- VGA switcheroo was broken for some users as a result of the ACPI-based
PCI hotplug (ACPIPHP) changes in 3.12, because some previously ignored
hotplug events started to be handled. The fix causes them to be
ignored again.
- There are two more issues related to cpufreq's suspend/resume handling
changes from the 3.12 cycle addressed by Viresh Kumar's fixes.
- intel_pstate triggers a divide error in a timer function if the P-state
information it needs is missing during initialization. This leads to
kernel panics on nested KVM clients and is fixed by failing the
initialization cleanly in those cases.
- PCI initalization code changes during the 3.9 cycle uncovered BIOS
issues related to ACPI wakeup notifications (some BIOSes send them
for devices that aren't supposed to support ACPI wakeup). Work around
them by installing an ACPI wakeup notify handler for all PCI devices
with ACPI support.
- The Calxeda cpuilde driver's probe function is tagged as __init, which
is incorrect and causes a section mismatch to occur during build. Fix
from Andre Przywara removes the __init tag from there.
- During the 3.12 cycle ACPIPHP started to print warnings about missing
_ADR for devices that legitimately don't have it. Fix from Toshi Kani
makes it only print the warnings where they make sense.
/
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.19 (GNU/Linux)
iQIcBAABCAAGBQJSxrJgAAoJEILEb/54YlRxRPkP/ifzrrVhdzqXIEy44b93JeDx
oSmZW6yTO51GZlDx2bjt6CGJcIUDC4ExYV6S2tB44/DL19CYdIxi7oBaXtUvzGRs
oZ6B1wfvKOIxZ0RQguaGd1uerQU304CGwUXu/jpRZ/UuZZFKq5Uts6O3bilGzCfR
Y+MUH+qECwdBXaFHUISdWFsa3lxj0U0kglszh+DsxwS4gy/pLbCu5fKLgHLuNNQC
hhEEToQ6uF4o8hbkGJvgUPo3V3aUSXObgvJh4ntP09YE1AEJScLB4wKmqL0zN8Qj
pbBf1WC5OpGXv8zGM9ErrY64YaKA36uhJvOi6RtBGLbG+pYM6E6IM9zNf4Ku+T79
JNEulpq27aEx2JghNSgMFYQZEOGTH+q24iXZdZlOIvqWpMymATlqP/gAQQIpg3VC
OIIdocMFRsbgwFXf41uyUqs458fg5xREz5k6geWZeyriM45wFShR+JnMopQWc5OB
a3sbcWUShFBL1T0pqYR4SDLDvH4NdEP2NKO2jlqMXUewLXsVRRt/42etGoe0rI3C
cMWPQq7z0GNN+NboUviqwHdxUKqONWGt+pd/3u8FI/Y1IlXEeXQYGawhSu81uCpT
5gLaKDkwOrCSwOw68Msuod0Cce6TnoTowi6hP2aAEu8mDJwQY+toqA3+CPoO8nty
DdhZjP1afEgsVVyjErX4
=LXh0
-----END PGP SIGNATURE-----
Merge tag 'pm+acpi-3.13-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm
Pull ACPI and PM fixes and new device IDs from Rafael Wysocki:
"These commits, except for one, are regression fixes and the remaining
one fixes a divide error leading to a kernel panic. The majority of
the regressions fixed here were introduced during the 3.12 cycle, one
of them is from this cycle and one is older.
Specifics:
- VGA switcheroo was broken for some users as a result of the
ACPI-based PCI hotplug (ACPIPHP) changes in 3.12, because some
previously ignored hotplug events started to be handled. The fix
causes them to be ignored again.
- There are two more issues related to cpufreq's suspend/resume
handling changes from the 3.12 cycle addressed by Viresh Kumar's
fixes.
- intel_pstate triggers a divide error in a timer function if the
P-state information it needs is missing during initialization.
This leads to kernel panics on nested KVM clients and is fixed by
failing the initialization cleanly in those cases.
- PCI initalization code changes during the 3.9 cycle uncovered BIOS
issues related to ACPI wakeup notifications (some BIOSes send them
for devices that aren't supposed to support ACPI wakeup). Work
around them by installing an ACPI wakeup notify handler for all PCI
devices with ACPI support.
- The Calxeda cpuilde driver's probe function is tagged as __init,
which is incorrect and causes a section mismatch to occur during
build. Fix from Andre Przywara removes the __init tag from there.
- During the 3.12 cycle ACPIPHP started to print warnings about
missing _ADR for devices that legitimately don't have it. Fix from
Toshi Kani makes it only print the warnings where they make sense"
* tag 'pm+acpi-3.13-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
ACPIPHP / radeon / nouveau: Fix VGA switcheroo problem related to hotplug
intel_pstate: Fail initialization if P-state information is missing
ARM/cpuidle: remove __init tag from Calxeda cpuidle probe function
PCI / ACPI: Install wakeup notify handlers for all PCI devs with ACPI
cpufreq: preserve user_policy across suspend/resume
cpufreq: Clean up after a failing light-weight initialization
ACPI / PCI / hotplug: Avoid warning when _ADR not present
* acpi-pci-pm:
PCI / ACPI: Install wakeup notify handlers for all PCI devs with ACPI
* acpi-pci-hotplug:
ACPIPHP / radeon / nouveau: Fix VGA switcheroo problem related to hotplug
ACPI / PCI / hotplug: Avoid warning when _ADR not present
The changes in the ACPI-based PCI hotplug (ACPIPHP) subsystem made
during the 3.12 development cycle uncovered a problem with VGA
switcheroo that on some systems, when the device-specific method
(ATPX in the radeon case, _DSM in the nouveau case) is used to turn
off the discrete graphics, the BIOS generates ACPI hotplug events for
that device and those events cause ACPIPHP to attempt to remove the
device from the system (they are events for a device that was present
previously and is not present any more, so that's what should be done
according to the spec). Then, the system stops functioning correctly.
Since the hotplug events in question were simply silently ignored
previously, the least intrusive way to address that problem is to
make ACPIPHP ignore them again. For this purpose, introduce a new
ACPI device flag, no_hotplug, and modify ACPIPHP to ignore hotplug
events for PCI devices whose ACPI companions have that flag set.
Next, make the radeon and nouveau switcheroo detection code set the
no_hotplug flag for the discrete graphics' ACPI companion.
Fixes: bbd34fcdd1 (ACPI / hotplug / PCI: Register all devices under the given bridge)
References: https://bugzilla.kernel.org/show_bug.cgi?id=61891
References: https://bugzilla.kernel.org/show_bug.cgi?id=64891
Reported-and-tested-by: Mike Lothian <mike@fireburn.co.uk>
Reported-and-tested-by: <madcatx@atlas.cz>
Reported-and-tested-by: Joaquín Aramendía <samsagax@gmail.com>
Cc: Alex Deucher <alexdeucher@gmail.com>
Cc: Dave Airlie <airlied@linux.ie>
Cc: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: 3.12+ <stable@vger.kernel.org> # 3.12+
We don't have the NUM_BANKS parameter, so we have to calculate it
from the other parameters. NUM_BANKS is not constant on CIK.
This fixes 2D tiling for the display engine on CIK.
Signed-off-by: Marek Olšák <marek.olsak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This will allow userspace to correctly program the PA_SC_RASTER_CONFIG
register, so it can be considered a fix.
Signed-off-by: Marek Olšák <marek.olsak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Only the render backends of the first shader engine were enabled. The others
were erroneously disabled. Enabling the other render backends improves
performance a lot.
Unigine Sanctuary on Bonaire:
Before: 15 fps
After: 90 fps
Judging from the fan noise, the GPU was also underclocked when the other
render backends were disabled, resulting in horrible performance. The fan is
a lot noisy under load now.
Signed-off-by: Marek Olšák <marek.olsak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Otherwise the kernel might reject our decoding requests.
Signed-off-by: Christian König <christian.koenig@amd.com>
Cc: stable@vger.kernel.org
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Besides the 2 fixes for tricky corner cases in gem from Chris I've
promised already two patche from Paulo to fix pc8 warnings (both ported
from -next, bug report from Dave Jones) and one patch from to fix vga
enable/disable on snb+. That one is a really old bug, but apparently it
can cause machine hangs if you try hard enough with vgacon/efifb handover.
* tag 'drm-intel-fixes-2013-12-18' of git://people.freedesktop.org/~danvet/drm-intel:
drm/i915: Use the correct GMCH_CTRL register for Sandybridge+
drm/i915: get a PC8 reference when enabling the power well
drm/i915: change CRTC assertion on LCPLL disable
drm/i915: Fix erroneous dereference of batch_obj inside reset_status
drm/i915: Prevent double unref following alloc failure during execbuffer
- fix for a long standing corruption bug on some Trinity/Richland parts.
- Stability fix for cayman dpm
- audio fixes for dce6+
* 'drm-fixes-3.13' of git://people.freedesktop.org/~agd5f/linux:
drm/radeon: fix asic gfx values for scrapper asics
drm/radeon: check for 0 count in speaker allocation and SAD code
drm/radeon/dpm: disable ss on Cayman
drm/radeon/dce6: set correct number of audio pins
Commit "drm/ttm: Don't move non-existing data" didn't take the
swapped-out corner case into account. This patch corrects that.
Fixes blank screen after attempted suspend / hibernate on vmwgfx.
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Fix build error: qxl uses crc32 functions so it needs to select
CRC32.
Also use angle quotes around a kernel header file name.
drivers/built-in.o: In function `qxl_display_read_client_monitors_config':
(.text+0x19d754): undefined reference to `crc32_le'
Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
The GMCH_CTRL register (or MGCC in the spec) is at a different address
on Sandybridge, and the address to which we currently write to is
undefined. These stray writes appear to upset (hard hang) my Ivybridge
machine whilst it is in UEFI mode.
Note that the register is still marked as locked RO on Sandybridge, so
vgaarb is still dysfunctional.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: stable@vger.kernel.org
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Spread spectrum seems to cause hangs when dynamic clock
switching is enabled. Disable it for now. This does not
affect performance or the amount of power saved. Tracked
down by Martin Andersson.
bug:
https://bugs.freedesktop.org/show_bug.cgi?id=69723
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
This bug in EDID was exposed by:
commit eccea7920c
Author: Alex Deucher <alexander.deucher@amd.com>
Date: Mon Mar 26 15:12:54 2012 -0400
drm/radeon/kms: improve bpc handling (v2)
Which resulted in kind of regression in 3.5. This fixes
https://bugs.freedesktop.org/show_bug.cgi?id=70934
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Cc: stable@vger.kernel.org
Signed-off-by: Dave Airlie <airlied@redhat.com>
VMAs covering a bo but that didn't start at the same address space offset as
the bo they were mapping were incorrectly generating SEGFAULT errors in
the fault handler.
Reported-by: Joseph Dolinak <kanilo2@yahoo.com>
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>
Cc: stable@vger.kernel.org
The intent was to only enable it by default for optimus, e.g. see the
runtime_idle callback. The suspend callback may be called directly, e.g.
as a result of nouveau_crtc_set_config.
Reported-by: Stefan Lippers-Hollmann <s.l-h@gmx.de>
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Tested-by: Stefan Lippers-Hollmann <s.l-h@gmx.de>
Cc: stable@vger.kernel.org
Signed-off-by: Dave Airlie <airlied@redhat.com>
In the current code, at haswell_modeset_global_resources, first we
decide if we want to enable/disable the power well, then we decide if
we want to enable/disable PC8. On the case where we're enabling PC8
this works fine, but on the case where we disable PC8 due to a non-eDP
monitor being enabled, we first enable the power well and then disable
PC8. Although wrong, this doesn't seem to be causing any problems now,
and we don't even see anything in dmesg. But the patches for runtime
D3 turn this problem into a real bug, so we need to fix it.
This fixes the "modeset-non-lpsp" subtest from the "pm_pc8" test from
intel-gpu-tools.
v2: - Rebase (i915_disable_power_well).
v3: - More reabase.
v4: - Rebase on top of -fixes instead of -nightly.
This is commit d62292c8f7 in -next, but
we need it in -fixes to address Dave's report.
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@gmail.com>
Reported-by: Dave Jones <davej@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Currently, PC8 is enabled at modeset_global_resources, which is called
after intel_modeset_update_state. Due to this, there's a small race
condition on the case where we start enabling PC8, then do a modeset
while PC8 is still being enabled. The racing condition triggers a WARN
because intel_modeset_update_state will mark the CRTC as enabled, then
the thread that's still enabling PC8 might look at the data structure
and think that PC8 is being enabled while a pipe is enabled. Despite
the WARN, this is not really a bug since we'll wait for the
PC8-enabling thread to finish when we call modeset_global_resources.
The spec says the CRTC cannot be enabled when we disable LCPLL, so we
had a check for crtc->base.enabled. If we change to crtc->active we
will still prevent disabling LCPLL while the CRTC is enabled, and we
will also prevent the WARN above.
This is a replacement for the previous patch named
"drm/i915: get/put PC8 when we get/put a CRTC"
Testcase: igt/pm_pc8/modeset-lpsp-stress-no-wait
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
(cherry picked from commit 798183c547
from -next due to Dave's report.)
Reported-by: Dave Jones <davej@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
All instances of drm_dev_register are followed by drm_dev_free on
failure. Don't free dev->control/render/primary on failure, as they will
be freed by drm_dev_free since commit 8f6599da8e (drm: delay minor
destruction to drm_dev_free()). Instead unplug them.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reported-and-tested-by: Bruno Prémont <bonbons@linux-vserver.org>
Reviewed-by: David Herrmann <dh.herrmann@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Part of a driver stack fix that fixes surface overcommiting on single execbuf calls.
* 'vmwgfx-fixes-3.13' of git://people.freedesktop.org/~thomash/linux:
drm/vmwgfx: Add max surface memory param
This causes a race condition between drm_dev_unregister()
and pci_driver.shutdown at shutdown or driver unload time.
We need to revisit how to properly support kexec within
the drm.
This reverts commit 846ae41ae9.
Otherwise we end up with a rather strange looking result.
Signed-off-by: Christian König <christian.koenig@amd.com>
Tested-by: Tom Stellard <thomas.stellard@amd.com>
Cc: stable@vger.kernel.org
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
I implemented support for this, but forget to hook
up the callback so the driver can actually use it.
On asics with a dedicated DMA engine, we use the DMA
engine for buffer migration so this is just for testing
purposes.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Commit ec39f64bba (drm/radeon/dpm: Convert
to use devm_hwmon_register_with_groups) converted one usage of
dev_get_drvdata, but there were two more.
bug:
https://bugs.freedesktop.org/show_bug.cgi?id=72457
Signed-off-by: Martin Andersson <g02maran@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Some RS690 boards with 64MB of sideport memory show up as
having 128MB sideport + 256MB of UMA. In this case,
just skip the sideport memory and use UMA. This fixes
rendering corruption and should improve performance.
bug:
https://bugs.freedesktop.org/show_bug.cgi?id=35457
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
As the rings may be processed and their requests deallocated in a
different order to the natural retirement during a reset,
/* Whilst this request exists, batch_obj will be on the
* active_list, and so will hold the active reference. Only when this
* request is retired will the the batch_obj be moved onto the
* inactive_list and lose its active reference. Hence we do not need
* to explicitly hold another reference here.
*/
is violated, and the batch_obj may be dereferenced after it had been
freed on another ring. This can be simply avoided by processing the
status update prior to deallocating any requests.
Fixes regression (a possible OOPS following a GPU hang) from
commit aa60c664e6
Author: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Date: Wed Jun 12 15:13:20 2013 +0300
drm/i915: find guilty batch buffer on ring resets
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@intel.com>
Cc: stable@vger.kernel.org
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>
[danvet: Add the code comment Chris supplied.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Whilst looking up the objects required for an execbuffer, an untimely
allocation failure in creating the vma results in the object being
unreferenced from two lists. The ownership during the lookup is meant to
be moved from the list of objects being looked to the vma, and this
double unreference upon error results in a use-after-free.
Fixes regression from
commit 27173f1f95
Author: Ben Widawsky <ben@bwidawsk.net>
Date: Wed Aug 14 11:38:36 2013 +0200
drm/i915: Convert execbuf code to use vmas
Based on the fix by Ben Widawsky.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Ben Widawsky <ben@bwidawsk.net>
Cc: stable@vger.kernel.org
[danvet: Bikeshed the crucial comment above the ownership transfer as
discussed on irc.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
These four patches fix a few issues discovered since the initial merge,
which have been reviewed by Rob Clark and Thierry Reding.
* 'drm-tda998x-3.12-fixes' of git://ftp.arm.linux.org.uk/~rmk/linux-cubox:
DRM: Armada: prime refcounting bug fix
DRM: Armada: fix printing of phys_addr_t/dma_addr_t
DRM: Armada: destroy framebuffer after helper
DRM: Armada: implement lastclose() for fbhelper
Just a bunch of regression fixes plus a few patches for long-standing
issues in gem corner-cases that we've hunted down in the past weeks. Since
apparently people hit those in the wild (and we also have nice igts for
them) I've opted for -fixes and cc: stable.
There's 1-2 things oustanding on top of this where I'm still waiting on
confirmation from testing, but nothing really scary.
* tag 'drm-intel-fixes-2013-12-11' of git://people.freedesktop.org/~danvet/drm-intel:
drm/i915: don't update the dri1 breadcrumb with modesetting
drm/i915: Repeat eviction search after idling the GPU
drm/i915: Fix use-after-free in do_switch
drm/i915: fix pm init ordering
drm/i915: Hold mutex across i915_gem_release
drm/i915: Skip clock checks on BDW
drm/i915: Do not clobber config status after a forced restore of hw state
drm/i915: Take modeset locks around intel_modeset_setup_hw_state()
As promised bdw fixes come separate for now. Just a few minior things.
* 'bdw-fixes' of git://people.freedesktop.org/~danvet/drm-intel:
drm/i915/bdw: PIPE_[BC] I[ME]R moved to powerwell
drm/i915/bdw: Limit GTT to 2GB
drm/i915/bdw: Add comment about gen8 HWS PGA
drm/i915/bdw: Free correct number of ppgtt pages
drm/i915/bdw: Do gen6 style reset for gen8
drm/i915/bdw: GEN8 backlight support
drm/i915/bdw: Add BDW to ULT macro
The update is horribly racy since it doesn't protect at all against
concurrent closing of the master fd. And it can't really since that
requires us to grab a mutex.
Instead of jumping through hoops and offloading this to a worker
thread just block this bit of code for the modesetting driver.
Note that the race is fairly easy to hit since we call the breadcrumb
function for any interrupt. So the vblank interrupt (which usually
keeps going for a bit) is enough. But even if we'd block this and only
update the breadcrumb for user interrupts from the CS we could hit
this race with kms/gem userspace: If a non-master is waiting somewhere
(and hence has interrupts enabled) and the master closes its fd
(probably due to crashing).
v2: Add a code comment to explain why fixing this for real isn't
really worth it. Also improve the commit message a bit.
v3: Fix the spelling in the comment.
Reported-by: Eugene Shatokhin <eugene.shatokhin@rosalab.ru>
Cc: Eugene Shatokhin <eugene.shatokhin@rosalab.ru>
Cc: stable@vger.kernel.org
Acked-by: Chris Wilson <chris@chris-wilson.co.uk>
Tested-by: Eugene Shatokhin <eugene.shatokhin@rosalab.ru>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Commit 011c2282c7 changed the way refcounting on imported dma_bufs
works, and this hadn't been spotted while forward-porting Armada.
Reflect the changes in that commit into the Armada driver.
Reviewed-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
These can be 64-bit quantities, so fix them up appropriately.
Reviewed-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Destroy the framebuffer only after the helper, since the helper may
still be referencing the framebufer at this point.
Reviewed-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Call drm_fb_helper_restore_fbdev_mode() upon last close so that in the
event of the X server crashing, we have some kind of mode restored.
Reviewed-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
With the advent of hw context support, we gained some objects that are
pinned for the duration of their request. That is we can make aperture
space available by idling the GPU and in the process performing a
context switch back to the always-pinned default context. As such, we
should not conclude that there is no space in the aperture for the
current object until we have unpinned any such context objects.
Note that we also have the problem of outstanding pageflips preventing
eviction of their framebuffer objects to resolve.
Testcase: igt/gem_ctx_exec/eviction
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=72507
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Tested-by: lu hua <huax.lu@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Userspace uses this to workaround overcommit issues
by flushing the command stream early.
Signed-off-by: Jakob Bornecrantz <jakob@vmware.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
So apparently under ridiculous amounts of memory pressure we can get
into trouble in do_switch when we try to move the old hw context
backing storage object onto the active lists.
With list debugging enabled that usually results in us chasing a
poisoned pointer - which means we've hit upon a vma that has been
removed from all lrus with list_del (and then deallocated, so it's a
real use-after free).
Ian Lister has done some great callchain chasing and noticed that we
can reenter do_switch:
i915_gem_do_execbuffer()
i915_switch_context()
do_switch()
from = ring->last_context;
i915_gem_object_pin()
i915_gem_object_bind_to_gtt()
ret = drm_mm_insert_node_in_range_generic();
// If the above call fails then it will try i915_gem_evict_something()
// If that fails it will call i915_gem_evict_everything() ...
i915_gem_evict_everything()
i915_gpu_idle()
i915_switch_context(DEFAULT_CONTEXT)
Like with everything else where the shrinker or eviction code can
invalidate pointers we need to reload relevant state.
Note that there's no need to recheck whether a context switch is still
required because:
- Doing a switch to the same context is harmless (besides wasting a
bit of energy).
- This can only happen with the default context. But since that one's
pinned we'll never call down into evict_everything under normal
circumstances. Note that there's a little driver bringup fun
involved namely that we could recourse into do_switch for the
initial switch. Atm we're fine since we assign the context pointer
only after the call to do_switch at driver load or resume time. And
in the gpu reset case we skip the entire setup sequence (which might
be a bug on its own, but definitely not this one here).
Cc'ing stable since apparently ChromeOS guys are seeing this in the
wild (and not just on artificial stress tests), see the reference.
Note that in upstream code doesn't calle evict_everything directly
from evict_something, that's an extension in this product branch. But
we can still hit upon this bug (and apparently we do, see the linked
backtraces). I've noticed this while trying to construct a testcase
for this bug and utterly failed to provoke it. It looks like we need
to driver the system squarly into the lowmem wall and provoke the
shrinker to evict the context object by doing the last-ditch
evict_everything call.
Aside: There's currently no means to get a badly-fragmenting hw
context object away from a bad spot in the upstream code. We should
fix this by at least adding some code to evict_something to handle hw
contexts.
References: https://code.google.com/p/chromium/issues/detail?id=248191
Reported-by: Ian Lister <ian.lister@intel.com>
Cc: Ian Lister <ian.lister@intel.com>
Cc: stable@vger.kernel.org
Cc: Ben Widawsky <benjamin.widawsky@intel.com>
Cc: Stéphane Marchesin <marcheu@chromium.org>
Cc: Bloomfield, Jon <jon.bloomfield@intel.com>
Tested-by: Rafael Barbalho <rafael.barbalho@intel.com>
Reviewed-by: Ian Lister <ian.lister@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Shovel a bit more of the the code into the setup function, and call
it earlier. Otherwise lockdep is unhappy since we cancel the delayed
resume work before it's initialized.
While at it also shovel the pc8 setup code into the same functions.
I wanted to also ditch the header declaration of the hws pc8 functions,
but for unfathomable reasons that stuff is in intel_display.c instead
of intel_pm.c.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=71980
Tested-by: Guo Jinxian <jinxianx.guo@intel.com>
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Some additional fixes for 3.13. Regression fixes for audio and hw_i2c,
vram fix for some SI PX cards, race fix in the hwmon code, and a few other
odds and ends.
* 'drm-fixes-3.13' of git://people.freedesktop.org/~agd5f/linux:
drm/radeon/atom: fix bus probes when hw_i2c is set (v2)
drm/radeon: fix null pointer dereference in dce6+ audio code
drm/radeon: fixup bad vram size on SI
drm/radeon: fix VGT_GS_INSTANCE_CNT register
drm/radeon: Fix a typo in Cayman and Evergreen registers
drm/radeon/dpm: simplify state adjust logic for NI
drm/radeon: add radeon_vm_bo_update trace point
drm/radeon: add VMID allocation trace point
drm/radeon/dpm: Convert to use devm_hwmon_register_with_groups
drm/radeon: program DCE2 audio dto just like DCE3
drm/radeon: fix typo in fetching mpll params
Ensure the side-by-side (half) flag is added to any existing flags when
adding modes from 3D_Structure_ALL.
Signed-off-by: Thomas Wood <thomas.wood@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Thomas Wood <thomas.wood@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Linus Torvalds