So the problem with async commit (especially async modeset commit) is
that the legacy pointers only get updated after the point of no
return, in the async part of the modeset sequence. At least as
implemented by the current helper functions. This is done in the
set_routing_links function in drm_atomic_helper.c.
Which also means that access isn't protected by locks but only
coordinated by synchronizing with async workers. No problem thus far,
until we lock at the getconnector/encoder ioctls.
So fix this up by adding special cases for atomic drivers: For those
we need to look at state objects. Unfortunately digging out the
correct encoder->crtc link is a bit of work, so wrap this up in a
helper function.
Moving the assignments of connector->encoder and encoder->crtc earlier
isn't a good idea because the point of the atomic helpers is that we
stage the state updates. That way the disable functions can still
inspect the links and rely upon them.
v2: Extract full encoder->crtc lookup into helper (Rob).
v3: Extract drm_connector_get_encoder too since - we need to always
return state->best_encoder when there is a state otherwise we might
return stale data if there's a pending async disable (and chase
unlocked pointers, too). Same issue with encoder_get_crtc but there
it's a bit more tricky to handle.
Cc: Rob Clark <robdclark@gmail.com>
Cc: Sean Paul <seanpaul@chromium.org>
Reviewed-by: Sean Paul <seanpaul@chromium.org>
Lightly-Tested-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Chasing plane->state->crtc of planes that are *not* part of the same
atomic update is racy, making it incredibly awkward (or impossible) to
do something simple like iterate over all planes and figure out which
ones are attached to a crtc.
Solve this by adding a bitmask of currently attached planes in the
crtc-state.
Note that the transitional helpers do not maintain the plane_mask. But
they only support the legacy ioctls, which have sufficient brute-force
locking around plane updates that they can continue to loop over all
planes to see what is attached to a crtc the old way.
Signed-off-by: Rob Clark <robdclark@gmail.com>
[danvet:
- Drop comments about locking in set_crtc_for_plane since they're a
bit misleading - we already should hold lock for the current crtc.
- Also WARN_ON if get_state on the old crtc fails since that should
have been done already.
- Squash in fixup to check get_plane_state return value, reported by
Dan Carpenter and acked by Rob Clark.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The current state of CRTCs, planes and connectors currently leaks during
DRM driver ->unload() unless drivers explicitly clean it up. Since there
is nothing driver-specific about it, that cleanup can be done within the
DRM core.
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
In most situations it will be useful to have the old state passed to the
->atomic_update() callback. For example if a plane is being disabled the
new state's .crtc field will be NULL, but some drivers may rely on this
field to program the CRTCs registers.
v2: rename variable to old_plane_state and remove redundant comment as
suggested by Daniel Vetter, remove an Exynos hunk that doesn't apply to
drm-next and add a hunk for pending MSM mdp5 changes
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The drm core can call the plane disable hook multiple times, which
means it can get called when plane->crtc is already NULL. That in turn
means we can't get at the implicit acquire ctx we use in the atomic
helpers for legacy entries points.
We could try to pass drm_modeset_legacy_acquire_ctx a drm_device
pointer so that it can cope with a NULL crtc. But that still doesn't
work since the cursor ioctls (remapped with the universal cursor plane
support code) only grabs the crtc locks. So the global acquire context
isn't set eitehr.
The real solution here would be to bite the bullet and wire up
explicit acquire context parameters to all relevant functions. We need
to do that anyway (to be able to get rid of some small allocations
which we can't cope with failing). But that's a lot of work and better
done once atomic has settled a bit.
So meanwhile just catch this case in the helper and bail out.
Signed-off-by: Jasper St. Pierre <jstpierre@mecheye.net>
Reviewed-by: Rob Clark <robdclark@gmail.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
[danvet: Completely rewrite commit message and comment but keep
Jasper's logic and author credits since his patch is the only
short-term solution that works.]
Tested-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
I've forgotten to remove that in my per-plane locking patch.
Reported-by: Rob Clark <robdclark@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Especially with legacy cursor ioctls existing userspace assumes that
you can pile up lots of updates in one go. The super-proper way to
support this would be a special commit mode which overwrites the last
update. But getting there will be quite a bit of work.
Meanwhile do what pretty much all the drivers have done for the plane
update functions: Simply skip the vblank wait for the buffer cleanup
if the buffer is the same. Since the universal cursor plane code will
not recreate framebuffers needlessly this allows us to not slow down
legacy pageflip events while someone moves the cursor around.
v2: Drop the async plane update hunk from a previous attempt at this
issue.
v3: Fix up kerneldoc.
v4: Don't oops so badly. Reported by Jasper.
Cc: Rob Clark <robdclark@gmail.com>
Cc: "Jasper St. Pierre" <jstpierre@mecheye.net>
Reviewed-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Jasper St. Pierre <jstpierre@mecheye.net>
Tested-by: Jasper St. Pierre <jstpierre@mecheye.net>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
In disable_outputs() we need to shut down the outgoing encoder, not the
incoming one (we have already swapped-state at this point). Without
this, we end up telling the driver to crtc->dpms(OFF) without first
encoder->dpms(OFF), and that makes some hw quite unhappy.
v2: missing WARN_ON() hunk and comment
Reviewed-by: Daniel Vetter <daniel.vetter@intel.com>
Signed-off-by: Rob Clark <robdclark@gmail.com>
The intention is that drivers can set crtc_state->mode_changed in their
atomic_check() fxns if they encounter a scenario that requires full
modeset.
Signed-off-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
On mdp4 there is a single global LAYERMIXER_IN_CFG register. The
previous logic to share that between multiple crtcs didn't actually
handle plane-disable very well. Easier just to look at all of the
crtcs each time.
Signed-off-by: Rob Clark <robdclark@gmail.com>
When disabling the interface (INTF), the change doesn't latch until next
vblank, so we need to wait for vblank.
Also, to be pedantic, in the crtc, set all the mixer stages to unused.
It shouldn't really matter, since at this point we have already disabled
the INTF and waited for necessary vblank.
Signed-off-by: Rob Clark <robdclark@gmail.com>
Convert mdp5 over to atomic helpers. Extend/wrap drm_plane_state to
track plane zpos and to keep track of the needed when applying the
atomic update. In mdp5's plane->atomic_check() we also need to check
for updates which require SMP reallocation, in order to trigger full
modeset.
Signed-off-by: Rob Clark <robdclark@gmail.com>
For example, use 'struct mdp5_smp *' everywhere instead of 'void *', but
only declare it as 'struct mdp5_smp;' in common headers, so the struct
body is still private. The accomplishes the desired modularity while
still letting the compiler provide some type checking for us.
Signed-off-by: Rob Clark <robdclark@gmail.com>
MDP5 currently support one single CRTC with its private pipe.
This change allows the configuration of multiple CRTCs with
the possibility to attach several public planes to these CRTCs.
Signed-off-by: Stephane Viau <sviau@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
The hardware configuration modification from a version to another
is quite consequent. Introducing a configuration module
(mdp5_cfg) may make things more clear and easier to access when a
new hardware version comes up.
Signed-off-by: Stephane Viau <sviau@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
The Shared Memory Pool (SMP) has its own limitation, features and
state. Some examples are:
- the number of Memory Macro Block (MMB) and their size
- the number of lines that can be fetched
- the state of MMB currently allocated
- the computation of number of blocks required per plane
- client IDs ...
In order to avoid private data to be overwritten by other modules,
let's make these private to the SMP module.
Some of these depend on the hardware configuration, let's add them
to the mdp5_config struct.
In some hw configurations, some MMBs are statically tied to RGB
pipes and cannot be re-allocated dynamically. This change
introduces the concept of MMB static usage and makes sure that
dynamic MMB requests are dimensioned accordingly.
A note on passing a pipe pointer, instead of client IDs:
Client IDs are SMP-related information. Passing PIPE information
to SMP lets SMP module to find out which SMP client(s) are used.
This allows the SMP module to access the PIPE pointer, which can
be used for FIFO watermark configuration.
By the way, even though REG_MDP5_PIPE_REQPRIO_FIFO_WM_* registers
are part of the PIPE registers, their functionality is to reflect
the behavior of the SMP block. These registers access is now
restricted to the SMP module.
Signed-off-by: Stephane Viau <sviau@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
A left-over from prior to component framework. The original intent was
to deal with hdmi getting unloaded before the master component, but that
isn't really going to work anyways. These days with the component
framework taking care to unload the master component first, we don't
have to worry about this.
Signed-off-by: Rob Clark <robdclark@gmail.com>
The core clock rate depends on the hw configuration. Once we have
read the hardware revision, we can set the core clock to its
maximum value.
Before then, the clock is set at a rate supported by all MDP5
revisions.
Signed-off-by: Stephane Viau <sviau@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
For mdp5, the irqs of hdmi/eDP/dsi0/dsi1 blocks get routed through the
mdp block. In order to decouple hdmi/eDP/etc, register an irq domain
in mdp5. When hdmi/dsi/etc are used with mdp4, they can directly setup
their irqs in their DT nodes as normal. When used with mdp5, instead
set the mdp device as the interrupt-parent, as in:
mdp: qcom,mdss_mdp@fd900000 {
compatible = "qcom,mdss_mdp";
interrupt-controller;
#interrupt-cells = <1>;
...
};
hdmi: qcom,hdmi_tx@fd922100 {
compatible = "qcom,hdmi-tx-8074";
interrupt-parent = <&mdp>;
interrupts = <8 0>; /* MDP5_HW_INTR_STATUS.INTR_HDMI */
...
};
There is a slight awkwardness, in that we cannot disable child irqs
at the mdp level, they can only be cleared in the child block. So
you must not use threaded irq handlers in the child. I'm not sure
if there is a better way to deal with that.
Signed-off-by: Rob Clark <robdclark@gmail.com>
- More CI dpm fixes
- Initial DPM fan control for SI/CI (disabled by default)
- GPUVM multi-ring efficiency improvements
- Some cursor fixes
* 'drm-next-3.19' of git://people.freedesktop.org/~agd5f/linux: (22 commits)
drm/radeon: update the VM after setting BO address v4
drm/radeon: sync PT updates as shared v2
drm/radeon: sync PD updates as shared
drm/radeon: fence BO_VAs manually
drm/radeon: use one VMID for each ring
drm/radeon: track VM update fences separately
drm/radeon: fence PT updates manually v2
drm/radeon: split semaphore and sync object handling v2
drm/radeon: remove unnecessary VM syncs
drm/radeon: stop re-reserving the BO in radeon_vm_bo_set_addr
drm/radeon: rework vm_flush parameters
drm/radeon/ci: disable needless sclk changes
drm/radeon/ci: force pcie level before sclk and mclk
drm/radeon/ci: use different smc command for pcie dpm
drm/radeon/ci: apply disp voltage changes before clk changes
drm/radeon: fix PCC debugging message for CI DPM
drm/radeon/dpm: add thermal dpm support for CI
drm/radeon/dpm: add smc fan control for CI (v2)
drm/radeon/dpm: add smc fan control for SI (v2)
drm/radeon: work around a hw bug in MGCG on CIK
...
The vfree() function performes also input parameter validation. Thus the test
around the call is not needed.
This issue was detected by using the Coccinelle software.
Signed-off-by: Markus Elfring <elfring@users.sourceforge.net>
Reviewed-by: Thierry Reding <thierry.reding@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
The vunmap() function performes also input parameter validation. Thus the test
around the call is not needed.
This issue was detected by using the Coccinelle software.
Signed-off-by: Markus Elfring <elfring@users.sourceforge.net>
Reviewed-by: Thierry Reding <thierry.reding@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
The drm_fbdev_cma_hotplug_event() function tests whether its argument is NULL
and then returns immediately. Thus the test around the call is not needed.
This issue was detected by using the Coccinelle software.
Signed-off-by: Markus Elfring <elfring@users.sourceforge.net>
Reviewed-by: Thierry Reding <thierry.reding@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
The release_firmware() function tests whether its argument is NULL and then
returns immediately. Thus the test around the call is not needed.
This issue was detected by using the Coccinelle software.
Signed-off-by: Markus Elfring <elfring@users.sourceforge.net>
Reviewed-by: Thierry Reding <thierry.reding@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
This is an oversight from
commit f52b69f1ec
Author: Daniel Vetter <daniel.vetter@ffwll.ch>
Date: Wed Nov 19 18:38:08 2014 +0100
drm/atomic: Don't overrun the connector array when hotplugging
Cc: Dave Airlie <airlied@redhat.com>
Cc: Rob Clark <robdclark@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Reviewed-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
It happens on occasion that developers of generic user-space applications
abuse the dumb buffer API to get hold of drm buffers that they can both
mmap() and use for GPU acceleration, using the assumptions that dumb buffers
and buffers available for GPU are
a) The same type and can be aribtrarily type-casted.
b) fully coherent.
This patch makes the most widely used drivers warn nicely when that happens,
the next step will be to fail.
v2: Move drmP.h changes to drm_gem.h. Fix Radeon dumb mmap breakage.
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
This way the necessary VM update is kicked off immediately
if all BOs involved are in GPU accessible memory.
v2: fix vm lock
v3: immediately update unmaps as well
v4: use drm_free_large instead of kfree
Tested-by: Kai Wasserbäch <kai@dev.carbon-project.org>
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Only invalidating PTEs needs to be executed synchronized to using the PT.
v2: fix sync to uses
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
We never invalidate PD entries and making them valid can
run with other users in parallel.
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This allows us to finally remove the VM fence and
so allow concurrent use of it from different engines.
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Use multiple VMIDs for each VM, one for each ring. That allows
us to execute flushes separately on each ring, still not ideal
cause in a lot of cases rings can share IDs.
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Note for each fence if it's a VM page table update or not. This allows
us to determine the last VM update in a sync object and so to figure
out if we need to flush the TLB or not.
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This allows us to add the real execution fence as shared.
v2: fix typo
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Previously we just allocated space for four hardware semaphores
in each software semaphore object. Make software semaphore objects
represent only one hardware semaphore address again by splitting
the sync code into it's own object.
v2: fix typo in comment
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The PD/PTs reservation object now contains everything needed.
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
That's useless when all callers drop the reservation
immediately after calling the function.
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Use ring structure instead of index and provide vm_id and pd_addr separately.
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The current code always reprogrammed the sclk levels,
but we don't currently handle disp sclk requirements
so just skip it.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Enable smc fan control for CI boards. Should
reduce the fan noise on systems with a higher
default fan profile.
v2: disable by default, add additional fan setup, rpm control
bug:
https://bugs.freedesktop.org/show_bug.cgi?id=73338
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
