Back in 2013, runtime PM for GPUs with integrated HDA controller was
introduced with commits 0d69704ae3 ("gpu/vga_switcheroo: add driver
control power feature. (v3)") and 246efa4a07 ("snd/hda: add runtime
suspend/resume on optimus support (v4)").
Briefly, the idea was that the HDA controller is forced on and off in
unison with the GPU.
The original code is mostly still in place even though it was never a
100% perfect solution: E.g. on access to the HDA controller, the GPU
is powered up via vga_switcheroo_runtime_resume_hdmi_audio() but there
are no provisions to keep it resumed until access to the HDA controller
has ceased: The GPU autosuspends after 5 seconds, rendering the HDA
controller inaccessible.
Additionally, a kludge is required when hda_intel.c probes: It has to
check whether the GPU is powered down (check_hdmi_disabled()) and defer
probing if so.
However in the meantime (in v4.10) the driver core has gained a feature
called device links which promises to solve such issues in a clean way:
It allows us to declare a dependency from the HDA controller (consumer)
to the GPU (supplier). The PM core then automagically ensures that the
GPU is runtime resumed as long as the HDA controller's ->probe hook is
executed and whenever the HDA controller is accessed.
By default, the HDA controller has a dependency on its parent, a PCIe
Root Port. Adding a device link creates another dependency on its
sibling:
PCIe Root Port
^ ^
| |
| |
HDA ===> GPU
The device link is not only used for runtime PM, it also guarantees that
on system sleep, the HDA controller suspends before the GPU and resumes
after the GPU, and on system shutdown the HDA controller's ->shutdown
hook is executed before the one of the GPU. It is a complete solution.
Using this functionality is as simple as calling device_link_add(),
which results in a dmesg entry like this:
pci 0000:01:00.1: Linked as a consumer to 0000:01:00.0
The code for the GPU-governed audio power management can thus be removed
(except where it's still needed for legacy manual power control).
The device link is added in a PCI quirk rather than in hda_intel.c.
It is therefore legal for the GPU to runtime suspend to D3cold even if
the HDA controller is not bound to a driver or if CONFIG_SND_HDA_INTEL
is not enabled, for accesses to the HDA controller will cause the GPU to
wake up regardless if they're occurring outside of hda_intel.c (think
config space readout via sysfs).
Contrary to the previous implementation, the HDA controller's power
state is now self-governed, rather than GPU-governed, whereas the GPU's
power state is no longer fully self-governed. (The HDA controller needs
to runtime suspend before the GPU can.)
It is thus crucial that runtime PM is always activated on the HDA
controller even if CONFIG_SND_HDA_POWER_SAVE_DEFAULT is set to 0 (which
is the default), lest the GPU stays awake. This is achieved by setting
the auto_runtime_pm flag on every codec and the AZX_DCAPS_PM_RUNTIME
flag on the HDA controller.
A side effect is that power consumption might be reduced if the GPU is
in use but the HDA controller is not, because the HDA controller is now
allowed to go to D3hot. Before, it was forced to stay in D0 as long as
the GPU was in use. (There is no reduction in power consumption on my
Nvidia GK107, but there might be on other chips.)
The code paths for legacy manual power control are adjusted such that
runtime PM is disabled during power off, thereby preventing the PM core
from resuming the HDA controller.
Note that the device link is not only added on vga_switcheroo capable
systems, but for *any* GPU with integrated HDA controller. The idea is
that the HDA controller streams audio via connectors located on the GPU,
so the GPU needs to be on for the HDA controller to do anything useful.
This commit implicitly fixes an unbalanced runtime PM ref upon unbind of
hda_intel.c: On ->probe, a runtime PM ref was previously released under
the condition "azx_has_pm_runtime(chip) || hda->use_vga_switcheroo", but
on ->remove a runtime PM ref was only acquired under the first of those
conditions. Thus, binding and unbinding the driver twice on a
vga_switcheroo capable system caused the runtime PM refcount to drop
below zero. The issue is resolved because the AZX_DCAPS_PM_RUNTIME flag
is now always set if use_vga_switcheroo is true.
For more information on device links please refer to:
https://www.kernel.org/doc/html/latest/driver-api/device_link.html
Documentation/driver-api/device_link.rst
Cc: Dave Airlie <airlied@redhat.com>
Cc: Ben Skeggs <bskeggs@redhat.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Takashi Iwai <tiwai@suse.de>
Reviewed-by: Peter Wu <peter@lekensteyn.nl>
Tested-by: Kai Heng Feng <kai.heng.feng@canonical.com> # AMD PowerXpress
Tested-by: Mike Lothian <mike@fireburn.co.uk> # AMD PowerXpress
Tested-by: Denis Lisov <dennis.lissov@gmail.com> # Nvidia Optimus
Tested-by: Peter Wu <peter@lekensteyn.nl> # Nvidia Optimus
Tested-by: Lukas Wunner <lukas@wunner.de> # MacBook Pro
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Link: https://patchwork.freedesktop.org/patch/msgid/51bd38360ff502a8c42b1ebf4405ee1d3f27118d.1520068884.git.lukas@wunner.de
So far we've got one condition when DRM drivers need to defer probing
on a dual GPU system and it's coded separately into each of the relevant
drivers. As suggested by Daniel Vetter, deduplicate that code in the
drivers and move it to a new vga_switcheroo helper. This yields better
encapsulation of concepts and lets us add further checks in a central
place. (The existing check pertains to pre-retina MacBook Pros and an
additional check is expected to be needed for retinas.)
One might be tempted to check deferred probing conditions in
vga_switcheroo_register_client(), but this is usually called fairly late
during driver load. The GPU is fully brought up and ready for switching
at that point. On boot the ->probe hook is potentially called dozens of
times until it finally succeeds, and each time we'd repeat bringup and
teardown of the GPU, lengthening boot time considerably and cluttering
logfiles. A separate helper is therefore needed which can be called
right at the beginning of the ->probe hook.
Note that amdgpu currently does not call this helper as the AMD GPUs
built into MacBook Pros are only supported by radeon so far.
v2: This helper could eventually be used by audio clients as well,
so rephrase kerneldoc to refer to "client" instead of "GPU"
and move the single existing check in an if block specific
to PCI_CLASS_DISPLAY_VGA devices. Move documentation on
that check from kerneldoc to a comment. (Daniel Vetter)
v3: Mandate in kerneldoc that registration of client shall only
happen after calling this helper. (Daniel Vetter)
v4: Rebase on 412c8f7de0 ("drm/radeon: Return -EPROBE_DEFER when
amdkfd not loaded")
v5: Some Optimus GPUs use PCI_CLASS_DISPLAY_3D, make sure those are
matched as well. (Emil Velikov)
v6: The if-condition referring to PCI_BASE_CLASS_DISPLAY may be
considered a functional change. Move to a separate commit to
keep this a pure refactoring change. (Emil Velikov, Jani Nikula)
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Ben Skeggs <bskeggs@redhat.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/575885fd440c2b13c3f19ddf44360cfbbff35f50.1464685538.git.lukas@wunner.de
Originally by Seth Forshee <seth.forshee@canonical.com>, 2012-10-04:
During graphics driver initialization it's useful to be able to mux
only the DDC to the inactive client in order to read the EDID. Add
a switch_ddc callback to allow capable handlers to provide this
functionality, and add vga_switcheroo_switch_ddc() to allow DRM
to mux only the DDC.
Modified by Dave Airlie <airlied@gmail.com>, 2012-12-22:
I can't figure out why I didn't like this, but I rewrote this [...]
to lock/unlock the ddc lines [...]. I think I'd prefer something
like that otherwise the interface got really ugly.
Modified by Lukas Wunner <lukas@wunner.de>, 2015-04 - 2015-10:
Change semantics of ->switch_ddc handler callback to return previous
DDC owner. Original version tried to determine previous DDC owner
with find_active_client() but this fails if the inactive client
registers before the active client.
Don't lock vgasr_mutex in _lock_ddc() / _unlock_ddc(), it can cause
deadlocks because (a) during switch (with vgasr_mutex already held),
GPU is woken and probes its outputs, tries to re-acquire vgasr_mutex
to lock DDC lines; (b) Likewise during switch, GPU is suspended and
calls cancel_delayed_work_sync() to stop output polling, if poll
task is running at this moment we may wait forever for it to finish.
Instead, lock mux_hw_lock when unregistering the handler because
the only reason why we'd want to lock vgasr_mutex in _lock_ddc() /
_unlock_ddc() is to block the handler from disappearing while DDC
lines are switched.
Also acquire mux_hw_lock in stage2 to avoid race condition where
reading the EDID and switching happens simultaneously. Likewise on
MIGD / MDIS commands and on runtime suspend.
v2.1: Overhaul locking, squash commits (Daniel Vetter)
v2.2: Readability improvements (Thierry Reding)
v2.3: Overhaul locking once more
v2.4: Retain semantics of ->switchto handler callback to switch all
pins, including DDC (Daniel Vetter)
v5: Rename ddc_lock to mux_hw_lock: Since we acquire this both
when calling ->switch_ddc and ->switchto, it protects not just
access to the DDC lines but to the mux in general. This is in
line with the DRM convention to use low-level locks to avoid
concurrent hw access (e.g. i2c, dp_aux) which are often called
hw_lock (Daniel Vetter)
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=88861
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=61115
Tested-by: Lukas Wunner <lukas@wunner.de>
[MBP 9,1 2012 intel IVB + nvidia GK107 pre-retina 15"]
Cc: Seth Forshee <seth.forshee@canonical.com>
Cc: Dave Airlie <airlied@gmail.com>
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/e81ae9722b84c5ed591805fee3ea6dbf5dc6c4b3.1452525860.git.lukas@wunner.de
Allow handlers to declare their capabilities and allow clients to
obtain that information. So far we have these use cases:
* If the handler is able to switch DDC separately, clients need to
probe EDID with drm_get_edid_switcheroo(). We should allow them
to detect a capable handler to ensure this function only gets
called when needed.
* Likewise if the handler is unable to switch AUX separately, the active
client needs to communicate link training parameters to the inactive
client, which may then skip the AUX handshake and set up its output
with these pre-calibrated values (DisplayPort specification v1.1a,
section 2.5.3.3). Clients need a way to recognize such a situation.
The flags for the radeon_atpx_handler and amdgpu_atpx_handler are
initially set to 0, this can later on be amended with
handler_flags |= VGA_SWITCHEROO_CAN_SWITCH_DDC;
when a ->switch_ddc callback is added.
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=88861
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=61115
Tested-by: Lukas Wunner <lukas@wunner.de>
[MBP 9,1 2012 intel IVB + nvidia GK107 pre-retina 15"]
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Reviewed-by: Darren Hart <dvhart@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/2b0d93ed6e511ca09e95e45e0b35627f330fabce.1452525860.git.lukas@wunner.de
hda_intel.c:azx_probe() defers initialization of an audio controller
on the discrete GPU if the GPU is powered off. The power state of the
GPU is determined by calling vga_switcheroo_get_client_state().
vga_switcheroo_get_client_state() returns VGA_SWITCHEROO_INIT if
vga_switcheroo is not enabled, i.e. if no second GPU or no handler
has registered.
This can go wrong in the following scenario:
- Driver for the integrated GPU is not loaded.
- Driver for the discrete GPU registers with vga_switcheroo, uses driver
power control to power down the GPU, handler cuts power to the GPU.
- Driver for the audio controller gets loaded after the GPU was powered
down, calls vga_switcheroo_get_client_state() which returns
VGA_SWITCHEROO_INIT instead of VGA_SWITCHEROO_OFF.
- Consequence: azx_probe() tries to initialize the audio controller even
though the GPU is powered down.
The power state VGA_SWITCHEROO_INIT was introduced by c8e9cf7bb2
("vga_switcheroo: Add a helper function to get the client state").
It is not apparent what its benefit might be. The idea seems to
be to initialize the audio controller even if the power state is
VGA_SWITCHEROO_OFF (were vga_switcheroo enabled), but as shown
above this can fail.
Drop VGA_SWITCHEROO_INIT to solve this.
Acked-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Signed-off-by: Dave Airlie <airlied@redhat.com>
The active attribute in struct vga_switcheroo_client denotes whether
the outputs are currently switched to this client. The attribute is
only meaningful for vga clients. It is never used for audio clients.
The function vga_switcheroo_register_audio_client() misuses this
attribute to store whether the audio device is fully initialized.
Most likely there was a misunderstanding about the meaning of
"active" when this was added.
Comment from Takashi's review:
"Not really. The full initialization of audio was meant that the audio
is active indeed. Admittedly, though, the active flag for each audio
client doesn't play any role because the audio always follows the gfx
state changes, and the value passed there doesn't reflect the actual
state due to the later change. So, I agree with the removal of the
flag itself -- or let the audio active flag following the
corresponding gfx flag. The latter will make the proc output more
consistent while the former is certainly more reduction of code."
Set the active attribute to false for audio clients. Remove the
active parameter from vga_switcheroo_register_audio_client() and
its sole caller, hda_intel.c:register_vga_switcheroo().
vga_switcheroo_register_audio_client() was introduced by 3e9e63dbd3
("vga_switcheroo: Add the support for audio clients"). Its use in
hda_intel.c was introduced by a82d51ed24 ("ALSA: hda - Support
VGA-switcheroo").
v1.1: The changes above imply that in find_active_client() the call
to client_is_vga() is now superfluous. Drop it.
Cc: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Lukas Wunner <lukas@wunner.de>
[danvet: Add Takashi's clarification to the commit message.]
Reviewed-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
This adds an "Overview" DOC section plus two DOC sections for the modes
of use ("Manual switching and manual power control" and "Driver power
control").
Also included is kernel-doc for all public functions, structs and enums.
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Drivers should call this on unload to unregister pmops.
Bug:
https://bugzilla.kernel.org/show_bug.cgi?id=84431
Reviewed-by: Ben Skeggs <bskeggs@redhat.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Signed-off-by: Pali Rohár <pali.rohar@gmail.com>
Cc: Ben Skeggs <bskeggs@redhat.com>
For optimus and powerxpress muxless we really want the GPU
driver deciding when to power up/down the GPU, not userspace.
This adds the ability for a driver to dynamically power up/down
the GPU and remove the switcheroo from controlling it, the
switcheroo reports the dynamic state to userspace also.
It also adds 2 power domains, one for machine where the power
switch is controlled outside the GPU D3 state, so the powerdown
ordering is done correctly, and the second for the hdmi audio
device to make sure it can resume for PCI config space accesses.
v1.1: fix build with switcheroo off
v2: add power domain support for radeon and v1 nvidia dsms
v2.1: fix typo in off case
v3: add audio power domain for hdmi audio + misc audio fixes
v4: use PCI_SLOT macro, drop power reference on hdmi audio resume
failure also.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Fix warnings on some architectures/configs (not on x86):
include/linux/vga_switcheroo.h:28:30: warning: 'struct pci_dev' declared inside parameter list [enabled by default]
include/linux/vga_switcheroo.h:28:30: warning: its scope is only this definition or declaration, which is probably not what you want [enabled by default]
Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Cc: Takashi Iwai <tiwai@suse.de>
Reported-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Add vga_switcheroo_get_client_state() to get the current state of the
client. This is necessary to determine the proper initial state of
audio clients in HD-audio driver.
Acked-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Add the support for audio clients to VGA-switcheroo for handling the
HDMI audio controller together with VGA switching. The id of the
audio controller should be given explicitly at registration time
unlike the video controller.
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=43155
Signed-off-by: Takashi Iwai <tiwai@suse.de>
This changes the API as a clean-up. Instead of passing multiple
function pointers at each time, introduce a new struct holding the
whole callback functions and pass it to the registration.
The same struct will be used for the upcoming audio client
registration, too.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
radeon was always including the atpx code unnecessarily, also core
switcheroo was including acpi headers.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Many new laptops now come with 2 gpus, one to be used for low power
modes and one for gaming/on-ac applications. These GPUs are typically
wired to the laptop panel and VGA ports via a multiplexer unit which
is controlled via ACPI methods.
4 combinations of systems typically exist - with 2 ACPI methods.
Intel/ATI - Lenovo W500/T500 - use ATPX ACPI method
ATI/ATI - some ASUS - use ATPX ACPI Method
Intel/Nvidia - - use _DSM ACPI method
Nvidia/Nvidia - - use _DSM ACPI method.
TODO:
This patch adds support for the ATPX method and initial bits
for the _DSM methods that need to written by someone with
access to the hardware.
Add a proper non-debugfs interface - need to get some proper
testing first.
v2: add power up/down support for both devices
on W500 puts i915/radeon into D3 and cuts power to radeon.
v3: redo probing methods, no DMI list, drm devices call to
register with switcheroo, it tries to find an ATPX method on
any device and once there is two devices + ATPX it inits the
switcher.
v4: ATPX msg handling using buffers - should work on more machines
v5: rearchitect after more mjg59 discussion - move ATPX handling to
radeon driver.
v6: add file headers + initial nouveau bits (to be filled out).
v7: merge delayed switcher code.
v8: avoid suspend/resume of gpu that is off
v9: rearchitect - mjg59 is always right. - move all ATPX code to
radeon, should allow simpler DSM also proper ATRM handling
v10: add ATRM support for radeon BIOS, add mutex to lock vgasr_priv
v11: fix bug in resuming Intel for 2nd time.
v12: start fixing up nvidia code blindly.
v13: blindly guess at finishing nvidia code
v14: remove radeon audio hacks - fix up intel resume more like upstream
v15: clean up printks + remove unnecessary igd/dis pointers
mount debugfs
/sys/kernel/debug/vgaswitcheroo/switch - should exist if ATPX detected
+ 2 cards.
DIS - immediate change to discrete
IGD - immediate change to IGD
DDIS - delayed change to discrete
DIGD - delayed change to IGD
ON - turn on not in use
OFF - turn off not in use
Tested on W500 (Intel/ATI) and T500 (Intel/ATI)
Signed-off-by: Dave Airlie <airlied@redhat.com>
