
* Malloc refcount overview
* ~~~~~~~~~~~~~~~~~~~~~~~~
*
* Malloc references are used to keep a &struct drm_dp_mst_port or &struct
* drm_dp_mst_branch allocated even after all of its topology references have
* been dropped, so that the driver or MST helpers can safely access each
* branch's last known state before it was disconnected from the topology.
* When the malloc refcount of a port or branch reaches 0, the memory
* allocation containing the &struct drm_dp_mst_branch or &struct
* drm_dp_mst_port respectively will be freed.
*
* For &struct drm_dp_mst_branch, malloc refcounts are not currently exposed
* to drivers. As of writing this documentation, there are no drivers that
* have a usecase for accessing &struct drm_dp_mst_branch outside of the MST
* helpers. Exposing this API to drivers in a race-free manner would take more
* tweaking of the refcounting scheme, however patches are welcome provided
* there is a legitimate driver usecase for this.
*
* Refcount relationships in a topology
* ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
*
* Let's take a look at why the relationship between topology and malloc
* refcounts is designed the way it is.
*
* .. kernel-figure:: dp-mst/topology-figure-1.dot
*
* An example of topology and malloc refs in a DP MST topology with two
* active payloads. Topology refcount increments are indicated by solid
* lines, and malloc refcount increments are indicated by dashed lines.
* Each starts from the branch which incremented the refcount, and ends at
* the branch to which the refcount belongs to, i.e. the arrow points the
* same way as the C pointers used to reference a structure.
*
* As you can see in the above figure, every branch increments the topology
* refcount of its children, and increments the malloc refcount of its
* parent. Additionally, every payload increments the malloc refcount of its
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
* assigned port by 1.
*
* So, what would happen if MSTB #3 from the above figure was unplugged from
* the system, but the driver hadn't yet removed payload #2 from port #3? The
* topology would start to look like the figure below.
*
* .. kernel-figure:: dp-mst/topology-figure-2.dot
*
* Ports and branch devices which have been released from memory are
* colored grey, and references which have been removed are colored red.
*
* Whenever a port or branch device's topology refcount reaches zero, it will
* decrement the topology refcounts of all its children, the malloc refcount
* of its parent, and finally its own malloc refcount. For MSTB #4 and port
* #4, this means they both have been disconnected from the topology and freed
* from memory. But, because payload #2 is still holding a reference to port
* #3, port #3 is removed from the topology but its &struct drm_dp_mst_port
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
* is still accessible from memory. This also means port #3 has not yet
* decremented the malloc refcount of MSTB #3, so its &struct
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
* drm_dp_mst_branch will also stay allocated in memory until port #3's
* malloc refcount reaches 0.
*
* This relationship is necessary because in order to release payload #2, we
* need to be able to figure out the last relative of port #3 that's still
* connected to the topology. In this case, we would travel up the topology as
* shown below.
*
* .. kernel-figure:: dp-mst/topology-figure-3.dot
*
* And finally, remove payload #2 by communicating with port #2 through
* sideband transactions.
*/
/**
* drm_dp_mst_get_mstb_malloc() - Increment the malloc refcount of a branch
* device
* @mstb: The &struct drm_dp_mst_branch to increment the malloc refcount of
*
* Increments &drm_dp_mst_branch.malloc_kref. When
* &drm_dp_mst_branch.malloc_kref reaches 0, the memory allocation for @mstb
* will be released and @mstb may no longer be used.
*
* See also: drm_dp_mst_put_mstb_malloc()
*/
static void
drm_dp_mst_get_mstb_malloc(struct drm_dp_mst_branch *mstb)
{
kref_get(&mstb->malloc_kref);
DRM_DEBUG("mstb %p (%d)\n", mstb, kref_read(&mstb->malloc_kref));
}
/**
* drm_dp_mst_put_mstb_malloc() - Decrement the malloc refcount of a branch
* device
* @mstb: The &struct drm_dp_mst_branch to decrement the malloc refcount of
*
* Decrements &drm_dp_mst_branch.malloc_kref. When
* &drm_dp_mst_branch.malloc_kref reaches 0, the memory allocation for @mstb
* will be released and @mstb may no longer be used.
*
* See also: drm_dp_mst_get_mstb_malloc()
*/
static void
drm_dp_mst_put_mstb_malloc(struct drm_dp_mst_branch *mstb)
{
DRM_DEBUG("mstb %p (%d)\n", mstb, kref_read(&mstb->malloc_kref) - 1);
kref_put(&mstb->malloc_kref, drm_dp_free_mst_branch_device);
}
static void drm_dp_free_mst_port(struct kref *kref)
{
struct drm_dp_mst_port *port =
container_of(kref, struct drm_dp_mst_port, malloc_kref);
drm_dp_mst_put_mstb_malloc(port->parent);
kfree(port);
}
/**
* drm_dp_mst_get_port_malloc() - Increment the malloc refcount of an MST port
* @port: The &struct drm_dp_mst_port to increment the malloc refcount of
*
* Increments &drm_dp_mst_port.malloc_kref. When &drm_dp_mst_port.malloc_kref
* reaches 0, the memory allocation for @port will be released and @port may
* no longer be used.
*
* Because @port could potentially be freed at any time by the DP MST helpers
* if &drm_dp_mst_port.malloc_kref reaches 0, including during a call to this
* function, drivers that which to make use of &struct drm_dp_mst_port should
* ensure that they grab at least one main malloc reference to their MST ports
* in &drm_dp_mst_topology_cbs.add_connector. This callback is called before
* there is any chance for &drm_dp_mst_port.malloc_kref to reach 0.
*
* See also: drm_dp_mst_put_port_malloc()
*/
void
drm_dp_mst_get_port_malloc(struct drm_dp_mst_port *port)
{
kref_get(&port->malloc_kref);
DRM_DEBUG("port %p (%d)\n", port, kref_read(&port->malloc_kref));
}
EXPORT_SYMBOL(drm_dp_mst_get_port_malloc);
/**
* drm_dp_mst_put_port_malloc() - Decrement the malloc refcount of an MST port
* @port: The &struct drm_dp_mst_port to decrement the malloc refcount of
*
* Decrements &drm_dp_mst_port.malloc_kref. When &drm_dp_mst_port.malloc_kref
* reaches 0, the memory allocation for @port will be released and @port may
* no longer be used.
*
* See also: drm_dp_mst_get_port_malloc()
*/
void
drm_dp_mst_put_port_malloc(struct drm_dp_mst_port *port)
{
DRM_DEBUG("port %p (%d)\n", port, kref_read(&port->malloc_kref) - 1);
kref_put(&port->malloc_kref, drm_dp_free_mst_port);
}
EXPORT_SYMBOL(drm_dp_mst_put_port_malloc);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
static void drm_dp_destroy_mst_branch_device(struct kref *kref)
{
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
struct drm_dp_mst_branch *mstb =
container_of(kref, struct drm_dp_mst_branch, topology_kref);
struct drm_dp_mst_topology_mgr *mgr = mstb->mgr;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
struct drm_dp_mst_port *port, *tmp;
bool wake_tx = false;
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
mutex_lock(&mgr->lock);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
list_for_each_entry_safe(port, tmp, &mstb->ports, next) {
list_del(&port->next);
drm_dp_mst_topology_put_port(port);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
mutex_unlock(&mgr->lock);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
/* drop any tx slots msg */
mutex_lock(&mstb->mgr->qlock);
if (mstb->tx_slots[0]) {
mstb->tx_slots[0]->state = DRM_DP_SIDEBAND_TX_TIMEOUT;
mstb->tx_slots[0] = NULL;
wake_tx = true;
}
if (mstb->tx_slots[1]) {
mstb->tx_slots[1]->state = DRM_DP_SIDEBAND_TX_TIMEOUT;
mstb->tx_slots[1] = NULL;
wake_tx = true;
}
mutex_unlock(&mstb->mgr->qlock);
if (wake_tx)
wake_up_all(&mstb->mgr->tx_waitq);
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
drm_dp_mst_put_mstb_malloc(mstb);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
/**
* drm_dp_mst_topology_try_get_mstb() - Increment the topology refcount of a
* branch device unless it's zero
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
* @mstb: &struct drm_dp_mst_branch to increment the topology refcount of
*
* Attempts to grab a topology reference to @mstb, if it hasn't yet been
* removed from the topology (e.g. &drm_dp_mst_branch.topology_kref has
* reached 0). Holding a topology reference implies that a malloc reference
* will be held to @mstb as long as the user holds the topology reference.
*
* Care should be taken to ensure that the user has at least one malloc
* reference to @mstb. If you already have a topology reference to @mstb, you
* should use drm_dp_mst_topology_get_mstb() instead.
*
* See also:
* drm_dp_mst_topology_get_mstb()
* drm_dp_mst_topology_put_mstb()
*
* Returns:
* * 1: A topology reference was grabbed successfully
* * 0: @port is no longer in the topology, no reference was grabbed
*/
static int __must_check
drm_dp_mst_topology_try_get_mstb(struct drm_dp_mst_branch *mstb)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
{
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
int ret = kref_get_unless_zero(&mstb->topology_kref);
if (ret)
DRM_DEBUG("mstb %p (%d)\n", mstb,
kref_read(&mstb->topology_kref));
return ret;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
/**
* drm_dp_mst_topology_get_mstb() - Increment the topology refcount of a
* branch device
* @mstb: The &struct drm_dp_mst_branch to increment the topology refcount of
*
* Increments &drm_dp_mst_branch.topology_refcount without checking whether or
* not it's already reached 0. This is only valid to use in scenarios where
* you are already guaranteed to have at least one active topology reference
* to @mstb. Otherwise, drm_dp_mst_topology_try_get_mstb() must be used.
*
* See also:
* drm_dp_mst_topology_try_get_mstb()
* drm_dp_mst_topology_put_mstb()
*/
static void drm_dp_mst_topology_get_mstb(struct drm_dp_mst_branch *mstb)
{
WARN_ON(kref_read(&mstb->topology_kref) == 0);
kref_get(&mstb->topology_kref);
DRM_DEBUG("mstb %p (%d)\n", mstb, kref_read(&mstb->topology_kref));
}
/**
* drm_dp_mst_topology_put_mstb() - release a topology reference to a branch
* device
* @mstb: The &struct drm_dp_mst_branch to release the topology reference from
*
* Releases a topology reference from @mstb by decrementing
* &drm_dp_mst_branch.topology_kref.
*
* See also:
* drm_dp_mst_topology_try_get_mstb()
* drm_dp_mst_topology_get_mstb()
*/
static void
drm_dp_mst_topology_put_mstb(struct drm_dp_mst_branch *mstb)
{
DRM_DEBUG("mstb %p (%d)\n",
mstb, kref_read(&mstb->topology_kref) - 1);
kref_put(&mstb->topology_kref, drm_dp_destroy_mst_branch_device);
}
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
static void drm_dp_port_teardown_pdt(struct drm_dp_mst_port *port, int old_pdt)
{
drm/dp-mst: Remove branches before dropping the reference When we unplug a dp mst branch we unreference the entire tree from the root towards the leaves. Which is ok, since that's the way the pointers and so also the refcounts go. But when we drop the reference we must make sure that we remove the branches/ports from the lists/pointers before dropping the reference. Otherwise the get_validated functions will still return it instead of returning NULL (which indicates a potentially on-going unplug). The mst branch destroy gets this right for ports: First it deletes the port from the ports list, then it unrefs. But the ports destroy function gets it wrong: First it unrefs, then it drops the ref. Which means a zombie mst branch can still be validate with get_validated_mstb_ref when it shouldn't. Fix this. This should address a backtrace Dave dug out somewhere on unplug: [<ffffffffa00cc262>] drm_dp_mst_get_validated_mstb_ref_locked+0x92/0xa0 [drm_kms_helper] [<ffffffffa00cc211>] drm_dp_mst_get_validated_mstb_ref_locked+0x41/0xa0 [drm_kms_helper] [<ffffffffa00cc2aa>] drm_dp_get_validated_mstb_ref+0x3a/0x60 [drm_kms_helper] [<ffffffffa00cc2fb>] drm_dp_payload_send_msg.isra.14+0x2b/0x100 [drm_kms_helper] [<ffffffffa00cc547>] drm_dp_update_payload_part1+0x177/0x360 [drm_kms_helper] [<ffffffffa015c52e>] intel_mst_disable_dp+0x3e/0x80 [i915] [<ffffffffa013d60b>] haswell_crtc_disable+0x1cb/0x340 [i915] [<ffffffffa0136739>] intel_crtc_control+0x49/0x100 [i915] [<ffffffffa0136857>] intel_crtc_update_dpms+0x67/0x80 [i915] [<ffffffffa013fa59>] intel_connector_dpms+0x59/0x70 [i915] [<ffffffffa015c752>] intel_dp_destroy_mst_connector+0x32/0xc0 [i915] [<ffffffffa00cb44b>] drm_dp_destroy_port+0x6b/0xa0 [drm_kms_helper] [<ffffffffa00cb588>] drm_dp_destroy_mst_branch_device+0x108/0x130 [drm_kms_helper] [<ffffffffa00cb3cd>] drm_dp_port_teardown_pdt+0x3d/0x50 [drm_kms_helper] [<ffffffffa00cdb79>] drm_dp_mst_handle_up_req+0x499/0x540 [drm_kms_helper] [<ffffffff810d9ead>] ? trace_hardirqs_on_caller+0x15d/0x200 [<ffffffffa00cdc73>] drm_dp_mst_hpd_irq+0x53/0xa00 [drm_kms_helper] [<ffffffffa00c7dfb>] ? drm_dp_dpcd_read+0x1b/0x20 [drm_kms_helper] [<ffffffffa0153ed8>] ? intel_dp_dpcd_read_wake+0x38/0x70 [i915] [<ffffffffa015a225>] intel_dp_check_mst_status+0xb5/0x250 [i915] [<ffffffffa015ac71>] intel_dp_hpd_pulse+0x181/0x210 [i915] [<ffffffffa01104f6>] i915_digport_work_func+0x96/0x120 [i915] Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: stable@vger.kernel.org Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-12-09 04:55:22 +07:00
struct drm_dp_mst_branch *mstb;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
switch (old_pdt) {
case DP_PEER_DEVICE_DP_LEGACY_CONV:
case DP_PEER_DEVICE_SST_SINK:
/* remove i2c over sideband */
drm_dp_mst_unregister_i2c_bus(&port->aux);
break;
case DP_PEER_DEVICE_MST_BRANCHING:
drm/dp-mst: Remove branches before dropping the reference When we unplug a dp mst branch we unreference the entire tree from the root towards the leaves. Which is ok, since that's the way the pointers and so also the refcounts go. But when we drop the reference we must make sure that we remove the branches/ports from the lists/pointers before dropping the reference. Otherwise the get_validated functions will still return it instead of returning NULL (which indicates a potentially on-going unplug). The mst branch destroy gets this right for ports: First it deletes the port from the ports list, then it unrefs. But the ports destroy function gets it wrong: First it unrefs, then it drops the ref. Which means a zombie mst branch can still be validate with get_validated_mstb_ref when it shouldn't. Fix this. This should address a backtrace Dave dug out somewhere on unplug: [<ffffffffa00cc262>] drm_dp_mst_get_validated_mstb_ref_locked+0x92/0xa0 [drm_kms_helper] [<ffffffffa00cc211>] drm_dp_mst_get_validated_mstb_ref_locked+0x41/0xa0 [drm_kms_helper] [<ffffffffa00cc2aa>] drm_dp_get_validated_mstb_ref+0x3a/0x60 [drm_kms_helper] [<ffffffffa00cc2fb>] drm_dp_payload_send_msg.isra.14+0x2b/0x100 [drm_kms_helper] [<ffffffffa00cc547>] drm_dp_update_payload_part1+0x177/0x360 [drm_kms_helper] [<ffffffffa015c52e>] intel_mst_disable_dp+0x3e/0x80 [i915] [<ffffffffa013d60b>] haswell_crtc_disable+0x1cb/0x340 [i915] [<ffffffffa0136739>] intel_crtc_control+0x49/0x100 [i915] [<ffffffffa0136857>] intel_crtc_update_dpms+0x67/0x80 [i915] [<ffffffffa013fa59>] intel_connector_dpms+0x59/0x70 [i915] [<ffffffffa015c752>] intel_dp_destroy_mst_connector+0x32/0xc0 [i915] [<ffffffffa00cb44b>] drm_dp_destroy_port+0x6b/0xa0 [drm_kms_helper] [<ffffffffa00cb588>] drm_dp_destroy_mst_branch_device+0x108/0x130 [drm_kms_helper] [<ffffffffa00cb3cd>] drm_dp_port_teardown_pdt+0x3d/0x50 [drm_kms_helper] [<ffffffffa00cdb79>] drm_dp_mst_handle_up_req+0x499/0x540 [drm_kms_helper] [<ffffffff810d9ead>] ? trace_hardirqs_on_caller+0x15d/0x200 [<ffffffffa00cdc73>] drm_dp_mst_hpd_irq+0x53/0xa00 [drm_kms_helper] [<ffffffffa00c7dfb>] ? drm_dp_dpcd_read+0x1b/0x20 [drm_kms_helper] [<ffffffffa0153ed8>] ? intel_dp_dpcd_read_wake+0x38/0x70 [i915] [<ffffffffa015a225>] intel_dp_check_mst_status+0xb5/0x250 [i915] [<ffffffffa015ac71>] intel_dp_hpd_pulse+0x181/0x210 [i915] [<ffffffffa01104f6>] i915_digport_work_func+0x96/0x120 [i915] Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: stable@vger.kernel.org Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-12-09 04:55:22 +07:00
mstb = port->mstb;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
port->mstb = NULL;
drm_dp_mst_topology_put_mstb(mstb);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
break;
}
}
static void drm_dp_destroy_port(struct kref *kref)
{
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
struct drm_dp_mst_port *port =
container_of(kref, struct drm_dp_mst_port, topology_kref);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
struct drm_dp_mst_topology_mgr *mgr = port->mgr;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
if (!port->input) {
kfree(port->cached_edid);
/*
* The only time we don't have a connector
* on an output port is if the connector init
* fails.
*/
if (port->connector) {
/* we can't destroy the connector here, as
* we might be holding the mode_config.mutex
* from an EDID retrieval */
mutex_lock(&mgr->destroy_connector_lock);
list_add(&port->next, &mgr->destroy_connector_list);
mutex_unlock(&mgr->destroy_connector_lock);
schedule_work(&mgr->destroy_connector_work);
return;
}
/* no need to clean up vcpi
* as if we have no connector we never setup a vcpi */
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
drm_dp_port_teardown_pdt(port, port->pdt);
2016-10-26 20:30:33 +07:00
port->pdt = DP_PEER_DEVICE_NONE;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
drm_dp_mst_put_port_malloc(port);
}
/**
* drm_dp_mst_topology_try_get_port() - Increment the topology refcount of a
* port unless it's zero
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
* @port: &struct drm_dp_mst_port to increment the topology refcount of
*
* Attempts to grab a topology reference to @port, if it hasn't yet been
* removed from the topology (e.g. &drm_dp_mst_port.topology_kref has reached
* 0). Holding a topology reference implies that a malloc reference will be
* held to @port as long as the user holds the topology reference.
*
* Care should be taken to ensure that the user has at least one malloc
* reference to @port. If you already have a topology reference to @port, you
* should use drm_dp_mst_topology_get_port() instead.
*
* See also:
* drm_dp_mst_topology_get_port()
* drm_dp_mst_topology_put_port()
*
* Returns:
* * 1: A topology reference was grabbed successfully
* * 0: @port is no longer in the topology, no reference was grabbed
*/
static int __must_check
drm_dp_mst_topology_try_get_port(struct drm_dp_mst_port *port)
{
int ret = kref_get_unless_zero(&port->topology_kref);
if (ret)
DRM_DEBUG("port %p (%d)\n", port,
kref_read(&port->topology_kref));
return ret;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
/**
* drm_dp_mst_topology_get_port() - Increment the topology refcount of a port
* @port: The &struct drm_dp_mst_port to increment the topology refcount of
*
* Increments &drm_dp_mst_port.topology_refcount without checking whether or
* not it's already reached 0. This is only valid to use in scenarios where
* you are already guaranteed to have at least one active topology reference
* to @port. Otherwise, drm_dp_mst_topology_try_get_port() must be used.
*
* See also:
* drm_dp_mst_topology_try_get_port()
* drm_dp_mst_topology_put_port()
*/
static void drm_dp_mst_topology_get_port(struct drm_dp_mst_port *port)
{
WARN_ON(kref_read(&port->topology_kref) == 0);
kref_get(&port->topology_kref);
DRM_DEBUG("port %p (%d)\n", port, kref_read(&port->topology_kref));
}
/**
* drm_dp_mst_topology_put_port() - release a topology reference to a port
* @port: The &struct drm_dp_mst_port to release the topology reference from
*
* Releases a topology reference from @port by decrementing
* &drm_dp_mst_port.topology_kref.
*
* See also:
* drm_dp_mst_topology_try_get_port()
* drm_dp_mst_topology_get_port()
*/
static void drm_dp_mst_topology_put_port(struct drm_dp_mst_port *port)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
{
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
DRM_DEBUG("port %p (%d)\n",
port, kref_read(&port->topology_kref) - 1);
kref_put(&port->topology_kref, drm_dp_destroy_port);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
static struct drm_dp_mst_branch *
drm_dp_mst_topology_get_mstb_validated_locked(struct drm_dp_mst_branch *mstb,
struct drm_dp_mst_branch *to_find)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
{
struct drm_dp_mst_port *port;
struct drm_dp_mst_branch *rmstb;
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
if (to_find == mstb)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
return mstb;
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
list_for_each_entry(port, &mstb->ports, next) {
if (port->mstb) {
rmstb = drm_dp_mst_topology_get_mstb_validated_locked(
port->mstb, to_find);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
if (rmstb)
return rmstb;
}
}
return NULL;
}
static struct drm_dp_mst_branch *
drm_dp_mst_topology_get_mstb_validated(struct drm_dp_mst_topology_mgr *mgr,
struct drm_dp_mst_branch *mstb)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
{
struct drm_dp_mst_branch *rmstb = NULL;
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
mutex_lock(&mgr->lock);
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
if (mgr->mst_primary) {
rmstb = drm_dp_mst_topology_get_mstb_validated_locked(
mgr->mst_primary, mstb);
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
if (rmstb && !drm_dp_mst_topology_try_get_mstb(rmstb))
rmstb = NULL;
}
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
mutex_unlock(&mgr->lock);
return rmstb;
}
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
static struct drm_dp_mst_port *
drm_dp_mst_topology_get_port_validated_locked(struct drm_dp_mst_branch *mstb,
struct drm_dp_mst_port *to_find)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
{
struct drm_dp_mst_port *port, *mport;
list_for_each_entry(port, &mstb->ports, next) {
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
if (port == to_find)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
return port;
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
if (port->mstb) {
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
mport = drm_dp_mst_topology_get_port_validated_locked(
port->mstb, to_find);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
if (mport)
return mport;
}
}
return NULL;
}
static struct drm_dp_mst_port *
drm_dp_mst_topology_get_port_validated(struct drm_dp_mst_topology_mgr *mgr,
struct drm_dp_mst_port *port)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
{
struct drm_dp_mst_port *rport = NULL;
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
mutex_lock(&mgr->lock);
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
if (mgr->mst_primary) {
rport = drm_dp_mst_topology_get_port_validated_locked(
mgr->mst_primary, port);
if (rport && !drm_dp_mst_topology_try_get_port(rport))
rport = NULL;
}
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
mutex_unlock(&mgr->lock);
return rport;
}
static struct drm_dp_mst_port *drm_dp_get_port(struct drm_dp_mst_branch *mstb, u8 port_num)
{
struct drm_dp_mst_port *port;
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
int ret;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
list_for_each_entry(port, &mstb->ports, next) {
if (port->port_num == port_num) {
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
ret = drm_dp_mst_topology_try_get_port(port);
return ret ? port : NULL;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
}
return NULL;
}
/*
* calculate a new RAD for this MST branch device
* if parent has an LCT of 2 then it has 1 nibble of RAD,
* if parent has an LCT of 3 then it has 2 nibbles of RAD,
*/
static u8 drm_dp_calculate_rad(struct drm_dp_mst_port *port,
u8 *rad)
{
int parent_lct = port->parent->lct;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
int shift = 4;
int idx = (parent_lct - 1) / 2;
if (parent_lct > 1) {
memcpy(rad, port->parent->rad, idx + 1);
shift = (parent_lct % 2) ? 4 : 0;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
} else
rad[0] = 0;
rad[idx] |= port->port_num << shift;
return parent_lct + 1;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
/*
* return sends link address for new mstb
*/
static bool drm_dp_port_setup_pdt(struct drm_dp_mst_port *port)
{
int ret;
u8 rad[6], lct;
bool send_link = false;
switch (port->pdt) {
case DP_PEER_DEVICE_DP_LEGACY_CONV:
case DP_PEER_DEVICE_SST_SINK:
/* add i2c over sideband */
ret = drm_dp_mst_register_i2c_bus(&port->aux);
break;
case DP_PEER_DEVICE_MST_BRANCHING:
lct = drm_dp_calculate_rad(port, rad);
port->mstb = drm_dp_add_mst_branch_device(lct, rad);
if (port->mstb) {
port->mstb->mgr = port->mgr;
port->mstb->port_parent = port;
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
/*
* Make sure this port's memory allocation stays
* around until its child MSTB releases it
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
*/
drm_dp_mst_get_port_malloc(port);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
send_link = true;
}
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
break;
}
return send_link;
}
drm/dp_mst: Enable registration of AUX devices for MST ports All available downstream ports - physical and logical - are exposed for each MST device. They are listed in /dev/, following the same naming scheme as SST devices by appending an incremental ID. Although all downstream ports are exposed, only some will work as expected. Consider the following topology: +---------+ | ASIC | +---------+ Conn-0| | +----v----+ +----| MST HUB |----+ | +---------+ | | | |Port-1 Port-2| +-----v-----+ +-----v-----+ | MST | | SST | | Display | | Display | +-----------+ +-----------+ |Port-1 x MST Path | MST Device ----------+---------------------------------- sst:0 | MST Hub mst:0-1 | MST Display mst:0-1-1 | MST Display's disconnected DP out mst:0-1-8 | MST Display's internal sink mst:0-2 | SST Display On certain MST displays, the upstream physical port will ACK DPCD reads. However, reads on the local logical port to the internal sink will *NAK*. i.e. reading mst:0-1 ACKs, but mst:0-1-8 NAKs. There may also be duplicates. Some displays will return the same GUID when reading DPCD from both mst:0-1 and mst:0-1-8. There are some device-dependent behavior as well. The MST hub used during testing will actually *ACK* read requests on a disconnected physical port, whereas the MST displays will NAK. In light of these discrepancies, it's simpler to expose all downstream ports - both physical and logical - and let the user decide what to use. v3 changes: * Change WARN_ON_ONCE -> DRM_ERROR on dpcd read errors * Docstring and cosmetic fixes v2 changes: Moved remote aux device (un)registration to new mst connector late register and early unregister helpers. Drivers should call these from their own mst connector function hooks. This is to solve an issue during driver unload, where mst connector devices are unregistered before the remote aux devices are. In a setup where aux devices are created as children of connector devices, the aux device would be removed too early, and uncleanly. Doing so in early_unregister solves this issue, as that is called before connector unregistration. Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Signed-off-by: Leo Li <sunpeng.li@amd.com> Reviewed-by: Lyude Paul <lyude@redhat.com> Signed-off-by: Harry Wentland <harry.wentland@amd.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190723232808.28128-3-sunpeng.li@amd.com
2019-07-24 06:28:01 +07:00
/**
* drm_dp_mst_dpcd_read() - read a series of bytes from the DPCD via sideband
* @aux: Fake sideband AUX CH
* @offset: address of the (first) register to read
* @buffer: buffer to store the register values
* @size: number of bytes in @buffer
*
* Performs the same functionality for remote devices via
* sideband messaging as drm_dp_dpcd_read() does for local
* devices via actual AUX CH.
*
* Return: Number of bytes read, or negative error code on failure.
*/
ssize_t drm_dp_mst_dpcd_read(struct drm_dp_aux *aux,
unsigned int offset, void *buffer, size_t size)
{
struct drm_dp_mst_port *port = container_of(aux, struct drm_dp_mst_port,
aux);
return drm_dp_send_dpcd_read(port->mgr, port,
offset, size, buffer);
}
/**
* drm_dp_mst_dpcd_write() - write a series of bytes to the DPCD via sideband
* @aux: Fake sideband AUX CH
* @offset: address of the (first) register to write
* @buffer: buffer containing the values to write
* @size: number of bytes in @buffer
*
* Performs the same functionality for remote devices via
* sideband messaging as drm_dp_dpcd_write() does for local
* devices via actual AUX CH.
*
* Return: 0 on success, negative error code on failure.
*/
ssize_t drm_dp_mst_dpcd_write(struct drm_dp_aux *aux,
unsigned int offset, void *buffer, size_t size)
{
struct drm_dp_mst_port *port = container_of(aux, struct drm_dp_mst_port,
aux);
return drm_dp_send_dpcd_write(port->mgr, port,
offset, size, buffer);
}
static void drm_dp_check_mstb_guid(struct drm_dp_mst_branch *mstb, u8 *guid)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
{
int ret;
memcpy(mstb->guid, guid, 16);
if (!drm_dp_validate_guid(mstb->mgr, mstb->guid)) {
if (mstb->port_parent) {
ret = drm_dp_send_dpcd_write(
mstb->mgr,
mstb->port_parent,
DP_GUID,
16,
mstb->guid);
} else {
ret = drm_dp_dpcd_write(
mstb->mgr->aux,
DP_GUID,
mstb->guid,
16);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
}
}
static void build_mst_prop_path(const struct drm_dp_mst_branch *mstb,
int pnum,
char *proppath,
size_t proppath_size)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
{
int i;
char temp[8];
snprintf(proppath, proppath_size, "mst:%d", mstb->mgr->conn_base_id);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
for (i = 0; i < (mstb->lct - 1); i++) {
int shift = (i % 2) ? 0 : 4;
int port_num = (mstb->rad[i / 2] >> shift) & 0xf;
snprintf(temp, sizeof(temp), "-%d", port_num);
strlcat(proppath, temp, proppath_size);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
snprintf(temp, sizeof(temp), "-%d", pnum);
strlcat(proppath, temp, proppath_size);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
drm/dp_mst: Enable registration of AUX devices for MST ports All available downstream ports - physical and logical - are exposed for each MST device. They are listed in /dev/, following the same naming scheme as SST devices by appending an incremental ID. Although all downstream ports are exposed, only some will work as expected. Consider the following topology: +---------+ | ASIC | +---------+ Conn-0| | +----v----+ +----| MST HUB |----+ | +---------+ | | | |Port-1 Port-2| +-----v-----+ +-----v-----+ | MST | | SST | | Display | | Display | +-----------+ +-----------+ |Port-1 x MST Path | MST Device ----------+---------------------------------- sst:0 | MST Hub mst:0-1 | MST Display mst:0-1-1 | MST Display's disconnected DP out mst:0-1-8 | MST Display's internal sink mst:0-2 | SST Display On certain MST displays, the upstream physical port will ACK DPCD reads. However, reads on the local logical port to the internal sink will *NAK*. i.e. reading mst:0-1 ACKs, but mst:0-1-8 NAKs. There may also be duplicates. Some displays will return the same GUID when reading DPCD from both mst:0-1 and mst:0-1-8. There are some device-dependent behavior as well. The MST hub used during testing will actually *ACK* read requests on a disconnected physical port, whereas the MST displays will NAK. In light of these discrepancies, it's simpler to expose all downstream ports - both physical and logical - and let the user decide what to use. v3 changes: * Change WARN_ON_ONCE -> DRM_ERROR on dpcd read errors * Docstring and cosmetic fixes v2 changes: Moved remote aux device (un)registration to new mst connector late register and early unregister helpers. Drivers should call these from their own mst connector function hooks. This is to solve an issue during driver unload, where mst connector devices are unregistered before the remote aux devices are. In a setup where aux devices are created as children of connector devices, the aux device would be removed too early, and uncleanly. Doing so in early_unregister solves this issue, as that is called before connector unregistration. Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Signed-off-by: Leo Li <sunpeng.li@amd.com> Reviewed-by: Lyude Paul <lyude@redhat.com> Signed-off-by: Harry Wentland <harry.wentland@amd.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190723232808.28128-3-sunpeng.li@amd.com
2019-07-24 06:28:01 +07:00
/**
* drm_dp_mst_connector_late_register() - Late MST connector registration
* @connector: The MST connector
drm/dp_mst: Enable registration of AUX devices for MST ports All available downstream ports - physical and logical - are exposed for each MST device. They are listed in /dev/, following the same naming scheme as SST devices by appending an incremental ID. Although all downstream ports are exposed, only some will work as expected. Consider the following topology: +---------+ | ASIC | +---------+ Conn-0| | +----v----+ +----| MST HUB |----+ | +---------+ | | | |Port-1 Port-2| +-----v-----+ +-----v-----+ | MST | | SST | | Display | | Display | +-----------+ +-----------+ |Port-1 x MST Path | MST Device ----------+---------------------------------- sst:0 | MST Hub mst:0-1 | MST Display mst:0-1-1 | MST Display's disconnected DP out mst:0-1-8 | MST Display's internal sink mst:0-2 | SST Display On certain MST displays, the upstream physical port will ACK DPCD reads. However, reads on the local logical port to the internal sink will *NAK*. i.e. reading mst:0-1 ACKs, but mst:0-1-8 NAKs. There may also be duplicates. Some displays will return the same GUID when reading DPCD from both mst:0-1 and mst:0-1-8. There are some device-dependent behavior as well. The MST hub used during testing will actually *ACK* read requests on a disconnected physical port, whereas the MST displays will NAK. In light of these discrepancies, it's simpler to expose all downstream ports - both physical and logical - and let the user decide what to use. v3 changes: * Change WARN_ON_ONCE -> DRM_ERROR on dpcd read errors * Docstring and cosmetic fixes v2 changes: Moved remote aux device (un)registration to new mst connector late register and early unregister helpers. Drivers should call these from their own mst connector function hooks. This is to solve an issue during driver unload, where mst connector devices are unregistered before the remote aux devices are. In a setup where aux devices are created as children of connector devices, the aux device would be removed too early, and uncleanly. Doing so in early_unregister solves this issue, as that is called before connector unregistration. Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Signed-off-by: Leo Li <sunpeng.li@amd.com> Reviewed-by: Lyude Paul <lyude@redhat.com> Signed-off-by: Harry Wentland <harry.wentland@amd.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190723232808.28128-3-sunpeng.li@amd.com
2019-07-24 06:28:01 +07:00
* @port: The MST port for this connector
*
* Helper to register the remote aux device for this MST port. Drivers should
* call this from their mst connector's late_register hook to enable MST aux
* devices.
*
* Return: 0 on success, negative error code on failure.
*/
int drm_dp_mst_connector_late_register(struct drm_connector *connector,
struct drm_dp_mst_port *port)
{
DRM_DEBUG_KMS("registering %s remote bus for %s\n",
port->aux.name, connector->kdev->kobj.name);
port->aux.dev = connector->kdev;
return drm_dp_aux_register_devnode(&port->aux);
}
EXPORT_SYMBOL(drm_dp_mst_connector_late_register);
/**
* drm_dp_mst_connector_early_unregister() - Early MST connector unregistration
* @connector: The MST connector
drm/dp_mst: Enable registration of AUX devices for MST ports All available downstream ports - physical and logical - are exposed for each MST device. They are listed in /dev/, following the same naming scheme as SST devices by appending an incremental ID. Although all downstream ports are exposed, only some will work as expected. Consider the following topology: +---------+ | ASIC | +---------+ Conn-0| | +----v----+ +----| MST HUB |----+ | +---------+ | | | |Port-1 Port-2| +-----v-----+ +-----v-----+ | MST | | SST | | Display | | Display | +-----------+ +-----------+ |Port-1 x MST Path | MST Device ----------+---------------------------------- sst:0 | MST Hub mst:0-1 | MST Display mst:0-1-1 | MST Display's disconnected DP out mst:0-1-8 | MST Display's internal sink mst:0-2 | SST Display On certain MST displays, the upstream physical port will ACK DPCD reads. However, reads on the local logical port to the internal sink will *NAK*. i.e. reading mst:0-1 ACKs, but mst:0-1-8 NAKs. There may also be duplicates. Some displays will return the same GUID when reading DPCD from both mst:0-1 and mst:0-1-8. There are some device-dependent behavior as well. The MST hub used during testing will actually *ACK* read requests on a disconnected physical port, whereas the MST displays will NAK. In light of these discrepancies, it's simpler to expose all downstream ports - both physical and logical - and let the user decide what to use. v3 changes: * Change WARN_ON_ONCE -> DRM_ERROR on dpcd read errors * Docstring and cosmetic fixes v2 changes: Moved remote aux device (un)registration to new mst connector late register and early unregister helpers. Drivers should call these from their own mst connector function hooks. This is to solve an issue during driver unload, where mst connector devices are unregistered before the remote aux devices are. In a setup where aux devices are created as children of connector devices, the aux device would be removed too early, and uncleanly. Doing so in early_unregister solves this issue, as that is called before connector unregistration. Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Signed-off-by: Leo Li <sunpeng.li@amd.com> Reviewed-by: Lyude Paul <lyude@redhat.com> Signed-off-by: Harry Wentland <harry.wentland@amd.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190723232808.28128-3-sunpeng.li@amd.com
2019-07-24 06:28:01 +07:00
* @port: The MST port for this connector
*
* Helper to unregister the remote aux device for this MST port, registered by
* drm_dp_mst_connector_late_register(). Drivers should call this from their mst
* connector's early_unregister hook.
*/
void drm_dp_mst_connector_early_unregister(struct drm_connector *connector,
struct drm_dp_mst_port *port)
{
DRM_DEBUG_KMS("unregistering %s remote bus for %s\n",
port->aux.name, connector->kdev->kobj.name);
drm_dp_aux_unregister_devnode(&port->aux);
}
EXPORT_SYMBOL(drm_dp_mst_connector_early_unregister);
static void
drm_dp_mst_handle_link_address_port(struct drm_dp_mst_branch *mstb,
struct drm_device *dev,
struct drm_dp_link_addr_reply_port *port_msg)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
{
struct drm_dp_mst_port *port;
bool ret;
bool created = false;
int old_pdt = 0;
int old_ddps = 0;
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
port = drm_dp_get_port(mstb, port_msg->port_number);
if (!port) {
port = kzalloc(sizeof(*port), GFP_KERNEL);
if (!port)
return;
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
kref_init(&port->topology_kref);
kref_init(&port->malloc_kref);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
port->parent = mstb;
port->port_num = port_msg->port_number;
port->mgr = mstb->mgr;
port->aux.name = "DPMST";
port->aux.dev = dev->dev;
drm/dp_mst: Enable registration of AUX devices for MST ports All available downstream ports - physical and logical - are exposed for each MST device. They are listed in /dev/, following the same naming scheme as SST devices by appending an incremental ID. Although all downstream ports are exposed, only some will work as expected. Consider the following topology: +---------+ | ASIC | +---------+ Conn-0| | +----v----+ +----| MST HUB |----+ | +---------+ | | | |Port-1 Port-2| +-----v-----+ +-----v-----+ | MST | | SST | | Display | | Display | +-----------+ +-----------+ |Port-1 x MST Path | MST Device ----------+---------------------------------- sst:0 | MST Hub mst:0-1 | MST Display mst:0-1-1 | MST Display's disconnected DP out mst:0-1-8 | MST Display's internal sink mst:0-2 | SST Display On certain MST displays, the upstream physical port will ACK DPCD reads. However, reads on the local logical port to the internal sink will *NAK*. i.e. reading mst:0-1 ACKs, but mst:0-1-8 NAKs. There may also be duplicates. Some displays will return the same GUID when reading DPCD from both mst:0-1 and mst:0-1-8. There are some device-dependent behavior as well. The MST hub used during testing will actually *ACK* read requests on a disconnected physical port, whereas the MST displays will NAK. In light of these discrepancies, it's simpler to expose all downstream ports - both physical and logical - and let the user decide what to use. v3 changes: * Change WARN_ON_ONCE -> DRM_ERROR on dpcd read errors * Docstring and cosmetic fixes v2 changes: Moved remote aux device (un)registration to new mst connector late register and early unregister helpers. Drivers should call these from their own mst connector function hooks. This is to solve an issue during driver unload, where mst connector devices are unregistered before the remote aux devices are. In a setup where aux devices are created as children of connector devices, the aux device would be removed too early, and uncleanly. Doing so in early_unregister solves this issue, as that is called before connector unregistration. Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Signed-off-by: Leo Li <sunpeng.li@amd.com> Reviewed-by: Lyude Paul <lyude@redhat.com> Signed-off-by: Harry Wentland <harry.wentland@amd.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190723232808.28128-3-sunpeng.li@amd.com
2019-07-24 06:28:01 +07:00
port->aux.is_remote = true;
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
/*
* Make sure the memory allocation for our parent branch stays
* around until our own memory allocation is released
*/
drm_dp_mst_get_mstb_malloc(mstb);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
created = true;
} else {
old_pdt = port->pdt;
old_ddps = port->ddps;
}
port->pdt = port_msg->peer_device_type;
port->input = port_msg->input_port;
port->mcs = port_msg->mcs;
port->ddps = port_msg->ddps;
port->ldps = port_msg->legacy_device_plug_status;
port->dpcd_rev = port_msg->dpcd_revision;
port->num_sdp_streams = port_msg->num_sdp_streams;
port->num_sdp_stream_sinks = port_msg->num_sdp_stream_sinks;
/* manage mstb port lists with mgr lock - take a reference
for this list */
if (created) {
mutex_lock(&mstb->mgr->lock);
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
drm_dp_mst_topology_get_port(port);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
list_add(&port->next, &mstb->ports);
mutex_unlock(&mstb->mgr->lock);
}
if (old_ddps != port->ddps) {
if (port->ddps) {
if (!port->input) {
drm_dp_send_enum_path_resources(mstb->mgr,
mstb, port);
}
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
} else {
port->available_pbn = 0;
}
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
if (old_pdt != port->pdt && !port->input) {
drm_dp_port_teardown_pdt(port, old_pdt);
ret = drm_dp_port_setup_pdt(port);
if (ret == true)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
drm_dp_send_link_address(mstb->mgr, port->mstb);
}
if (created && !port->input) {
char proppath[255];
build_mst_prop_path(mstb, port->port_num, proppath,
sizeof(proppath));
port->connector = (*mstb->mgr->cbs->add_connector)(mstb->mgr,
port,
proppath);
if (!port->connector) {
/* remove it from the port list */
mutex_lock(&mstb->mgr->lock);
list_del(&port->next);
mutex_unlock(&mstb->mgr->lock);
/* drop port list reference */
drm_dp_mst_topology_put_port(port);
goto out;
}
if ((port->pdt == DP_PEER_DEVICE_DP_LEGACY_CONV ||
port->pdt == DP_PEER_DEVICE_SST_SINK) &&
port->port_num >= DP_MST_LOGICAL_PORT_0) {
port->cached_edid = drm_get_edid(port->connector,
&port->aux.ddc);
drm_connector_set_tile_property(port->connector);
}
(*mstb->mgr->cbs->register_connector)(port->connector);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
out:
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
/* put reference to this port */
drm_dp_mst_topology_put_port(port);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
static void
drm_dp_mst_handle_conn_stat(struct drm_dp_mst_branch *mstb,
struct drm_dp_connection_status_notify *conn_stat)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
{
struct drm_dp_mst_port *port;
int old_pdt;
int old_ddps;
bool dowork = false;
port = drm_dp_get_port(mstb, conn_stat->port_number);
if (!port)
return;
old_ddps = port->ddps;
old_pdt = port->pdt;
port->pdt = conn_stat->peer_device_type;
port->mcs = conn_stat->message_capability_status;
port->ldps = conn_stat->legacy_device_plug_status;
port->ddps = conn_stat->displayport_device_plug_status;
if (old_ddps != port->ddps) {
if (port->ddps) {
dowork = true;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
} else {
port->available_pbn = 0;
}
}
if (old_pdt != port->pdt && !port->input) {
drm_dp_port_teardown_pdt(port, old_pdt);
if (drm_dp_port_setup_pdt(port))
dowork = true;
}
drm_dp_mst_topology_put_port(port);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
if (dowork)
queue_work(system_long_wq, &mstb->mgr->work);
}
static struct drm_dp_mst_branch *drm_dp_get_mst_branch_device(struct drm_dp_mst_topology_mgr *mgr,
u8 lct, u8 *rad)
{
struct drm_dp_mst_branch *mstb;
struct drm_dp_mst_port *port;
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
int i, ret;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
/* find the port by iterating down */
mutex_lock(&mgr->lock);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
mstb = mgr->mst_primary;
if (!mstb)
goto out;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
for (i = 0; i < lct - 1; i++) {
int shift = (i % 2) ? 0 : 4;
int port_num = (rad[i / 2] >> shift) & 0xf;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
list_for_each_entry(port, &mstb->ports, next) {
if (port->port_num == port_num) {
mstb = port->mstb;
if (!mstb) {
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
DRM_ERROR("failed to lookup MSTB with lct %d, rad %02x\n", lct, rad[0]);
goto out;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
break;
}
}
}
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
ret = drm_dp_mst_topology_try_get_mstb(mstb);
if (!ret)
mstb = NULL;
out:
mutex_unlock(&mgr->lock);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
return mstb;
}
static struct drm_dp_mst_branch *get_mst_branch_device_by_guid_helper(
struct drm_dp_mst_branch *mstb,
const uint8_t *guid)
{
struct drm_dp_mst_branch *found_mstb;
struct drm_dp_mst_port *port;
if (memcmp(mstb->guid, guid, 16) == 0)
return mstb;
list_for_each_entry(port, &mstb->ports, next) {
if (!port->mstb)
continue;
found_mstb = get_mst_branch_device_by_guid_helper(port->mstb, guid);
if (found_mstb)
return found_mstb;
}
return NULL;
}
static struct drm_dp_mst_branch *
drm_dp_get_mst_branch_device_by_guid(struct drm_dp_mst_topology_mgr *mgr,
const uint8_t *guid)
{
struct drm_dp_mst_branch *mstb;
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
int ret;
/* find the port by iterating down */
mutex_lock(&mgr->lock);
mstb = get_mst_branch_device_by_guid_helper(mgr->mst_primary, guid);
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
if (mstb) {
ret = drm_dp_mst_topology_try_get_mstb(mstb);
if (!ret)
mstb = NULL;
}
mutex_unlock(&mgr->lock);
return mstb;
}
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
static void drm_dp_check_and_send_link_address(struct drm_dp_mst_topology_mgr *mgr,
struct drm_dp_mst_branch *mstb)
{
struct drm_dp_mst_port *port;
struct drm_dp_mst_branch *mstb_child;
if (!mstb->link_address_sent)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
drm_dp_send_link_address(mgr, mstb);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
list_for_each_entry(port, &mstb->ports, next) {
if (port->input)
continue;
if (!port->ddps)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
continue;
if (!port->available_pbn)
drm_dp_send_enum_path_resources(mgr, mstb, port);
if (port->mstb) {
mstb_child = drm_dp_mst_topology_get_mstb_validated(
mgr, port->mstb);
if (mstb_child) {
drm_dp_check_and_send_link_address(mgr, mstb_child);
drm_dp_mst_topology_put_mstb(mstb_child);
}
}
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
}
static void drm_dp_mst_link_probe_work(struct work_struct *work)
{
struct drm_dp_mst_topology_mgr *mgr = container_of(work, struct drm_dp_mst_topology_mgr, work);
struct drm_dp_mst_branch *mstb;
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
int ret;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
mutex_lock(&mgr->lock);
mstb = mgr->mst_primary;
if (mstb) {
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
ret = drm_dp_mst_topology_try_get_mstb(mstb);
if (!ret)
mstb = NULL;
}
mutex_unlock(&mgr->lock);
if (mstb) {
drm_dp_check_and_send_link_address(mgr, mstb);
drm_dp_mst_topology_put_mstb(mstb);
}
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
static bool drm_dp_validate_guid(struct drm_dp_mst_topology_mgr *mgr,
u8 *guid)
{
u64 salt;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
if (memchr_inv(guid, 0, 16))
return true;
salt = get_jiffies_64();
memcpy(&guid[0], &salt, sizeof(u64));
memcpy(&guid[8], &salt, sizeof(u64));
return false;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
static int build_dpcd_read(struct drm_dp_sideband_msg_tx *msg, u8 port_num, u32 offset, u8 num_bytes)
{
struct drm_dp_sideband_msg_req_body req;
req.req_type = DP_REMOTE_DPCD_READ;
req.u.dpcd_read.port_number = port_num;
req.u.dpcd_read.dpcd_address = offset;
req.u.dpcd_read.num_bytes = num_bytes;
drm_dp_encode_sideband_req(&req, msg);
return 0;
}
static int drm_dp_send_sideband_msg(struct drm_dp_mst_topology_mgr *mgr,
bool up, u8 *msg, int len)
{
int ret;
int regbase = up ? DP_SIDEBAND_MSG_UP_REP_BASE : DP_SIDEBAND_MSG_DOWN_REQ_BASE;
int tosend, total, offset;
int retries = 0;
retry:
total = len;
offset = 0;
do {
tosend = min3(mgr->max_dpcd_transaction_bytes, 16, total);
ret = drm_dp_dpcd_write(mgr->aux, regbase + offset,
&msg[offset],
tosend);
if (ret != tosend) {
if (ret == -EIO && retries < 5) {
retries++;
goto retry;
}
DRM_DEBUG_KMS("failed to dpcd write %d %d\n", tosend, ret);
return -EIO;
}
offset += tosend;
total -= tosend;
} while (total > 0);
return 0;
}
static int set_hdr_from_dst_qlock(struct drm_dp_sideband_msg_hdr *hdr,
struct drm_dp_sideband_msg_tx *txmsg)
{
struct drm_dp_mst_branch *mstb = txmsg->dst;
u8 req_type;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
/* both msg slots are full */
if (txmsg->seqno == -1) {
if (mstb->tx_slots[0] && mstb->tx_slots[1]) {
DRM_DEBUG_KMS("%s: failed to find slot\n", __func__);
return -EAGAIN;
}
if (mstb->tx_slots[0] == NULL && mstb->tx_slots[1] == NULL) {
txmsg->seqno = mstb->last_seqno;
mstb->last_seqno ^= 1;
} else if (mstb->tx_slots[0] == NULL)
txmsg->seqno = 0;
else
txmsg->seqno = 1;
mstb->tx_slots[txmsg->seqno] = txmsg;
}
req_type = txmsg->msg[0] & 0x7f;
if (req_type == DP_CONNECTION_STATUS_NOTIFY ||
req_type == DP_RESOURCE_STATUS_NOTIFY)
hdr->broadcast = 1;
else
hdr->broadcast = 0;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
hdr->path_msg = txmsg->path_msg;
hdr->lct = mstb->lct;
hdr->lcr = mstb->lct - 1;
if (mstb->lct > 1)
memcpy(hdr->rad, mstb->rad, mstb->lct / 2);
hdr->seqno = txmsg->seqno;
return 0;
}
/*
* process a single block of the next message in the sideband queue
*/
static int process_single_tx_qlock(struct drm_dp_mst_topology_mgr *mgr,
struct drm_dp_sideband_msg_tx *txmsg,
bool up)
{
u8 chunk[48];
struct drm_dp_sideband_msg_hdr hdr;
int len, space, idx, tosend;
int ret;
memset(&hdr, 0, sizeof(struct drm_dp_sideband_msg_hdr));
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
if (txmsg->state == DRM_DP_SIDEBAND_TX_QUEUED) {
txmsg->seqno = -1;
txmsg->state = DRM_DP_SIDEBAND_TX_START_SEND;
}
/* make hdr from dst mst - for replies use seqno
otherwise assign one */
ret = set_hdr_from_dst_qlock(&hdr, txmsg);
if (ret < 0)
return ret;
/* amount left to send in this message */
len = txmsg->cur_len - txmsg->cur_offset;
/* 48 - sideband msg size - 1 byte for data CRC, x header bytes */
space = 48 - 1 - drm_dp_calc_sb_hdr_size(&hdr);
tosend = min(len, space);
if (len == txmsg->cur_len)
hdr.somt = 1;
if (space >= len)
hdr.eomt = 1;
hdr.msg_len = tosend + 1;
drm_dp_encode_sideband_msg_hdr(&hdr, chunk, &idx);
memcpy(&chunk[idx], &txmsg->msg[txmsg->cur_offset], tosend);
/* add crc at end */
drm_dp_crc_sideband_chunk_req(&chunk[idx], tosend);
idx += tosend + 1;
ret = drm_dp_send_sideband_msg(mgr, up, chunk, idx);
if (unlikely(ret) && drm_debug_enabled(DRM_UT_DP)) {
struct drm_printer p = drm_debug_printer(DBG_PREFIX);
drm_printf(&p, "sideband msg failed to send\n");
drm_dp_mst_dump_sideband_msg_tx(&p, txmsg);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
return ret;
}
txmsg->cur_offset += tosend;
if (txmsg->cur_offset == txmsg->cur_len) {
txmsg->state = DRM_DP_SIDEBAND_TX_SENT;
return 1;
}
return 0;
}
static void process_single_down_tx_qlock(struct drm_dp_mst_topology_mgr *mgr)
{
struct drm_dp_sideband_msg_tx *txmsg;
int ret;
WARN_ON(!mutex_is_locked(&mgr->qlock));
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
/* construct a chunk from the first msg in the tx_msg queue */
if (list_empty(&mgr->tx_msg_downq))
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
return;
txmsg = list_first_entry(&mgr->tx_msg_downq, struct drm_dp_sideband_msg_tx, next);
ret = process_single_tx_qlock(mgr, txmsg, false);
if (ret == 1) {
/* txmsg is sent it should be in the slots now */
list_del(&txmsg->next);
} else if (ret) {
DRM_DEBUG_KMS("failed to send msg in q %d\n", ret);
list_del(&txmsg->next);
if (txmsg->seqno != -1)
txmsg->dst->tx_slots[txmsg->seqno] = NULL;
txmsg->state = DRM_DP_SIDEBAND_TX_TIMEOUT;
wake_up_all(&mgr->tx_waitq);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
}
/* called holding qlock */
static void process_single_up_tx_qlock(struct drm_dp_mst_topology_mgr *mgr,
struct drm_dp_sideband_msg_tx *txmsg)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
{
int ret;
/* construct a chunk from the first msg in the tx_msg queue */
ret = process_single_tx_qlock(mgr, txmsg, true);
if (ret != 1)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
DRM_DEBUG_KMS("failed to send msg in q %d\n", ret);
drm/mst: Fix MST sideband up-reply failure handling Fix the breakage resulting in the stacktrace below, due to tx queue being full when trying to send an up-reply. txmsg->seqno is -1 in this case leading to a corruption of the mstb object by txmsg->dst->tx_slots[txmsg->seqno] = NULL; in process_single_up_tx_qlock(). [ +0,005162] [drm:process_single_tx_qlock [drm_kms_helper]] set_hdr_from_dst_qlock: failed to find slot [ +0,000015] [drm:drm_dp_send_up_ack_reply.constprop.19 [drm_kms_helper]] failed to send msg in q -11 [ +0,000939] BUG: kernel NULL pointer dereference, address: 00000000000005a0 [ +0,006982] #PF: supervisor write access in kernel mode [ +0,005223] #PF: error_code(0x0002) - not-present page [ +0,005135] PGD 0 P4D 0 [ +0,002581] Oops: 0002 [#1] PREEMPT SMP NOPTI [ +0,004359] CPU: 1 PID: 1200 Comm: kworker/u16:3 Tainted: G U 5.2.0-rc1+ #410 [ +0,008433] Hardware name: Intel Corporation Ice Lake Client Platform/IceLake U DDR4 SODIMM PD RVP, BIOS ICLSFWR1.R00.3175.A00.1904261428 04/26/2019 [ +0,013323] Workqueue: i915-dp i915_digport_work_func [i915] [ +0,005676] RIP: 0010:queue_work_on+0x19/0x70 [ +0,004372] Code: ff ff ff 0f 1f 40 00 66 2e 0f 1f 84 00 00 00 00 00 41 56 49 89 f6 41 55 41 89 fd 41 54 55 53 48 89 d3 9c 5d fa e8 e7 81 0c 00 <f0> 48 0f ba 2b 00 73 31 45 31 e4 f7 c5 00 02 00 00 74 13 e8 cf 7f [ +0,018750] RSP: 0018:ffffc900007dfc50 EFLAGS: 00010006 [ +0,005222] RAX: 0000000000000046 RBX: 00000000000005a0 RCX: 0000000000000001 [ +0,007133] RDX: 000000000001b608 RSI: 0000000000000000 RDI: ffffffff82121972 [ +0,007129] RBP: 0000000000000202 R08: 0000000000000000 R09: 0000000000000001 [ +0,007129] R10: 0000000000000000 R11: 0000000000000000 R12: ffff88847bfa5096 [ +0,007131] R13: 0000000000000010 R14: ffff88849c08f3f8 R15: 0000000000000000 [ +0,007128] FS: 0000000000000000(0000) GS:ffff88849dc80000(0000) knlGS:0000000000000000 [ +0,008083] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ +0,005749] CR2: 00000000000005a0 CR3: 0000000005210006 CR4: 0000000000760ee0 [ +0,007128] PKRU: 55555554 [ +0,002722] Call Trace: [ +0,002458] drm_dp_mst_handle_up_req+0x517/0x540 [drm_kms_helper] [ +0,006197] ? drm_dp_mst_hpd_irq+0x5b/0x9c0 [drm_kms_helper] [ +0,005764] drm_dp_mst_hpd_irq+0x5b/0x9c0 [drm_kms_helper] [ +0,005623] ? intel_dp_hpd_pulse+0x205/0x370 [i915] [ +0,005018] intel_dp_hpd_pulse+0x205/0x370 [i915] [ +0,004836] i915_digport_work_func+0xbb/0x140 [i915] [ +0,005108] process_one_work+0x245/0x610 [ +0,004027] worker_thread+0x37/0x380 [ +0,003684] ? process_one_work+0x610/0x610 [ +0,004184] kthread+0x119/0x130 [ +0,003240] ? kthread_park+0x80/0x80 [ +0,003668] ret_from_fork+0x24/0x50 Cc: Lyude Paul <lyude@redhat.com> Cc: Dave Airlie <airlied@redhat.com> Signed-off-by: Imre Deak <imre.deak@intel.com> Reviewed-by: Lyude Paul <lyude@redhat.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190523212433.9058-1-imre.deak@intel.com
2019-05-24 04:24:33 +07:00
if (txmsg->seqno != -1) {
WARN_ON((unsigned int)txmsg->seqno >
ARRAY_SIZE(txmsg->dst->tx_slots));
txmsg->dst->tx_slots[txmsg->seqno] = NULL;
}
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
static void drm_dp_queue_down_tx(struct drm_dp_mst_topology_mgr *mgr,
struct drm_dp_sideband_msg_tx *txmsg)
{
mutex_lock(&mgr->qlock);
list_add_tail(&txmsg->next, &mgr->tx_msg_downq);
if (drm_debug_enabled(DRM_UT_DP)) {
struct drm_printer p = drm_debug_printer(DBG_PREFIX);
drm_dp_mst_dump_sideband_msg_tx(&p, txmsg);
}
if (list_is_singular(&mgr->tx_msg_downq))
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
process_single_down_tx_qlock(mgr);
mutex_unlock(&mgr->qlock);
}
static void
drm_dp_dump_link_address(struct drm_dp_link_address_ack_reply *reply)
{
struct drm_dp_link_addr_reply_port *port_reply;
int i;
for (i = 0; i < reply->nports; i++) {
port_reply = &reply->ports[i];
DRM_DEBUG_KMS("port %d: input %d, pdt: %d, pn: %d, dpcd_rev: %02x, mcs: %d, ddps: %d, ldps %d, sdp %d/%d\n",
i,
port_reply->input_port,
port_reply->peer_device_type,
port_reply->port_number,
port_reply->dpcd_revision,
port_reply->mcs,
port_reply->ddps,
port_reply->legacy_device_plug_status,
port_reply->num_sdp_streams,
port_reply->num_sdp_stream_sinks);
}
}
static void drm_dp_send_link_address(struct drm_dp_mst_topology_mgr *mgr,
struct drm_dp_mst_branch *mstb)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
{
struct drm_dp_sideband_msg_tx *txmsg;
struct drm_dp_link_address_ack_reply *reply;
int i, len, ret;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
txmsg = kzalloc(sizeof(*txmsg), GFP_KERNEL);
if (!txmsg)
return;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
txmsg->dst = mstb;
len = build_link_address(txmsg);
mstb->link_address_sent = true;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
drm_dp_queue_down_tx(mgr, txmsg);
/* FIXME: Actually do some real error handling here */
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
ret = drm_dp_mst_wait_tx_reply(mstb, txmsg);
if (ret <= 0) {
DRM_ERROR("Sending link address failed with %d\n", ret);
goto out;
}
if (txmsg->reply.reply_type == DP_SIDEBAND_REPLY_NAK) {
DRM_ERROR("link address NAK received\n");
ret = -EIO;
goto out;
}
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
reply = &txmsg->reply.u.link_addr;
DRM_DEBUG_KMS("link address reply: %d\n", reply->nports);
drm_dp_dump_link_address(reply);
drm_dp_check_mstb_guid(mstb, reply->guid);
for (i = 0; i < reply->nports; i++)
drm_dp_mst_handle_link_address_port(mstb, mgr->dev,
&reply->ports[i]);
drm_kms_helper_hotplug_event(mgr->dev);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
out:
if (ret <= 0)
mstb->link_address_sent = false;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
kfree(txmsg);
}
static int
drm_dp_send_enum_path_resources(struct drm_dp_mst_topology_mgr *mgr,
struct drm_dp_mst_branch *mstb,
struct drm_dp_mst_port *port)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
{
struct drm_dp_enum_path_resources_ack_reply *path_res;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
struct drm_dp_sideband_msg_tx *txmsg;
int len;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
int ret;
txmsg = kzalloc(sizeof(*txmsg), GFP_KERNEL);
if (!txmsg)
return -ENOMEM;
txmsg->dst = mstb;
len = build_enum_path_resources(txmsg, port->port_num);
drm_dp_queue_down_tx(mgr, txmsg);
ret = drm_dp_mst_wait_tx_reply(mstb, txmsg);
if (ret > 0) {
path_res = &txmsg->reply.u.path_resources;
if (txmsg->reply.reply_type == DP_SIDEBAND_REPLY_NAK) {
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
DRM_DEBUG_KMS("enum path resources nak received\n");
} else {
if (port->port_num != path_res->port_number)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
DRM_ERROR("got incorrect port in response\n");
DRM_DEBUG_KMS("enum path resources %d: %d %d\n",
path_res->port_number,
path_res->full_payload_bw_number,
path_res->avail_payload_bw_number);
port->available_pbn =
path_res->avail_payload_bw_number;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
}
kfree(txmsg);
return 0;
}
static struct drm_dp_mst_port *drm_dp_get_last_connected_port_to_mstb(struct drm_dp_mst_branch *mstb)
{
if (!mstb->port_parent)
return NULL;
if (mstb->port_parent->mstb != mstb)
return mstb->port_parent;
return drm_dp_get_last_connected_port_to_mstb(mstb->port_parent->parent);
}
drm/dp_mst: Restart last_connected_port_and_mstb() if topology ref fails While this isn't a complete fix, this will improve the reliability of drm_dp_get_last_connected_port_and_mstb() pretty significantly during hotplug events, since there's a chance that the in-memory topology tree may not be fully updated when drm_dp_get_last_connected_port_and_mstb() is called and thus might end up causing our search to fail on an mstb whose topology refcount has reached 0, but has not yet been removed from it's parent. Ideally, we should further fix this problem by ensuring that we deal with the potential for racing with a hotplug event, which would look like this: * drm_dp_payload_send_msg() retrieves the last living relative of mstb with drm_dp_get_last_connected_port_and_mstb() * drm_dp_payload_send_msg() starts building payload message At the same time, mstb gets unplugged from the topology and is no longer the actual last living relative of the original mstb * drm_dp_payload_send_msg() tries sending the payload message, hub times out * Hub timed out, we give up and run away-resulting in the payload being leaked This could be fixed by restarting the drm_dp_get_last_connected_port_and_mstb() search whenever we get a timeout, sending the payload to the new mstb, then repeating until either the entire topology is removed from the system or drm_dp_get_last_connected_port_and_mstb() fails. But since the above race condition is not terribly likely, we'll address that in a later patch series once we've improved the recovery handling for VCPI allocations in the rest of the DP MST helpers. Changes since v1: * Convert kerneldoc for drm_dp_get_last_connected_port_and_mstb to normal comment - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-8-lyude@redhat.com
2019-01-11 07:53:30 +07:00
/*
* Searches upwards in the topology starting from mstb to try to find the
* closest available parent of mstb that's still connected to the rest of the
* topology. This can be used in order to perform operations like releasing
* payloads, where the branch device which owned the payload may no longer be
* around and thus would require that the payload on the last living relative
* be freed instead.
*/
static struct drm_dp_mst_branch *
drm_dp_get_last_connected_port_and_mstb(struct drm_dp_mst_topology_mgr *mgr,
struct drm_dp_mst_branch *mstb,
int *port_num)
{
struct drm_dp_mst_branch *rmstb = NULL;
struct drm_dp_mst_port *found_port;
drm/dp_mst: Restart last_connected_port_and_mstb() if topology ref fails While this isn't a complete fix, this will improve the reliability of drm_dp_get_last_connected_port_and_mstb() pretty significantly during hotplug events, since there's a chance that the in-memory topology tree may not be fully updated when drm_dp_get_last_connected_port_and_mstb() is called and thus might end up causing our search to fail on an mstb whose topology refcount has reached 0, but has not yet been removed from it's parent. Ideally, we should further fix this problem by ensuring that we deal with the potential for racing with a hotplug event, which would look like this: * drm_dp_payload_send_msg() retrieves the last living relative of mstb with drm_dp_get_last_connected_port_and_mstb() * drm_dp_payload_send_msg() starts building payload message At the same time, mstb gets unplugged from the topology and is no longer the actual last living relative of the original mstb * drm_dp_payload_send_msg() tries sending the payload message, hub times out * Hub timed out, we give up and run away-resulting in the payload being leaked This could be fixed by restarting the drm_dp_get_last_connected_port_and_mstb() search whenever we get a timeout, sending the payload to the new mstb, then repeating until either the entire topology is removed from the system or drm_dp_get_last_connected_port_and_mstb() fails. But since the above race condition is not terribly likely, we'll address that in a later patch series once we've improved the recovery handling for VCPI allocations in the rest of the DP MST helpers. Changes since v1: * Convert kerneldoc for drm_dp_get_last_connected_port_and_mstb to normal comment - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-8-lyude@redhat.com
2019-01-11 07:53:30 +07:00
mutex_lock(&mgr->lock);
drm/dp_mst: Restart last_connected_port_and_mstb() if topology ref fails While this isn't a complete fix, this will improve the reliability of drm_dp_get_last_connected_port_and_mstb() pretty significantly during hotplug events, since there's a chance that the in-memory topology tree may not be fully updated when drm_dp_get_last_connected_port_and_mstb() is called and thus might end up causing our search to fail on an mstb whose topology refcount has reached 0, but has not yet been removed from it's parent. Ideally, we should further fix this problem by ensuring that we deal with the potential for racing with a hotplug event, which would look like this: * drm_dp_payload_send_msg() retrieves the last living relative of mstb with drm_dp_get_last_connected_port_and_mstb() * drm_dp_payload_send_msg() starts building payload message At the same time, mstb gets unplugged from the topology and is no longer the actual last living relative of the original mstb * drm_dp_payload_send_msg() tries sending the payload message, hub times out * Hub timed out, we give up and run away-resulting in the payload being leaked This could be fixed by restarting the drm_dp_get_last_connected_port_and_mstb() search whenever we get a timeout, sending the payload to the new mstb, then repeating until either the entire topology is removed from the system or drm_dp_get_last_connected_port_and_mstb() fails. But since the above race condition is not terribly likely, we'll address that in a later patch series once we've improved the recovery handling for VCPI allocations in the rest of the DP MST helpers. Changes since v1: * Convert kerneldoc for drm_dp_get_last_connected_port_and_mstb to normal comment - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-8-lyude@redhat.com
2019-01-11 07:53:30 +07:00
if (!mgr->mst_primary)
goto out;
do {
found_port = drm_dp_get_last_connected_port_to_mstb(mstb);
drm/dp_mst: Restart last_connected_port_and_mstb() if topology ref fails While this isn't a complete fix, this will improve the reliability of drm_dp_get_last_connected_port_and_mstb() pretty significantly during hotplug events, since there's a chance that the in-memory topology tree may not be fully updated when drm_dp_get_last_connected_port_and_mstb() is called and thus might end up causing our search to fail on an mstb whose topology refcount has reached 0, but has not yet been removed from it's parent. Ideally, we should further fix this problem by ensuring that we deal with the potential for racing with a hotplug event, which would look like this: * drm_dp_payload_send_msg() retrieves the last living relative of mstb with drm_dp_get_last_connected_port_and_mstb() * drm_dp_payload_send_msg() starts building payload message At the same time, mstb gets unplugged from the topology and is no longer the actual last living relative of the original mstb * drm_dp_payload_send_msg() tries sending the payload message, hub times out * Hub timed out, we give up and run away-resulting in the payload being leaked This could be fixed by restarting the drm_dp_get_last_connected_port_and_mstb() search whenever we get a timeout, sending the payload to the new mstb, then repeating until either the entire topology is removed from the system or drm_dp_get_last_connected_port_and_mstb() fails. But since the above race condition is not terribly likely, we'll address that in a later patch series once we've improved the recovery handling for VCPI allocations in the rest of the DP MST helpers. Changes since v1: * Convert kerneldoc for drm_dp_get_last_connected_port_and_mstb to normal comment - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-8-lyude@redhat.com
2019-01-11 07:53:30 +07:00
if (!found_port)
break;
drm/dp_mst: Restart last_connected_port_and_mstb() if topology ref fails While this isn't a complete fix, this will improve the reliability of drm_dp_get_last_connected_port_and_mstb() pretty significantly during hotplug events, since there's a chance that the in-memory topology tree may not be fully updated when drm_dp_get_last_connected_port_and_mstb() is called and thus might end up causing our search to fail on an mstb whose topology refcount has reached 0, but has not yet been removed from it's parent. Ideally, we should further fix this problem by ensuring that we deal with the potential for racing with a hotplug event, which would look like this: * drm_dp_payload_send_msg() retrieves the last living relative of mstb with drm_dp_get_last_connected_port_and_mstb() * drm_dp_payload_send_msg() starts building payload message At the same time, mstb gets unplugged from the topology and is no longer the actual last living relative of the original mstb * drm_dp_payload_send_msg() tries sending the payload message, hub times out * Hub timed out, we give up and run away-resulting in the payload being leaked This could be fixed by restarting the drm_dp_get_last_connected_port_and_mstb() search whenever we get a timeout, sending the payload to the new mstb, then repeating until either the entire topology is removed from the system or drm_dp_get_last_connected_port_and_mstb() fails. But since the above race condition is not terribly likely, we'll address that in a later patch series once we've improved the recovery handling for VCPI allocations in the rest of the DP MST helpers. Changes since v1: * Convert kerneldoc for drm_dp_get_last_connected_port_and_mstb to normal comment - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-8-lyude@redhat.com
2019-01-11 07:53:30 +07:00
if (drm_dp_mst_topology_try_get_mstb(found_port->parent)) {
rmstb = found_port->parent;
drm/dp_mst: Restart last_connected_port_and_mstb() if topology ref fails While this isn't a complete fix, this will improve the reliability of drm_dp_get_last_connected_port_and_mstb() pretty significantly during hotplug events, since there's a chance that the in-memory topology tree may not be fully updated when drm_dp_get_last_connected_port_and_mstb() is called and thus might end up causing our search to fail on an mstb whose topology refcount has reached 0, but has not yet been removed from it's parent. Ideally, we should further fix this problem by ensuring that we deal with the potential for racing with a hotplug event, which would look like this: * drm_dp_payload_send_msg() retrieves the last living relative of mstb with drm_dp_get_last_connected_port_and_mstb() * drm_dp_payload_send_msg() starts building payload message At the same time, mstb gets unplugged from the topology and is no longer the actual last living relative of the original mstb * drm_dp_payload_send_msg() tries sending the payload message, hub times out * Hub timed out, we give up and run away-resulting in the payload being leaked This could be fixed by restarting the drm_dp_get_last_connected_port_and_mstb() search whenever we get a timeout, sending the payload to the new mstb, then repeating until either the entire topology is removed from the system or drm_dp_get_last_connected_port_and_mstb() fails. But since the above race condition is not terribly likely, we'll address that in a later patch series once we've improved the recovery handling for VCPI allocations in the rest of the DP MST helpers. Changes since v1: * Convert kerneldoc for drm_dp_get_last_connected_port_and_mstb to normal comment - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-8-lyude@redhat.com
2019-01-11 07:53:30 +07:00
*port_num = found_port->port_num;
} else {
/* Search again, starting from this parent */
mstb = found_port->parent;
}
drm/dp_mst: Restart last_connected_port_and_mstb() if topology ref fails While this isn't a complete fix, this will improve the reliability of drm_dp_get_last_connected_port_and_mstb() pretty significantly during hotplug events, since there's a chance that the in-memory topology tree may not be fully updated when drm_dp_get_last_connected_port_and_mstb() is called and thus might end up causing our search to fail on an mstb whose topology refcount has reached 0, but has not yet been removed from it's parent. Ideally, we should further fix this problem by ensuring that we deal with the potential for racing with a hotplug event, which would look like this: * drm_dp_payload_send_msg() retrieves the last living relative of mstb with drm_dp_get_last_connected_port_and_mstb() * drm_dp_payload_send_msg() starts building payload message At the same time, mstb gets unplugged from the topology and is no longer the actual last living relative of the original mstb * drm_dp_payload_send_msg() tries sending the payload message, hub times out * Hub timed out, we give up and run away-resulting in the payload being leaked This could be fixed by restarting the drm_dp_get_last_connected_port_and_mstb() search whenever we get a timeout, sending the payload to the new mstb, then repeating until either the entire topology is removed from the system or drm_dp_get_last_connected_port_and_mstb() fails. But since the above race condition is not terribly likely, we'll address that in a later patch series once we've improved the recovery handling for VCPI allocations in the rest of the DP MST helpers. Changes since v1: * Convert kerneldoc for drm_dp_get_last_connected_port_and_mstb to normal comment - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-8-lyude@redhat.com
2019-01-11 07:53:30 +07:00
} while (!rmstb);
out:
mutex_unlock(&mgr->lock);
return rmstb;
}
static int drm_dp_payload_send_msg(struct drm_dp_mst_topology_mgr *mgr,
struct drm_dp_mst_port *port,
int id,
int pbn)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
{
struct drm_dp_sideband_msg_tx *txmsg;
struct drm_dp_mst_branch *mstb;
int len, ret, port_num;
u8 sinks[DRM_DP_MAX_SDP_STREAMS];
int i;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
port_num = port->port_num;
mstb = drm_dp_mst_topology_get_mstb_validated(mgr, port->parent);
if (!mstb) {
mstb = drm_dp_get_last_connected_port_and_mstb(mgr,
port->parent,
&port_num);
if (!mstb)
return -EINVAL;
}
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
txmsg = kzalloc(sizeof(*txmsg), GFP_KERNEL);
if (!txmsg) {
ret = -ENOMEM;
goto fail_put;
}
for (i = 0; i < port->num_sdp_streams; i++)
sinks[i] = i;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
txmsg->dst = mstb;
len = build_allocate_payload(txmsg, port_num,
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
id,
pbn, port->num_sdp_streams, sinks);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
drm_dp_queue_down_tx(mgr, txmsg);
drm/dp_mst: Restart last_connected_port_and_mstb() if topology ref fails While this isn't a complete fix, this will improve the reliability of drm_dp_get_last_connected_port_and_mstb() pretty significantly during hotplug events, since there's a chance that the in-memory topology tree may not be fully updated when drm_dp_get_last_connected_port_and_mstb() is called and thus might end up causing our search to fail on an mstb whose topology refcount has reached 0, but has not yet been removed from it's parent. Ideally, we should further fix this problem by ensuring that we deal with the potential for racing with a hotplug event, which would look like this: * drm_dp_payload_send_msg() retrieves the last living relative of mstb with drm_dp_get_last_connected_port_and_mstb() * drm_dp_payload_send_msg() starts building payload message At the same time, mstb gets unplugged from the topology and is no longer the actual last living relative of the original mstb * drm_dp_payload_send_msg() tries sending the payload message, hub times out * Hub timed out, we give up and run away-resulting in the payload being leaked This could be fixed by restarting the drm_dp_get_last_connected_port_and_mstb() search whenever we get a timeout, sending the payload to the new mstb, then repeating until either the entire topology is removed from the system or drm_dp_get_last_connected_port_and_mstb() fails. But since the above race condition is not terribly likely, we'll address that in a later patch series once we've improved the recovery handling for VCPI allocations in the rest of the DP MST helpers. Changes since v1: * Convert kerneldoc for drm_dp_get_last_connected_port_and_mstb to normal comment - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-8-lyude@redhat.com
2019-01-11 07:53:30 +07:00
/*
* FIXME: there is a small chance that between getting the last
* connected mstb and sending the payload message, the last connected
* mstb could also be removed from the topology. In the future, this
* needs to be fixed by restarting the
* drm_dp_get_last_connected_port_and_mstb() search in the event of a
* timeout if the topology is still connected to the system.
*/
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
ret = drm_dp_mst_wait_tx_reply(mstb, txmsg);
if (ret > 0) {
if (txmsg->reply.reply_type == DP_SIDEBAND_REPLY_NAK)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
ret = -EINVAL;
else
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
ret = 0;
}
kfree(txmsg);
fail_put:
drm_dp_mst_topology_put_mstb(mstb);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
return ret;
}
int drm_dp_send_power_updown_phy(struct drm_dp_mst_topology_mgr *mgr,
struct drm_dp_mst_port *port, bool power_up)
{
struct drm_dp_sideband_msg_tx *txmsg;
int len, ret;
port = drm_dp_mst_topology_get_port_validated(mgr, port);
if (!port)
return -EINVAL;
txmsg = kzalloc(sizeof(*txmsg), GFP_KERNEL);
if (!txmsg) {
drm_dp_mst_topology_put_port(port);
return -ENOMEM;
}
txmsg->dst = port->parent;
len = build_power_updown_phy(txmsg, port->port_num, power_up);
drm_dp_queue_down_tx(mgr, txmsg);
ret = drm_dp_mst_wait_tx_reply(port->parent, txmsg);
if (ret > 0) {
if (txmsg->reply.reply_type == DP_SIDEBAND_REPLY_NAK)
ret = -EINVAL;
else
ret = 0;
}
kfree(txmsg);
drm_dp_mst_topology_put_port(port);
return ret;
}
EXPORT_SYMBOL(drm_dp_send_power_updown_phy);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
static int drm_dp_create_payload_step1(struct drm_dp_mst_topology_mgr *mgr,
int id,
struct drm_dp_payload *payload)
{
int ret;
ret = drm_dp_dpcd_write_payload(mgr, id, payload);
if (ret < 0) {
payload->payload_state = 0;
return ret;
}
payload->payload_state = DP_PAYLOAD_LOCAL;
return 0;
}
static int drm_dp_create_payload_step2(struct drm_dp_mst_topology_mgr *mgr,
struct drm_dp_mst_port *port,
int id,
struct drm_dp_payload *payload)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
{
int ret;
ret = drm_dp_payload_send_msg(mgr, port, id, port->vcpi.pbn);
if (ret < 0)
return ret;
payload->payload_state = DP_PAYLOAD_REMOTE;
return ret;
}
static int drm_dp_destroy_payload_step1(struct drm_dp_mst_topology_mgr *mgr,
struct drm_dp_mst_port *port,
int id,
struct drm_dp_payload *payload)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
{
DRM_DEBUG_KMS("\n");
/* it's okay for these to fail */
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
if (port) {
drm_dp_payload_send_msg(mgr, port, id, 0);
}
drm_dp_dpcd_write_payload(mgr, id, payload);
payload->payload_state = DP_PAYLOAD_DELETE_LOCAL;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
return 0;
}
static int drm_dp_destroy_payload_step2(struct drm_dp_mst_topology_mgr *mgr,
int id,
struct drm_dp_payload *payload)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
{
payload->payload_state = 0;
return 0;
}
/**
* drm_dp_update_payload_part1() - Execute payload update part 1
* @mgr: manager to use.
*
* This iterates over all proposed virtual channels, and tries to
* allocate space in the link for them. For 0->slots transitions,
* this step just writes the VCPI to the MST device. For slots->0
* transitions, this writes the updated VCPIs and removes the
* remote VC payloads.
*
* after calling this the driver should generate ACT and payload
* packets.
*/
int drm_dp_update_payload_part1(struct drm_dp_mst_topology_mgr *mgr)
{
struct drm_dp_payload req_payload;
struct drm_dp_mst_port *port;
int i, j;
int cur_slots = 1;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
mutex_lock(&mgr->payload_lock);
for (i = 0; i < mgr->max_payloads; i++) {
struct drm_dp_vcpi *vcpi = mgr->proposed_vcpis[i];
struct drm_dp_payload *payload = &mgr->payloads[i];
bool put_port = false;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
/* solve the current payloads - compare to the hw ones
- update the hw view */
req_payload.start_slot = cur_slots;
if (vcpi) {
port = container_of(vcpi, struct drm_dp_mst_port,
vcpi);
/* Validated ports don't matter if we're releasing
* VCPI
*/
if (vcpi->num_slots) {
port = drm_dp_mst_topology_get_port_validated(
mgr, port);
if (!port) {
mutex_unlock(&mgr->payload_lock);
return -EINVAL;
}
put_port = true;
drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1() We can thank KASAN for finding this, otherwise I probably would have spent hours on it. This fixes a somewhat harder to trigger kernel panic, occuring while enabling MST where the port we were currently updating the payload on would have all of it's refs dropped before we finished what we were doing: ================================================================== BUG: KASAN: use-after-free in drm_dp_update_payload_part1+0xb3f/0xdb0 [drm_kms_helper] at addr ffff8800d29de018 Read of size 4 by task Xorg/973 ============================================================================= BUG kmalloc-2048 (Tainted: G B W ): kasan: bad access detected ----------------------------------------------------------------------------- INFO: Allocated in drm_dp_add_port+0x1aa/0x1ed0 [drm_kms_helper] age=16477 cpu=0 pid=2175 ___slab_alloc+0x472/0x490 __slab_alloc+0x20/0x40 kmem_cache_alloc_trace+0x151/0x190 drm_dp_add_port+0x1aa/0x1ed0 [drm_kms_helper] drm_dp_send_link_address+0x526/0x960 [drm_kms_helper] drm_dp_check_and_send_link_address+0x1ac/0x210 [drm_kms_helper] drm_dp_mst_link_probe_work+0x77/0xd0 [drm_kms_helper] process_one_work+0x562/0x1350 worker_thread+0xd9/0x1390 kthread+0x1c5/0x260 ret_from_fork+0x22/0x40 INFO: Freed in drm_dp_free_mst_port+0x50/0x60 [drm_kms_helper] age=7521 cpu=0 pid=2175 __slab_free+0x17f/0x2d0 kfree+0x169/0x180 drm_dp_free_mst_port+0x50/0x60 [drm_kms_helper] drm_dp_destroy_connector_work+0x2b8/0x490 [drm_kms_helper] process_one_work+0x562/0x1350 worker_thread+0xd9/0x1390 kthread+0x1c5/0x260 ret_from_fork+0x22/0x40 which on this T460s, would eventually lead to kernel panics in somewhat random places later in intel_mst_enable_dp() if we got lucky enough. Signed-off-by: Lyude <cpaul@redhat.com> Cc: stable@vger.kernel.org Signed-off-by: Dave Airlie <airlied@redhat.com>
2016-04-23 03:08:46 +07:00
}
req_payload.num_slots = vcpi->num_slots;
req_payload.vcpi = vcpi->vcpi;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
} else {
port = NULL;
req_payload.num_slots = 0;
}
payload->start_slot = req_payload.start_slot;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
/* work out what is required to happen with this payload */
if (payload->num_slots != req_payload.num_slots) {
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
/* need to push an update for this payload */
if (req_payload.num_slots) {
drm_dp_create_payload_step1(mgr, vcpi->vcpi,
&req_payload);
payload->num_slots = req_payload.num_slots;
payload->vcpi = req_payload.vcpi;
} else if (payload->num_slots) {
payload->num_slots = 0;
drm_dp_destroy_payload_step1(mgr, port,
payload->vcpi,
payload);
req_payload.payload_state =
payload->payload_state;
payload->start_slot = 0;
}
payload->payload_state = req_payload.payload_state;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
cur_slots += req_payload.num_slots;
drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1() We can thank KASAN for finding this, otherwise I probably would have spent hours on it. This fixes a somewhat harder to trigger kernel panic, occuring while enabling MST where the port we were currently updating the payload on would have all of it's refs dropped before we finished what we were doing: ================================================================== BUG: KASAN: use-after-free in drm_dp_update_payload_part1+0xb3f/0xdb0 [drm_kms_helper] at addr ffff8800d29de018 Read of size 4 by task Xorg/973 ============================================================================= BUG kmalloc-2048 (Tainted: G B W ): kasan: bad access detected ----------------------------------------------------------------------------- INFO: Allocated in drm_dp_add_port+0x1aa/0x1ed0 [drm_kms_helper] age=16477 cpu=0 pid=2175 ___slab_alloc+0x472/0x490 __slab_alloc+0x20/0x40 kmem_cache_alloc_trace+0x151/0x190 drm_dp_add_port+0x1aa/0x1ed0 [drm_kms_helper] drm_dp_send_link_address+0x526/0x960 [drm_kms_helper] drm_dp_check_and_send_link_address+0x1ac/0x210 [drm_kms_helper] drm_dp_mst_link_probe_work+0x77/0xd0 [drm_kms_helper] process_one_work+0x562/0x1350 worker_thread+0xd9/0x1390 kthread+0x1c5/0x260 ret_from_fork+0x22/0x40 INFO: Freed in drm_dp_free_mst_port+0x50/0x60 [drm_kms_helper] age=7521 cpu=0 pid=2175 __slab_free+0x17f/0x2d0 kfree+0x169/0x180 drm_dp_free_mst_port+0x50/0x60 [drm_kms_helper] drm_dp_destroy_connector_work+0x2b8/0x490 [drm_kms_helper] process_one_work+0x562/0x1350 worker_thread+0xd9/0x1390 kthread+0x1c5/0x260 ret_from_fork+0x22/0x40 which on this T460s, would eventually lead to kernel panics in somewhat random places later in intel_mst_enable_dp() if we got lucky enough. Signed-off-by: Lyude <cpaul@redhat.com> Cc: stable@vger.kernel.org Signed-off-by: Dave Airlie <airlied@redhat.com>
2016-04-23 03:08:46 +07:00
if (put_port)
drm_dp_mst_topology_put_port(port);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
for (i = 0; i < mgr->max_payloads; i++) {
if (mgr->payloads[i].payload_state != DP_PAYLOAD_DELETE_LOCAL)
continue;
DRM_DEBUG_KMS("removing payload %d\n", i);
for (j = i; j < mgr->max_payloads - 1; j++) {
mgr->payloads[j] = mgr->payloads[j + 1];
mgr->proposed_vcpis[j] = mgr->proposed_vcpis[j + 1];
if (mgr->proposed_vcpis[j] &&
mgr->proposed_vcpis[j]->num_slots) {
set_bit(j + 1, &mgr->payload_mask);
} else {
clear_bit(j + 1, &mgr->payload_mask);
}
}
memset(&mgr->payloads[mgr->max_payloads - 1], 0,
sizeof(struct drm_dp_payload));
mgr->proposed_vcpis[mgr->max_payloads - 1] = NULL;
clear_bit(mgr->max_payloads, &mgr->payload_mask);
}
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
mutex_unlock(&mgr->payload_lock);
return 0;
}
EXPORT_SYMBOL(drm_dp_update_payload_part1);
/**
* drm_dp_update_payload_part2() - Execute payload update part 2
* @mgr: manager to use.
*
* This iterates over all proposed virtual channels, and tries to
* allocate space in the link for them. For 0->slots transitions,
* this step writes the remote VC payload commands. For slots->0
* this just resets some internal state.
*/
int drm_dp_update_payload_part2(struct drm_dp_mst_topology_mgr *mgr)
{
struct drm_dp_mst_port *port;
int i;
int ret = 0;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
mutex_lock(&mgr->payload_lock);
for (i = 0; i < mgr->max_payloads; i++) {
if (!mgr->proposed_vcpis[i])
continue;
port = container_of(mgr->proposed_vcpis[i], struct drm_dp_mst_port, vcpi);
DRM_DEBUG_KMS("payload %d %d\n", i, mgr->payloads[i].payload_state);
if (mgr->payloads[i].payload_state == DP_PAYLOAD_LOCAL) {
ret = drm_dp_create_payload_step2(mgr, port, mgr->proposed_vcpis[i]->vcpi, &mgr->payloads[i]);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
} else if (mgr->payloads[i].payload_state == DP_PAYLOAD_DELETE_LOCAL) {
ret = drm_dp_destroy_payload_step2(mgr, mgr->proposed_vcpis[i]->vcpi, &mgr->payloads[i]);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
if (ret) {
mutex_unlock(&mgr->payload_lock);
return ret;
}
}
mutex_unlock(&mgr->payload_lock);
return 0;
}
EXPORT_SYMBOL(drm_dp_update_payload_part2);
static int drm_dp_send_dpcd_read(struct drm_dp_mst_topology_mgr *mgr,
struct drm_dp_mst_port *port,
drm/dp_mst: Enable registration of AUX devices for MST ports All available downstream ports - physical and logical - are exposed for each MST device. They are listed in /dev/, following the same naming scheme as SST devices by appending an incremental ID. Although all downstream ports are exposed, only some will work as expected. Consider the following topology: +---------+ | ASIC | +---------+ Conn-0| | +----v----+ +----| MST HUB |----+ | +---------+ | | | |Port-1 Port-2| +-----v-----+ +-----v-----+ | MST | | SST | | Display | | Display | +-----------+ +-----------+ |Port-1 x MST Path | MST Device ----------+---------------------------------- sst:0 | MST Hub mst:0-1 | MST Display mst:0-1-1 | MST Display's disconnected DP out mst:0-1-8 | MST Display's internal sink mst:0-2 | SST Display On certain MST displays, the upstream physical port will ACK DPCD reads. However, reads on the local logical port to the internal sink will *NAK*. i.e. reading mst:0-1 ACKs, but mst:0-1-8 NAKs. There may also be duplicates. Some displays will return the same GUID when reading DPCD from both mst:0-1 and mst:0-1-8. There are some device-dependent behavior as well. The MST hub used during testing will actually *ACK* read requests on a disconnected physical port, whereas the MST displays will NAK. In light of these discrepancies, it's simpler to expose all downstream ports - both physical and logical - and let the user decide what to use. v3 changes: * Change WARN_ON_ONCE -> DRM_ERROR on dpcd read errors * Docstring and cosmetic fixes v2 changes: Moved remote aux device (un)registration to new mst connector late register and early unregister helpers. Drivers should call these from their own mst connector function hooks. This is to solve an issue during driver unload, where mst connector devices are unregistered before the remote aux devices are. In a setup where aux devices are created as children of connector devices, the aux device would be removed too early, and uncleanly. Doing so in early_unregister solves this issue, as that is called before connector unregistration. Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Signed-off-by: Leo Li <sunpeng.li@amd.com> Reviewed-by: Lyude Paul <lyude@redhat.com> Signed-off-by: Harry Wentland <harry.wentland@amd.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190723232808.28128-3-sunpeng.li@amd.com
2019-07-24 06:28:01 +07:00
int offset, int size, u8 *bytes)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
{
int len;
drm/dp_mst: Enable registration of AUX devices for MST ports All available downstream ports - physical and logical - are exposed for each MST device. They are listed in /dev/, following the same naming scheme as SST devices by appending an incremental ID. Although all downstream ports are exposed, only some will work as expected. Consider the following topology: +---------+ | ASIC | +---------+ Conn-0| | +----v----+ +----| MST HUB |----+ | +---------+ | | | |Port-1 Port-2| +-----v-----+ +-----v-----+ | MST | | SST | | Display | | Display | +-----------+ +-----------+ |Port-1 x MST Path | MST Device ----------+---------------------------------- sst:0 | MST Hub mst:0-1 | MST Display mst:0-1-1 | MST Display's disconnected DP out mst:0-1-8 | MST Display's internal sink mst:0-2 | SST Display On certain MST displays, the upstream physical port will ACK DPCD reads. However, reads on the local logical port to the internal sink will *NAK*. i.e. reading mst:0-1 ACKs, but mst:0-1-8 NAKs. There may also be duplicates. Some displays will return the same GUID when reading DPCD from both mst:0-1 and mst:0-1-8. There are some device-dependent behavior as well. The MST hub used during testing will actually *ACK* read requests on a disconnected physical port, whereas the MST displays will NAK. In light of these discrepancies, it's simpler to expose all downstream ports - both physical and logical - and let the user decide what to use. v3 changes: * Change WARN_ON_ONCE -> DRM_ERROR on dpcd read errors * Docstring and cosmetic fixes v2 changes: Moved remote aux device (un)registration to new mst connector late register and early unregister helpers. Drivers should call these from their own mst connector function hooks. This is to solve an issue during driver unload, where mst connector devices are unregistered before the remote aux devices are. In a setup where aux devices are created as children of connector devices, the aux device would be removed too early, and uncleanly. Doing so in early_unregister solves this issue, as that is called before connector unregistration. Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Signed-off-by: Leo Li <sunpeng.li@amd.com> Reviewed-by: Lyude Paul <lyude@redhat.com> Signed-off-by: Harry Wentland <harry.wentland@amd.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190723232808.28128-3-sunpeng.li@amd.com
2019-07-24 06:28:01 +07:00
int ret = 0;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
struct drm_dp_sideband_msg_tx *txmsg;
drm/dp_mst: Enable registration of AUX devices for MST ports All available downstream ports - physical and logical - are exposed for each MST device. They are listed in /dev/, following the same naming scheme as SST devices by appending an incremental ID. Although all downstream ports are exposed, only some will work as expected. Consider the following topology: +---------+ | ASIC | +---------+ Conn-0| | +----v----+ +----| MST HUB |----+ | +---------+ | | | |Port-1 Port-2| +-----v-----+ +-----v-----+ | MST | | SST | | Display | | Display | +-----------+ +-----------+ |Port-1 x MST Path | MST Device ----------+---------------------------------- sst:0 | MST Hub mst:0-1 | MST Display mst:0-1-1 | MST Display's disconnected DP out mst:0-1-8 | MST Display's internal sink mst:0-2 | SST Display On certain MST displays, the upstream physical port will ACK DPCD reads. However, reads on the local logical port to the internal sink will *NAK*. i.e. reading mst:0-1 ACKs, but mst:0-1-8 NAKs. There may also be duplicates. Some displays will return the same GUID when reading DPCD from both mst:0-1 and mst:0-1-8. There are some device-dependent behavior as well. The MST hub used during testing will actually *ACK* read requests on a disconnected physical port, whereas the MST displays will NAK. In light of these discrepancies, it's simpler to expose all downstream ports - both physical and logical - and let the user decide what to use. v3 changes: * Change WARN_ON_ONCE -> DRM_ERROR on dpcd read errors * Docstring and cosmetic fixes v2 changes: Moved remote aux device (un)registration to new mst connector late register and early unregister helpers. Drivers should call these from their own mst connector function hooks. This is to solve an issue during driver unload, where mst connector devices are unregistered before the remote aux devices are. In a setup where aux devices are created as children of connector devices, the aux device would be removed too early, and uncleanly. Doing so in early_unregister solves this issue, as that is called before connector unregistration. Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Signed-off-by: Leo Li <sunpeng.li@amd.com> Reviewed-by: Lyude Paul <lyude@redhat.com> Signed-off-by: Harry Wentland <harry.wentland@amd.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190723232808.28128-3-sunpeng.li@amd.com
2019-07-24 06:28:01 +07:00
struct drm_dp_mst_branch *mstb;
mstb = drm_dp_mst_topology_get_mstb_validated(mgr, port->parent);
if (!mstb)
return -EINVAL;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
txmsg = kzalloc(sizeof(*txmsg), GFP_KERNEL);
drm/dp_mst: Enable registration of AUX devices for MST ports All available downstream ports - physical and logical - are exposed for each MST device. They are listed in /dev/, following the same naming scheme as SST devices by appending an incremental ID. Although all downstream ports are exposed, only some will work as expected. Consider the following topology: +---------+ | ASIC | +---------+ Conn-0| | +----v----+ +----| MST HUB |----+ | +---------+ | | | |Port-1 Port-2| +-----v-----+ +-----v-----+ | MST | | SST | | Display | | Display | +-----------+ +-----------+ |Port-1 x MST Path | MST Device ----------+---------------------------------- sst:0 | MST Hub mst:0-1 | MST Display mst:0-1-1 | MST Display's disconnected DP out mst:0-1-8 | MST Display's internal sink mst:0-2 | SST Display On certain MST displays, the upstream physical port will ACK DPCD reads. However, reads on the local logical port to the internal sink will *NAK*. i.e. reading mst:0-1 ACKs, but mst:0-1-8 NAKs. There may also be duplicates. Some displays will return the same GUID when reading DPCD from both mst:0-1 and mst:0-1-8. There are some device-dependent behavior as well. The MST hub used during testing will actually *ACK* read requests on a disconnected physical port, whereas the MST displays will NAK. In light of these discrepancies, it's simpler to expose all downstream ports - both physical and logical - and let the user decide what to use. v3 changes: * Change WARN_ON_ONCE -> DRM_ERROR on dpcd read errors * Docstring and cosmetic fixes v2 changes: Moved remote aux device (un)registration to new mst connector late register and early unregister helpers. Drivers should call these from their own mst connector function hooks. This is to solve an issue during driver unload, where mst connector devices are unregistered before the remote aux devices are. In a setup where aux devices are created as children of connector devices, the aux device would be removed too early, and uncleanly. Doing so in early_unregister solves this issue, as that is called before connector unregistration. Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Signed-off-by: Leo Li <sunpeng.li@amd.com> Reviewed-by: Lyude Paul <lyude@redhat.com> Signed-off-by: Harry Wentland <harry.wentland@amd.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190723232808.28128-3-sunpeng.li@amd.com
2019-07-24 06:28:01 +07:00
if (!txmsg) {
ret = -ENOMEM;
goto fail_put;
}
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
drm/dp_mst: Enable registration of AUX devices for MST ports All available downstream ports - physical and logical - are exposed for each MST device. They are listed in /dev/, following the same naming scheme as SST devices by appending an incremental ID. Although all downstream ports are exposed, only some will work as expected. Consider the following topology: +---------+ | ASIC | +---------+ Conn-0| | +----v----+ +----| MST HUB |----+ | +---------+ | | | |Port-1 Port-2| +-----v-----+ +-----v-----+ | MST | | SST | | Display | | Display | +-----------+ +-----------+ |Port-1 x MST Path | MST Device ----------+---------------------------------- sst:0 | MST Hub mst:0-1 | MST Display mst:0-1-1 | MST Display's disconnected DP out mst:0-1-8 | MST Display's internal sink mst:0-2 | SST Display On certain MST displays, the upstream physical port will ACK DPCD reads. However, reads on the local logical port to the internal sink will *NAK*. i.e. reading mst:0-1 ACKs, but mst:0-1-8 NAKs. There may also be duplicates. Some displays will return the same GUID when reading DPCD from both mst:0-1 and mst:0-1-8. There are some device-dependent behavior as well. The MST hub used during testing will actually *ACK* read requests on a disconnected physical port, whereas the MST displays will NAK. In light of these discrepancies, it's simpler to expose all downstream ports - both physical and logical - and let the user decide what to use. v3 changes: * Change WARN_ON_ONCE -> DRM_ERROR on dpcd read errors * Docstring and cosmetic fixes v2 changes: Moved remote aux device (un)registration to new mst connector late register and early unregister helpers. Drivers should call these from their own mst connector function hooks. This is to solve an issue during driver unload, where mst connector devices are unregistered before the remote aux devices are. In a setup where aux devices are created as children of connector devices, the aux device would be removed too early, and uncleanly. Doing so in early_unregister solves this issue, as that is called before connector unregistration. Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Signed-off-by: Leo Li <sunpeng.li@amd.com> Reviewed-by: Lyude Paul <lyude@redhat.com> Signed-off-by: Harry Wentland <harry.wentland@amd.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190723232808.28128-3-sunpeng.li@amd.com
2019-07-24 06:28:01 +07:00
len = build_dpcd_read(txmsg, port->port_num, offset, size);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
txmsg->dst = port->parent;
drm_dp_queue_down_tx(mgr, txmsg);
drm/dp_mst: Enable registration of AUX devices for MST ports All available downstream ports - physical and logical - are exposed for each MST device. They are listed in /dev/, following the same naming scheme as SST devices by appending an incremental ID. Although all downstream ports are exposed, only some will work as expected. Consider the following topology: +---------+ | ASIC | +---------+ Conn-0| | +----v----+ +----| MST HUB |----+ | +---------+ | | | |Port-1 Port-2| +-----v-----+ +-----v-----+ | MST | | SST | | Display | | Display | +-----------+ +-----------+ |Port-1 x MST Path | MST Device ----------+---------------------------------- sst:0 | MST Hub mst:0-1 | MST Display mst:0-1-1 | MST Display's disconnected DP out mst:0-1-8 | MST Display's internal sink mst:0-2 | SST Display On certain MST displays, the upstream physical port will ACK DPCD reads. However, reads on the local logical port to the internal sink will *NAK*. i.e. reading mst:0-1 ACKs, but mst:0-1-8 NAKs. There may also be duplicates. Some displays will return the same GUID when reading DPCD from both mst:0-1 and mst:0-1-8. There are some device-dependent behavior as well. The MST hub used during testing will actually *ACK* read requests on a disconnected physical port, whereas the MST displays will NAK. In light of these discrepancies, it's simpler to expose all downstream ports - both physical and logical - and let the user decide what to use. v3 changes: * Change WARN_ON_ONCE -> DRM_ERROR on dpcd read errors * Docstring and cosmetic fixes v2 changes: Moved remote aux device (un)registration to new mst connector late register and early unregister helpers. Drivers should call these from their own mst connector function hooks. This is to solve an issue during driver unload, where mst connector devices are unregistered before the remote aux devices are. In a setup where aux devices are created as children of connector devices, the aux device would be removed too early, and uncleanly. Doing so in early_unregister solves this issue, as that is called before connector unregistration. Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Signed-off-by: Leo Li <sunpeng.li@amd.com> Reviewed-by: Lyude Paul <lyude@redhat.com> Signed-off-by: Harry Wentland <harry.wentland@amd.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190723232808.28128-3-sunpeng.li@amd.com
2019-07-24 06:28:01 +07:00
ret = drm_dp_mst_wait_tx_reply(mstb, txmsg);
if (ret < 0)
goto fail_free;
/* DPCD read should never be NACKed */
if (txmsg->reply.reply_type == 1) {
DRM_ERROR("mstb %p port %d: DPCD read on addr 0x%x for %d bytes NAKed\n",
mstb, port->port_num, offset, size);
ret = -EIO;
goto fail_free;
}
if (txmsg->reply.u.remote_dpcd_read_ack.num_bytes != size) {
ret = -EPROTO;
goto fail_free;
}
ret = min_t(size_t, txmsg->reply.u.remote_dpcd_read_ack.num_bytes,
size);
memcpy(bytes, txmsg->reply.u.remote_dpcd_read_ack.bytes, ret);
fail_free:
kfree(txmsg);
fail_put:
drm_dp_mst_topology_put_mstb(mstb);
return ret;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
static int drm_dp_send_dpcd_write(struct drm_dp_mst_topology_mgr *mgr,
struct drm_dp_mst_port *port,
int offset, int size, u8 *bytes)
{
int len;
int ret;
struct drm_dp_sideband_msg_tx *txmsg;
struct drm_dp_mst_branch *mstb;
mstb = drm_dp_mst_topology_get_mstb_validated(mgr, port->parent);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
if (!mstb)
return -EINVAL;
txmsg = kzalloc(sizeof(*txmsg), GFP_KERNEL);
if (!txmsg) {
ret = -ENOMEM;
goto fail_put;
}
len = build_dpcd_write(txmsg, port->port_num, offset, size, bytes);
txmsg->dst = mstb;
drm_dp_queue_down_tx(mgr, txmsg);
ret = drm_dp_mst_wait_tx_reply(mstb, txmsg);
if (ret > 0) {
if (txmsg->reply.reply_type == DP_SIDEBAND_REPLY_NAK)
drm/dp_mst: Enable registration of AUX devices for MST ports All available downstream ports - physical and logical - are exposed for each MST device. They are listed in /dev/, following the same naming scheme as SST devices by appending an incremental ID. Although all downstream ports are exposed, only some will work as expected. Consider the following topology: +---------+ | ASIC | +---------+ Conn-0| | +----v----+ +----| MST HUB |----+ | +---------+ | | | |Port-1 Port-2| +-----v-----+ +-----v-----+ | MST | | SST | | Display | | Display | +-----------+ +-----------+ |Port-1 x MST Path | MST Device ----------+---------------------------------- sst:0 | MST Hub mst:0-1 | MST Display mst:0-1-1 | MST Display's disconnected DP out mst:0-1-8 | MST Display's internal sink mst:0-2 | SST Display On certain MST displays, the upstream physical port will ACK DPCD reads. However, reads on the local logical port to the internal sink will *NAK*. i.e. reading mst:0-1 ACKs, but mst:0-1-8 NAKs. There may also be duplicates. Some displays will return the same GUID when reading DPCD from both mst:0-1 and mst:0-1-8. There are some device-dependent behavior as well. The MST hub used during testing will actually *ACK* read requests on a disconnected physical port, whereas the MST displays will NAK. In light of these discrepancies, it's simpler to expose all downstream ports - both physical and logical - and let the user decide what to use. v3 changes: * Change WARN_ON_ONCE -> DRM_ERROR on dpcd read errors * Docstring and cosmetic fixes v2 changes: Moved remote aux device (un)registration to new mst connector late register and early unregister helpers. Drivers should call these from their own mst connector function hooks. This is to solve an issue during driver unload, where mst connector devices are unregistered before the remote aux devices are. In a setup where aux devices are created as children of connector devices, the aux device would be removed too early, and uncleanly. Doing so in early_unregister solves this issue, as that is called before connector unregistration. Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Signed-off-by: Leo Li <sunpeng.li@amd.com> Reviewed-by: Lyude Paul <lyude@redhat.com> Signed-off-by: Harry Wentland <harry.wentland@amd.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190723232808.28128-3-sunpeng.li@amd.com
2019-07-24 06:28:01 +07:00
ret = -EIO;
else
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
ret = 0;
}
kfree(txmsg);
fail_put:
drm_dp_mst_topology_put_mstb(mstb);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
return ret;
}
static int drm_dp_encode_up_ack_reply(struct drm_dp_sideband_msg_tx *msg, u8 req_type)
{
struct drm_dp_sideband_msg_reply_body reply;
reply.reply_type = DP_SIDEBAND_REPLY_ACK;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
reply.req_type = req_type;
drm_dp_encode_sideband_reply(&reply, msg);
return 0;
}
static int drm_dp_send_up_ack_reply(struct drm_dp_mst_topology_mgr *mgr,
struct drm_dp_mst_branch *mstb,
int req_type, int seqno, bool broadcast)
{
struct drm_dp_sideband_msg_tx *txmsg;
txmsg = kzalloc(sizeof(*txmsg), GFP_KERNEL);
if (!txmsg)
return -ENOMEM;
txmsg->dst = mstb;
txmsg->seqno = seqno;
drm_dp_encode_up_ack_reply(txmsg, req_type);
mutex_lock(&mgr->qlock);
process_single_up_tx_qlock(mgr, txmsg);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
mutex_unlock(&mgr->qlock);
kfree(txmsg);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
return 0;
}
static int drm_dp_get_vc_payload_bw(u8 dp_link_bw, u8 dp_link_count)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
{
if (dp_link_bw == 0 || dp_link_count == 0)
DRM_DEBUG_KMS("invalid link bandwidth in DPCD: %x (link count: %d)\n",
dp_link_bw, dp_link_count);
return dp_link_bw * dp_link_count / 2;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
/**
* drm_dp_mst_topology_mgr_set_mst() - Set the MST state for a topology manager
* @mgr: manager to set state for
* @mst_state: true to enable MST on this connector - false to disable.
*
* This is called by the driver when it detects an MST capable device plugged
* into a DP MST capable port, or when a DP MST capable device is unplugged.
*/
int drm_dp_mst_topology_mgr_set_mst(struct drm_dp_mst_topology_mgr *mgr, bool mst_state)
{
int ret = 0;
struct drm_dp_mst_branch *mstb = NULL;
mutex_lock(&mgr->lock);
if (mst_state == mgr->mst_state)
goto out_unlock;
mgr->mst_state = mst_state;
/* set the device into MST mode */
if (mst_state) {
WARN_ON(mgr->mst_primary);
/* get dpcd info */
ret = drm_dp_dpcd_read(mgr->aux, DP_DPCD_REV, mgr->dpcd, DP_RECEIVER_CAP_SIZE);
if (ret != DP_RECEIVER_CAP_SIZE) {
DRM_DEBUG_KMS("failed to read DPCD\n");
goto out_unlock;
}
mgr->pbn_div = drm_dp_get_vc_payload_bw(mgr->dpcd[1],
mgr->dpcd[2] & DP_MAX_LANE_COUNT_MASK);
if (mgr->pbn_div == 0) {
ret = -EINVAL;
goto out_unlock;
}
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
/* add initial branch device at LCT 1 */
mstb = drm_dp_add_mst_branch_device(1, NULL);
if (mstb == NULL) {
ret = -ENOMEM;
goto out_unlock;
}
mstb->mgr = mgr;
/* give this the main reference */
mgr->mst_primary = mstb;
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
drm_dp_mst_topology_get_mstb(mgr->mst_primary);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
ret = drm_dp_dpcd_writeb(mgr->aux, DP_MSTM_CTRL,
DP_MST_EN | DP_UP_REQ_EN | DP_UPSTREAM_IS_SRC);
if (ret < 0) {
goto out_unlock;
}
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
{
struct drm_dp_payload reset_pay;
reset_pay.start_slot = 0;
reset_pay.num_slots = 0x3f;
drm_dp_dpcd_write_payload(mgr, 0, &reset_pay);
}
queue_work(system_long_wq, &mgr->work);
ret = 0;
} else {
/* disable MST on the device */
mstb = mgr->mst_primary;
mgr->mst_primary = NULL;
/* this can fail if the device is gone */
drm_dp_dpcd_writeb(mgr->aux, DP_MSTM_CTRL, 0);
ret = 0;
memset(mgr->payloads, 0, mgr->max_payloads * sizeof(struct drm_dp_payload));
mgr->payload_mask = 0;
set_bit(0, &mgr->payload_mask);
mgr->vcpi_mask = 0;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
out_unlock:
mutex_unlock(&mgr->lock);
if (mstb)
drm_dp_mst_topology_put_mstb(mstb);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
return ret;
}
EXPORT_SYMBOL(drm_dp_mst_topology_mgr_set_mst);
/**
* drm_dp_mst_topology_mgr_suspend() - suspend the MST manager
* @mgr: manager to suspend
*
* This function tells the MST device that we can't handle UP messages
* anymore. This should stop it from sending any since we are suspended.
*/
void drm_dp_mst_topology_mgr_suspend(struct drm_dp_mst_topology_mgr *mgr)
{
mutex_lock(&mgr->lock);
drm_dp_dpcd_writeb(mgr->aux, DP_MSTM_CTRL,
DP_MST_EN | DP_UPSTREAM_IS_SRC);
mutex_unlock(&mgr->lock);
drm/dp/mst: drop cancel work sync in the mstb destroy path (v2) Since 9eb1e57f564d4e6e10991402726cc83fe0b9172f drm/dp/mst: make sure mst_primary mstb is valid in work function we validate the mstb structs in the work function, and doing that takes a reference. So we should never get here with the work function running using the mstb device, only if the work function hasn't run yet or is running for another mstb. So we don't need to sync the work here, this was causing lockdep spew as below. [ +0.000160] ============================================= [ +0.000001] [ INFO: possible recursive locking detected ] [ +0.000002] 3.10.0-320.el7.rhel72.stable.backport.3.x86_64.debug #1 Tainted: G W ------------ [ +0.000001] --------------------------------------------- [ +0.000001] kworker/4:2/1262 is trying to acquire lock: [ +0.000001] ((&mgr->work)){+.+.+.}, at: [<ffffffff810b29a5>] flush_work+0x5/0x2e0 [ +0.000007] but task is already holding lock: [ +0.000001] ((&mgr->work)){+.+.+.}, at: [<ffffffff810b57e4>] process_one_work+0x1b4/0x710 [ +0.000004] other info that might help us debug this: [ +0.000001] Possible unsafe locking scenario: [ +0.000002] CPU0 [ +0.000000] ---- [ +0.000001] lock((&mgr->work)); [ +0.000002] lock((&mgr->work)); [ +0.000001] *** DEADLOCK *** [ +0.000001] May be due to missing lock nesting notation [ +0.000002] 2 locks held by kworker/4:2/1262: [ +0.000001] #0: (events_long){.+.+.+}, at: [<ffffffff810b57e4>] process_one_work+0x1b4/0x710 [ +0.000004] #1: ((&mgr->work)){+.+.+.}, at: [<ffffffff810b57e4>] process_one_work+0x1b4/0x710 [ +0.000003] stack backtrace: [ +0.000003] CPU: 4 PID: 1262 Comm: kworker/4:2 Tainted: G W ------------ 3.10.0-320.el7.rhel72.stable.backport.3.x86_64.debug #1 [ +0.000001] Hardware name: LENOVO 20EGS0R600/20EGS0R600, BIOS GNET71WW (2.19 ) 02/05/2015 [ +0.000008] Workqueue: events_long drm_dp_mst_link_probe_work [drm_kms_helper] [ +0.000001] ffffffff82c26c90 00000000a527b914 ffff88046399bae8 ffffffff816fe04d [ +0.000004] ffff88046399bb58 ffffffff8110f47f ffff880461438000 0001009b840fc003 [ +0.000002] ffff880461438a98 0000000000000000 0000000804dc26e1 ffffffff824a2c00 [ +0.000003] Call Trace: [ +0.000004] [<ffffffff816fe04d>] dump_stack+0x19/0x1b [ +0.000004] [<ffffffff8110f47f>] __lock_acquire+0x115f/0x1250 [ +0.000002] [<ffffffff8110fd49>] lock_acquire+0x99/0x1e0 [ +0.000002] [<ffffffff810b29a5>] ? flush_work+0x5/0x2e0 [ +0.000002] [<ffffffff810b29ee>] flush_work+0x4e/0x2e0 [ +0.000002] [<ffffffff810b29a5>] ? flush_work+0x5/0x2e0 [ +0.000004] [<ffffffff81025905>] ? native_sched_clock+0x35/0x80 [ +0.000002] [<ffffffff81025959>] ? sched_clock+0x9/0x10 [ +0.000002] [<ffffffff810da1f5>] ? local_clock+0x25/0x30 [ +0.000002] [<ffffffff8110dca9>] ? mark_held_locks+0xb9/0x140 [ +0.000003] [<ffffffff810b4ed5>] ? __cancel_work_timer+0x95/0x160 [ +0.000002] [<ffffffff810b4ee8>] __cancel_work_timer+0xa8/0x160 [ +0.000002] [<ffffffff810b4fb0>] cancel_work_sync+0x10/0x20 [ +0.000007] [<ffffffffa0160d17>] drm_dp_destroy_mst_branch_device+0x27/0x120 [drm_kms_helper] [ +0.000006] [<ffffffffa0163968>] drm_dp_mst_link_probe_work+0x78/0xa0 [drm_kms_helper] [ +0.000002] [<ffffffff810b5850>] process_one_work+0x220/0x710 [ +0.000002] [<ffffffff810b57e4>] ? process_one_work+0x1b4/0x710 [ +0.000005] [<ffffffff810b5e5b>] worker_thread+0x11b/0x3a0 [ +0.000003] [<ffffffff810b5d40>] ? process_one_work+0x710/0x710 [ +0.000002] [<ffffffff810beced>] kthread+0xed/0x100 [ +0.000003] [<ffffffff810bec00>] ? insert_kthread_work+0x80/0x80 [ +0.000003] [<ffffffff817121d8>] ret_from_fork+0x58/0x90 v2: add flush_work. Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: stable@vger.kernel.org Signed-off-by: Dave Airlie <airlied@redhat.com>
2015-09-30 07:39:42 +07:00
flush_work(&mgr->work);
flush_work(&mgr->destroy_connector_work);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
EXPORT_SYMBOL(drm_dp_mst_topology_mgr_suspend);
/**
* drm_dp_mst_topology_mgr_resume() - resume the MST manager
* @mgr: manager to resume
*
* This will fetch DPCD and see if the device is still there,
* if it is, it will rewrite the MSTM control bits, and return.
*
* if the device fails this returns -1, and the driver should do
* a full MST reprobe, in case we were undocked.
*/
int drm_dp_mst_topology_mgr_resume(struct drm_dp_mst_topology_mgr *mgr)
{
int ret = 0;
mutex_lock(&mgr->lock);
if (mgr->mst_primary) {
int sret;
u8 guid[16];
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
sret = drm_dp_dpcd_read(mgr->aux, DP_DPCD_REV, mgr->dpcd, DP_RECEIVER_CAP_SIZE);
if (sret != DP_RECEIVER_CAP_SIZE) {
DRM_DEBUG_KMS("dpcd read failed - undocked during suspend?\n");
ret = -1;
goto out_unlock;
}
ret = drm_dp_dpcd_writeb(mgr->aux, DP_MSTM_CTRL,
DP_MST_EN | DP_UP_REQ_EN | DP_UPSTREAM_IS_SRC);
if (ret < 0) {
DRM_DEBUG_KMS("mst write failed - undocked during suspend?\n");
ret = -1;
goto out_unlock;
}
/* Some hubs forget their guids after they resume */
sret = drm_dp_dpcd_read(mgr->aux, DP_GUID, guid, 16);
if (sret != 16) {
DRM_DEBUG_KMS("dpcd read failed - undocked during suspend?\n");
ret = -1;
goto out_unlock;
}
drm_dp_check_mstb_guid(mgr->mst_primary, guid);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
ret = 0;
} else
ret = -1;
out_unlock:
mutex_unlock(&mgr->lock);
return ret;
}
EXPORT_SYMBOL(drm_dp_mst_topology_mgr_resume);
static bool drm_dp_get_one_sb_msg(struct drm_dp_mst_topology_mgr *mgr, bool up)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
{
int len;
u8 replyblock[32];
int replylen, origlen, curreply;
int ret;
struct drm_dp_sideband_msg_rx *msg;
int basereg = up ? DP_SIDEBAND_MSG_UP_REQ_BASE : DP_SIDEBAND_MSG_DOWN_REP_BASE;
msg = up ? &mgr->up_req_recv : &mgr->down_rep_recv;
len = min(mgr->max_dpcd_transaction_bytes, 16);
ret = drm_dp_dpcd_read(mgr->aux, basereg,
replyblock, len);
if (ret != len) {
DRM_DEBUG_KMS("failed to read DPCD down rep %d %d\n", len, ret);
return false;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
ret = drm_dp_sideband_msg_build(msg, replyblock, len, true);
if (!ret) {
DRM_DEBUG_KMS("sideband msg build failed %d\n", replyblock[0]);
return false;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
replylen = msg->curchunk_len + msg->curchunk_hdrlen;
origlen = replylen;
replylen -= len;
curreply = len;
while (replylen > 0) {
len = min3(replylen, mgr->max_dpcd_transaction_bytes, 16);
ret = drm_dp_dpcd_read(mgr->aux, basereg + curreply,
replyblock, len);
if (ret != len) {
DRM_DEBUG_KMS("failed to read a chunk (len %d, ret %d)\n",
len, ret);
return false;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
ret = drm_dp_sideband_msg_build(msg, replyblock, len, false);
if (!ret) {
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
DRM_DEBUG_KMS("failed to build sideband msg\n");
return false;
}
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
curreply += len;
replylen -= len;
}
return true;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
static int drm_dp_mst_handle_down_rep(struct drm_dp_mst_topology_mgr *mgr)
{
struct drm_dp_sideband_msg_tx *txmsg;
struct drm_dp_mst_branch *mstb;
struct drm_dp_sideband_msg_hdr *hdr = &mgr->down_rep_recv.initial_hdr;
int slot = -1;
if (!drm_dp_get_one_sb_msg(mgr, false))
goto clear_down_rep_recv;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
if (!mgr->down_rep_recv.have_eomt)
return 0;
mstb = drm_dp_get_mst_branch_device(mgr, hdr->lct, hdr->rad);
if (!mstb) {
DRM_DEBUG_KMS("Got MST reply from unknown device %d\n",
hdr->lct);
goto clear_down_rep_recv;
}
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
/* find the message */
slot = hdr->seqno;
mutex_lock(&mgr->qlock);
txmsg = mstb->tx_slots[slot];
/* remove from slots */
mutex_unlock(&mgr->qlock);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
if (!txmsg) {
DRM_DEBUG_KMS("Got MST reply with no msg %p %d %d %02x %02x\n",
mstb, hdr->seqno, hdr->lct, hdr->rad[0],
mgr->down_rep_recv.msg[0]);
goto no_msg;
}
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
drm_dp_sideband_parse_reply(&mgr->down_rep_recv, &txmsg->reply);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
if (txmsg->reply.reply_type == DP_SIDEBAND_REPLY_NAK)
DRM_DEBUG_KMS("Got NAK reply: req 0x%02x (%s), reason 0x%02x (%s), nak data 0x%02x\n",
txmsg->reply.req_type,
drm_dp_mst_req_type_str(txmsg->reply.req_type),
txmsg->reply.u.nak.reason,
drm_dp_mst_nak_reason_str(txmsg->reply.u.nak.reason),
txmsg->reply.u.nak.nak_data);
memset(&mgr->down_rep_recv, 0, sizeof(struct drm_dp_sideband_msg_rx));
drm_dp_mst_topology_put_mstb(mstb);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
mutex_lock(&mgr->qlock);
txmsg->state = DRM_DP_SIDEBAND_TX_RX;
mstb->tx_slots[slot] = NULL;
mutex_unlock(&mgr->qlock);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
wake_up_all(&mgr->tx_waitq);
return 0;
no_msg:
drm_dp_mst_topology_put_mstb(mstb);
clear_down_rep_recv:
memset(&mgr->down_rep_recv, 0, sizeof(struct drm_dp_sideband_msg_rx));
return 0;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
static int drm_dp_mst_handle_up_req(struct drm_dp_mst_topology_mgr *mgr)
{
struct drm_dp_sideband_msg_req_body msg;
struct drm_dp_sideband_msg_hdr *hdr = &mgr->up_req_recv.initial_hdr;
struct drm_dp_mst_branch *mstb = NULL;
const u8 *guid;
bool seqno;
if (!drm_dp_get_one_sb_msg(mgr, true))
goto out;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
if (!mgr->up_req_recv.have_eomt)
return 0;
if (!hdr->broadcast) {
mstb = drm_dp_get_mst_branch_device(mgr, hdr->lct, hdr->rad);
if (!mstb) {
DRM_DEBUG_KMS("Got MST reply from unknown device %d\n",
hdr->lct);
goto out;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
}
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
seqno = hdr->seqno;
drm_dp_sideband_parse_req(&mgr->up_req_recv, &msg);
if (msg.req_type == DP_CONNECTION_STATUS_NOTIFY)
guid = msg.u.conn_stat.guid;
else if (msg.req_type == DP_RESOURCE_STATUS_NOTIFY)
guid = msg.u.resource_stat.guid;
else
goto out;
drm_dp_send_up_ack_reply(mgr, mgr->mst_primary, msg.req_type, seqno,
false);
if (!mstb) {
mstb = drm_dp_get_mst_branch_device_by_guid(mgr, guid);
if (!mstb) {
DRM_DEBUG_KMS("Got MST reply from unknown device %d\n",
hdr->lct);
goto out;
}
}
if (msg.req_type == DP_CONNECTION_STATUS_NOTIFY) {
drm_dp_mst_handle_conn_stat(mstb, &msg.u.conn_stat);
DRM_DEBUG_KMS("Got CSN: pn: %d ldps:%d ddps: %d mcs: %d ip: %d pdt: %d\n",
msg.u.conn_stat.port_number,
msg.u.conn_stat.legacy_device_plug_status,
msg.u.conn_stat.displayport_device_plug_status,
msg.u.conn_stat.message_capability_status,
msg.u.conn_stat.input_port,
msg.u.conn_stat.peer_device_type);
drm_kms_helper_hotplug_event(mgr->dev);
} else if (msg.req_type == DP_RESOURCE_STATUS_NOTIFY) {
DRM_DEBUG_KMS("Got RSN: pn: %d avail_pbn %d\n",
msg.u.resource_stat.port_number,
msg.u.resource_stat.available_pbn);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
drm_dp_mst_topology_put_mstb(mstb);
out:
memset(&mgr->up_req_recv, 0, sizeof(struct drm_dp_sideband_msg_rx));
return 0;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
/**
* drm_dp_mst_hpd_irq() - MST hotplug IRQ notify
* @mgr: manager to notify irq for.
* @esi: 4 bytes from SINK_COUNT_ESI
* @handled: whether the hpd interrupt was consumed or not
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
*
* This should be called from the driver when it detects a short IRQ,
* along with the value of the DEVICE_SERVICE_IRQ_VECTOR_ESI0. The
* topology manager will process the sideband messages received as a result
* of this.
*/
int drm_dp_mst_hpd_irq(struct drm_dp_mst_topology_mgr *mgr, u8 *esi, bool *handled)
{
int ret = 0;
int sc;
*handled = false;
sc = esi[0] & 0x3f;
if (sc != mgr->sink_count) {
mgr->sink_count = sc;
*handled = true;
}
if (esi[1] & DP_DOWN_REP_MSG_RDY) {
ret = drm_dp_mst_handle_down_rep(mgr);
*handled = true;
}
if (esi[1] & DP_UP_REQ_MSG_RDY) {
ret |= drm_dp_mst_handle_up_req(mgr);
*handled = true;
}
drm_dp_mst_kick_tx(mgr);
return ret;
}
EXPORT_SYMBOL(drm_dp_mst_hpd_irq);
/**
* drm_dp_mst_detect_port() - get connection status for an MST port
* @connector: DRM connector for this port
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
* @mgr: manager for this port
* @port: unverified pointer to a port
*
* This returns the current connection state for a port. It validates the
* port pointer still exists so the caller doesn't require a reference
*/
enum drm_connector_status drm_dp_mst_detect_port(struct drm_connector *connector,
struct drm_dp_mst_topology_mgr *mgr, struct drm_dp_mst_port *port)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
{
enum drm_connector_status status = connector_status_disconnected;
/* we need to search for the port in the mgr in case it's gone */
port = drm_dp_mst_topology_get_port_validated(mgr, port);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
if (!port)
return connector_status_disconnected;
if (!port->ddps)
goto out;
switch (port->pdt) {
case DP_PEER_DEVICE_NONE:
case DP_PEER_DEVICE_MST_BRANCHING:
break;
case DP_PEER_DEVICE_SST_SINK:
status = connector_status_connected;
/* for logical ports - cache the EDID */
if (port->port_num >= 8 && !port->cached_edid) {
port->cached_edid = drm_get_edid(connector, &port->aux.ddc);
}
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
break;
case DP_PEER_DEVICE_DP_LEGACY_CONV:
if (port->ldps)
status = connector_status_connected;
break;
}
out:
drm_dp_mst_topology_put_port(port);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
return status;
}
EXPORT_SYMBOL(drm_dp_mst_detect_port);
/**
* drm_dp_mst_port_has_audio() - Check whether port has audio capability or not
* @mgr: manager for this port
* @port: unverified pointer to a port.
*
* This returns whether the port supports audio or not.
*/
bool drm_dp_mst_port_has_audio(struct drm_dp_mst_topology_mgr *mgr,
struct drm_dp_mst_port *port)
{
bool ret = false;
port = drm_dp_mst_topology_get_port_validated(mgr, port);
if (!port)
return ret;
ret = port->has_audio;
drm_dp_mst_topology_put_port(port);
return ret;
}
EXPORT_SYMBOL(drm_dp_mst_port_has_audio);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
/**
* drm_dp_mst_get_edid() - get EDID for an MST port
* @connector: toplevel connector to get EDID for
* @mgr: manager for this port
* @port: unverified pointer to a port.
*
* This returns an EDID for the port connected to a connector,
* It validates the pointer still exists so the caller doesn't require a
* reference.
*/
struct edid *drm_dp_mst_get_edid(struct drm_connector *connector, struct drm_dp_mst_topology_mgr *mgr, struct drm_dp_mst_port *port)
{
struct edid *edid = NULL;
/* we need to search for the port in the mgr in case it's gone */
port = drm_dp_mst_topology_get_port_validated(mgr, port);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
if (!port)
return NULL;
if (port->cached_edid)
edid = drm_edid_duplicate(port->cached_edid);
else {
edid = drm_get_edid(connector, &port->aux.ddc);
}
port->has_audio = drm_detect_monitor_audio(edid);
drm_dp_mst_topology_put_port(port);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
return edid;
}
EXPORT_SYMBOL(drm_dp_mst_get_edid);
/**
* drm_dp_find_vcpi_slots() - Find VCPI slots for this PBN value
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
* @mgr: manager to use
* @pbn: payload bandwidth to convert into slots.
*
* Calculate the number of VCPI slots that will be required for the given PBN
* value. This function is deprecated, and should not be used in atomic
* drivers.
*
* RETURNS:
* The total slots required for this port, or error.
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
*/
int drm_dp_find_vcpi_slots(struct drm_dp_mst_topology_mgr *mgr,
int pbn)
{
int num_slots;
num_slots = DIV_ROUND_UP(pbn, mgr->pbn_div);
/* max. time slots - one slot for MTP header */
if (num_slots > 63)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
return -ENOSPC;
return num_slots;
}
EXPORT_SYMBOL(drm_dp_find_vcpi_slots);
static int drm_dp_init_vcpi(struct drm_dp_mst_topology_mgr *mgr,
struct drm_dp_vcpi *vcpi, int pbn, int slots)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
{
int ret;
/* max. time slots - one slot for MTP header */
if (slots > 63)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
return -ENOSPC;
vcpi->pbn = pbn;
vcpi->aligned_pbn = slots * mgr->pbn_div;
vcpi->num_slots = slots;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
ret = drm_dp_mst_assign_payload_id(mgr, vcpi);
if (ret < 0)
return ret;
return 0;
}
drm/dp: Add DP MST helpers to atomically find and release vcpi slots drm_dp_atomic_find_vcpi_slots() should be called from ->atomic_check() to check there are sufficient vcpi slots for a mode and to add that to the state. This should be followed by a call to drm_dp_mst_allocate_vcpi() in ->atomic_commit() to initialize a struct vcpi for the port. drm_dp_atomic_release_vcpi_slots() should be called from ->atomic_check() to release a port's vcpi slot allocation from the state. Drivers that do not make use of this atomic helper are expected to call drm_dp_find_vcpi_slots() instead before calling drm_dp_mst_allocate_vcpi(). v3: drm_dp_atomic_release_vcpi_slots() now needs to know how many slots to release as we may not have a valid reference to port. v2: Added checks for verifying the port reference is valid Moved get_mst_topology_state() into the helpers (Daniel) Changed find_vcpi_slots() to not depend on current allocation Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Cc: Archit Taneja <architt@codeaurora.org> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Harry Wentland <Harry.wentland@amd.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Signed-off-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com> Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1492753893-3748-4-git-send-email-dhinakaran.pandiyan@intel.com
2017-04-21 12:51:32 +07:00
/**
drm/dp_mst: Start tracking per-port VCPI allocations There has been a TODO waiting for quite a long time in drm_dp_mst_topology.c: /* We cannot rely on port->vcpi.num_slots to update * topology_state->avail_slots as the port may not exist if the parent * branch device was unplugged. This should be fixed by tracking * per-port slot allocation in drm_dp_mst_topology_state instead of * depending on the caller to tell us how many slots to release. */ That's not the only reason we should fix this: forcing the driver to track the VCPI allocations throughout a state's atomic check is error prone, because it means that extra care has to be taken with the order that drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() are called in in order to ensure idempotency. Currently the only driver actually using these helpers, i915, doesn't even do this correctly: multiple ->best_encoder() checks with i915's current implementation would not be idempotent and would over-allocate VCPI slots, something I learned trying to implement fallback retraining in MST. So: simplify this whole mess, and teach drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() to track the VCPI allocations for each port. This allows us to ensure idempotency without having to rely on the driver as much. Additionally: the driver doesn't need to do any kind of VCPI slot tracking anymore if it doesn't need it for it's own internal state. Additionally; this adds a new drm_dp_mst_atomic_check() helper which must be used by atomic drivers to perform validity checks for the new VCPI allocations incurred by a state. Also: update the documentation and make it more obvious that these /must/ be called by /all/ atomic drivers supporting MST. Changes since v9: * Add some missing changes that were requested by danvet that I forgot about after I redid all of the kref stuff: * Remove unnecessary state changes in intel_dp_mst_atomic_check * Cleanup atomic check logic for VCPI allocations - all we need to check in compute_config is whether or not this state disables a CRTC, then free VCPI based off that Changes since v8: * Fix compile errors, whoops! Changes since v7: - Don't check for mixed stale/valid VCPI allocations, just rely on connector registration to stop such erroneous modesets Changes since v6: - Keep a kref to all of the ports we have allocations on. This required a good bit of changing to when we call drm_dp_find_vcpi_slots(), mainly that we need to ensure that we only redo VCPI allocations on actual mode or CRTC changes, not crtc_state->active changes. Additionally, we no longer take the registration of the DRM connector for each port into account because so long as we have a kref to the port in the new or previous atomic state, the connector will stay registered. - Use the small changes to drm_dp_put_port() to add even more error checking to make misusage of the helpers more obvious. I added this after having to chase down various use-after-free conditions that started popping up from the new helpers so no one else has to troubleshoot that. - Move some accidental DRM_DEBUG_KMS() calls to DRM_DEBUG_ATOMIC() - Update documentation again, note that find/release() should both not be called on the same port in a single atomic check phase (but multiple calls to one or the other is OK) Changes since v4: - Don't skip the atomic checks for VCPI allocations if no new VCPI allocations happen in a state. This makes the next change I'm about to list here a lot easier to implement. - Don't ignore VCPI allocations on destroyed ports, instead ensure that when ports are destroyed and still have VCPI allocations in the topology state, the only state changes allowed are releasing said ports' VCPI. This prevents a state with a mix of VCPI allocations from destroyed ports, and allocations from valid ports. Changes since v3: - Don't release VCPI allocations in the topology state immediately in drm_dp_atomic_release_vcpi_slots(), instead mark them as 0 and skip over them in drm_dp_mst_duplicate_state(). This makes it so drm_dp_atomic_release_vcpi_slots() is still idempotent while also throwing warnings if the driver messes up it's book keeping and tries to release VCPI slots on a port that doesn't have any pre-existing VCPI allocation - danvet - Change mst_state/state in some debugging messages to "mst state" Changes since v2: - Use kmemdup() for duplicating MST state - danvet - Move port validation out of duplicate state callback - danvet - Handle looping through MST topology states in drm_dp_mst_atomic_check() so the driver doesn't have to do it - Fix documentation in drm_dp_atomic_find_vcpi_slots() - Move the atomic check for each individual topology state into it's own function, reduces indenting - Don't consider "stale" MST ports when calculating the bandwidth requirements. This is needed because originally we relied on the state duplication functions to prune any stale ports from the new state, which would prevent us from incorrectly considering their bandwidth requirements alongside legitimate new payloads. - Add function references in drm_dp_atomic_release_vcpi_slots() - danvet - Annotate atomic VCPI and atomic check functions with __must_check - danvet Changes since v1: - Don't use the now-removed ->atomic_check() for private objects hook, just give drivers a function to call themselves Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Harry Wentland <harry.wentland@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-19-lyude@redhat.com
2019-01-11 07:53:41 +07:00
* drm_dp_atomic_find_vcpi_slots() - Find and add VCPI slots to the state
drm/dp: Add DP MST helpers to atomically find and release vcpi slots drm_dp_atomic_find_vcpi_slots() should be called from ->atomic_check() to check there are sufficient vcpi slots for a mode and to add that to the state. This should be followed by a call to drm_dp_mst_allocate_vcpi() in ->atomic_commit() to initialize a struct vcpi for the port. drm_dp_atomic_release_vcpi_slots() should be called from ->atomic_check() to release a port's vcpi slot allocation from the state. Drivers that do not make use of this atomic helper are expected to call drm_dp_find_vcpi_slots() instead before calling drm_dp_mst_allocate_vcpi(). v3: drm_dp_atomic_release_vcpi_slots() now needs to know how many slots to release as we may not have a valid reference to port. v2: Added checks for verifying the port reference is valid Moved get_mst_topology_state() into the helpers (Daniel) Changed find_vcpi_slots() to not depend on current allocation Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Cc: Archit Taneja <architt@codeaurora.org> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Harry Wentland <Harry.wentland@amd.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Signed-off-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com> Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1492753893-3748-4-git-send-email-dhinakaran.pandiyan@intel.com
2017-04-21 12:51:32 +07:00
* @state: global atomic state
* @mgr: MST topology manager for the port
* @port: port to find vcpi slots for
* @pbn: bandwidth required for the mode in PBN
*
drm/dp_mst: Start tracking per-port VCPI allocations There has been a TODO waiting for quite a long time in drm_dp_mst_topology.c: /* We cannot rely on port->vcpi.num_slots to update * topology_state->avail_slots as the port may not exist if the parent * branch device was unplugged. This should be fixed by tracking * per-port slot allocation in drm_dp_mst_topology_state instead of * depending on the caller to tell us how many slots to release. */ That's not the only reason we should fix this: forcing the driver to track the VCPI allocations throughout a state's atomic check is error prone, because it means that extra care has to be taken with the order that drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() are called in in order to ensure idempotency. Currently the only driver actually using these helpers, i915, doesn't even do this correctly: multiple ->best_encoder() checks with i915's current implementation would not be idempotent and would over-allocate VCPI slots, something I learned trying to implement fallback retraining in MST. So: simplify this whole mess, and teach drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() to track the VCPI allocations for each port. This allows us to ensure idempotency without having to rely on the driver as much. Additionally: the driver doesn't need to do any kind of VCPI slot tracking anymore if it doesn't need it for it's own internal state. Additionally; this adds a new drm_dp_mst_atomic_check() helper which must be used by atomic drivers to perform validity checks for the new VCPI allocations incurred by a state. Also: update the documentation and make it more obvious that these /must/ be called by /all/ atomic drivers supporting MST. Changes since v9: * Add some missing changes that were requested by danvet that I forgot about after I redid all of the kref stuff: * Remove unnecessary state changes in intel_dp_mst_atomic_check * Cleanup atomic check logic for VCPI allocations - all we need to check in compute_config is whether or not this state disables a CRTC, then free VCPI based off that Changes since v8: * Fix compile errors, whoops! Changes since v7: - Don't check for mixed stale/valid VCPI allocations, just rely on connector registration to stop such erroneous modesets Changes since v6: - Keep a kref to all of the ports we have allocations on. This required a good bit of changing to when we call drm_dp_find_vcpi_slots(), mainly that we need to ensure that we only redo VCPI allocations on actual mode or CRTC changes, not crtc_state->active changes. Additionally, we no longer take the registration of the DRM connector for each port into account because so long as we have a kref to the port in the new or previous atomic state, the connector will stay registered. - Use the small changes to drm_dp_put_port() to add even more error checking to make misusage of the helpers more obvious. I added this after having to chase down various use-after-free conditions that started popping up from the new helpers so no one else has to troubleshoot that. - Move some accidental DRM_DEBUG_KMS() calls to DRM_DEBUG_ATOMIC() - Update documentation again, note that find/release() should both not be called on the same port in a single atomic check phase (but multiple calls to one or the other is OK) Changes since v4: - Don't skip the atomic checks for VCPI allocations if no new VCPI allocations happen in a state. This makes the next change I'm about to list here a lot easier to implement. - Don't ignore VCPI allocations on destroyed ports, instead ensure that when ports are destroyed and still have VCPI allocations in the topology state, the only state changes allowed are releasing said ports' VCPI. This prevents a state with a mix of VCPI allocations from destroyed ports, and allocations from valid ports. Changes since v3: - Don't release VCPI allocations in the topology state immediately in drm_dp_atomic_release_vcpi_slots(), instead mark them as 0 and skip over them in drm_dp_mst_duplicate_state(). This makes it so drm_dp_atomic_release_vcpi_slots() is still idempotent while also throwing warnings if the driver messes up it's book keeping and tries to release VCPI slots on a port that doesn't have any pre-existing VCPI allocation - danvet - Change mst_state/state in some debugging messages to "mst state" Changes since v2: - Use kmemdup() for duplicating MST state - danvet - Move port validation out of duplicate state callback - danvet - Handle looping through MST topology states in drm_dp_mst_atomic_check() so the driver doesn't have to do it - Fix documentation in drm_dp_atomic_find_vcpi_slots() - Move the atomic check for each individual topology state into it's own function, reduces indenting - Don't consider "stale" MST ports when calculating the bandwidth requirements. This is needed because originally we relied on the state duplication functions to prune any stale ports from the new state, which would prevent us from incorrectly considering their bandwidth requirements alongside legitimate new payloads. - Add function references in drm_dp_atomic_release_vcpi_slots() - danvet - Annotate atomic VCPI and atomic check functions with __must_check - danvet Changes since v1: - Don't use the now-removed ->atomic_check() for private objects hook, just give drivers a function to call themselves Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Harry Wentland <harry.wentland@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-19-lyude@redhat.com
2019-01-11 07:53:41 +07:00
* Allocates VCPI slots to @port, replacing any previous VCPI allocations it
* may have had. Any atomic drivers which support MST must call this function
* in their &drm_encoder_helper_funcs.atomic_check() callback to change the
* current VCPI allocation for the new state, but only when
* &drm_crtc_state.mode_changed or &drm_crtc_state.connectors_changed is set
* to ensure compatibility with userspace applications that still use the
* legacy modesetting UAPI.
*
* Allocations set by this function are not checked against the bandwidth
* restraints of @mgr until the driver calls drm_dp_mst_atomic_check().
*
* Additionally, it is OK to call this function multiple times on the same
* @port as needed. It is not OK however, to call this function and
* drm_dp_atomic_release_vcpi_slots() in the same atomic check phase.
*
* See also:
* drm_dp_atomic_release_vcpi_slots()
* drm_dp_mst_atomic_check()
*
* Returns:
* Total slots in the atomic state assigned for this port, or a negative error
* code if the port no longer exists
drm/dp: Add DP MST helpers to atomically find and release vcpi slots drm_dp_atomic_find_vcpi_slots() should be called from ->atomic_check() to check there are sufficient vcpi slots for a mode and to add that to the state. This should be followed by a call to drm_dp_mst_allocate_vcpi() in ->atomic_commit() to initialize a struct vcpi for the port. drm_dp_atomic_release_vcpi_slots() should be called from ->atomic_check() to release a port's vcpi slot allocation from the state. Drivers that do not make use of this atomic helper are expected to call drm_dp_find_vcpi_slots() instead before calling drm_dp_mst_allocate_vcpi(). v3: drm_dp_atomic_release_vcpi_slots() now needs to know how many slots to release as we may not have a valid reference to port. v2: Added checks for verifying the port reference is valid Moved get_mst_topology_state() into the helpers (Daniel) Changed find_vcpi_slots() to not depend on current allocation Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Cc: Archit Taneja <architt@codeaurora.org> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Harry Wentland <Harry.wentland@amd.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Signed-off-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com> Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1492753893-3748-4-git-send-email-dhinakaran.pandiyan@intel.com
2017-04-21 12:51:32 +07:00
*/
int drm_dp_atomic_find_vcpi_slots(struct drm_atomic_state *state,
struct drm_dp_mst_topology_mgr *mgr,
struct drm_dp_mst_port *port, int pbn)
{
struct drm_dp_mst_topology_state *topology_state;
drm/dp_mst: Start tracking per-port VCPI allocations There has been a TODO waiting for quite a long time in drm_dp_mst_topology.c: /* We cannot rely on port->vcpi.num_slots to update * topology_state->avail_slots as the port may not exist if the parent * branch device was unplugged. This should be fixed by tracking * per-port slot allocation in drm_dp_mst_topology_state instead of * depending on the caller to tell us how many slots to release. */ That's not the only reason we should fix this: forcing the driver to track the VCPI allocations throughout a state's atomic check is error prone, because it means that extra care has to be taken with the order that drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() are called in in order to ensure idempotency. Currently the only driver actually using these helpers, i915, doesn't even do this correctly: multiple ->best_encoder() checks with i915's current implementation would not be idempotent and would over-allocate VCPI slots, something I learned trying to implement fallback retraining in MST. So: simplify this whole mess, and teach drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() to track the VCPI allocations for each port. This allows us to ensure idempotency without having to rely on the driver as much. Additionally: the driver doesn't need to do any kind of VCPI slot tracking anymore if it doesn't need it for it's own internal state. Additionally; this adds a new drm_dp_mst_atomic_check() helper which must be used by atomic drivers to perform validity checks for the new VCPI allocations incurred by a state. Also: update the documentation and make it more obvious that these /must/ be called by /all/ atomic drivers supporting MST. Changes since v9: * Add some missing changes that were requested by danvet that I forgot about after I redid all of the kref stuff: * Remove unnecessary state changes in intel_dp_mst_atomic_check * Cleanup atomic check logic for VCPI allocations - all we need to check in compute_config is whether or not this state disables a CRTC, then free VCPI based off that Changes since v8: * Fix compile errors, whoops! Changes since v7: - Don't check for mixed stale/valid VCPI allocations, just rely on connector registration to stop such erroneous modesets Changes since v6: - Keep a kref to all of the ports we have allocations on. This required a good bit of changing to when we call drm_dp_find_vcpi_slots(), mainly that we need to ensure that we only redo VCPI allocations on actual mode or CRTC changes, not crtc_state->active changes. Additionally, we no longer take the registration of the DRM connector for each port into account because so long as we have a kref to the port in the new or previous atomic state, the connector will stay registered. - Use the small changes to drm_dp_put_port() to add even more error checking to make misusage of the helpers more obvious. I added this after having to chase down various use-after-free conditions that started popping up from the new helpers so no one else has to troubleshoot that. - Move some accidental DRM_DEBUG_KMS() calls to DRM_DEBUG_ATOMIC() - Update documentation again, note that find/release() should both not be called on the same port in a single atomic check phase (but multiple calls to one or the other is OK) Changes since v4: - Don't skip the atomic checks for VCPI allocations if no new VCPI allocations happen in a state. This makes the next change I'm about to list here a lot easier to implement. - Don't ignore VCPI allocations on destroyed ports, instead ensure that when ports are destroyed and still have VCPI allocations in the topology state, the only state changes allowed are releasing said ports' VCPI. This prevents a state with a mix of VCPI allocations from destroyed ports, and allocations from valid ports. Changes since v3: - Don't release VCPI allocations in the topology state immediately in drm_dp_atomic_release_vcpi_slots(), instead mark them as 0 and skip over them in drm_dp_mst_duplicate_state(). This makes it so drm_dp_atomic_release_vcpi_slots() is still idempotent while also throwing warnings if the driver messes up it's book keeping and tries to release VCPI slots on a port that doesn't have any pre-existing VCPI allocation - danvet - Change mst_state/state in some debugging messages to "mst state" Changes since v2: - Use kmemdup() for duplicating MST state - danvet - Move port validation out of duplicate state callback - danvet - Handle looping through MST topology states in drm_dp_mst_atomic_check() so the driver doesn't have to do it - Fix documentation in drm_dp_atomic_find_vcpi_slots() - Move the atomic check for each individual topology state into it's own function, reduces indenting - Don't consider "stale" MST ports when calculating the bandwidth requirements. This is needed because originally we relied on the state duplication functions to prune any stale ports from the new state, which would prevent us from incorrectly considering their bandwidth requirements alongside legitimate new payloads. - Add function references in drm_dp_atomic_release_vcpi_slots() - danvet - Annotate atomic VCPI and atomic check functions with __must_check - danvet Changes since v1: - Don't use the now-removed ->atomic_check() for private objects hook, just give drivers a function to call themselves Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Harry Wentland <harry.wentland@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-19-lyude@redhat.com
2019-01-11 07:53:41 +07:00
struct drm_dp_vcpi_allocation *pos, *vcpi = NULL;
int prev_slots, req_slots, ret;
drm/dp: Add DP MST helpers to atomically find and release vcpi slots drm_dp_atomic_find_vcpi_slots() should be called from ->atomic_check() to check there are sufficient vcpi slots for a mode and to add that to the state. This should be followed by a call to drm_dp_mst_allocate_vcpi() in ->atomic_commit() to initialize a struct vcpi for the port. drm_dp_atomic_release_vcpi_slots() should be called from ->atomic_check() to release a port's vcpi slot allocation from the state. Drivers that do not make use of this atomic helper are expected to call drm_dp_find_vcpi_slots() instead before calling drm_dp_mst_allocate_vcpi(). v3: drm_dp_atomic_release_vcpi_slots() now needs to know how many slots to release as we may not have a valid reference to port. v2: Added checks for verifying the port reference is valid Moved get_mst_topology_state() into the helpers (Daniel) Changed find_vcpi_slots() to not depend on current allocation Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Cc: Archit Taneja <architt@codeaurora.org> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Harry Wentland <Harry.wentland@amd.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Signed-off-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com> Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1492753893-3748-4-git-send-email-dhinakaran.pandiyan@intel.com
2017-04-21 12:51:32 +07:00
topology_state = drm_atomic_get_mst_topology_state(state, mgr);
if (IS_ERR(topology_state))
return PTR_ERR(topology_state);
drm/dp: Add DP MST helpers to atomically find and release vcpi slots drm_dp_atomic_find_vcpi_slots() should be called from ->atomic_check() to check there are sufficient vcpi slots for a mode and to add that to the state. This should be followed by a call to drm_dp_mst_allocate_vcpi() in ->atomic_commit() to initialize a struct vcpi for the port. drm_dp_atomic_release_vcpi_slots() should be called from ->atomic_check() to release a port's vcpi slot allocation from the state. Drivers that do not make use of this atomic helper are expected to call drm_dp_find_vcpi_slots() instead before calling drm_dp_mst_allocate_vcpi(). v3: drm_dp_atomic_release_vcpi_slots() now needs to know how many slots to release as we may not have a valid reference to port. v2: Added checks for verifying the port reference is valid Moved get_mst_topology_state() into the helpers (Daniel) Changed find_vcpi_slots() to not depend on current allocation Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Cc: Archit Taneja <architt@codeaurora.org> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Harry Wentland <Harry.wentland@amd.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Signed-off-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com> Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1492753893-3748-4-git-send-email-dhinakaran.pandiyan@intel.com
2017-04-21 12:51:32 +07:00
drm/dp_mst: Start tracking per-port VCPI allocations There has been a TODO waiting for quite a long time in drm_dp_mst_topology.c: /* We cannot rely on port->vcpi.num_slots to update * topology_state->avail_slots as the port may not exist if the parent * branch device was unplugged. This should be fixed by tracking * per-port slot allocation in drm_dp_mst_topology_state instead of * depending on the caller to tell us how many slots to release. */ That's not the only reason we should fix this: forcing the driver to track the VCPI allocations throughout a state's atomic check is error prone, because it means that extra care has to be taken with the order that drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() are called in in order to ensure idempotency. Currently the only driver actually using these helpers, i915, doesn't even do this correctly: multiple ->best_encoder() checks with i915's current implementation would not be idempotent and would over-allocate VCPI slots, something I learned trying to implement fallback retraining in MST. So: simplify this whole mess, and teach drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() to track the VCPI allocations for each port. This allows us to ensure idempotency without having to rely on the driver as much. Additionally: the driver doesn't need to do any kind of VCPI slot tracking anymore if it doesn't need it for it's own internal state. Additionally; this adds a new drm_dp_mst_atomic_check() helper which must be used by atomic drivers to perform validity checks for the new VCPI allocations incurred by a state. Also: update the documentation and make it more obvious that these /must/ be called by /all/ atomic drivers supporting MST. Changes since v9: * Add some missing changes that were requested by danvet that I forgot about after I redid all of the kref stuff: * Remove unnecessary state changes in intel_dp_mst_atomic_check * Cleanup atomic check logic for VCPI allocations - all we need to check in compute_config is whether or not this state disables a CRTC, then free VCPI based off that Changes since v8: * Fix compile errors, whoops! Changes since v7: - Don't check for mixed stale/valid VCPI allocations, just rely on connector registration to stop such erroneous modesets Changes since v6: - Keep a kref to all of the ports we have allocations on. This required a good bit of changing to when we call drm_dp_find_vcpi_slots(), mainly that we need to ensure that we only redo VCPI allocations on actual mode or CRTC changes, not crtc_state->active changes. Additionally, we no longer take the registration of the DRM connector for each port into account because so long as we have a kref to the port in the new or previous atomic state, the connector will stay registered. - Use the small changes to drm_dp_put_port() to add even more error checking to make misusage of the helpers more obvious. I added this after having to chase down various use-after-free conditions that started popping up from the new helpers so no one else has to troubleshoot that. - Move some accidental DRM_DEBUG_KMS() calls to DRM_DEBUG_ATOMIC() - Update documentation again, note that find/release() should both not be called on the same port in a single atomic check phase (but multiple calls to one or the other is OK) Changes since v4: - Don't skip the atomic checks for VCPI allocations if no new VCPI allocations happen in a state. This makes the next change I'm about to list here a lot easier to implement. - Don't ignore VCPI allocations on destroyed ports, instead ensure that when ports are destroyed and still have VCPI allocations in the topology state, the only state changes allowed are releasing said ports' VCPI. This prevents a state with a mix of VCPI allocations from destroyed ports, and allocations from valid ports. Changes since v3: - Don't release VCPI allocations in the topology state immediately in drm_dp_atomic_release_vcpi_slots(), instead mark them as 0 and skip over them in drm_dp_mst_duplicate_state(). This makes it so drm_dp_atomic_release_vcpi_slots() is still idempotent while also throwing warnings if the driver messes up it's book keeping and tries to release VCPI slots on a port that doesn't have any pre-existing VCPI allocation - danvet - Change mst_state/state in some debugging messages to "mst state" Changes since v2: - Use kmemdup() for duplicating MST state - danvet - Move port validation out of duplicate state callback - danvet - Handle looping through MST topology states in drm_dp_mst_atomic_check() so the driver doesn't have to do it - Fix documentation in drm_dp_atomic_find_vcpi_slots() - Move the atomic check for each individual topology state into it's own function, reduces indenting - Don't consider "stale" MST ports when calculating the bandwidth requirements. This is needed because originally we relied on the state duplication functions to prune any stale ports from the new state, which would prevent us from incorrectly considering their bandwidth requirements alongside legitimate new payloads. - Add function references in drm_dp_atomic_release_vcpi_slots() - danvet - Annotate atomic VCPI and atomic check functions with __must_check - danvet Changes since v1: - Don't use the now-removed ->atomic_check() for private objects hook, just give drivers a function to call themselves Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Harry Wentland <harry.wentland@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-19-lyude@redhat.com
2019-01-11 07:53:41 +07:00
/* Find the current allocation for this port, if any */
list_for_each_entry(pos, &topology_state->vcpis, next) {
if (pos->port == port) {
vcpi = pos;
prev_slots = vcpi->vcpi;
/*
* This should never happen, unless the driver tries
* releasing and allocating the same VCPI allocation,
* which is an error
*/
if (WARN_ON(!prev_slots)) {
DRM_ERROR("cannot allocate and release VCPI on [MST PORT:%p] in the same state\n",
port);
return -EINVAL;
}
break;
}
drm/dp: Add DP MST helpers to atomically find and release vcpi slots drm_dp_atomic_find_vcpi_slots() should be called from ->atomic_check() to check there are sufficient vcpi slots for a mode and to add that to the state. This should be followed by a call to drm_dp_mst_allocate_vcpi() in ->atomic_commit() to initialize a struct vcpi for the port. drm_dp_atomic_release_vcpi_slots() should be called from ->atomic_check() to release a port's vcpi slot allocation from the state. Drivers that do not make use of this atomic helper are expected to call drm_dp_find_vcpi_slots() instead before calling drm_dp_mst_allocate_vcpi(). v3: drm_dp_atomic_release_vcpi_slots() now needs to know how many slots to release as we may not have a valid reference to port. v2: Added checks for verifying the port reference is valid Moved get_mst_topology_state() into the helpers (Daniel) Changed find_vcpi_slots() to not depend on current allocation Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Cc: Archit Taneja <architt@codeaurora.org> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Harry Wentland <Harry.wentland@amd.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Signed-off-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com> Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1492753893-3748-4-git-send-email-dhinakaran.pandiyan@intel.com
2017-04-21 12:51:32 +07:00
}
drm/dp_mst: Start tracking per-port VCPI allocations There has been a TODO waiting for quite a long time in drm_dp_mst_topology.c: /* We cannot rely on port->vcpi.num_slots to update * topology_state->avail_slots as the port may not exist if the parent * branch device was unplugged. This should be fixed by tracking * per-port slot allocation in drm_dp_mst_topology_state instead of * depending on the caller to tell us how many slots to release. */ That's not the only reason we should fix this: forcing the driver to track the VCPI allocations throughout a state's atomic check is error prone, because it means that extra care has to be taken with the order that drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() are called in in order to ensure idempotency. Currently the only driver actually using these helpers, i915, doesn't even do this correctly: multiple ->best_encoder() checks with i915's current implementation would not be idempotent and would over-allocate VCPI slots, something I learned trying to implement fallback retraining in MST. So: simplify this whole mess, and teach drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() to track the VCPI allocations for each port. This allows us to ensure idempotency without having to rely on the driver as much. Additionally: the driver doesn't need to do any kind of VCPI slot tracking anymore if it doesn't need it for it's own internal state. Additionally; this adds a new drm_dp_mst_atomic_check() helper which must be used by atomic drivers to perform validity checks for the new VCPI allocations incurred by a state. Also: update the documentation and make it more obvious that these /must/ be called by /all/ atomic drivers supporting MST. Changes since v9: * Add some missing changes that were requested by danvet that I forgot about after I redid all of the kref stuff: * Remove unnecessary state changes in intel_dp_mst_atomic_check * Cleanup atomic check logic for VCPI allocations - all we need to check in compute_config is whether or not this state disables a CRTC, then free VCPI based off that Changes since v8: * Fix compile errors, whoops! Changes since v7: - Don't check for mixed stale/valid VCPI allocations, just rely on connector registration to stop such erroneous modesets Changes since v6: - Keep a kref to all of the ports we have allocations on. This required a good bit of changing to when we call drm_dp_find_vcpi_slots(), mainly that we need to ensure that we only redo VCPI allocations on actual mode or CRTC changes, not crtc_state->active changes. Additionally, we no longer take the registration of the DRM connector for each port into account because so long as we have a kref to the port in the new or previous atomic state, the connector will stay registered. - Use the small changes to drm_dp_put_port() to add even more error checking to make misusage of the helpers more obvious. I added this after having to chase down various use-after-free conditions that started popping up from the new helpers so no one else has to troubleshoot that. - Move some accidental DRM_DEBUG_KMS() calls to DRM_DEBUG_ATOMIC() - Update documentation again, note that find/release() should both not be called on the same port in a single atomic check phase (but multiple calls to one or the other is OK) Changes since v4: - Don't skip the atomic checks for VCPI allocations if no new VCPI allocations happen in a state. This makes the next change I'm about to list here a lot easier to implement. - Don't ignore VCPI allocations on destroyed ports, instead ensure that when ports are destroyed and still have VCPI allocations in the topology state, the only state changes allowed are releasing said ports' VCPI. This prevents a state with a mix of VCPI allocations from destroyed ports, and allocations from valid ports. Changes since v3: - Don't release VCPI allocations in the topology state immediately in drm_dp_atomic_release_vcpi_slots(), instead mark them as 0 and skip over them in drm_dp_mst_duplicate_state(). This makes it so drm_dp_atomic_release_vcpi_slots() is still idempotent while also throwing warnings if the driver messes up it's book keeping and tries to release VCPI slots on a port that doesn't have any pre-existing VCPI allocation - danvet - Change mst_state/state in some debugging messages to "mst state" Changes since v2: - Use kmemdup() for duplicating MST state - danvet - Move port validation out of duplicate state callback - danvet - Handle looping through MST topology states in drm_dp_mst_atomic_check() so the driver doesn't have to do it - Fix documentation in drm_dp_atomic_find_vcpi_slots() - Move the atomic check for each individual topology state into it's own function, reduces indenting - Don't consider "stale" MST ports when calculating the bandwidth requirements. This is needed because originally we relied on the state duplication functions to prune any stale ports from the new state, which would prevent us from incorrectly considering their bandwidth requirements alongside legitimate new payloads. - Add function references in drm_dp_atomic_release_vcpi_slots() - danvet - Annotate atomic VCPI and atomic check functions with __must_check - danvet Changes since v1: - Don't use the now-removed ->atomic_check() for private objects hook, just give drivers a function to call themselves Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Harry Wentland <harry.wentland@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-19-lyude@redhat.com
2019-01-11 07:53:41 +07:00
if (!vcpi)
prev_slots = 0;
req_slots = DIV_ROUND_UP(pbn, mgr->pbn_div);
DRM_DEBUG_ATOMIC("[CONNECTOR:%d:%s] [MST PORT:%p] VCPI %d -> %d\n",
port->connector->base.id, port->connector->name,
port, prev_slots, req_slots);
/* Add the new allocation to the state */
if (!vcpi) {
vcpi = kzalloc(sizeof(*vcpi), GFP_KERNEL);
if (!vcpi)
return -ENOMEM;
drm/dp: Add DP MST helpers to atomically find and release vcpi slots drm_dp_atomic_find_vcpi_slots() should be called from ->atomic_check() to check there are sufficient vcpi slots for a mode and to add that to the state. This should be followed by a call to drm_dp_mst_allocate_vcpi() in ->atomic_commit() to initialize a struct vcpi for the port. drm_dp_atomic_release_vcpi_slots() should be called from ->atomic_check() to release a port's vcpi slot allocation from the state. Drivers that do not make use of this atomic helper are expected to call drm_dp_find_vcpi_slots() instead before calling drm_dp_mst_allocate_vcpi(). v3: drm_dp_atomic_release_vcpi_slots() now needs to know how many slots to release as we may not have a valid reference to port. v2: Added checks for verifying the port reference is valid Moved get_mst_topology_state() into the helpers (Daniel) Changed find_vcpi_slots() to not depend on current allocation Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Cc: Archit Taneja <architt@codeaurora.org> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Harry Wentland <Harry.wentland@amd.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Signed-off-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com> Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1492753893-3748-4-git-send-email-dhinakaran.pandiyan@intel.com
2017-04-21 12:51:32 +07:00
drm/dp_mst: Start tracking per-port VCPI allocations There has been a TODO waiting for quite a long time in drm_dp_mst_topology.c: /* We cannot rely on port->vcpi.num_slots to update * topology_state->avail_slots as the port may not exist if the parent * branch device was unplugged. This should be fixed by tracking * per-port slot allocation in drm_dp_mst_topology_state instead of * depending on the caller to tell us how many slots to release. */ That's not the only reason we should fix this: forcing the driver to track the VCPI allocations throughout a state's atomic check is error prone, because it means that extra care has to be taken with the order that drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() are called in in order to ensure idempotency. Currently the only driver actually using these helpers, i915, doesn't even do this correctly: multiple ->best_encoder() checks with i915's current implementation would not be idempotent and would over-allocate VCPI slots, something I learned trying to implement fallback retraining in MST. So: simplify this whole mess, and teach drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() to track the VCPI allocations for each port. This allows us to ensure idempotency without having to rely on the driver as much. Additionally: the driver doesn't need to do any kind of VCPI slot tracking anymore if it doesn't need it for it's own internal state. Additionally; this adds a new drm_dp_mst_atomic_check() helper which must be used by atomic drivers to perform validity checks for the new VCPI allocations incurred by a state. Also: update the documentation and make it more obvious that these /must/ be called by /all/ atomic drivers supporting MST. Changes since v9: * Add some missing changes that were requested by danvet that I forgot about after I redid all of the kref stuff: * Remove unnecessary state changes in intel_dp_mst_atomic_check * Cleanup atomic check logic for VCPI allocations - all we need to check in compute_config is whether or not this state disables a CRTC, then free VCPI based off that Changes since v8: * Fix compile errors, whoops! Changes since v7: - Don't check for mixed stale/valid VCPI allocations, just rely on connector registration to stop such erroneous modesets Changes since v6: - Keep a kref to all of the ports we have allocations on. This required a good bit of changing to when we call drm_dp_find_vcpi_slots(), mainly that we need to ensure that we only redo VCPI allocations on actual mode or CRTC changes, not crtc_state->active changes. Additionally, we no longer take the registration of the DRM connector for each port into account because so long as we have a kref to the port in the new or previous atomic state, the connector will stay registered. - Use the small changes to drm_dp_put_port() to add even more error checking to make misusage of the helpers more obvious. I added this after having to chase down various use-after-free conditions that started popping up from the new helpers so no one else has to troubleshoot that. - Move some accidental DRM_DEBUG_KMS() calls to DRM_DEBUG_ATOMIC() - Update documentation again, note that find/release() should both not be called on the same port in a single atomic check phase (but multiple calls to one or the other is OK) Changes since v4: - Don't skip the atomic checks for VCPI allocations if no new VCPI allocations happen in a state. This makes the next change I'm about to list here a lot easier to implement. - Don't ignore VCPI allocations on destroyed ports, instead ensure that when ports are destroyed and still have VCPI allocations in the topology state, the only state changes allowed are releasing said ports' VCPI. This prevents a state with a mix of VCPI allocations from destroyed ports, and allocations from valid ports. Changes since v3: - Don't release VCPI allocations in the topology state immediately in drm_dp_atomic_release_vcpi_slots(), instead mark them as 0 and skip over them in drm_dp_mst_duplicate_state(). This makes it so drm_dp_atomic_release_vcpi_slots() is still idempotent while also throwing warnings if the driver messes up it's book keeping and tries to release VCPI slots on a port that doesn't have any pre-existing VCPI allocation - danvet - Change mst_state/state in some debugging messages to "mst state" Changes since v2: - Use kmemdup() for duplicating MST state - danvet - Move port validation out of duplicate state callback - danvet - Handle looping through MST topology states in drm_dp_mst_atomic_check() so the driver doesn't have to do it - Fix documentation in drm_dp_atomic_find_vcpi_slots() - Move the atomic check for each individual topology state into it's own function, reduces indenting - Don't consider "stale" MST ports when calculating the bandwidth requirements. This is needed because originally we relied on the state duplication functions to prune any stale ports from the new state, which would prevent us from incorrectly considering their bandwidth requirements alongside legitimate new payloads. - Add function references in drm_dp_atomic_release_vcpi_slots() - danvet - Annotate atomic VCPI and atomic check functions with __must_check - danvet Changes since v1: - Don't use the now-removed ->atomic_check() for private objects hook, just give drivers a function to call themselves Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Harry Wentland <harry.wentland@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-19-lyude@redhat.com
2019-01-11 07:53:41 +07:00
drm_dp_mst_get_port_malloc(port);
vcpi->port = port;
list_add(&vcpi->next, &topology_state->vcpis);
}
vcpi->vcpi = req_slots;
drm/dp: Add DP MST helpers to atomically find and release vcpi slots drm_dp_atomic_find_vcpi_slots() should be called from ->atomic_check() to check there are sufficient vcpi slots for a mode and to add that to the state. This should be followed by a call to drm_dp_mst_allocate_vcpi() in ->atomic_commit() to initialize a struct vcpi for the port. drm_dp_atomic_release_vcpi_slots() should be called from ->atomic_check() to release a port's vcpi slot allocation from the state. Drivers that do not make use of this atomic helper are expected to call drm_dp_find_vcpi_slots() instead before calling drm_dp_mst_allocate_vcpi(). v3: drm_dp_atomic_release_vcpi_slots() now needs to know how many slots to release as we may not have a valid reference to port. v2: Added checks for verifying the port reference is valid Moved get_mst_topology_state() into the helpers (Daniel) Changed find_vcpi_slots() to not depend on current allocation Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Cc: Archit Taneja <architt@codeaurora.org> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Harry Wentland <Harry.wentland@amd.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Signed-off-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com> Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1492753893-3748-4-git-send-email-dhinakaran.pandiyan@intel.com
2017-04-21 12:51:32 +07:00
drm/dp_mst: Start tracking per-port VCPI allocations There has been a TODO waiting for quite a long time in drm_dp_mst_topology.c: /* We cannot rely on port->vcpi.num_slots to update * topology_state->avail_slots as the port may not exist if the parent * branch device was unplugged. This should be fixed by tracking * per-port slot allocation in drm_dp_mst_topology_state instead of * depending on the caller to tell us how many slots to release. */ That's not the only reason we should fix this: forcing the driver to track the VCPI allocations throughout a state's atomic check is error prone, because it means that extra care has to be taken with the order that drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() are called in in order to ensure idempotency. Currently the only driver actually using these helpers, i915, doesn't even do this correctly: multiple ->best_encoder() checks with i915's current implementation would not be idempotent and would over-allocate VCPI slots, something I learned trying to implement fallback retraining in MST. So: simplify this whole mess, and teach drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() to track the VCPI allocations for each port. This allows us to ensure idempotency without having to rely on the driver as much. Additionally: the driver doesn't need to do any kind of VCPI slot tracking anymore if it doesn't need it for it's own internal state. Additionally; this adds a new drm_dp_mst_atomic_check() helper which must be used by atomic drivers to perform validity checks for the new VCPI allocations incurred by a state. Also: update the documentation and make it more obvious that these /must/ be called by /all/ atomic drivers supporting MST. Changes since v9: * Add some missing changes that were requested by danvet that I forgot about after I redid all of the kref stuff: * Remove unnecessary state changes in intel_dp_mst_atomic_check * Cleanup atomic check logic for VCPI allocations - all we need to check in compute_config is whether or not this state disables a CRTC, then free VCPI based off that Changes since v8: * Fix compile errors, whoops! Changes since v7: - Don't check for mixed stale/valid VCPI allocations, just rely on connector registration to stop such erroneous modesets Changes since v6: - Keep a kref to all of the ports we have allocations on. This required a good bit of changing to when we call drm_dp_find_vcpi_slots(), mainly that we need to ensure that we only redo VCPI allocations on actual mode or CRTC changes, not crtc_state->active changes. Additionally, we no longer take the registration of the DRM connector for each port into account because so long as we have a kref to the port in the new or previous atomic state, the connector will stay registered. - Use the small changes to drm_dp_put_port() to add even more error checking to make misusage of the helpers more obvious. I added this after having to chase down various use-after-free conditions that started popping up from the new helpers so no one else has to troubleshoot that. - Move some accidental DRM_DEBUG_KMS() calls to DRM_DEBUG_ATOMIC() - Update documentation again, note that find/release() should both not be called on the same port in a single atomic check phase (but multiple calls to one or the other is OK) Changes since v4: - Don't skip the atomic checks for VCPI allocations if no new VCPI allocations happen in a state. This makes the next change I'm about to list here a lot easier to implement. - Don't ignore VCPI allocations on destroyed ports, instead ensure that when ports are destroyed and still have VCPI allocations in the topology state, the only state changes allowed are releasing said ports' VCPI. This prevents a state with a mix of VCPI allocations from destroyed ports, and allocations from valid ports. Changes since v3: - Don't release VCPI allocations in the topology state immediately in drm_dp_atomic_release_vcpi_slots(), instead mark them as 0 and skip over them in drm_dp_mst_duplicate_state(). This makes it so drm_dp_atomic_release_vcpi_slots() is still idempotent while also throwing warnings if the driver messes up it's book keeping and tries to release VCPI slots on a port that doesn't have any pre-existing VCPI allocation - danvet - Change mst_state/state in some debugging messages to "mst state" Changes since v2: - Use kmemdup() for duplicating MST state - danvet - Move port validation out of duplicate state callback - danvet - Handle looping through MST topology states in drm_dp_mst_atomic_check() so the driver doesn't have to do it - Fix documentation in drm_dp_atomic_find_vcpi_slots() - Move the atomic check for each individual topology state into it's own function, reduces indenting - Don't consider "stale" MST ports when calculating the bandwidth requirements. This is needed because originally we relied on the state duplication functions to prune any stale ports from the new state, which would prevent us from incorrectly considering their bandwidth requirements alongside legitimate new payloads. - Add function references in drm_dp_atomic_release_vcpi_slots() - danvet - Annotate atomic VCPI and atomic check functions with __must_check - danvet Changes since v1: - Don't use the now-removed ->atomic_check() for private objects hook, just give drivers a function to call themselves Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Harry Wentland <harry.wentland@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-19-lyude@redhat.com
2019-01-11 07:53:41 +07:00
ret = req_slots;
return ret;
drm/dp: Add DP MST helpers to atomically find and release vcpi slots drm_dp_atomic_find_vcpi_slots() should be called from ->atomic_check() to check there are sufficient vcpi slots for a mode and to add that to the state. This should be followed by a call to drm_dp_mst_allocate_vcpi() in ->atomic_commit() to initialize a struct vcpi for the port. drm_dp_atomic_release_vcpi_slots() should be called from ->atomic_check() to release a port's vcpi slot allocation from the state. Drivers that do not make use of this atomic helper are expected to call drm_dp_find_vcpi_slots() instead before calling drm_dp_mst_allocate_vcpi(). v3: drm_dp_atomic_release_vcpi_slots() now needs to know how many slots to release as we may not have a valid reference to port. v2: Added checks for verifying the port reference is valid Moved get_mst_topology_state() into the helpers (Daniel) Changed find_vcpi_slots() to not depend on current allocation Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Cc: Archit Taneja <architt@codeaurora.org> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Harry Wentland <Harry.wentland@amd.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Signed-off-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com> Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1492753893-3748-4-git-send-email-dhinakaran.pandiyan@intel.com
2017-04-21 12:51:32 +07:00
}
EXPORT_SYMBOL(drm_dp_atomic_find_vcpi_slots);
/**
* drm_dp_atomic_release_vcpi_slots() - Release allocated vcpi slots
* @state: global atomic state
* @mgr: MST topology manager for the port
drm/dp_mst: Start tracking per-port VCPI allocations There has been a TODO waiting for quite a long time in drm_dp_mst_topology.c: /* We cannot rely on port->vcpi.num_slots to update * topology_state->avail_slots as the port may not exist if the parent * branch device was unplugged. This should be fixed by tracking * per-port slot allocation in drm_dp_mst_topology_state instead of * depending on the caller to tell us how many slots to release. */ That's not the only reason we should fix this: forcing the driver to track the VCPI allocations throughout a state's atomic check is error prone, because it means that extra care has to be taken with the order that drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() are called in in order to ensure idempotency. Currently the only driver actually using these helpers, i915, doesn't even do this correctly: multiple ->best_encoder() checks with i915's current implementation would not be idempotent and would over-allocate VCPI slots, something I learned trying to implement fallback retraining in MST. So: simplify this whole mess, and teach drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() to track the VCPI allocations for each port. This allows us to ensure idempotency without having to rely on the driver as much. Additionally: the driver doesn't need to do any kind of VCPI slot tracking anymore if it doesn't need it for it's own internal state. Additionally; this adds a new drm_dp_mst_atomic_check() helper which must be used by atomic drivers to perform validity checks for the new VCPI allocations incurred by a state. Also: update the documentation and make it more obvious that these /must/ be called by /all/ atomic drivers supporting MST. Changes since v9: * Add some missing changes that were requested by danvet that I forgot about after I redid all of the kref stuff: * Remove unnecessary state changes in intel_dp_mst_atomic_check * Cleanup atomic check logic for VCPI allocations - all we need to check in compute_config is whether or not this state disables a CRTC, then free VCPI based off that Changes since v8: * Fix compile errors, whoops! Changes since v7: - Don't check for mixed stale/valid VCPI allocations, just rely on connector registration to stop such erroneous modesets Changes since v6: - Keep a kref to all of the ports we have allocations on. This required a good bit of changing to when we call drm_dp_find_vcpi_slots(), mainly that we need to ensure that we only redo VCPI allocations on actual mode or CRTC changes, not crtc_state->active changes. Additionally, we no longer take the registration of the DRM connector for each port into account because so long as we have a kref to the port in the new or previous atomic state, the connector will stay registered. - Use the small changes to drm_dp_put_port() to add even more error checking to make misusage of the helpers more obvious. I added this after having to chase down various use-after-free conditions that started popping up from the new helpers so no one else has to troubleshoot that. - Move some accidental DRM_DEBUG_KMS() calls to DRM_DEBUG_ATOMIC() - Update documentation again, note that find/release() should both not be called on the same port in a single atomic check phase (but multiple calls to one or the other is OK) Changes since v4: - Don't skip the atomic checks for VCPI allocations if no new VCPI allocations happen in a state. This makes the next change I'm about to list here a lot easier to implement. - Don't ignore VCPI allocations on destroyed ports, instead ensure that when ports are destroyed and still have VCPI allocations in the topology state, the only state changes allowed are releasing said ports' VCPI. This prevents a state with a mix of VCPI allocations from destroyed ports, and allocations from valid ports. Changes since v3: - Don't release VCPI allocations in the topology state immediately in drm_dp_atomic_release_vcpi_slots(), instead mark them as 0 and skip over them in drm_dp_mst_duplicate_state(). This makes it so drm_dp_atomic_release_vcpi_slots() is still idempotent while also throwing warnings if the driver messes up it's book keeping and tries to release VCPI slots on a port that doesn't have any pre-existing VCPI allocation - danvet - Change mst_state/state in some debugging messages to "mst state" Changes since v2: - Use kmemdup() for duplicating MST state - danvet - Move port validation out of duplicate state callback - danvet - Handle looping through MST topology states in drm_dp_mst_atomic_check() so the driver doesn't have to do it - Fix documentation in drm_dp_atomic_find_vcpi_slots() - Move the atomic check for each individual topology state into it's own function, reduces indenting - Don't consider "stale" MST ports when calculating the bandwidth requirements. This is needed because originally we relied on the state duplication functions to prune any stale ports from the new state, which would prevent us from incorrectly considering their bandwidth requirements alongside legitimate new payloads. - Add function references in drm_dp_atomic_release_vcpi_slots() - danvet - Annotate atomic VCPI and atomic check functions with __must_check - danvet Changes since v1: - Don't use the now-removed ->atomic_check() for private objects hook, just give drivers a function to call themselves Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Harry Wentland <harry.wentland@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-19-lyude@redhat.com
2019-01-11 07:53:41 +07:00
* @port: The port to release the VCPI slots from
drm/dp: Add DP MST helpers to atomically find and release vcpi slots drm_dp_atomic_find_vcpi_slots() should be called from ->atomic_check() to check there are sufficient vcpi slots for a mode and to add that to the state. This should be followed by a call to drm_dp_mst_allocate_vcpi() in ->atomic_commit() to initialize a struct vcpi for the port. drm_dp_atomic_release_vcpi_slots() should be called from ->atomic_check() to release a port's vcpi slot allocation from the state. Drivers that do not make use of this atomic helper are expected to call drm_dp_find_vcpi_slots() instead before calling drm_dp_mst_allocate_vcpi(). v3: drm_dp_atomic_release_vcpi_slots() now needs to know how many slots to release as we may not have a valid reference to port. v2: Added checks for verifying the port reference is valid Moved get_mst_topology_state() into the helpers (Daniel) Changed find_vcpi_slots() to not depend on current allocation Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Cc: Archit Taneja <architt@codeaurora.org> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Harry Wentland <Harry.wentland@amd.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Signed-off-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com> Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1492753893-3748-4-git-send-email-dhinakaran.pandiyan@intel.com
2017-04-21 12:51:32 +07:00
*
drm/dp_mst: Start tracking per-port VCPI allocations There has been a TODO waiting for quite a long time in drm_dp_mst_topology.c: /* We cannot rely on port->vcpi.num_slots to update * topology_state->avail_slots as the port may not exist if the parent * branch device was unplugged. This should be fixed by tracking * per-port slot allocation in drm_dp_mst_topology_state instead of * depending on the caller to tell us how many slots to release. */ That's not the only reason we should fix this: forcing the driver to track the VCPI allocations throughout a state's atomic check is error prone, because it means that extra care has to be taken with the order that drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() are called in in order to ensure idempotency. Currently the only driver actually using these helpers, i915, doesn't even do this correctly: multiple ->best_encoder() checks with i915's current implementation would not be idempotent and would over-allocate VCPI slots, something I learned trying to implement fallback retraining in MST. So: simplify this whole mess, and teach drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() to track the VCPI allocations for each port. This allows us to ensure idempotency without having to rely on the driver as much. Additionally: the driver doesn't need to do any kind of VCPI slot tracking anymore if it doesn't need it for it's own internal state. Additionally; this adds a new drm_dp_mst_atomic_check() helper which must be used by atomic drivers to perform validity checks for the new VCPI allocations incurred by a state. Also: update the documentation and make it more obvious that these /must/ be called by /all/ atomic drivers supporting MST. Changes since v9: * Add some missing changes that were requested by danvet that I forgot about after I redid all of the kref stuff: * Remove unnecessary state changes in intel_dp_mst_atomic_check * Cleanup atomic check logic for VCPI allocations - all we need to check in compute_config is whether or not this state disables a CRTC, then free VCPI based off that Changes since v8: * Fix compile errors, whoops! Changes since v7: - Don't check for mixed stale/valid VCPI allocations, just rely on connector registration to stop such erroneous modesets Changes since v6: - Keep a kref to all of the ports we have allocations on. This required a good bit of changing to when we call drm_dp_find_vcpi_slots(), mainly that we need to ensure that we only redo VCPI allocations on actual mode or CRTC changes, not crtc_state->active changes. Additionally, we no longer take the registration of the DRM connector for each port into account because so long as we have a kref to the port in the new or previous atomic state, the connector will stay registered. - Use the small changes to drm_dp_put_port() to add even more error checking to make misusage of the helpers more obvious. I added this after having to chase down various use-after-free conditions that started popping up from the new helpers so no one else has to troubleshoot that. - Move some accidental DRM_DEBUG_KMS() calls to DRM_DEBUG_ATOMIC() - Update documentation again, note that find/release() should both not be called on the same port in a single atomic check phase (but multiple calls to one or the other is OK) Changes since v4: - Don't skip the atomic checks for VCPI allocations if no new VCPI allocations happen in a state. This makes the next change I'm about to list here a lot easier to implement. - Don't ignore VCPI allocations on destroyed ports, instead ensure that when ports are destroyed and still have VCPI allocations in the topology state, the only state changes allowed are releasing said ports' VCPI. This prevents a state with a mix of VCPI allocations from destroyed ports, and allocations from valid ports. Changes since v3: - Don't release VCPI allocations in the topology state immediately in drm_dp_atomic_release_vcpi_slots(), instead mark them as 0 and skip over them in drm_dp_mst_duplicate_state(). This makes it so drm_dp_atomic_release_vcpi_slots() is still idempotent while also throwing warnings if the driver messes up it's book keeping and tries to release VCPI slots on a port that doesn't have any pre-existing VCPI allocation - danvet - Change mst_state/state in some debugging messages to "mst state" Changes since v2: - Use kmemdup() for duplicating MST state - danvet - Move port validation out of duplicate state callback - danvet - Handle looping through MST topology states in drm_dp_mst_atomic_check() so the driver doesn't have to do it - Fix documentation in drm_dp_atomic_find_vcpi_slots() - Move the atomic check for each individual topology state into it's own function, reduces indenting - Don't consider "stale" MST ports when calculating the bandwidth requirements. This is needed because originally we relied on the state duplication functions to prune any stale ports from the new state, which would prevent us from incorrectly considering their bandwidth requirements alongside legitimate new payloads. - Add function references in drm_dp_atomic_release_vcpi_slots() - danvet - Annotate atomic VCPI and atomic check functions with __must_check - danvet Changes since v1: - Don't use the now-removed ->atomic_check() for private objects hook, just give drivers a function to call themselves Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Harry Wentland <harry.wentland@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-19-lyude@redhat.com
2019-01-11 07:53:41 +07:00
* Releases any VCPI slots that have been allocated to a port in the atomic
* state. Any atomic drivers which support MST must call this function in
* their &drm_connector_helper_funcs.atomic_check() callback when the
* connector will no longer have VCPI allocated (e.g. because its CRTC was
drm/dp_mst: Start tracking per-port VCPI allocations There has been a TODO waiting for quite a long time in drm_dp_mst_topology.c: /* We cannot rely on port->vcpi.num_slots to update * topology_state->avail_slots as the port may not exist if the parent * branch device was unplugged. This should be fixed by tracking * per-port slot allocation in drm_dp_mst_topology_state instead of * depending on the caller to tell us how many slots to release. */ That's not the only reason we should fix this: forcing the driver to track the VCPI allocations throughout a state's atomic check is error prone, because it means that extra care has to be taken with the order that drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() are called in in order to ensure idempotency. Currently the only driver actually using these helpers, i915, doesn't even do this correctly: multiple ->best_encoder() checks with i915's current implementation would not be idempotent and would over-allocate VCPI slots, something I learned trying to implement fallback retraining in MST. So: simplify this whole mess, and teach drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() to track the VCPI allocations for each port. This allows us to ensure idempotency without having to rely on the driver as much. Additionally: the driver doesn't need to do any kind of VCPI slot tracking anymore if it doesn't need it for it's own internal state. Additionally; this adds a new drm_dp_mst_atomic_check() helper which must be used by atomic drivers to perform validity checks for the new VCPI allocations incurred by a state. Also: update the documentation and make it more obvious that these /must/ be called by /all/ atomic drivers supporting MST. Changes since v9: * Add some missing changes that were requested by danvet that I forgot about after I redid all of the kref stuff: * Remove unnecessary state changes in intel_dp_mst_atomic_check * Cleanup atomic check logic for VCPI allocations - all we need to check in compute_config is whether or not this state disables a CRTC, then free VCPI based off that Changes since v8: * Fix compile errors, whoops! Changes since v7: - Don't check for mixed stale/valid VCPI allocations, just rely on connector registration to stop such erroneous modesets Changes since v6: - Keep a kref to all of the ports we have allocations on. This required a good bit of changing to when we call drm_dp_find_vcpi_slots(), mainly that we need to ensure that we only redo VCPI allocations on actual mode or CRTC changes, not crtc_state->active changes. Additionally, we no longer take the registration of the DRM connector for each port into account because so long as we have a kref to the port in the new or previous atomic state, the connector will stay registered. - Use the small changes to drm_dp_put_port() to add even more error checking to make misusage of the helpers more obvious. I added this after having to chase down various use-after-free conditions that started popping up from the new helpers so no one else has to troubleshoot that. - Move some accidental DRM_DEBUG_KMS() calls to DRM_DEBUG_ATOMIC() - Update documentation again, note that find/release() should both not be called on the same port in a single atomic check phase (but multiple calls to one or the other is OK) Changes since v4: - Don't skip the atomic checks for VCPI allocations if no new VCPI allocations happen in a state. This makes the next change I'm about to list here a lot easier to implement. - Don't ignore VCPI allocations on destroyed ports, instead ensure that when ports are destroyed and still have VCPI allocations in the topology state, the only state changes allowed are releasing said ports' VCPI. This prevents a state with a mix of VCPI allocations from destroyed ports, and allocations from valid ports. Changes since v3: - Don't release VCPI allocations in the topology state immediately in drm_dp_atomic_release_vcpi_slots(), instead mark them as 0 and skip over them in drm_dp_mst_duplicate_state(). This makes it so drm_dp_atomic_release_vcpi_slots() is still idempotent while also throwing warnings if the driver messes up it's book keeping and tries to release VCPI slots on a port that doesn't have any pre-existing VCPI allocation - danvet - Change mst_state/state in some debugging messages to "mst state" Changes since v2: - Use kmemdup() for duplicating MST state - danvet - Move port validation out of duplicate state callback - danvet - Handle looping through MST topology states in drm_dp_mst_atomic_check() so the driver doesn't have to do it - Fix documentation in drm_dp_atomic_find_vcpi_slots() - Move the atomic check for each individual topology state into it's own function, reduces indenting - Don't consider "stale" MST ports when calculating the bandwidth requirements. This is needed because originally we relied on the state duplication functions to prune any stale ports from the new state, which would prevent us from incorrectly considering their bandwidth requirements alongside legitimate new payloads. - Add function references in drm_dp_atomic_release_vcpi_slots() - danvet - Annotate atomic VCPI and atomic check functions with __must_check - danvet Changes since v1: - Don't use the now-removed ->atomic_check() for private objects hook, just give drivers a function to call themselves Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Harry Wentland <harry.wentland@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-19-lyude@redhat.com
2019-01-11 07:53:41 +07:00
* removed) when it had VCPI allocated in the previous atomic state.
*
* It is OK to call this even if @port has been removed from the system.
* Additionally, it is OK to call this function multiple times on the same
* @port as needed. It is not OK however, to call this function and
* drm_dp_atomic_find_vcpi_slots() on the same @port in a single atomic check
* phase.
*
* See also:
* drm_dp_atomic_find_vcpi_slots()
* drm_dp_mst_atomic_check()
*
* Returns:
* 0 if all slots for this port were added back to
* &drm_dp_mst_topology_state.avail_slots or negative error code
drm/dp: Add DP MST helpers to atomically find and release vcpi slots drm_dp_atomic_find_vcpi_slots() should be called from ->atomic_check() to check there are sufficient vcpi slots for a mode and to add that to the state. This should be followed by a call to drm_dp_mst_allocate_vcpi() in ->atomic_commit() to initialize a struct vcpi for the port. drm_dp_atomic_release_vcpi_slots() should be called from ->atomic_check() to release a port's vcpi slot allocation from the state. Drivers that do not make use of this atomic helper are expected to call drm_dp_find_vcpi_slots() instead before calling drm_dp_mst_allocate_vcpi(). v3: drm_dp_atomic_release_vcpi_slots() now needs to know how many slots to release as we may not have a valid reference to port. v2: Added checks for verifying the port reference is valid Moved get_mst_topology_state() into the helpers (Daniel) Changed find_vcpi_slots() to not depend on current allocation Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Cc: Archit Taneja <architt@codeaurora.org> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Harry Wentland <Harry.wentland@amd.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Signed-off-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com> Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1492753893-3748-4-git-send-email-dhinakaran.pandiyan@intel.com
2017-04-21 12:51:32 +07:00
*/
int drm_dp_atomic_release_vcpi_slots(struct drm_atomic_state *state,
struct drm_dp_mst_topology_mgr *mgr,
drm/dp_mst: Start tracking per-port VCPI allocations There has been a TODO waiting for quite a long time in drm_dp_mst_topology.c: /* We cannot rely on port->vcpi.num_slots to update * topology_state->avail_slots as the port may not exist if the parent * branch device was unplugged. This should be fixed by tracking * per-port slot allocation in drm_dp_mst_topology_state instead of * depending on the caller to tell us how many slots to release. */ That's not the only reason we should fix this: forcing the driver to track the VCPI allocations throughout a state's atomic check is error prone, because it means that extra care has to be taken with the order that drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() are called in in order to ensure idempotency. Currently the only driver actually using these helpers, i915, doesn't even do this correctly: multiple ->best_encoder() checks with i915's current implementation would not be idempotent and would over-allocate VCPI slots, something I learned trying to implement fallback retraining in MST. So: simplify this whole mess, and teach drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() to track the VCPI allocations for each port. This allows us to ensure idempotency without having to rely on the driver as much. Additionally: the driver doesn't need to do any kind of VCPI slot tracking anymore if it doesn't need it for it's own internal state. Additionally; this adds a new drm_dp_mst_atomic_check() helper which must be used by atomic drivers to perform validity checks for the new VCPI allocations incurred by a state. Also: update the documentation and make it more obvious that these /must/ be called by /all/ atomic drivers supporting MST. Changes since v9: * Add some missing changes that were requested by danvet that I forgot about after I redid all of the kref stuff: * Remove unnecessary state changes in intel_dp_mst_atomic_check * Cleanup atomic check logic for VCPI allocations - all we need to check in compute_config is whether or not this state disables a CRTC, then free VCPI based off that Changes since v8: * Fix compile errors, whoops! Changes since v7: - Don't check for mixed stale/valid VCPI allocations, just rely on connector registration to stop such erroneous modesets Changes since v6: - Keep a kref to all of the ports we have allocations on. This required a good bit of changing to when we call drm_dp_find_vcpi_slots(), mainly that we need to ensure that we only redo VCPI allocations on actual mode or CRTC changes, not crtc_state->active changes. Additionally, we no longer take the registration of the DRM connector for each port into account because so long as we have a kref to the port in the new or previous atomic state, the connector will stay registered. - Use the small changes to drm_dp_put_port() to add even more error checking to make misusage of the helpers more obvious. I added this after having to chase down various use-after-free conditions that started popping up from the new helpers so no one else has to troubleshoot that. - Move some accidental DRM_DEBUG_KMS() calls to DRM_DEBUG_ATOMIC() - Update documentation again, note that find/release() should both not be called on the same port in a single atomic check phase (but multiple calls to one or the other is OK) Changes since v4: - Don't skip the atomic checks for VCPI allocations if no new VCPI allocations happen in a state. This makes the next change I'm about to list here a lot easier to implement. - Don't ignore VCPI allocations on destroyed ports, instead ensure that when ports are destroyed and still have VCPI allocations in the topology state, the only state changes allowed are releasing said ports' VCPI. This prevents a state with a mix of VCPI allocations from destroyed ports, and allocations from valid ports. Changes since v3: - Don't release VCPI allocations in the topology state immediately in drm_dp_atomic_release_vcpi_slots(), instead mark them as 0 and skip over them in drm_dp_mst_duplicate_state(). This makes it so drm_dp_atomic_release_vcpi_slots() is still idempotent while also throwing warnings if the driver messes up it's book keeping and tries to release VCPI slots on a port that doesn't have any pre-existing VCPI allocation - danvet - Change mst_state/state in some debugging messages to "mst state" Changes since v2: - Use kmemdup() for duplicating MST state - danvet - Move port validation out of duplicate state callback - danvet - Handle looping through MST topology states in drm_dp_mst_atomic_check() so the driver doesn't have to do it - Fix documentation in drm_dp_atomic_find_vcpi_slots() - Move the atomic check for each individual topology state into it's own function, reduces indenting - Don't consider "stale" MST ports when calculating the bandwidth requirements. This is needed because originally we relied on the state duplication functions to prune any stale ports from the new state, which would prevent us from incorrectly considering their bandwidth requirements alongside legitimate new payloads. - Add function references in drm_dp_atomic_release_vcpi_slots() - danvet - Annotate atomic VCPI and atomic check functions with __must_check - danvet Changes since v1: - Don't use the now-removed ->atomic_check() for private objects hook, just give drivers a function to call themselves Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Harry Wentland <harry.wentland@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-19-lyude@redhat.com
2019-01-11 07:53:41 +07:00
struct drm_dp_mst_port *port)
drm/dp: Add DP MST helpers to atomically find and release vcpi slots drm_dp_atomic_find_vcpi_slots() should be called from ->atomic_check() to check there are sufficient vcpi slots for a mode and to add that to the state. This should be followed by a call to drm_dp_mst_allocate_vcpi() in ->atomic_commit() to initialize a struct vcpi for the port. drm_dp_atomic_release_vcpi_slots() should be called from ->atomic_check() to release a port's vcpi slot allocation from the state. Drivers that do not make use of this atomic helper are expected to call drm_dp_find_vcpi_slots() instead before calling drm_dp_mst_allocate_vcpi(). v3: drm_dp_atomic_release_vcpi_slots() now needs to know how many slots to release as we may not have a valid reference to port. v2: Added checks for verifying the port reference is valid Moved get_mst_topology_state() into the helpers (Daniel) Changed find_vcpi_slots() to not depend on current allocation Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Cc: Archit Taneja <architt@codeaurora.org> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Harry Wentland <Harry.wentland@amd.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Signed-off-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com> Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1492753893-3748-4-git-send-email-dhinakaran.pandiyan@intel.com
2017-04-21 12:51:32 +07:00
{
struct drm_dp_mst_topology_state *topology_state;
drm/dp_mst: Start tracking per-port VCPI allocations There has been a TODO waiting for quite a long time in drm_dp_mst_topology.c: /* We cannot rely on port->vcpi.num_slots to update * topology_state->avail_slots as the port may not exist if the parent * branch device was unplugged. This should be fixed by tracking * per-port slot allocation in drm_dp_mst_topology_state instead of * depending on the caller to tell us how many slots to release. */ That's not the only reason we should fix this: forcing the driver to track the VCPI allocations throughout a state's atomic check is error prone, because it means that extra care has to be taken with the order that drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() are called in in order to ensure idempotency. Currently the only driver actually using these helpers, i915, doesn't even do this correctly: multiple ->best_encoder() checks with i915's current implementation would not be idempotent and would over-allocate VCPI slots, something I learned trying to implement fallback retraining in MST. So: simplify this whole mess, and teach drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() to track the VCPI allocations for each port. This allows us to ensure idempotency without having to rely on the driver as much. Additionally: the driver doesn't need to do any kind of VCPI slot tracking anymore if it doesn't need it for it's own internal state. Additionally; this adds a new drm_dp_mst_atomic_check() helper which must be used by atomic drivers to perform validity checks for the new VCPI allocations incurred by a state. Also: update the documentation and make it more obvious that these /must/ be called by /all/ atomic drivers supporting MST. Changes since v9: * Add some missing changes that were requested by danvet that I forgot about after I redid all of the kref stuff: * Remove unnecessary state changes in intel_dp_mst_atomic_check * Cleanup atomic check logic for VCPI allocations - all we need to check in compute_config is whether or not this state disables a CRTC, then free VCPI based off that Changes since v8: * Fix compile errors, whoops! Changes since v7: - Don't check for mixed stale/valid VCPI allocations, just rely on connector registration to stop such erroneous modesets Changes since v6: - Keep a kref to all of the ports we have allocations on. This required a good bit of changing to when we call drm_dp_find_vcpi_slots(), mainly that we need to ensure that we only redo VCPI allocations on actual mode or CRTC changes, not crtc_state->active changes. Additionally, we no longer take the registration of the DRM connector for each port into account because so long as we have a kref to the port in the new or previous atomic state, the connector will stay registered. - Use the small changes to drm_dp_put_port() to add even more error checking to make misusage of the helpers more obvious. I added this after having to chase down various use-after-free conditions that started popping up from the new helpers so no one else has to troubleshoot that. - Move some accidental DRM_DEBUG_KMS() calls to DRM_DEBUG_ATOMIC() - Update documentation again, note that find/release() should both not be called on the same port in a single atomic check phase (but multiple calls to one or the other is OK) Changes since v4: - Don't skip the atomic checks for VCPI allocations if no new VCPI allocations happen in a state. This makes the next change I'm about to list here a lot easier to implement. - Don't ignore VCPI allocations on destroyed ports, instead ensure that when ports are destroyed and still have VCPI allocations in the topology state, the only state changes allowed are releasing said ports' VCPI. This prevents a state with a mix of VCPI allocations from destroyed ports, and allocations from valid ports. Changes since v3: - Don't release VCPI allocations in the topology state immediately in drm_dp_atomic_release_vcpi_slots(), instead mark them as 0 and skip over them in drm_dp_mst_duplicate_state(). This makes it so drm_dp_atomic_release_vcpi_slots() is still idempotent while also throwing warnings if the driver messes up it's book keeping and tries to release VCPI slots on a port that doesn't have any pre-existing VCPI allocation - danvet - Change mst_state/state in some debugging messages to "mst state" Changes since v2: - Use kmemdup() for duplicating MST state - danvet - Move port validation out of duplicate state callback - danvet - Handle looping through MST topology states in drm_dp_mst_atomic_check() so the driver doesn't have to do it - Fix documentation in drm_dp_atomic_find_vcpi_slots() - Move the atomic check for each individual topology state into it's own function, reduces indenting - Don't consider "stale" MST ports when calculating the bandwidth requirements. This is needed because originally we relied on the state duplication functions to prune any stale ports from the new state, which would prevent us from incorrectly considering their bandwidth requirements alongside legitimate new payloads. - Add function references in drm_dp_atomic_release_vcpi_slots() - danvet - Annotate atomic VCPI and atomic check functions with __must_check - danvet Changes since v1: - Don't use the now-removed ->atomic_check() for private objects hook, just give drivers a function to call themselves Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Harry Wentland <harry.wentland@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-19-lyude@redhat.com
2019-01-11 07:53:41 +07:00
struct drm_dp_vcpi_allocation *pos;
bool found = false;
drm/dp: Add DP MST helpers to atomically find and release vcpi slots drm_dp_atomic_find_vcpi_slots() should be called from ->atomic_check() to check there are sufficient vcpi slots for a mode and to add that to the state. This should be followed by a call to drm_dp_mst_allocate_vcpi() in ->atomic_commit() to initialize a struct vcpi for the port. drm_dp_atomic_release_vcpi_slots() should be called from ->atomic_check() to release a port's vcpi slot allocation from the state. Drivers that do not make use of this atomic helper are expected to call drm_dp_find_vcpi_slots() instead before calling drm_dp_mst_allocate_vcpi(). v3: drm_dp_atomic_release_vcpi_slots() now needs to know how many slots to release as we may not have a valid reference to port. v2: Added checks for verifying the port reference is valid Moved get_mst_topology_state() into the helpers (Daniel) Changed find_vcpi_slots() to not depend on current allocation Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Cc: Archit Taneja <architt@codeaurora.org> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Harry Wentland <Harry.wentland@amd.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Signed-off-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com> Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1492753893-3748-4-git-send-email-dhinakaran.pandiyan@intel.com
2017-04-21 12:51:32 +07:00
topology_state = drm_atomic_get_mst_topology_state(state, mgr);
if (IS_ERR(topology_state))
return PTR_ERR(topology_state);
drm/dp: Add DP MST helpers to atomically find and release vcpi slots drm_dp_atomic_find_vcpi_slots() should be called from ->atomic_check() to check there are sufficient vcpi slots for a mode and to add that to the state. This should be followed by a call to drm_dp_mst_allocate_vcpi() in ->atomic_commit() to initialize a struct vcpi for the port. drm_dp_atomic_release_vcpi_slots() should be called from ->atomic_check() to release a port's vcpi slot allocation from the state. Drivers that do not make use of this atomic helper are expected to call drm_dp_find_vcpi_slots() instead before calling drm_dp_mst_allocate_vcpi(). v3: drm_dp_atomic_release_vcpi_slots() now needs to know how many slots to release as we may not have a valid reference to port. v2: Added checks for verifying the port reference is valid Moved get_mst_topology_state() into the helpers (Daniel) Changed find_vcpi_slots() to not depend on current allocation Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Cc: Archit Taneja <architt@codeaurora.org> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Harry Wentland <Harry.wentland@amd.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Signed-off-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com> Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1492753893-3748-4-git-send-email-dhinakaran.pandiyan@intel.com
2017-04-21 12:51:32 +07:00
drm/dp_mst: Start tracking per-port VCPI allocations There has been a TODO waiting for quite a long time in drm_dp_mst_topology.c: /* We cannot rely on port->vcpi.num_slots to update * topology_state->avail_slots as the port may not exist if the parent * branch device was unplugged. This should be fixed by tracking * per-port slot allocation in drm_dp_mst_topology_state instead of * depending on the caller to tell us how many slots to release. */ That's not the only reason we should fix this: forcing the driver to track the VCPI allocations throughout a state's atomic check is error prone, because it means that extra care has to be taken with the order that drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() are called in in order to ensure idempotency. Currently the only driver actually using these helpers, i915, doesn't even do this correctly: multiple ->best_encoder() checks with i915's current implementation would not be idempotent and would over-allocate VCPI slots, something I learned trying to implement fallback retraining in MST. So: simplify this whole mess, and teach drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() to track the VCPI allocations for each port. This allows us to ensure idempotency without having to rely on the driver as much. Additionally: the driver doesn't need to do any kind of VCPI slot tracking anymore if it doesn't need it for it's own internal state. Additionally; this adds a new drm_dp_mst_atomic_check() helper which must be used by atomic drivers to perform validity checks for the new VCPI allocations incurred by a state. Also: update the documentation and make it more obvious that these /must/ be called by /all/ atomic drivers supporting MST. Changes since v9: * Add some missing changes that were requested by danvet that I forgot about after I redid all of the kref stuff: * Remove unnecessary state changes in intel_dp_mst_atomic_check * Cleanup atomic check logic for VCPI allocations - all we need to check in compute_config is whether or not this state disables a CRTC, then free VCPI based off that Changes since v8: * Fix compile errors, whoops! Changes since v7: - Don't check for mixed stale/valid VCPI allocations, just rely on connector registration to stop such erroneous modesets Changes since v6: - Keep a kref to all of the ports we have allocations on. This required a good bit of changing to when we call drm_dp_find_vcpi_slots(), mainly that we need to ensure that we only redo VCPI allocations on actual mode or CRTC changes, not crtc_state->active changes. Additionally, we no longer take the registration of the DRM connector for each port into account because so long as we have a kref to the port in the new or previous atomic state, the connector will stay registered. - Use the small changes to drm_dp_put_port() to add even more error checking to make misusage of the helpers more obvious. I added this after having to chase down various use-after-free conditions that started popping up from the new helpers so no one else has to troubleshoot that. - Move some accidental DRM_DEBUG_KMS() calls to DRM_DEBUG_ATOMIC() - Update documentation again, note that find/release() should both not be called on the same port in a single atomic check phase (but multiple calls to one or the other is OK) Changes since v4: - Don't skip the atomic checks for VCPI allocations if no new VCPI allocations happen in a state. This makes the next change I'm about to list here a lot easier to implement. - Don't ignore VCPI allocations on destroyed ports, instead ensure that when ports are destroyed and still have VCPI allocations in the topology state, the only state changes allowed are releasing said ports' VCPI. This prevents a state with a mix of VCPI allocations from destroyed ports, and allocations from valid ports. Changes since v3: - Don't release VCPI allocations in the topology state immediately in drm_dp_atomic_release_vcpi_slots(), instead mark them as 0 and skip over them in drm_dp_mst_duplicate_state(). This makes it so drm_dp_atomic_release_vcpi_slots() is still idempotent while also throwing warnings if the driver messes up it's book keeping and tries to release VCPI slots on a port that doesn't have any pre-existing VCPI allocation - danvet - Change mst_state/state in some debugging messages to "mst state" Changes since v2: - Use kmemdup() for duplicating MST state - danvet - Move port validation out of duplicate state callback - danvet - Handle looping through MST topology states in drm_dp_mst_atomic_check() so the driver doesn't have to do it - Fix documentation in drm_dp_atomic_find_vcpi_slots() - Move the atomic check for each individual topology state into it's own function, reduces indenting - Don't consider "stale" MST ports when calculating the bandwidth requirements. This is needed because originally we relied on the state duplication functions to prune any stale ports from the new state, which would prevent us from incorrectly considering their bandwidth requirements alongside legitimate new payloads. - Add function references in drm_dp_atomic_release_vcpi_slots() - danvet - Annotate atomic VCPI and atomic check functions with __must_check - danvet Changes since v1: - Don't use the now-removed ->atomic_check() for private objects hook, just give drivers a function to call themselves Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Harry Wentland <harry.wentland@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-19-lyude@redhat.com
2019-01-11 07:53:41 +07:00
list_for_each_entry(pos, &topology_state->vcpis, next) {
if (pos->port == port) {
found = true;
break;
}
}
if (WARN_ON(!found)) {
DRM_ERROR("no VCPI for [MST PORT:%p] found in mst state %p\n",
port, &topology_state->base);
return -EINVAL;
}
DRM_DEBUG_ATOMIC("[MST PORT:%p] VCPI %d -> 0\n", port, pos->vcpi);
if (pos->vcpi) {
drm_dp_mst_put_port_malloc(port);
pos->vcpi = 0;
}
drm/dp: Add DP MST helpers to atomically find and release vcpi slots drm_dp_atomic_find_vcpi_slots() should be called from ->atomic_check() to check there are sufficient vcpi slots for a mode and to add that to the state. This should be followed by a call to drm_dp_mst_allocate_vcpi() in ->atomic_commit() to initialize a struct vcpi for the port. drm_dp_atomic_release_vcpi_slots() should be called from ->atomic_check() to release a port's vcpi slot allocation from the state. Drivers that do not make use of this atomic helper are expected to call drm_dp_find_vcpi_slots() instead before calling drm_dp_mst_allocate_vcpi(). v3: drm_dp_atomic_release_vcpi_slots() now needs to know how many slots to release as we may not have a valid reference to port. v2: Added checks for verifying the port reference is valid Moved get_mst_topology_state() into the helpers (Daniel) Changed find_vcpi_slots() to not depend on current allocation Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Cc: Archit Taneja <architt@codeaurora.org> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Harry Wentland <Harry.wentland@amd.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Signed-off-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com> Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1492753893-3748-4-git-send-email-dhinakaran.pandiyan@intel.com
2017-04-21 12:51:32 +07:00
return 0;
}
EXPORT_SYMBOL(drm_dp_atomic_release_vcpi_slots);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
/**
* drm_dp_mst_allocate_vcpi() - Allocate a virtual channel
* @mgr: manager for this port
* @port: port to allocate a virtual channel for.
* @pbn: payload bandwidth number to request
* @slots: returned number of slots for this PBN.
*/
bool drm_dp_mst_allocate_vcpi(struct drm_dp_mst_topology_mgr *mgr,
struct drm_dp_mst_port *port, int pbn, int slots)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
{
int ret;
port = drm_dp_mst_topology_get_port_validated(mgr, port);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
if (!port)
return false;
if (slots < 0)
return false;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
if (port->vcpi.vcpi > 0) {
DRM_DEBUG_KMS("payload: vcpi %d already allocated for pbn %d - requested pbn %d\n",
port->vcpi.vcpi, port->vcpi.pbn, pbn);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
if (pbn == port->vcpi.pbn) {
drm_dp_mst_topology_put_port(port);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
return true;
}
}
ret = drm_dp_init_vcpi(mgr, &port->vcpi, pbn, slots);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
if (ret) {
DRM_DEBUG_KMS("failed to init vcpi slots=%d max=63 ret=%d\n",
DIV_ROUND_UP(pbn, mgr->pbn_div), ret);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
goto out;
}
DRM_DEBUG_KMS("initing vcpi for pbn=%d slots=%d\n",
pbn, port->vcpi.num_slots);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
/* Keep port allocated until its payload has been removed */
drm_dp_mst_get_port_malloc(port);
drm_dp_mst_topology_put_port(port);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
return true;
out:
return false;
}
EXPORT_SYMBOL(drm_dp_mst_allocate_vcpi);
int drm_dp_mst_get_vcpi_slots(struct drm_dp_mst_topology_mgr *mgr, struct drm_dp_mst_port *port)
{
int slots = 0;
port = drm_dp_mst_topology_get_port_validated(mgr, port);
if (!port)
return slots;
slots = port->vcpi.num_slots;
drm_dp_mst_topology_put_port(port);
return slots;
}
EXPORT_SYMBOL(drm_dp_mst_get_vcpi_slots);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
/**
* drm_dp_mst_reset_vcpi_slots() - Reset number of slots to 0 for VCPI
* @mgr: manager for this port
* @port: unverified pointer to a port.
*
* This just resets the number of slots for the ports VCPI for later programming.
*/
void drm_dp_mst_reset_vcpi_slots(struct drm_dp_mst_topology_mgr *mgr, struct drm_dp_mst_port *port)
{
/*
* A port with VCPI will remain allocated until its VCPI is
* released, no verified ref needed
*/
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
port->vcpi.num_slots = 0;
}
EXPORT_SYMBOL(drm_dp_mst_reset_vcpi_slots);
/**
* drm_dp_mst_deallocate_vcpi() - deallocate a VCPI
* @mgr: manager for this port
drm/dp_mst: Fix unbalanced malloc ref in drm_dp_mst_deallocate_vcpi() In drm_dp_mst_deallocate_vcpi(), we currently unconditionally call drm_dp_mst_put_port_malloc() on the port that's passed to us, even if we never successfully allocated VCPI to it. This is contrary to what we do in drm_dp_mst_allocate_vcpi(), where we only call drm_dp_mst_get_port_malloc() on the passed port if we successfully allocated VCPI to it. As a result, if drm_dp_mst_allocate_vcpi() fails during a modeset and another successive modeset calls drm_dp_mst_deallocate_vcpi() we will end up dropping someone else's malloc reference to the port. Example: [ 962.309260] ================================================================== [ 962.309290] BUG: KASAN: use-after-free in drm_dp_mst_put_port_malloc+0x72/0x180 [drm_kms_helper] [ 962.309296] Read of size 4 at addr ffff888416c30004 by task kworker/0:1H/500 [ 962.309308] CPU: 0 PID: 500 Comm: kworker/0:1H Tainted: G W O 5.0.0-rc2Lyude-Test+ #1 [ 962.309313] Hardware name: LENOVO 20L8S2N800/20L8S2N800, BIOS N22ET35W (1.12 ) 04/09/2018 [ 962.309428] Workqueue: events_highpri intel_atomic_cleanup_work [i915] [ 962.309434] Call Trace: [ 962.309452] dump_stack+0xad/0x150 [ 962.309462] ? dump_stack_print_info.cold.0+0x1b/0x1b [ 962.309472] ? kmsg_dump_rewind_nolock+0xd9/0xd9 [ 962.309504] ? drm_dp_mst_put_port_malloc+0x72/0x180 [drm_kms_helper] [ 962.309515] print_address_description+0x6c/0x23c [ 962.309542] ? drm_dp_mst_put_port_malloc+0x72/0x180 [drm_kms_helper] [ 962.309568] ? drm_dp_mst_put_port_malloc+0x72/0x180 [drm_kms_helper] [ 962.309577] kasan_report.cold.3+0x1a/0x32 [ 962.309605] ? drm_dp_mst_put_port_malloc+0x72/0x180 [drm_kms_helper] [ 962.309631] drm_dp_mst_put_port_malloc+0x72/0x180 [drm_kms_helper] [ 962.309658] ? drm_dp_mst_put_mstb_malloc+0x180/0x180 [drm_kms_helper] [ 962.309687] drm_dp_mst_destroy_state+0xcd/0x120 [drm_kms_helper] [ 962.309745] drm_atomic_state_default_clear+0x6ee/0xcc0 [drm] [ 962.309864] intel_atomic_state_clear+0xe/0x80 [i915] [ 962.309928] __drm_atomic_state_free+0x35/0xd0 [drm] [ 962.310044] intel_atomic_cleanup_work+0x56/0x70 [i915] [ 962.310057] process_one_work+0x884/0x1400 [ 962.310067] ? drain_workqueue+0x5a0/0x5a0 [ 962.310075] ? __schedule+0x87f/0x1e80 [ 962.310086] ? __sched_text_start+0x8/0x8 [ 962.310095] ? run_rebalance_domains+0x400/0x400 [ 962.310110] ? deref_stack_reg+0xb4/0x120 [ 962.310117] ? __read_once_size_nocheck.constprop.7+0x10/0x10 [ 962.310124] ? worker_enter_idle+0x47f/0x6a0 [ 962.310134] ? schedule+0xd7/0x2e0 [ 962.310141] ? __schedule+0x1e80/0x1e80 [ 962.310148] ? _raw_spin_lock_irq+0x9f/0x130 [ 962.310155] ? _raw_write_unlock_irqrestore+0x110/0x110 [ 962.310164] worker_thread+0x196/0x11e0 [ 962.310175] ? set_load_weight+0x2e0/0x2e0 [ 962.310181] ? __switch_to_asm+0x34/0x70 [ 962.310187] ? __switch_to_asm+0x40/0x70 [ 962.310194] ? process_one_work+0x1400/0x1400 [ 962.310199] ? __switch_to_asm+0x40/0x70 [ 962.310205] ? __switch_to_asm+0x34/0x70 [ 962.310211] ? __switch_to_asm+0x34/0x70 [ 962.310216] ? __switch_to_asm+0x40/0x70 [ 962.310221] ? __switch_to_asm+0x34/0x70 [ 962.310226] ? __switch_to_asm+0x40/0x70 [ 962.310231] ? __switch_to_asm+0x34/0x70 [ 962.310236] ? __switch_to_asm+0x40/0x70 [ 962.310242] ? syscall_return_via_sysret+0xf/0x7f [ 962.310248] ? __switch_to_asm+0x34/0x70 [ 962.310253] ? __switch_to_asm+0x40/0x70 [ 962.310258] ? __switch_to_asm+0x34/0x70 [ 962.310263] ? __switch_to_asm+0x40/0x70 [ 962.310268] ? __switch_to_asm+0x34/0x70 [ 962.310273] ? __switch_to_asm+0x40/0x70 [ 962.310281] ? __schedule+0x87f/0x1e80 [ 962.310292] ? __sched_text_start+0x8/0x8 [ 962.310300] ? save_stack+0x8c/0xb0 [ 962.310308] ? __kasan_kmalloc.constprop.6+0xc6/0xd0 [ 962.310313] ? kthread+0x98/0x3a0 [ 962.310318] ? ret_from_fork+0x35/0x40 [ 962.310334] ? __wake_up_common+0x178/0x6f0 [ 962.310343] ? _raw_spin_lock_irqsave+0xa4/0x140 [ 962.310349] ? __lock_text_start+0x8/0x8 [ 962.310355] ? _raw_write_lock_irqsave+0x70/0x130 [ 962.310360] ? __lock_text_start+0x8/0x8 [ 962.310371] ? process_one_work+0x1400/0x1400 [ 962.310376] kthread+0x2e2/0x3a0 [ 962.310383] ? kthread_create_on_node+0xc0/0xc0 [ 962.310389] ret_from_fork+0x35/0x40 [ 962.310401] Allocated by task 1462: [ 962.310410] __kasan_kmalloc.constprop.6+0xc6/0xd0 [ 962.310437] drm_dp_add_port+0xd60/0x1960 [drm_kms_helper] [ 962.310464] drm_dp_send_link_address+0x4b0/0x770 [drm_kms_helper] [ 962.310491] drm_dp_check_and_send_link_address+0x197/0x1f0 [drm_kms_helper] [ 962.310515] drm_dp_mst_link_probe_work+0x2b6/0x330 [drm_kms_helper] [ 962.310522] process_one_work+0x884/0x1400 [ 962.310529] worker_thread+0x196/0x11e0 [ 962.310533] kthread+0x2e2/0x3a0 [ 962.310538] ret_from_fork+0x35/0x40 [ 962.310543] Freed by task 500: [ 962.310550] __kasan_slab_free+0x133/0x180 [ 962.310555] kfree+0x92/0x1a0 [ 962.310581] drm_dp_mst_put_port_malloc+0x14d/0x180 [drm_kms_helper] [ 962.310693] intel_connector_destroy+0xb2/0xe0 [i915] [ 962.310747] drm_mode_object_put.part.0+0x12b/0x1a0 [drm] [ 962.310802] drm_atomic_state_default_clear+0x1f2/0xcc0 [drm] [ 962.310916] intel_atomic_state_clear+0xe/0x80 [i915] [ 962.310972] __drm_atomic_state_free+0x35/0xd0 [drm] [ 962.311083] intel_atomic_cleanup_work+0x56/0x70 [i915] [ 962.311092] process_one_work+0x884/0x1400 [ 962.311098] worker_thread+0x196/0x11e0 [ 962.311103] kthread+0x2e2/0x3a0 [ 962.311108] ret_from_fork+0x35/0x40 [ 962.311116] The buggy address belongs to the object at ffff888416c30000 which belongs to the cache kmalloc-2k of size 2048 [ 962.311122] The buggy address is located 4 bytes inside of 2048-byte region [ffff888416c30000, ffff888416c30800) [ 962.311124] The buggy address belongs to the page: [ 962.311132] page:ffffea00105b0c00 count:1 mapcount:0 mapping:ffff88841d003040 index:0x0 compound_mapcount: 0 [ 962.311142] flags: 0x8000000000010200(slab|head) [ 962.311152] raw: 8000000000010200 dead000000000100 dead000000000200 ffff88841d003040 [ 962.311159] raw: 0000000000000000 00000000000f000f 00000001ffffffff 0000000000000000 [ 962.311162] page dumped because: kasan: bad access detected So, bail early if drm_dp_mst_deallocate_vcpi() is called on a port with no VCPI allocation. Additionally, clean up the surrounding kerneldoc while we're at it since the port is assumed to be kept around because the DRM driver is expected to hold a malloc reference to it, not just us. Changes since v1: * Doc changes - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Fixes: eceae1472467 ("drm/dp_mst: Start tracking per-port VCPI allocations") Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/msgid/20190202002023.29665-2-lyude@redhat.com
2019-02-02 07:20:01 +07:00
* @port: port to deallocate vcpi for
*
* This can be called unconditionally, regardless of whether
* drm_dp_mst_allocate_vcpi() succeeded or not.
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
*/
void drm_dp_mst_deallocate_vcpi(struct drm_dp_mst_topology_mgr *mgr,
struct drm_dp_mst_port *port)
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
{
drm/dp_mst: Fix unbalanced malloc ref in drm_dp_mst_deallocate_vcpi() In drm_dp_mst_deallocate_vcpi(), we currently unconditionally call drm_dp_mst_put_port_malloc() on the port that's passed to us, even if we never successfully allocated VCPI to it. This is contrary to what we do in drm_dp_mst_allocate_vcpi(), where we only call drm_dp_mst_get_port_malloc() on the passed port if we successfully allocated VCPI to it. As a result, if drm_dp_mst_allocate_vcpi() fails during a modeset and another successive modeset calls drm_dp_mst_deallocate_vcpi() we will end up dropping someone else's malloc reference to the port. Example: [ 962.309260] ================================================================== [ 962.309290] BUG: KASAN: use-after-free in drm_dp_mst_put_port_malloc+0x72/0x180 [drm_kms_helper] [ 962.309296] Read of size 4 at addr ffff888416c30004 by task kworker/0:1H/500 [ 962.309308] CPU: 0 PID: 500 Comm: kworker/0:1H Tainted: G W O 5.0.0-rc2Lyude-Test+ #1 [ 962.309313] Hardware name: LENOVO 20L8S2N800/20L8S2N800, BIOS N22ET35W (1.12 ) 04/09/2018 [ 962.309428] Workqueue: events_highpri intel_atomic_cleanup_work [i915] [ 962.309434] Call Trace: [ 962.309452] dump_stack+0xad/0x150 [ 962.309462] ? dump_stack_print_info.cold.0+0x1b/0x1b [ 962.309472] ? kmsg_dump_rewind_nolock+0xd9/0xd9 [ 962.309504] ? drm_dp_mst_put_port_malloc+0x72/0x180 [drm_kms_helper] [ 962.309515] print_address_description+0x6c/0x23c [ 962.309542] ? drm_dp_mst_put_port_malloc+0x72/0x180 [drm_kms_helper] [ 962.309568] ? drm_dp_mst_put_port_malloc+0x72/0x180 [drm_kms_helper] [ 962.309577] kasan_report.cold.3+0x1a/0x32 [ 962.309605] ? drm_dp_mst_put_port_malloc+0x72/0x180 [drm_kms_helper] [ 962.309631] drm_dp_mst_put_port_malloc+0x72/0x180 [drm_kms_helper] [ 962.309658] ? drm_dp_mst_put_mstb_malloc+0x180/0x180 [drm_kms_helper] [ 962.309687] drm_dp_mst_destroy_state+0xcd/0x120 [drm_kms_helper] [ 962.309745] drm_atomic_state_default_clear+0x6ee/0xcc0 [drm] [ 962.309864] intel_atomic_state_clear+0xe/0x80 [i915] [ 962.309928] __drm_atomic_state_free+0x35/0xd0 [drm] [ 962.310044] intel_atomic_cleanup_work+0x56/0x70 [i915] [ 962.310057] process_one_work+0x884/0x1400 [ 962.310067] ? drain_workqueue+0x5a0/0x5a0 [ 962.310075] ? __schedule+0x87f/0x1e80 [ 962.310086] ? __sched_text_start+0x8/0x8 [ 962.310095] ? run_rebalance_domains+0x400/0x400 [ 962.310110] ? deref_stack_reg+0xb4/0x120 [ 962.310117] ? __read_once_size_nocheck.constprop.7+0x10/0x10 [ 962.310124] ? worker_enter_idle+0x47f/0x6a0 [ 962.310134] ? schedule+0xd7/0x2e0 [ 962.310141] ? __schedule+0x1e80/0x1e80 [ 962.310148] ? _raw_spin_lock_irq+0x9f/0x130 [ 962.310155] ? _raw_write_unlock_irqrestore+0x110/0x110 [ 962.310164] worker_thread+0x196/0x11e0 [ 962.310175] ? set_load_weight+0x2e0/0x2e0 [ 962.310181] ? __switch_to_asm+0x34/0x70 [ 962.310187] ? __switch_to_asm+0x40/0x70 [ 962.310194] ? process_one_work+0x1400/0x1400 [ 962.310199] ? __switch_to_asm+0x40/0x70 [ 962.310205] ? __switch_to_asm+0x34/0x70 [ 962.310211] ? __switch_to_asm+0x34/0x70 [ 962.310216] ? __switch_to_asm+0x40/0x70 [ 962.310221] ? __switch_to_asm+0x34/0x70 [ 962.310226] ? __switch_to_asm+0x40/0x70 [ 962.310231] ? __switch_to_asm+0x34/0x70 [ 962.310236] ? __switch_to_asm+0x40/0x70 [ 962.310242] ? syscall_return_via_sysret+0xf/0x7f [ 962.310248] ? __switch_to_asm+0x34/0x70 [ 962.310253] ? __switch_to_asm+0x40/0x70 [ 962.310258] ? __switch_to_asm+0x34/0x70 [ 962.310263] ? __switch_to_asm+0x40/0x70 [ 962.310268] ? __switch_to_asm+0x34/0x70 [ 962.310273] ? __switch_to_asm+0x40/0x70 [ 962.310281] ? __schedule+0x87f/0x1e80 [ 962.310292] ? __sched_text_start+0x8/0x8 [ 962.310300] ? save_stack+0x8c/0xb0 [ 962.310308] ? __kasan_kmalloc.constprop.6+0xc6/0xd0 [ 962.310313] ? kthread+0x98/0x3a0 [ 962.310318] ? ret_from_fork+0x35/0x40 [ 962.310334] ? __wake_up_common+0x178/0x6f0 [ 962.310343] ? _raw_spin_lock_irqsave+0xa4/0x140 [ 962.310349] ? __lock_text_start+0x8/0x8 [ 962.310355] ? _raw_write_lock_irqsave+0x70/0x130 [ 962.310360] ? __lock_text_start+0x8/0x8 [ 962.310371] ? process_one_work+0x1400/0x1400 [ 962.310376] kthread+0x2e2/0x3a0 [ 962.310383] ? kthread_create_on_node+0xc0/0xc0 [ 962.310389] ret_from_fork+0x35/0x40 [ 962.310401] Allocated by task 1462: [ 962.310410] __kasan_kmalloc.constprop.6+0xc6/0xd0 [ 962.310437] drm_dp_add_port+0xd60/0x1960 [drm_kms_helper] [ 962.310464] drm_dp_send_link_address+0x4b0/0x770 [drm_kms_helper] [ 962.310491] drm_dp_check_and_send_link_address+0x197/0x1f0 [drm_kms_helper] [ 962.310515] drm_dp_mst_link_probe_work+0x2b6/0x330 [drm_kms_helper] [ 962.310522] process_one_work+0x884/0x1400 [ 962.310529] worker_thread+0x196/0x11e0 [ 962.310533] kthread+0x2e2/0x3a0 [ 962.310538] ret_from_fork+0x35/0x40 [ 962.310543] Freed by task 500: [ 962.310550] __kasan_slab_free+0x133/0x180 [ 962.310555] kfree+0x92/0x1a0 [ 962.310581] drm_dp_mst_put_port_malloc+0x14d/0x180 [drm_kms_helper] [ 962.310693] intel_connector_destroy+0xb2/0xe0 [i915] [ 962.310747] drm_mode_object_put.part.0+0x12b/0x1a0 [drm] [ 962.310802] drm_atomic_state_default_clear+0x1f2/0xcc0 [drm] [ 962.310916] intel_atomic_state_clear+0xe/0x80 [i915] [ 962.310972] __drm_atomic_state_free+0x35/0xd0 [drm] [ 962.311083] intel_atomic_cleanup_work+0x56/0x70 [i915] [ 962.311092] process_one_work+0x884/0x1400 [ 962.311098] worker_thread+0x196/0x11e0 [ 962.311103] kthread+0x2e2/0x3a0 [ 962.311108] ret_from_fork+0x35/0x40 [ 962.311116] The buggy address belongs to the object at ffff888416c30000 which belongs to the cache kmalloc-2k of size 2048 [ 962.311122] The buggy address is located 4 bytes inside of 2048-byte region [ffff888416c30000, ffff888416c30800) [ 962.311124] The buggy address belongs to the page: [ 962.311132] page:ffffea00105b0c00 count:1 mapcount:0 mapping:ffff88841d003040 index:0x0 compound_mapcount: 0 [ 962.311142] flags: 0x8000000000010200(slab|head) [ 962.311152] raw: 8000000000010200 dead000000000100 dead000000000200 ffff88841d003040 [ 962.311159] raw: 0000000000000000 00000000000f000f 00000001ffffffff 0000000000000000 [ 962.311162] page dumped because: kasan: bad access detected So, bail early if drm_dp_mst_deallocate_vcpi() is called on a port with no VCPI allocation. Additionally, clean up the surrounding kerneldoc while we're at it since the port is assumed to be kept around because the DRM driver is expected to hold a malloc reference to it, not just us. Changes since v1: * Doc changes - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Fixes: eceae1472467 ("drm/dp_mst: Start tracking per-port VCPI allocations") Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/msgid/20190202002023.29665-2-lyude@redhat.com
2019-02-02 07:20:01 +07:00
if (!port->vcpi.vcpi)
return;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
drm_dp_mst_put_payload_id(mgr, port->vcpi.vcpi);
port->vcpi.num_slots = 0;
port->vcpi.pbn = 0;
port->vcpi.aligned_pbn = 0;
port->vcpi.vcpi = 0;
drm_dp_mst_put_port_malloc(port);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
EXPORT_SYMBOL(drm_dp_mst_deallocate_vcpi);
static int drm_dp_dpcd_write_payload(struct drm_dp_mst_topology_mgr *mgr,
int id, struct drm_dp_payload *payload)
{
u8 payload_alloc[3], status;
int ret;
int retries = 0;
drm_dp_dpcd_writeb(mgr->aux, DP_PAYLOAD_TABLE_UPDATE_STATUS,
DP_PAYLOAD_TABLE_UPDATED);
payload_alloc[0] = id;
payload_alloc[1] = payload->start_slot;
payload_alloc[2] = payload->num_slots;
ret = drm_dp_dpcd_write(mgr->aux, DP_PAYLOAD_ALLOCATE_SET, payload_alloc, 3);
if (ret != 3) {
DRM_DEBUG_KMS("failed to write payload allocation %d\n", ret);
goto fail;
}
retry:
ret = drm_dp_dpcd_readb(mgr->aux, DP_PAYLOAD_TABLE_UPDATE_STATUS, &status);
if (ret < 0) {
DRM_DEBUG_KMS("failed to read payload table status %d\n", ret);
goto fail;
}
if (!(status & DP_PAYLOAD_TABLE_UPDATED)) {
retries++;
if (retries < 20) {
usleep_range(10000, 20000);
goto retry;
}
DRM_DEBUG_KMS("status not set after read payload table status %d\n", status);
ret = -EINVAL;
goto fail;
}
ret = 0;
fail:
return ret;
}
/**
* drm_dp_check_act_status() - Check ACT handled status.
* @mgr: manager to use
*
* Check the payload status bits in the DPCD for ACT handled completion.
*/
int drm_dp_check_act_status(struct drm_dp_mst_topology_mgr *mgr)
{
u8 status;
int ret;
int count = 0;
do {
ret = drm_dp_dpcd_readb(mgr->aux, DP_PAYLOAD_TABLE_UPDATE_STATUS, &status);
if (ret < 0) {
DRM_DEBUG_KMS("failed to read payload table status %d\n", ret);
goto fail;
}
if (status & DP_PAYLOAD_ACT_HANDLED)
break;
count++;
udelay(100);
} while (count < 30);
if (!(status & DP_PAYLOAD_ACT_HANDLED)) {
DRM_DEBUG_KMS("failed to get ACT bit %d after %d retries\n", status, count);
ret = -EINVAL;
goto fail;
}
return 0;
fail:
return ret;
}
EXPORT_SYMBOL(drm_dp_check_act_status);
/**
* drm_dp_calc_pbn_mode() - Calculate the PBN for a mode.
* @clock: dot clock for the mode
* @bpp: bpp for the mode.
*
* This uses the formula in the spec to calculate the PBN value for a mode.
*/
int drm_dp_calc_pbn_mode(int clock, int bpp)
{
/*
* margin 5300ppm + 300ppm ~ 0.6% as per spec, factor is 1.006
* The unit of 54/64Mbytes/sec is an arbitrary unit chosen based on
* common multiplier to render an integer PBN for all link rate/lane
* counts combinations
* calculate
* peak_kbps *= (1006/1000)
* peak_kbps *= (64/54)
* peak_kbps *= 8 convert to bytes
*/
return DIV_ROUND_UP_ULL(mul_u32_u32(clock * bpp, 64 * 1006),
8 * 54 * 1000 * 1000);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
EXPORT_SYMBOL(drm_dp_calc_pbn_mode);
/* we want to kick the TX after we've ack the up/down IRQs. */
static void drm_dp_mst_kick_tx(struct drm_dp_mst_topology_mgr *mgr)
{
queue_work(system_long_wq, &mgr->tx_work);
}
static void drm_dp_mst_dump_mstb(struct seq_file *m,
struct drm_dp_mst_branch *mstb)
{
struct drm_dp_mst_port *port;
int tabs = mstb->lct;
char prefix[10];
int i;
for (i = 0; i < tabs; i++)
prefix[i] = '\t';
prefix[i] = '\0';
seq_printf(m, "%smst: %p, %d\n", prefix, mstb, mstb->num_ports);
list_for_each_entry(port, &mstb->ports, next) {
seq_printf(m, "%sport: %d: input: %d: pdt: %d, ddps: %d ldps: %d, sdp: %d/%d, %p, conn: %p\n", prefix, port->port_num, port->input, port->pdt, port->ddps, port->ldps, port->num_sdp_streams, port->num_sdp_stream_sinks, port, port->connector);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
if (port->mstb)
drm_dp_mst_dump_mstb(m, port->mstb);
}
}
#define DP_PAYLOAD_TABLE_SIZE 64
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
static bool dump_dp_payload_table(struct drm_dp_mst_topology_mgr *mgr,
char *buf)
{
int i;
for (i = 0; i < DP_PAYLOAD_TABLE_SIZE; i += 16) {
if (drm_dp_dpcd_read(mgr->aux,
DP_PAYLOAD_TABLE_UPDATE_STATUS + i,
&buf[i], 16) != 16)
return false;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
return true;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
static void fetch_monitor_name(struct drm_dp_mst_topology_mgr *mgr,
struct drm_dp_mst_port *port, char *name,
int namelen)
{
struct edid *mst_edid;
mst_edid = drm_dp_mst_get_edid(port->connector, mgr, port);
drm_edid_get_monitor_name(mst_edid, name, namelen);
}
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
/**
* drm_dp_mst_dump_topology(): dump topology to seq file.
* @m: seq_file to dump output to
* @mgr: manager to dump current topology for.
*
* helper to dump MST topology to a seq file for debugfs.
*/
void drm_dp_mst_dump_topology(struct seq_file *m,
struct drm_dp_mst_topology_mgr *mgr)
{
int i;
struct drm_dp_mst_port *port;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
mutex_lock(&mgr->lock);
if (mgr->mst_primary)
drm_dp_mst_dump_mstb(m, mgr->mst_primary);
/* dump VCPIs */
mutex_unlock(&mgr->lock);
mutex_lock(&mgr->payload_lock);
seq_printf(m, "vcpi: %lx %lx %d\n", mgr->payload_mask, mgr->vcpi_mask,
mgr->max_payloads);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
for (i = 0; i < mgr->max_payloads; i++) {
if (mgr->proposed_vcpis[i]) {
char name[14];
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
port = container_of(mgr->proposed_vcpis[i], struct drm_dp_mst_port, vcpi);
fetch_monitor_name(mgr, port, name, sizeof(name));
seq_printf(m, "vcpi %d: %d %d %d sink name: %s\n", i,
port->port_num, port->vcpi.vcpi,
port->vcpi.num_slots,
(*name != 0) ? name : "Unknown");
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
} else
seq_printf(m, "vcpi %d:unused\n", i);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
for (i = 0; i < mgr->max_payloads; i++) {
seq_printf(m, "payload %d: %d, %d, %d\n",
i,
mgr->payloads[i].payload_state,
mgr->payloads[i].start_slot,
mgr->payloads[i].num_slots);
}
mutex_unlock(&mgr->payload_lock);
mutex_lock(&mgr->lock);
if (mgr->mst_primary) {
u8 buf[DP_PAYLOAD_TABLE_SIZE];
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
int ret;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
ret = drm_dp_dpcd_read(mgr->aux, DP_DPCD_REV, buf, DP_RECEIVER_CAP_SIZE);
seq_printf(m, "dpcd: %*ph\n", DP_RECEIVER_CAP_SIZE, buf);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
ret = drm_dp_dpcd_read(mgr->aux, DP_FAUX_CAP, buf, 2);
seq_printf(m, "faux/mst: %*ph\n", 2, buf);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
ret = drm_dp_dpcd_read(mgr->aux, DP_MSTM_CTRL, buf, 1);
seq_printf(m, "mst ctrl: %*ph\n", 1, buf);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
/* dump the standard OUI branch header */
ret = drm_dp_dpcd_read(mgr->aux, DP_BRANCH_OUI, buf, DP_BRANCH_OUI_HEADER_SIZE);
seq_printf(m, "branch oui: %*phN devid: ", 3, buf);
for (i = 0x3; i < 0x8 && buf[i]; i++)
seq_printf(m, "%c", buf[i]);
seq_printf(m, " revision: hw: %x.%x sw: %x.%x\n",
buf[0x9] >> 4, buf[0x9] & 0xf, buf[0xa], buf[0xb]);
if (dump_dp_payload_table(mgr, buf))
seq_printf(m, "payload table: %*ph\n", DP_PAYLOAD_TABLE_SIZE, buf);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
mutex_unlock(&mgr->lock);
}
EXPORT_SYMBOL(drm_dp_mst_dump_topology);
static void drm_dp_tx_work(struct work_struct *work)
{
struct drm_dp_mst_topology_mgr *mgr = container_of(work, struct drm_dp_mst_topology_mgr, tx_work);
mutex_lock(&mgr->qlock);
if (!list_empty(&mgr->tx_msg_downq))
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
process_single_down_tx_qlock(mgr);
mutex_unlock(&mgr->qlock);
}
static void drm_dp_destroy_connector_work(struct work_struct *work)
{
struct drm_dp_mst_topology_mgr *mgr = container_of(work, struct drm_dp_mst_topology_mgr, destroy_connector_work);
struct drm_dp_mst_port *port;
bool send_hotplug = false;
/*
* Not a regular list traverse as we have to drop the destroy
* connector lock before destroying the connector, to avoid AB->BA
* ordering between this lock and the config mutex.
*/
for (;;) {
mutex_lock(&mgr->destroy_connector_lock);
port = list_first_entry_or_null(&mgr->destroy_connector_list, struct drm_dp_mst_port, next);
if (!port) {
mutex_unlock(&mgr->destroy_connector_lock);
break;
}
list_del(&port->next);
mutex_unlock(&mgr->destroy_connector_lock);
mgr->cbs->destroy_connector(mgr, port->connector);
drm_dp_port_teardown_pdt(port, port->pdt);
2016-10-26 20:30:33 +07:00
port->pdt = DP_PEER_DEVICE_NONE;
drm/dp_mst: Introduce new refcounting scheme for mstbs and ports The current way of handling refcounting in the DP MST helpers is really confusing and probably just plain wrong because it's been hacked up many times over the years without anyone actually going over the code and seeing if things could be simplified. To the best of my understanding, the current scheme works like this: drm_dp_mst_port and drm_dp_mst_branch both have a single refcount. When this refcount hits 0 for either of the two, they're removed from the topology state, but not immediately freed. Both ports and branch devices will reinitialize their kref once it's hit 0 before actually destroying themselves. The intended purpose behind this is so that we can avoid problems like not being able to free a remote payload that might still be active, due to us having removed all of the port/branch device structures in memory, as per: commit 91a25e463130 ("drm/dp/mst: deallocate payload on port destruction") Which may have worked, but then it caused use-after-free errors. Being new to MST at the time, I tried fixing it; commit 263efde31f97 ("drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1()") But, that was broken: both drm_dp_mst_port and drm_dp_mst_branch structs are validated in almost every DP MST helper function. Simply put, this means we go through the topology and try to see if the given drm_dp_mst_branch or drm_dp_mst_port is still attached to something before trying to use it in order to avoid dereferencing freed memory (something that has happened a LOT in the past with this library). Because of this it doesn't actually matter whether or not we keep keep the ports and branches around in memory as that's not enough, because any function that validates the branches and ports passed to it will still reject them anyway since they're no longer in the topology structure. So, use-after-free errors were fixed but payload deallocation was completely broken. Two years later, AMD informed me about this issue and I attempted to come up with a temporary fix, pending a long-overdue cleanup of this library: commit c54c7374ff44 ("drm/dp_mst: Skip validating ports during destruction, just ref") But then that introduced use-after-free errors, so I quickly reverted it: commit 9765635b3075 ("Revert "drm/dp_mst: Skip validating ports during destruction, just ref"") And in the process, learned that there is just no simple fix for this: the design is just broken. Unfortunately, the usage of these helpers are quite broken as well. Some drivers like i915 have been smart enough to avoid accessing any kind of information from MST port structures, but others like nouveau have assumed, understandably so, that drm_dp_mst_port structures are normal and can just be accessed at any time without worrying about use-after-free errors. After a lot of discussion, me and Daniel Vetter came up with a better idea to replace all of this. To summarize, since this is documented far more indepth in the documentation this patch introduces, we make it so that drm_dp_mst_port and drm_dp_mst_branch structures have two different classes of refcounts: topology_kref, and malloc_kref. topology_kref corresponds to the lifetime of the given drm_dp_mst_port or drm_dp_mst_branch in it's given topology. Once it hits zero, any associated connectors are removed and the branch or port can no longer be validated. malloc_kref corresponds to the lifetime of the memory allocation for the actual structure, and will always be non-zero so long as the topology_kref is non-zero. This gives us a way to allow callers to hold onto port and branch device structures past their topology lifetime, and dramatically simplifies the lifetimes of both structures. This also finally fixes the port deallocation problem, properly. Additionally: since this now means that we can keep ports and branch devices allocated in memory for however long we need, we no longer need a significant amount of the port validation that we currently do. Additionally, there is one last scenario that this fixes, which couldn't have been fixed properly beforehand: - CPU1 unrefs port from topology (refcount 1->0) - CPU2 refs port in topology(refcount 0->1) Since we now can guarantee memory safety for ports and branches as-needed, we also can make our main reference counting functions fix this problem by using kref_get_unless_zero() internally so that topology refcounts can only ever reach 0 once. Changes since v4: * Change the kernel-figure summary for dp-mst/topology-figure-1.dot a bit - danvet * Remove figure numbers - danvet Changes since v3: * Remove rebase detritus - danvet * Split out purely style changes into separate patches - hwentlan Changes since v2: * Fix commit message - checkpatch * s/)-1/) - 1/g - checkpatch Changes since v1: * Remove forward declarations - danvet * Move "Branch device and port refcounting" section from documentation into kernel-doc comments - danvet * Export internal topology lifetime functions into their own section in the kernel-docs - danvet * s/@/&/g for struct references in kernel-docs - danvet * Drop the "when they are no longer being used" bits from the kernel docs - danvet * Modify diagrams to show how the DRM driver interacts with the topology and payloads - danvet * Make suggested documentation changes for drm_dp_mst_topology_get_mstb() and drm_dp_mst_topology_get_port() - danvet * Better explain the relationship between malloc refs and topology krefs in the documentation for drm_dp_mst_topology_get_port() and drm_dp_mst_topology_get_mstb() - danvet * Fix "See also" in drm_dp_mst_topology_get_mstb() - danvet * Rename drm_dp_mst_topology_get_(port|mstb)() -> drm_dp_mst_topology_try_get_(port|mstb)() and drm_dp_mst_topology_ref_(port|mstb)() -> drm_dp_mst_topology_get_(port|mstb)() - danvet * s/should/must in docs - danvet * WARN_ON(refcount == 0) in topology_get_(mstb|port) - danvet * Move kdocs for mstb/port structs inline - danvet * Split drm_dp_get_last_connected_port_and_mstb() changes into their own commit - danvet Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-7-lyude@redhat.com
2019-01-11 07:53:29 +07:00
drm_dp_mst_put_port_malloc(port);
send_hotplug = true;
}
if (send_hotplug)
drm_kms_helper_hotplug_event(mgr->dev);
}
static struct drm_private_state *
drm_dp_mst_duplicate_state(struct drm_private_obj *obj)
{
drm/dp_mst: Start tracking per-port VCPI allocations There has been a TODO waiting for quite a long time in drm_dp_mst_topology.c: /* We cannot rely on port->vcpi.num_slots to update * topology_state->avail_slots as the port may not exist if the parent * branch device was unplugged. This should be fixed by tracking * per-port slot allocation in drm_dp_mst_topology_state instead of * depending on the caller to tell us how many slots to release. */ That's not the only reason we should fix this: forcing the driver to track the VCPI allocations throughout a state's atomic check is error prone, because it means that extra care has to be taken with the order that drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() are called in in order to ensure idempotency. Currently the only driver actually using these helpers, i915, doesn't even do this correctly: multiple ->best_encoder() checks with i915's current implementation would not be idempotent and would over-allocate VCPI slots, something I learned trying to implement fallback retraining in MST. So: simplify this whole mess, and teach drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() to track the VCPI allocations for each port. This allows us to ensure idempotency without having to rely on the driver as much. Additionally: the driver doesn't need to do any kind of VCPI slot tracking anymore if it doesn't need it for it's own internal state. Additionally; this adds a new drm_dp_mst_atomic_check() helper which must be used by atomic drivers to perform validity checks for the new VCPI allocations incurred by a state. Also: update the documentation and make it more obvious that these /must/ be called by /all/ atomic drivers supporting MST. Changes since v9: * Add some missing changes that were requested by danvet that I forgot about after I redid all of the kref stuff: * Remove unnecessary state changes in intel_dp_mst_atomic_check * Cleanup atomic check logic for VCPI allocations - all we need to check in compute_config is whether or not this state disables a CRTC, then free VCPI based off that Changes since v8: * Fix compile errors, whoops! Changes since v7: - Don't check for mixed stale/valid VCPI allocations, just rely on connector registration to stop such erroneous modesets Changes since v6: - Keep a kref to all of the ports we have allocations on. This required a good bit of changing to when we call drm_dp_find_vcpi_slots(), mainly that we need to ensure that we only redo VCPI allocations on actual mode or CRTC changes, not crtc_state->active changes. Additionally, we no longer take the registration of the DRM connector for each port into account because so long as we have a kref to the port in the new or previous atomic state, the connector will stay registered. - Use the small changes to drm_dp_put_port() to add even more error checking to make misusage of the helpers more obvious. I added this after having to chase down various use-after-free conditions that started popping up from the new helpers so no one else has to troubleshoot that. - Move some accidental DRM_DEBUG_KMS() calls to DRM_DEBUG_ATOMIC() - Update documentation again, note that find/release() should both not be called on the same port in a single atomic check phase (but multiple calls to one or the other is OK) Changes since v4: - Don't skip the atomic checks for VCPI allocations if no new VCPI allocations happen in a state. This makes the next change I'm about to list here a lot easier to implement. - Don't ignore VCPI allocations on destroyed ports, instead ensure that when ports are destroyed and still have VCPI allocations in the topology state, the only state changes allowed are releasing said ports' VCPI. This prevents a state with a mix of VCPI allocations from destroyed ports, and allocations from valid ports. Changes since v3: - Don't release VCPI allocations in the topology state immediately in drm_dp_atomic_release_vcpi_slots(), instead mark them as 0 and skip over them in drm_dp_mst_duplicate_state(). This makes it so drm_dp_atomic_release_vcpi_slots() is still idempotent while also throwing warnings if the driver messes up it's book keeping and tries to release VCPI slots on a port that doesn't have any pre-existing VCPI allocation - danvet - Change mst_state/state in some debugging messages to "mst state" Changes since v2: - Use kmemdup() for duplicating MST state - danvet - Move port validation out of duplicate state callback - danvet - Handle looping through MST topology states in drm_dp_mst_atomic_check() so the driver doesn't have to do it - Fix documentation in drm_dp_atomic_find_vcpi_slots() - Move the atomic check for each individual topology state into it's own function, reduces indenting - Don't consider "stale" MST ports when calculating the bandwidth requirements. This is needed because originally we relied on the state duplication functions to prune any stale ports from the new state, which would prevent us from incorrectly considering their bandwidth requirements alongside legitimate new payloads. - Add function references in drm_dp_atomic_release_vcpi_slots() - danvet - Annotate atomic VCPI and atomic check functions with __must_check - danvet Changes since v1: - Don't use the now-removed ->atomic_check() for private objects hook, just give drivers a function to call themselves Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Harry Wentland <harry.wentland@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-19-lyude@redhat.com
2019-01-11 07:53:41 +07:00
struct drm_dp_mst_topology_state *state, *old_state =
to_dp_mst_topology_state(obj->state);
struct drm_dp_vcpi_allocation *pos, *vcpi;
drm/dp_mst: Start tracking per-port VCPI allocations There has been a TODO waiting for quite a long time in drm_dp_mst_topology.c: /* We cannot rely on port->vcpi.num_slots to update * topology_state->avail_slots as the port may not exist if the parent * branch device was unplugged. This should be fixed by tracking * per-port slot allocation in drm_dp_mst_topology_state instead of * depending on the caller to tell us how many slots to release. */ That's not the only reason we should fix this: forcing the driver to track the VCPI allocations throughout a state's atomic check is error prone, because it means that extra care has to be taken with the order that drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() are called in in order to ensure idempotency. Currently the only driver actually using these helpers, i915, doesn't even do this correctly: multiple ->best_encoder() checks with i915's current implementation would not be idempotent and would over-allocate VCPI slots, something I learned trying to implement fallback retraining in MST. So: simplify this whole mess, and teach drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() to track the VCPI allocations for each port. This allows us to ensure idempotency without having to rely on the driver as much. Additionally: the driver doesn't need to do any kind of VCPI slot tracking anymore if it doesn't need it for it's own internal state. Additionally; this adds a new drm_dp_mst_atomic_check() helper which must be used by atomic drivers to perform validity checks for the new VCPI allocations incurred by a state. Also: update the documentation and make it more obvious that these /must/ be called by /all/ atomic drivers supporting MST. Changes since v9: * Add some missing changes that were requested by danvet that I forgot about after I redid all of the kref stuff: * Remove unnecessary state changes in intel_dp_mst_atomic_check * Cleanup atomic check logic for VCPI allocations - all we need to check in compute_config is whether or not this state disables a CRTC, then free VCPI based off that Changes since v8: * Fix compile errors, whoops! Changes since v7: - Don't check for mixed stale/valid VCPI allocations, just rely on connector registration to stop such erroneous modesets Changes since v6: - Keep a kref to all of the ports we have allocations on. This required a good bit of changing to when we call drm_dp_find_vcpi_slots(), mainly that we need to ensure that we only redo VCPI allocations on actual mode or CRTC changes, not crtc_state->active changes. Additionally, we no longer take the registration of the DRM connector for each port into account because so long as we have a kref to the port in the new or previous atomic state, the connector will stay registered. - Use the small changes to drm_dp_put_port() to add even more error checking to make misusage of the helpers more obvious. I added this after having to chase down various use-after-free conditions that started popping up from the new helpers so no one else has to troubleshoot that. - Move some accidental DRM_DEBUG_KMS() calls to DRM_DEBUG_ATOMIC() - Update documentation again, note that find/release() should both not be called on the same port in a single atomic check phase (but multiple calls to one or the other is OK) Changes since v4: - Don't skip the atomic checks for VCPI allocations if no new VCPI allocations happen in a state. This makes the next change I'm about to list here a lot easier to implement. - Don't ignore VCPI allocations on destroyed ports, instead ensure that when ports are destroyed and still have VCPI allocations in the topology state, the only state changes allowed are releasing said ports' VCPI. This prevents a state with a mix of VCPI allocations from destroyed ports, and allocations from valid ports. Changes since v3: - Don't release VCPI allocations in the topology state immediately in drm_dp_atomic_release_vcpi_slots(), instead mark them as 0 and skip over them in drm_dp_mst_duplicate_state(). This makes it so drm_dp_atomic_release_vcpi_slots() is still idempotent while also throwing warnings if the driver messes up it's book keeping and tries to release VCPI slots on a port that doesn't have any pre-existing VCPI allocation - danvet - Change mst_state/state in some debugging messages to "mst state" Changes since v2: - Use kmemdup() for duplicating MST state - danvet - Move port validation out of duplicate state callback - danvet - Handle looping through MST topology states in drm_dp_mst_atomic_check() so the driver doesn't have to do it - Fix documentation in drm_dp_atomic_find_vcpi_slots() - Move the atomic check for each individual topology state into it's own function, reduces indenting - Don't consider "stale" MST ports when calculating the bandwidth requirements. This is needed because originally we relied on the state duplication functions to prune any stale ports from the new state, which would prevent us from incorrectly considering their bandwidth requirements alongside legitimate new payloads. - Add function references in drm_dp_atomic_release_vcpi_slots() - danvet - Annotate atomic VCPI and atomic check functions with __must_check - danvet Changes since v1: - Don't use the now-removed ->atomic_check() for private objects hook, just give drivers a function to call themselves Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Harry Wentland <harry.wentland@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-19-lyude@redhat.com
2019-01-11 07:53:41 +07:00
state = kmemdup(old_state, sizeof(*state), GFP_KERNEL);
if (!state)
return NULL;
__drm_atomic_helper_private_obj_duplicate_state(obj, &state->base);
drm/dp_mst: Start tracking per-port VCPI allocations There has been a TODO waiting for quite a long time in drm_dp_mst_topology.c: /* We cannot rely on port->vcpi.num_slots to update * topology_state->avail_slots as the port may not exist if the parent * branch device was unplugged. This should be fixed by tracking * per-port slot allocation in drm_dp_mst_topology_state instead of * depending on the caller to tell us how many slots to release. */ That's not the only reason we should fix this: forcing the driver to track the VCPI allocations throughout a state's atomic check is error prone, because it means that extra care has to be taken with the order that drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() are called in in order to ensure idempotency. Currently the only driver actually using these helpers, i915, doesn't even do this correctly: multiple ->best_encoder() checks with i915's current implementation would not be idempotent and would over-allocate VCPI slots, something I learned trying to implement fallback retraining in MST. So: simplify this whole mess, and teach drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() to track the VCPI allocations for each port. This allows us to ensure idempotency without having to rely on the driver as much. Additionally: the driver doesn't need to do any kind of VCPI slot tracking anymore if it doesn't need it for it's own internal state. Additionally; this adds a new drm_dp_mst_atomic_check() helper which must be used by atomic drivers to perform validity checks for the new VCPI allocations incurred by a state. Also: update the documentation and make it more obvious that these /must/ be called by /all/ atomic drivers supporting MST. Changes since v9: * Add some missing changes that were requested by danvet that I forgot about after I redid all of the kref stuff: * Remove unnecessary state changes in intel_dp_mst_atomic_check * Cleanup atomic check logic for VCPI allocations - all we need to check in compute_config is whether or not this state disables a CRTC, then free VCPI based off that Changes since v8: * Fix compile errors, whoops! Changes since v7: - Don't check for mixed stale/valid VCPI allocations, just rely on connector registration to stop such erroneous modesets Changes since v6: - Keep a kref to all of the ports we have allocations on. This required a good bit of changing to when we call drm_dp_find_vcpi_slots(), mainly that we need to ensure that we only redo VCPI allocations on actual mode or CRTC changes, not crtc_state->active changes. Additionally, we no longer take the registration of the DRM connector for each port into account because so long as we have a kref to the port in the new or previous atomic state, the connector will stay registered. - Use the small changes to drm_dp_put_port() to add even more error checking to make misusage of the helpers more obvious. I added this after having to chase down various use-after-free conditions that started popping up from the new helpers so no one else has to troubleshoot that. - Move some accidental DRM_DEBUG_KMS() calls to DRM_DEBUG_ATOMIC() - Update documentation again, note that find/release() should both not be called on the same port in a single atomic check phase (but multiple calls to one or the other is OK) Changes since v4: - Don't skip the atomic checks for VCPI allocations if no new VCPI allocations happen in a state. This makes the next change I'm about to list here a lot easier to implement. - Don't ignore VCPI allocations on destroyed ports, instead ensure that when ports are destroyed and still have VCPI allocations in the topology state, the only state changes allowed are releasing said ports' VCPI. This prevents a state with a mix of VCPI allocations from destroyed ports, and allocations from valid ports. Changes since v3: - Don't release VCPI allocations in the topology state immediately in drm_dp_atomic_release_vcpi_slots(), instead mark them as 0 and skip over them in drm_dp_mst_duplicate_state(). This makes it so drm_dp_atomic_release_vcpi_slots() is still idempotent while also throwing warnings if the driver messes up it's book keeping and tries to release VCPI slots on a port that doesn't have any pre-existing VCPI allocation - danvet - Change mst_state/state in some debugging messages to "mst state" Changes since v2: - Use kmemdup() for duplicating MST state - danvet - Move port validation out of duplicate state callback - danvet - Handle looping through MST topology states in drm_dp_mst_atomic_check() so the driver doesn't have to do it - Fix documentation in drm_dp_atomic_find_vcpi_slots() - Move the atomic check for each individual topology state into it's own function, reduces indenting - Don't consider "stale" MST ports when calculating the bandwidth requirements. This is needed because originally we relied on the state duplication functions to prune any stale ports from the new state, which would prevent us from incorrectly considering their bandwidth requirements alongside legitimate new payloads. - Add function references in drm_dp_atomic_release_vcpi_slots() - danvet - Annotate atomic VCPI and atomic check functions with __must_check - danvet Changes since v1: - Don't use the now-removed ->atomic_check() for private objects hook, just give drivers a function to call themselves Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Harry Wentland <harry.wentland@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-19-lyude@redhat.com
2019-01-11 07:53:41 +07:00
INIT_LIST_HEAD(&state->vcpis);
list_for_each_entry(pos, &old_state->vcpis, next) {
/* Prune leftover freed VCPI allocations */
if (!pos->vcpi)
continue;
vcpi = kmemdup(pos, sizeof(*vcpi), GFP_KERNEL);
if (!vcpi)
goto fail;
drm_dp_mst_get_port_malloc(vcpi->port);
list_add(&vcpi->next, &state->vcpis);
}
return &state->base;
drm/dp_mst: Start tracking per-port VCPI allocations There has been a TODO waiting for quite a long time in drm_dp_mst_topology.c: /* We cannot rely on port->vcpi.num_slots to update * topology_state->avail_slots as the port may not exist if the parent * branch device was unplugged. This should be fixed by tracking * per-port slot allocation in drm_dp_mst_topology_state instead of * depending on the caller to tell us how many slots to release. */ That's not the only reason we should fix this: forcing the driver to track the VCPI allocations throughout a state's atomic check is error prone, because it means that extra care has to be taken with the order that drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() are called in in order to ensure idempotency. Currently the only driver actually using these helpers, i915, doesn't even do this correctly: multiple ->best_encoder() checks with i915's current implementation would not be idempotent and would over-allocate VCPI slots, something I learned trying to implement fallback retraining in MST. So: simplify this whole mess, and teach drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() to track the VCPI allocations for each port. This allows us to ensure idempotency without having to rely on the driver as much. Additionally: the driver doesn't need to do any kind of VCPI slot tracking anymore if it doesn't need it for it's own internal state. Additionally; this adds a new drm_dp_mst_atomic_check() helper which must be used by atomic drivers to perform validity checks for the new VCPI allocations incurred by a state. Also: update the documentation and make it more obvious that these /must/ be called by /all/ atomic drivers supporting MST. Changes since v9: * Add some missing changes that were requested by danvet that I forgot about after I redid all of the kref stuff: * Remove unnecessary state changes in intel_dp_mst_atomic_check * Cleanup atomic check logic for VCPI allocations - all we need to check in compute_config is whether or not this state disables a CRTC, then free VCPI based off that Changes since v8: * Fix compile errors, whoops! Changes since v7: - Don't check for mixed stale/valid VCPI allocations, just rely on connector registration to stop such erroneous modesets Changes since v6: - Keep a kref to all of the ports we have allocations on. This required a good bit of changing to when we call drm_dp_find_vcpi_slots(), mainly that we need to ensure that we only redo VCPI allocations on actual mode or CRTC changes, not crtc_state->active changes. Additionally, we no longer take the registration of the DRM connector for each port into account because so long as we have a kref to the port in the new or previous atomic state, the connector will stay registered. - Use the small changes to drm_dp_put_port() to add even more error checking to make misusage of the helpers more obvious. I added this after having to chase down various use-after-free conditions that started popping up from the new helpers so no one else has to troubleshoot that. - Move some accidental DRM_DEBUG_KMS() calls to DRM_DEBUG_ATOMIC() - Update documentation again, note that find/release() should both not be called on the same port in a single atomic check phase (but multiple calls to one or the other is OK) Changes since v4: - Don't skip the atomic checks for VCPI allocations if no new VCPI allocations happen in a state. This makes the next change I'm about to list here a lot easier to implement. - Don't ignore VCPI allocations on destroyed ports, instead ensure that when ports are destroyed and still have VCPI allocations in the topology state, the only state changes allowed are releasing said ports' VCPI. This prevents a state with a mix of VCPI allocations from destroyed ports, and allocations from valid ports. Changes since v3: - Don't release VCPI allocations in the topology state immediately in drm_dp_atomic_release_vcpi_slots(), instead mark them as 0 and skip over them in drm_dp_mst_duplicate_state(). This makes it so drm_dp_atomic_release_vcpi_slots() is still idempotent while also throwing warnings if the driver messes up it's book keeping and tries to release VCPI slots on a port that doesn't have any pre-existing VCPI allocation - danvet - Change mst_state/state in some debugging messages to "mst state" Changes since v2: - Use kmemdup() for duplicating MST state - danvet - Move port validation out of duplicate state callback - danvet - Handle looping through MST topology states in drm_dp_mst_atomic_check() so the driver doesn't have to do it - Fix documentation in drm_dp_atomic_find_vcpi_slots() - Move the atomic check for each individual topology state into it's own function, reduces indenting - Don't consider "stale" MST ports when calculating the bandwidth requirements. This is needed because originally we relied on the state duplication functions to prune any stale ports from the new state, which would prevent us from incorrectly considering their bandwidth requirements alongside legitimate new payloads. - Add function references in drm_dp_atomic_release_vcpi_slots() - danvet - Annotate atomic VCPI and atomic check functions with __must_check - danvet Changes since v1: - Don't use the now-removed ->atomic_check() for private objects hook, just give drivers a function to call themselves Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Harry Wentland <harry.wentland@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-19-lyude@redhat.com
2019-01-11 07:53:41 +07:00
fail:
list_for_each_entry_safe(pos, vcpi, &state->vcpis, next) {
drm_dp_mst_put_port_malloc(pos->port);
kfree(pos);
}
kfree(state);
return NULL;
}
static void drm_dp_mst_destroy_state(struct drm_private_obj *obj,
struct drm_private_state *state)
{
struct drm_dp_mst_topology_state *mst_state =
to_dp_mst_topology_state(state);
drm/dp_mst: Start tracking per-port VCPI allocations There has been a TODO waiting for quite a long time in drm_dp_mst_topology.c: /* We cannot rely on port->vcpi.num_slots to update * topology_state->avail_slots as the port may not exist if the parent * branch device was unplugged. This should be fixed by tracking * per-port slot allocation in drm_dp_mst_topology_state instead of * depending on the caller to tell us how many slots to release. */ That's not the only reason we should fix this: forcing the driver to track the VCPI allocations throughout a state's atomic check is error prone, because it means that extra care has to be taken with the order that drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() are called in in order to ensure idempotency. Currently the only driver actually using these helpers, i915, doesn't even do this correctly: multiple ->best_encoder() checks with i915's current implementation would not be idempotent and would over-allocate VCPI slots, something I learned trying to implement fallback retraining in MST. So: simplify this whole mess, and teach drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() to track the VCPI allocations for each port. This allows us to ensure idempotency without having to rely on the driver as much. Additionally: the driver doesn't need to do any kind of VCPI slot tracking anymore if it doesn't need it for it's own internal state. Additionally; this adds a new drm_dp_mst_atomic_check() helper which must be used by atomic drivers to perform validity checks for the new VCPI allocations incurred by a state. Also: update the documentation and make it more obvious that these /must/ be called by /all/ atomic drivers supporting MST. Changes since v9: * Add some missing changes that were requested by danvet that I forgot about after I redid all of the kref stuff: * Remove unnecessary state changes in intel_dp_mst_atomic_check * Cleanup atomic check logic for VCPI allocations - all we need to check in compute_config is whether or not this state disables a CRTC, then free VCPI based off that Changes since v8: * Fix compile errors, whoops! Changes since v7: - Don't check for mixed stale/valid VCPI allocations, just rely on connector registration to stop such erroneous modesets Changes since v6: - Keep a kref to all of the ports we have allocations on. This required a good bit of changing to when we call drm_dp_find_vcpi_slots(), mainly that we need to ensure that we only redo VCPI allocations on actual mode or CRTC changes, not crtc_state->active changes. Additionally, we no longer take the registration of the DRM connector for each port into account because so long as we have a kref to the port in the new or previous atomic state, the connector will stay registered. - Use the small changes to drm_dp_put_port() to add even more error checking to make misusage of the helpers more obvious. I added this after having to chase down various use-after-free conditions that started popping up from the new helpers so no one else has to troubleshoot that. - Move some accidental DRM_DEBUG_KMS() calls to DRM_DEBUG_ATOMIC() - Update documentation again, note that find/release() should both not be called on the same port in a single atomic check phase (but multiple calls to one or the other is OK) Changes since v4: - Don't skip the atomic checks for VCPI allocations if no new VCPI allocations happen in a state. This makes the next change I'm about to list here a lot easier to implement. - Don't ignore VCPI allocations on destroyed ports, instead ensure that when ports are destroyed and still have VCPI allocations in the topology state, the only state changes allowed are releasing said ports' VCPI. This prevents a state with a mix of VCPI allocations from destroyed ports, and allocations from valid ports. Changes since v3: - Don't release VCPI allocations in the topology state immediately in drm_dp_atomic_release_vcpi_slots(), instead mark them as 0 and skip over them in drm_dp_mst_duplicate_state(). This makes it so drm_dp_atomic_release_vcpi_slots() is still idempotent while also throwing warnings if the driver messes up it's book keeping and tries to release VCPI slots on a port that doesn't have any pre-existing VCPI allocation - danvet - Change mst_state/state in some debugging messages to "mst state" Changes since v2: - Use kmemdup() for duplicating MST state - danvet - Move port validation out of duplicate state callback - danvet - Handle looping through MST topology states in drm_dp_mst_atomic_check() so the driver doesn't have to do it - Fix documentation in drm_dp_atomic_find_vcpi_slots() - Move the atomic check for each individual topology state into it's own function, reduces indenting - Don't consider "stale" MST ports when calculating the bandwidth requirements. This is needed because originally we relied on the state duplication functions to prune any stale ports from the new state, which would prevent us from incorrectly considering their bandwidth requirements alongside legitimate new payloads. - Add function references in drm_dp_atomic_release_vcpi_slots() - danvet - Annotate atomic VCPI and atomic check functions with __must_check - danvet Changes since v1: - Don't use the now-removed ->atomic_check() for private objects hook, just give drivers a function to call themselves Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Harry Wentland <harry.wentland@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-19-lyude@redhat.com
2019-01-11 07:53:41 +07:00
struct drm_dp_vcpi_allocation *pos, *tmp;
list_for_each_entry_safe(pos, tmp, &mst_state->vcpis, next) {
/* We only keep references to ports with non-zero VCPIs */
if (pos->vcpi)
drm_dp_mst_put_port_malloc(pos->port);
kfree(pos);
}
kfree(mst_state);
}
drm/dp_mst: Start tracking per-port VCPI allocations There has been a TODO waiting for quite a long time in drm_dp_mst_topology.c: /* We cannot rely on port->vcpi.num_slots to update * topology_state->avail_slots as the port may not exist if the parent * branch device was unplugged. This should be fixed by tracking * per-port slot allocation in drm_dp_mst_topology_state instead of * depending on the caller to tell us how many slots to release. */ That's not the only reason we should fix this: forcing the driver to track the VCPI allocations throughout a state's atomic check is error prone, because it means that extra care has to be taken with the order that drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() are called in in order to ensure idempotency. Currently the only driver actually using these helpers, i915, doesn't even do this correctly: multiple ->best_encoder() checks with i915's current implementation would not be idempotent and would over-allocate VCPI slots, something I learned trying to implement fallback retraining in MST. So: simplify this whole mess, and teach drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() to track the VCPI allocations for each port. This allows us to ensure idempotency without having to rely on the driver as much. Additionally: the driver doesn't need to do any kind of VCPI slot tracking anymore if it doesn't need it for it's own internal state. Additionally; this adds a new drm_dp_mst_atomic_check() helper which must be used by atomic drivers to perform validity checks for the new VCPI allocations incurred by a state. Also: update the documentation and make it more obvious that these /must/ be called by /all/ atomic drivers supporting MST. Changes since v9: * Add some missing changes that were requested by danvet that I forgot about after I redid all of the kref stuff: * Remove unnecessary state changes in intel_dp_mst_atomic_check * Cleanup atomic check logic for VCPI allocations - all we need to check in compute_config is whether or not this state disables a CRTC, then free VCPI based off that Changes since v8: * Fix compile errors, whoops! Changes since v7: - Don't check for mixed stale/valid VCPI allocations, just rely on connector registration to stop such erroneous modesets Changes since v6: - Keep a kref to all of the ports we have allocations on. This required a good bit of changing to when we call drm_dp_find_vcpi_slots(), mainly that we need to ensure that we only redo VCPI allocations on actual mode or CRTC changes, not crtc_state->active changes. Additionally, we no longer take the registration of the DRM connector for each port into account because so long as we have a kref to the port in the new or previous atomic state, the connector will stay registered. - Use the small changes to drm_dp_put_port() to add even more error checking to make misusage of the helpers more obvious. I added this after having to chase down various use-after-free conditions that started popping up from the new helpers so no one else has to troubleshoot that. - Move some accidental DRM_DEBUG_KMS() calls to DRM_DEBUG_ATOMIC() - Update documentation again, note that find/release() should both not be called on the same port in a single atomic check phase (but multiple calls to one or the other is OK) Changes since v4: - Don't skip the atomic checks for VCPI allocations if no new VCPI allocations happen in a state. This makes the next change I'm about to list here a lot easier to implement. - Don't ignore VCPI allocations on destroyed ports, instead ensure that when ports are destroyed and still have VCPI allocations in the topology state, the only state changes allowed are releasing said ports' VCPI. This prevents a state with a mix of VCPI allocations from destroyed ports, and allocations from valid ports. Changes since v3: - Don't release VCPI allocations in the topology state immediately in drm_dp_atomic_release_vcpi_slots(), instead mark them as 0 and skip over them in drm_dp_mst_duplicate_state(). This makes it so drm_dp_atomic_release_vcpi_slots() is still idempotent while also throwing warnings if the driver messes up it's book keeping and tries to release VCPI slots on a port that doesn't have any pre-existing VCPI allocation - danvet - Change mst_state/state in some debugging messages to "mst state" Changes since v2: - Use kmemdup() for duplicating MST state - danvet - Move port validation out of duplicate state callback - danvet - Handle looping through MST topology states in drm_dp_mst_atomic_check() so the driver doesn't have to do it - Fix documentation in drm_dp_atomic_find_vcpi_slots() - Move the atomic check for each individual topology state into it's own function, reduces indenting - Don't consider "stale" MST ports when calculating the bandwidth requirements. This is needed because originally we relied on the state duplication functions to prune any stale ports from the new state, which would prevent us from incorrectly considering their bandwidth requirements alongside legitimate new payloads. - Add function references in drm_dp_atomic_release_vcpi_slots() - danvet - Annotate atomic VCPI and atomic check functions with __must_check - danvet Changes since v1: - Don't use the now-removed ->atomic_check() for private objects hook, just give drivers a function to call themselves Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Harry Wentland <harry.wentland@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-19-lyude@redhat.com
2019-01-11 07:53:41 +07:00
static inline int
drm_dp_mst_atomic_check_topology_state(struct drm_dp_mst_topology_mgr *mgr,
struct drm_dp_mst_topology_state *mst_state)
{
struct drm_dp_vcpi_allocation *vcpi;
int avail_slots = 63, payload_count = 0;
drm/dp_mst: Start tracking per-port VCPI allocations There has been a TODO waiting for quite a long time in drm_dp_mst_topology.c: /* We cannot rely on port->vcpi.num_slots to update * topology_state->avail_slots as the port may not exist if the parent * branch device was unplugged. This should be fixed by tracking * per-port slot allocation in drm_dp_mst_topology_state instead of * depending on the caller to tell us how many slots to release. */ That's not the only reason we should fix this: forcing the driver to track the VCPI allocations throughout a state's atomic check is error prone, because it means that extra care has to be taken with the order that drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() are called in in order to ensure idempotency. Currently the only driver actually using these helpers, i915, doesn't even do this correctly: multiple ->best_encoder() checks with i915's current implementation would not be idempotent and would over-allocate VCPI slots, something I learned trying to implement fallback retraining in MST. So: simplify this whole mess, and teach drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() to track the VCPI allocations for each port. This allows us to ensure idempotency without having to rely on the driver as much. Additionally: the driver doesn't need to do any kind of VCPI slot tracking anymore if it doesn't need it for it's own internal state. Additionally; this adds a new drm_dp_mst_atomic_check() helper which must be used by atomic drivers to perform validity checks for the new VCPI allocations incurred by a state. Also: update the documentation and make it more obvious that these /must/ be called by /all/ atomic drivers supporting MST. Changes since v9: * Add some missing changes that were requested by danvet that I forgot about after I redid all of the kref stuff: * Remove unnecessary state changes in intel_dp_mst_atomic_check * Cleanup atomic check logic for VCPI allocations - all we need to check in compute_config is whether or not this state disables a CRTC, then free VCPI based off that Changes since v8: * Fix compile errors, whoops! Changes since v7: - Don't check for mixed stale/valid VCPI allocations, just rely on connector registration to stop such erroneous modesets Changes since v6: - Keep a kref to all of the ports we have allocations on. This required a good bit of changing to when we call drm_dp_find_vcpi_slots(), mainly that we need to ensure that we only redo VCPI allocations on actual mode or CRTC changes, not crtc_state->active changes. Additionally, we no longer take the registration of the DRM connector for each port into account because so long as we have a kref to the port in the new or previous atomic state, the connector will stay registered. - Use the small changes to drm_dp_put_port() to add even more error checking to make misusage of the helpers more obvious. I added this after having to chase down various use-after-free conditions that started popping up from the new helpers so no one else has to troubleshoot that. - Move some accidental DRM_DEBUG_KMS() calls to DRM_DEBUG_ATOMIC() - Update documentation again, note that find/release() should both not be called on the same port in a single atomic check phase (but multiple calls to one or the other is OK) Changes since v4: - Don't skip the atomic checks for VCPI allocations if no new VCPI allocations happen in a state. This makes the next change I'm about to list here a lot easier to implement. - Don't ignore VCPI allocations on destroyed ports, instead ensure that when ports are destroyed and still have VCPI allocations in the topology state, the only state changes allowed are releasing said ports' VCPI. This prevents a state with a mix of VCPI allocations from destroyed ports, and allocations from valid ports. Changes since v3: - Don't release VCPI allocations in the topology state immediately in drm_dp_atomic_release_vcpi_slots(), instead mark them as 0 and skip over them in drm_dp_mst_duplicate_state(). This makes it so drm_dp_atomic_release_vcpi_slots() is still idempotent while also throwing warnings if the driver messes up it's book keeping and tries to release VCPI slots on a port that doesn't have any pre-existing VCPI allocation - danvet - Change mst_state/state in some debugging messages to "mst state" Changes since v2: - Use kmemdup() for duplicating MST state - danvet - Move port validation out of duplicate state callback - danvet - Handle looping through MST topology states in drm_dp_mst_atomic_check() so the driver doesn't have to do it - Fix documentation in drm_dp_atomic_find_vcpi_slots() - Move the atomic check for each individual topology state into it's own function, reduces indenting - Don't consider "stale" MST ports when calculating the bandwidth requirements. This is needed because originally we relied on the state duplication functions to prune any stale ports from the new state, which would prevent us from incorrectly considering their bandwidth requirements alongside legitimate new payloads. - Add function references in drm_dp_atomic_release_vcpi_slots() - danvet - Annotate atomic VCPI and atomic check functions with __must_check - danvet Changes since v1: - Don't use the now-removed ->atomic_check() for private objects hook, just give drivers a function to call themselves Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Harry Wentland <harry.wentland@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-19-lyude@redhat.com
2019-01-11 07:53:41 +07:00
list_for_each_entry(vcpi, &mst_state->vcpis, next) {
/* Releasing VCPI is always OK-even if the port is gone */
if (!vcpi->vcpi) {
DRM_DEBUG_ATOMIC("[MST PORT:%p] releases all VCPI slots\n",
vcpi->port);
continue;
}
DRM_DEBUG_ATOMIC("[MST PORT:%p] requires %d vcpi slots\n",
vcpi->port, vcpi->vcpi);
avail_slots -= vcpi->vcpi;
if (avail_slots < 0) {
DRM_DEBUG_ATOMIC("[MST PORT:%p] not enough VCPI slots in mst state %p (avail=%d)\n",
vcpi->port, mst_state,
avail_slots + vcpi->vcpi);
return -ENOSPC;
}
if (++payload_count > mgr->max_payloads) {
DRM_DEBUG_ATOMIC("[MST MGR:%p] state %p has too many payloads (max=%d)\n",
mgr, mst_state, mgr->max_payloads);
return -EINVAL;
}
drm/dp_mst: Start tracking per-port VCPI allocations There has been a TODO waiting for quite a long time in drm_dp_mst_topology.c: /* We cannot rely on port->vcpi.num_slots to update * topology_state->avail_slots as the port may not exist if the parent * branch device was unplugged. This should be fixed by tracking * per-port slot allocation in drm_dp_mst_topology_state instead of * depending on the caller to tell us how many slots to release. */ That's not the only reason we should fix this: forcing the driver to track the VCPI allocations throughout a state's atomic check is error prone, because it means that extra care has to be taken with the order that drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() are called in in order to ensure idempotency. Currently the only driver actually using these helpers, i915, doesn't even do this correctly: multiple ->best_encoder() checks with i915's current implementation would not be idempotent and would over-allocate VCPI slots, something I learned trying to implement fallback retraining in MST. So: simplify this whole mess, and teach drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() to track the VCPI allocations for each port. This allows us to ensure idempotency without having to rely on the driver as much. Additionally: the driver doesn't need to do any kind of VCPI slot tracking anymore if it doesn't need it for it's own internal state. Additionally; this adds a new drm_dp_mst_atomic_check() helper which must be used by atomic drivers to perform validity checks for the new VCPI allocations incurred by a state. Also: update the documentation and make it more obvious that these /must/ be called by /all/ atomic drivers supporting MST. Changes since v9: * Add some missing changes that were requested by danvet that I forgot about after I redid all of the kref stuff: * Remove unnecessary state changes in intel_dp_mst_atomic_check * Cleanup atomic check logic for VCPI allocations - all we need to check in compute_config is whether or not this state disables a CRTC, then free VCPI based off that Changes since v8: * Fix compile errors, whoops! Changes since v7: - Don't check for mixed stale/valid VCPI allocations, just rely on connector registration to stop such erroneous modesets Changes since v6: - Keep a kref to all of the ports we have allocations on. This required a good bit of changing to when we call drm_dp_find_vcpi_slots(), mainly that we need to ensure that we only redo VCPI allocations on actual mode or CRTC changes, not crtc_state->active changes. Additionally, we no longer take the registration of the DRM connector for each port into account because so long as we have a kref to the port in the new or previous atomic state, the connector will stay registered. - Use the small changes to drm_dp_put_port() to add even more error checking to make misusage of the helpers more obvious. I added this after having to chase down various use-after-free conditions that started popping up from the new helpers so no one else has to troubleshoot that. - Move some accidental DRM_DEBUG_KMS() calls to DRM_DEBUG_ATOMIC() - Update documentation again, note that find/release() should both not be called on the same port in a single atomic check phase (but multiple calls to one or the other is OK) Changes since v4: - Don't skip the atomic checks for VCPI allocations if no new VCPI allocations happen in a state. This makes the next change I'm about to list here a lot easier to implement. - Don't ignore VCPI allocations on destroyed ports, instead ensure that when ports are destroyed and still have VCPI allocations in the topology state, the only state changes allowed are releasing said ports' VCPI. This prevents a state with a mix of VCPI allocations from destroyed ports, and allocations from valid ports. Changes since v3: - Don't release VCPI allocations in the topology state immediately in drm_dp_atomic_release_vcpi_slots(), instead mark them as 0 and skip over them in drm_dp_mst_duplicate_state(). This makes it so drm_dp_atomic_release_vcpi_slots() is still idempotent while also throwing warnings if the driver messes up it's book keeping and tries to release VCPI slots on a port that doesn't have any pre-existing VCPI allocation - danvet - Change mst_state/state in some debugging messages to "mst state" Changes since v2: - Use kmemdup() for duplicating MST state - danvet - Move port validation out of duplicate state callback - danvet - Handle looping through MST topology states in drm_dp_mst_atomic_check() so the driver doesn't have to do it - Fix documentation in drm_dp_atomic_find_vcpi_slots() - Move the atomic check for each individual topology state into it's own function, reduces indenting - Don't consider "stale" MST ports when calculating the bandwidth requirements. This is needed because originally we relied on the state duplication functions to prune any stale ports from the new state, which would prevent us from incorrectly considering their bandwidth requirements alongside legitimate new payloads. - Add function references in drm_dp_atomic_release_vcpi_slots() - danvet - Annotate atomic VCPI and atomic check functions with __must_check - danvet Changes since v1: - Don't use the now-removed ->atomic_check() for private objects hook, just give drivers a function to call themselves Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Harry Wentland <harry.wentland@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-19-lyude@redhat.com
2019-01-11 07:53:41 +07:00
}
DRM_DEBUG_ATOMIC("[MST MGR:%p] mst state %p VCPI avail=%d used=%d\n",
mgr, mst_state, avail_slots,
63 - avail_slots);
return 0;
}
/**
* drm_dp_mst_atomic_check - Check that the new state of an MST topology in an
* atomic update is valid
* @state: Pointer to the new &struct drm_dp_mst_topology_state
*
* Checks the given topology state for an atomic update to ensure that it's
* valid. This includes checking whether there's enough bandwidth to support
* the new VCPI allocations in the atomic update.
*
* Any atomic drivers supporting DP MST must make sure to call this after
* checking the rest of their state in their
* &drm_mode_config_funcs.atomic_check() callback.
*
* See also:
* drm_dp_atomic_find_vcpi_slots()
* drm_dp_atomic_release_vcpi_slots()
*
* Returns:
*
* 0 if the new state is valid, negative error code otherwise.
*/
int drm_dp_mst_atomic_check(struct drm_atomic_state *state)
{
struct drm_dp_mst_topology_mgr *mgr;
struct drm_dp_mst_topology_state *mst_state;
int i, ret = 0;
for_each_new_mst_mgr_in_state(state, mgr, mst_state, i) {
ret = drm_dp_mst_atomic_check_topology_state(mgr, mst_state);
if (ret)
break;
}
return ret;
}
EXPORT_SYMBOL(drm_dp_mst_atomic_check);
const struct drm_private_state_funcs drm_dp_mst_topology_state_funcs = {
.atomic_duplicate_state = drm_dp_mst_duplicate_state,
.atomic_destroy_state = drm_dp_mst_destroy_state,
};
EXPORT_SYMBOL(drm_dp_mst_topology_state_funcs);
/**
* drm_atomic_get_mst_topology_state: get MST topology state
*
* @state: global atomic state
* @mgr: MST topology manager, also the private object in this case
*
* This function wraps drm_atomic_get_priv_obj_state() passing in the MST atomic
* state vtable so that the private object state returned is that of a MST
* topology object. Also, drm_atomic_get_private_obj_state() expects the caller
* to care of the locking, so warn if don't hold the connection_mutex.
*
* RETURNS:
*
* The MST topology state or error pointer.
*/
struct drm_dp_mst_topology_state *drm_atomic_get_mst_topology_state(struct drm_atomic_state *state,
struct drm_dp_mst_topology_mgr *mgr)
{
struct drm_device *dev = mgr->dev;
WARN_ON(!drm_modeset_is_locked(&dev->mode_config.connection_mutex));
return to_dp_mst_topology_state(drm_atomic_get_private_obj_state(state, &mgr->base));
}
EXPORT_SYMBOL(drm_atomic_get_mst_topology_state);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
/**
* drm_dp_mst_topology_mgr_init - initialise a topology manager
* @mgr: manager struct to initialise
* @dev: device providing this structure - for i2c addition.
* @aux: DP helper aux channel to talk to this device
* @max_dpcd_transaction_bytes: hw specific DPCD transaction limit
* @max_payloads: maximum number of payloads this GPU can source
* @conn_base_id: the connector object ID the MST device is connected to.
*
* Return 0 for success, or negative error code on failure
*/
int drm_dp_mst_topology_mgr_init(struct drm_dp_mst_topology_mgr *mgr,
struct drm_device *dev, struct drm_dp_aux *aux,
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
int max_dpcd_transaction_bytes,
int max_payloads, int conn_base_id)
{
struct drm_dp_mst_topology_state *mst_state;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
mutex_init(&mgr->lock);
mutex_init(&mgr->qlock);
mutex_init(&mgr->payload_lock);
mutex_init(&mgr->destroy_connector_lock);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
INIT_LIST_HEAD(&mgr->tx_msg_downq);
INIT_LIST_HEAD(&mgr->destroy_connector_list);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
INIT_WORK(&mgr->work, drm_dp_mst_link_probe_work);
INIT_WORK(&mgr->tx_work, drm_dp_tx_work);
INIT_WORK(&mgr->destroy_connector_work, drm_dp_destroy_connector_work);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
init_waitqueue_head(&mgr->tx_waitq);
mgr->dev = dev;
mgr->aux = aux;
mgr->max_dpcd_transaction_bytes = max_dpcd_transaction_bytes;
mgr->max_payloads = max_payloads;
mgr->conn_base_id = conn_base_id;
if (max_payloads + 1 > sizeof(mgr->payload_mask) * 8 ||
max_payloads + 1 > sizeof(mgr->vcpi_mask) * 8)
return -EINVAL;
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
mgr->payloads = kcalloc(max_payloads, sizeof(struct drm_dp_payload), GFP_KERNEL);
if (!mgr->payloads)
return -ENOMEM;
mgr->proposed_vcpis = kcalloc(max_payloads, sizeof(struct drm_dp_vcpi *), GFP_KERNEL);
if (!mgr->proposed_vcpis)
return -ENOMEM;
set_bit(0, &mgr->payload_mask);
mst_state = kzalloc(sizeof(*mst_state), GFP_KERNEL);
if (mst_state == NULL)
return -ENOMEM;
mst_state->mgr = mgr;
drm/dp_mst: Start tracking per-port VCPI allocations There has been a TODO waiting for quite a long time in drm_dp_mst_topology.c: /* We cannot rely on port->vcpi.num_slots to update * topology_state->avail_slots as the port may not exist if the parent * branch device was unplugged. This should be fixed by tracking * per-port slot allocation in drm_dp_mst_topology_state instead of * depending on the caller to tell us how many slots to release. */ That's not the only reason we should fix this: forcing the driver to track the VCPI allocations throughout a state's atomic check is error prone, because it means that extra care has to be taken with the order that drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() are called in in order to ensure idempotency. Currently the only driver actually using these helpers, i915, doesn't even do this correctly: multiple ->best_encoder() checks with i915's current implementation would not be idempotent and would over-allocate VCPI slots, something I learned trying to implement fallback retraining in MST. So: simplify this whole mess, and teach drm_dp_atomic_find_vcpi_slots() and drm_dp_atomic_release_vcpi_slots() to track the VCPI allocations for each port. This allows us to ensure idempotency without having to rely on the driver as much. Additionally: the driver doesn't need to do any kind of VCPI slot tracking anymore if it doesn't need it for it's own internal state. Additionally; this adds a new drm_dp_mst_atomic_check() helper which must be used by atomic drivers to perform validity checks for the new VCPI allocations incurred by a state. Also: update the documentation and make it more obvious that these /must/ be called by /all/ atomic drivers supporting MST. Changes since v9: * Add some missing changes that were requested by danvet that I forgot about after I redid all of the kref stuff: * Remove unnecessary state changes in intel_dp_mst_atomic_check * Cleanup atomic check logic for VCPI allocations - all we need to check in compute_config is whether or not this state disables a CRTC, then free VCPI based off that Changes since v8: * Fix compile errors, whoops! Changes since v7: - Don't check for mixed stale/valid VCPI allocations, just rely on connector registration to stop such erroneous modesets Changes since v6: - Keep a kref to all of the ports we have allocations on. This required a good bit of changing to when we call drm_dp_find_vcpi_slots(), mainly that we need to ensure that we only redo VCPI allocations on actual mode or CRTC changes, not crtc_state->active changes. Additionally, we no longer take the registration of the DRM connector for each port into account because so long as we have a kref to the port in the new or previous atomic state, the connector will stay registered. - Use the small changes to drm_dp_put_port() to add even more error checking to make misusage of the helpers more obvious. I added this after having to chase down various use-after-free conditions that started popping up from the new helpers so no one else has to troubleshoot that. - Move some accidental DRM_DEBUG_KMS() calls to DRM_DEBUG_ATOMIC() - Update documentation again, note that find/release() should both not be called on the same port in a single atomic check phase (but multiple calls to one or the other is OK) Changes since v4: - Don't skip the atomic checks for VCPI allocations if no new VCPI allocations happen in a state. This makes the next change I'm about to list here a lot easier to implement. - Don't ignore VCPI allocations on destroyed ports, instead ensure that when ports are destroyed and still have VCPI allocations in the topology state, the only state changes allowed are releasing said ports' VCPI. This prevents a state with a mix of VCPI allocations from destroyed ports, and allocations from valid ports. Changes since v3: - Don't release VCPI allocations in the topology state immediately in drm_dp_atomic_release_vcpi_slots(), instead mark them as 0 and skip over them in drm_dp_mst_duplicate_state(). This makes it so drm_dp_atomic_release_vcpi_slots() is still idempotent while also throwing warnings if the driver messes up it's book keeping and tries to release VCPI slots on a port that doesn't have any pre-existing VCPI allocation - danvet - Change mst_state/state in some debugging messages to "mst state" Changes since v2: - Use kmemdup() for duplicating MST state - danvet - Move port validation out of duplicate state callback - danvet - Handle looping through MST topology states in drm_dp_mst_atomic_check() so the driver doesn't have to do it - Fix documentation in drm_dp_atomic_find_vcpi_slots() - Move the atomic check for each individual topology state into it's own function, reduces indenting - Don't consider "stale" MST ports when calculating the bandwidth requirements. This is needed because originally we relied on the state duplication functions to prune any stale ports from the new state, which would prevent us from incorrectly considering their bandwidth requirements alongside legitimate new payloads. - Add function references in drm_dp_atomic_release_vcpi_slots() - danvet - Annotate atomic VCPI and atomic check functions with __must_check - danvet Changes since v1: - Don't use the now-removed ->atomic_check() for private objects hook, just give drivers a function to call themselves Signed-off-by: Lyude Paul <lyude@redhat.com> Reviewed-by: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@redhat.com> Cc: Jerry Zuo <Jerry.Zuo@amd.com> Cc: Harry Wentland <harry.wentland@amd.com> Cc: Juston Li <juston.li@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190111005343.17443-19-lyude@redhat.com
2019-01-11 07:53:41 +07:00
INIT_LIST_HEAD(&mst_state->vcpis);
drm_atomic_private_obj_init(dev, &mgr->base,
&mst_state->base,
&drm_dp_mst_topology_state_funcs);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
return 0;
}
EXPORT_SYMBOL(drm_dp_mst_topology_mgr_init);
/**
* drm_dp_mst_topology_mgr_destroy() - destroy topology manager.
* @mgr: manager to destroy
*/
void drm_dp_mst_topology_mgr_destroy(struct drm_dp_mst_topology_mgr *mgr)
{
drm_dp_mst_topology_mgr_set_mst(mgr, false);
drm/dp/mst: drop cancel work sync in the mstb destroy path (v2) Since 9eb1e57f564d4e6e10991402726cc83fe0b9172f drm/dp/mst: make sure mst_primary mstb is valid in work function we validate the mstb structs in the work function, and doing that takes a reference. So we should never get here with the work function running using the mstb device, only if the work function hasn't run yet or is running for another mstb. So we don't need to sync the work here, this was causing lockdep spew as below. [ +0.000160] ============================================= [ +0.000001] [ INFO: possible recursive locking detected ] [ +0.000002] 3.10.0-320.el7.rhel72.stable.backport.3.x86_64.debug #1 Tainted: G W ------------ [ +0.000001] --------------------------------------------- [ +0.000001] kworker/4:2/1262 is trying to acquire lock: [ +0.000001] ((&mgr->work)){+.+.+.}, at: [<ffffffff810b29a5>] flush_work+0x5/0x2e0 [ +0.000007] but task is already holding lock: [ +0.000001] ((&mgr->work)){+.+.+.}, at: [<ffffffff810b57e4>] process_one_work+0x1b4/0x710 [ +0.000004] other info that might help us debug this: [ +0.000001] Possible unsafe locking scenario: [ +0.000002] CPU0 [ +0.000000] ---- [ +0.000001] lock((&mgr->work)); [ +0.000002] lock((&mgr->work)); [ +0.000001] *** DEADLOCK *** [ +0.000001] May be due to missing lock nesting notation [ +0.000002] 2 locks held by kworker/4:2/1262: [ +0.000001] #0: (events_long){.+.+.+}, at: [<ffffffff810b57e4>] process_one_work+0x1b4/0x710 [ +0.000004] #1: ((&mgr->work)){+.+.+.}, at: [<ffffffff810b57e4>] process_one_work+0x1b4/0x710 [ +0.000003] stack backtrace: [ +0.000003] CPU: 4 PID: 1262 Comm: kworker/4:2 Tainted: G W ------------ 3.10.0-320.el7.rhel72.stable.backport.3.x86_64.debug #1 [ +0.000001] Hardware name: LENOVO 20EGS0R600/20EGS0R600, BIOS GNET71WW (2.19 ) 02/05/2015 [ +0.000008] Workqueue: events_long drm_dp_mst_link_probe_work [drm_kms_helper] [ +0.000001] ffffffff82c26c90 00000000a527b914 ffff88046399bae8 ffffffff816fe04d [ +0.000004] ffff88046399bb58 ffffffff8110f47f ffff880461438000 0001009b840fc003 [ +0.000002] ffff880461438a98 0000000000000000 0000000804dc26e1 ffffffff824a2c00 [ +0.000003] Call Trace: [ +0.000004] [<ffffffff816fe04d>] dump_stack+0x19/0x1b [ +0.000004] [<ffffffff8110f47f>] __lock_acquire+0x115f/0x1250 [ +0.000002] [<ffffffff8110fd49>] lock_acquire+0x99/0x1e0 [ +0.000002] [<ffffffff810b29a5>] ? flush_work+0x5/0x2e0 [ +0.000002] [<ffffffff810b29ee>] flush_work+0x4e/0x2e0 [ +0.000002] [<ffffffff810b29a5>] ? flush_work+0x5/0x2e0 [ +0.000004] [<ffffffff81025905>] ? native_sched_clock+0x35/0x80 [ +0.000002] [<ffffffff81025959>] ? sched_clock+0x9/0x10 [ +0.000002] [<ffffffff810da1f5>] ? local_clock+0x25/0x30 [ +0.000002] [<ffffffff8110dca9>] ? mark_held_locks+0xb9/0x140 [ +0.000003] [<ffffffff810b4ed5>] ? __cancel_work_timer+0x95/0x160 [ +0.000002] [<ffffffff810b4ee8>] __cancel_work_timer+0xa8/0x160 [ +0.000002] [<ffffffff810b4fb0>] cancel_work_sync+0x10/0x20 [ +0.000007] [<ffffffffa0160d17>] drm_dp_destroy_mst_branch_device+0x27/0x120 [drm_kms_helper] [ +0.000006] [<ffffffffa0163968>] drm_dp_mst_link_probe_work+0x78/0xa0 [drm_kms_helper] [ +0.000002] [<ffffffff810b5850>] process_one_work+0x220/0x710 [ +0.000002] [<ffffffff810b57e4>] ? process_one_work+0x1b4/0x710 [ +0.000005] [<ffffffff810b5e5b>] worker_thread+0x11b/0x3a0 [ +0.000003] [<ffffffff810b5d40>] ? process_one_work+0x710/0x710 [ +0.000002] [<ffffffff810beced>] kthread+0xed/0x100 [ +0.000003] [<ffffffff810bec00>] ? insert_kthread_work+0x80/0x80 [ +0.000003] [<ffffffff817121d8>] ret_from_fork+0x58/0x90 v2: add flush_work. Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: stable@vger.kernel.org Signed-off-by: Dave Airlie <airlied@redhat.com>
2015-09-30 07:39:42 +07:00
flush_work(&mgr->work);
flush_work(&mgr->destroy_connector_work);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
mutex_lock(&mgr->payload_lock);
kfree(mgr->payloads);
mgr->payloads = NULL;
kfree(mgr->proposed_vcpis);
mgr->proposed_vcpis = NULL;
mutex_unlock(&mgr->payload_lock);
mgr->dev = NULL;
mgr->aux = NULL;
drm_atomic_private_obj_fini(&mgr->base);
mgr->funcs = NULL;
mutex_destroy(&mgr->destroy_connector_lock);
mutex_destroy(&mgr->payload_lock);
mutex_destroy(&mgr->qlock);
mutex_destroy(&mgr->lock);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
EXPORT_SYMBOL(drm_dp_mst_topology_mgr_destroy);
static bool remote_i2c_read_ok(const struct i2c_msg msgs[], int num)
{
int i;
if (num - 1 > DP_REMOTE_I2C_READ_MAX_TRANSACTIONS)
return false;
for (i = 0; i < num - 1; i++) {
if (msgs[i].flags & I2C_M_RD ||
msgs[i].len > 0xff)
return false;
}
return msgs[num - 1].flags & I2C_M_RD &&
msgs[num - 1].len <= 0xff;
}
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
/* I2C device */
static int drm_dp_mst_i2c_xfer(struct i2c_adapter *adapter, struct i2c_msg *msgs,
int num)
{
struct drm_dp_aux *aux = adapter->algo_data;
struct drm_dp_mst_port *port = container_of(aux, struct drm_dp_mst_port, aux);
struct drm_dp_mst_branch *mstb;
struct drm_dp_mst_topology_mgr *mgr = port->mgr;
unsigned int i;
struct drm_dp_sideband_msg_req_body msg;
struct drm_dp_sideband_msg_tx *txmsg = NULL;
int ret;
mstb = drm_dp_mst_topology_get_mstb_validated(mgr, port->parent);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
if (!mstb)
return -EREMOTEIO;
if (!remote_i2c_read_ok(msgs, num)) {
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
DRM_DEBUG_KMS("Unsupported I2C transaction for MST device\n");
ret = -EIO;
goto out;
}
memset(&msg, 0, sizeof(msg));
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
msg.req_type = DP_REMOTE_I2C_READ;
msg.u.i2c_read.num_transactions = num - 1;
msg.u.i2c_read.port_number = port->port_num;
for (i = 0; i < num - 1; i++) {
msg.u.i2c_read.transactions[i].i2c_dev_id = msgs[i].addr;
msg.u.i2c_read.transactions[i].num_bytes = msgs[i].len;
msg.u.i2c_read.transactions[i].bytes = msgs[i].buf;
msg.u.i2c_read.transactions[i].no_stop_bit = !(msgs[i].flags & I2C_M_STOP);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
}
msg.u.i2c_read.read_i2c_device_id = msgs[num - 1].addr;
msg.u.i2c_read.num_bytes_read = msgs[num - 1].len;
txmsg = kzalloc(sizeof(*txmsg), GFP_KERNEL);
if (!txmsg) {
ret = -ENOMEM;
goto out;
}
txmsg->dst = mstb;
drm_dp_encode_sideband_req(&msg, txmsg);
drm_dp_queue_down_tx(mgr, txmsg);
ret = drm_dp_mst_wait_tx_reply(mstb, txmsg);
if (ret > 0) {
if (txmsg->reply.reply_type == DP_SIDEBAND_REPLY_NAK) {
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
ret = -EREMOTEIO;
goto out;
}
if (txmsg->reply.u.remote_i2c_read_ack.num_bytes != msgs[num - 1].len) {
ret = -EIO;
goto out;
}
memcpy(msgs[num - 1].buf, txmsg->reply.u.remote_i2c_read_ack.bytes, msgs[num - 1].len);
ret = num;
}
out:
kfree(txmsg);
drm_dp_mst_topology_put_mstb(mstb);
drm/helper: add Displayport multi-stream helper (v0.6) This is the initial import of the helper for displayport multistream. It consists of a topology manager, init/destroy/set mst state It supports DP 1.2 MST sideband msg protocol handler - via hpd irqs connector detect and edid retrieval interface. It supports i2c device over DP 1.2 sideband msg protocol (EDID reads only) bandwidth manager API via vcpi allocation and payload updating, along with a helper to check the ACT status. Objects: MST topology manager - one per toplevel MST capable GPU port - not sure if this should be higher level again MST branch unit - one instance per plugged branching unit - one at top of hierarchy - others hanging from ports MST port - one port per port reported by branching units, can have MST units hanging from them as well. Changes since initial posting: a) add a mutex responsbile for the queues, it locks the sideband and msg slots, and msgs to transmit state b) add worker to handle connection state change events, for MST device chaining and hotplug c) add a payload spinlock d) add path sideband msg support e) fixup enum path resources transmit f) reduce max dpcd msg to 16, as per DP1.2 spec. g) separate tx queue kicking from irq processing and move irq acking back to drivers. Changes since v0.2: a) reorganise code, b) drop ACT forcing code c) add connector naming interface using path property d) add topology dumper helper e) proper reference counting and lookup for ports and mstbs. f) move tx kicking into a workq g) add aux locking - this should be redone h) split teardown into two parts i) start working on documentation on interface. Changes since v0.3: a) vc payload locking and tracking fixes b) add hotplug callback into driver - replaces crazy return 1 scheme c) txmsg + mst branch device refcount fixes d) don't bail on mst shutdown if device is gone e) change irq handler to take all 4 bytes of SINK_COUNT + ESI vectors f) make DP payload updates timeout longer - observed on docking station redock g) add more info to debugfs dumper Changes since v0.4: a) suspend/resume support b) more debugging in debugfs Changes since v0.5: a) use byte * to avoid unnecessary stack usage b) fix num_sdp_streams interpretation. c) init payload state for unplug events d) remove lenovo dock sink count hack e) drop aux lock - post rebase f) call hotplug on port destroy TODO: misc features Reviewed-by: Todd Previte <tprevite@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2014-06-05 11:01:32 +07:00
return ret;
}
static u32 drm_dp_mst_i2c_functionality(struct i2c_adapter *adapter)
{
return I2C_FUNC_I2C | I2C_FUNC_SMBUS_EMUL |
I2C_FUNC_SMBUS_READ_BLOCK_DATA |
I2C_FUNC_SMBUS_BLOCK_PROC_CALL |
I2C_FUNC_10BIT_ADDR;
}
static const struct i2c_algorithm drm_dp_mst_i2c_algo = {
.functionality = drm_dp_mst_i2c_functionality,
.master_xfer = drm_dp_mst_i2c_xfer,
};
/**
* drm_dp_mst_register_i2c_bus() - register an I2C adapter for I2C-over-AUX
* @aux: DisplayPort AUX channel
*
* Returns 0 on success or a negative error code on failure.
*/
static int drm_dp_mst_register_i2c_bus(struct drm_dp_aux *aux)
{
aux->ddc.algo = &drm_dp_mst_i2c_algo;
aux->ddc.algo_data = aux;
aux->ddc.retries = 3;
aux->ddc.class = I2C_CLASS_DDC;
aux->ddc.owner = THIS_MODULE;
aux->ddc.dev.parent = aux->dev;
aux->ddc.dev.of_node = aux->dev->of_node;
strlcpy(aux->ddc.name, aux->name ? aux->name : dev_name(aux->dev),
sizeof(aux->ddc.name));
return i2c_add_adapter(&aux->ddc);
}
/**
* drm_dp_mst_unregister_i2c_bus() - unregister an I2C-over-AUX adapter
* @aux: DisplayPort AUX channel
*/
static void drm_dp_mst_unregister_i2c_bus(struct drm_dp_aux *aux)
{
i2c_del_adapter(&aux->ddc);
}