We get a large number of bugs which have a, "hey I have that too"
because they see a GPU hang in dmesg. While two machines of the same
model having a GPU hang is indeed a coincidence, it is far from enough
evidence to suggest they are the same.
In order to reduce this effect, and hopefully get people to file new bug
reports, clearly the error message itself has been insufficient (see ref
at the bottom for a new bug report with this characteristic).
The algorithm is purposely pretty naive. I don't think we need much in
order to avoid the problem I am trying to solve, and keeping it naive
gives us some ability to make a decent test case.
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
References: https://bugs.freedesktop.org/show_bug.cgi?id=73276
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
We want to use the GTT for reading back objects upon an error so that we
have exactly the information that the GPU saw. However, it is verboten
to access snoopable pages through the GTT and causes my PineView GPU to
throw a page fault instead.
This has not been a problem in the past as we only dumped ringbuffers
and batchbuffers, both of which must be not snooped. However, the
introduction of HWS page dumping leads to a read of a snooped object
through the GTT. This was introduced by
commit f3ce382139
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date: Thu Jan 23 22:40:36 2014 +0000
drm/i915: Include HW status page in error capture
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
[danvet:s/uncached/not snooped/ for one case in the commit message as
requested by Chris.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Repeating the same information multiple times is just annoying.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
v2: Rebased upon cleaned up error state
v3: Make sure hangcheck info remains last (Chris)
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Chris:
Do we also want to capture?
GAC_ECO_BITS /* gen6,7 */
GAM_ECOCHK /* gen6,7 */
GAB_CTL /* gen6 */
GFX_MODE /* gen6 */
Requested-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
v2: Moved num_requests up (Chris)
Rebased on new hws page capture which required a rename since it made
two members named, 'hws' in the per ring error state. (Ben)
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Create logical sections in an attempt to clean up, and continue to keep
future additions clean.
v2: Reworded the comments. Added section headers (Chris)
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The code has become quite hairy. By relocating all the generic registers
it will become more obvious where future ones should go. There is still
admittedly a bit of confusion left for things like per ring registers.
A subsequent patch will clean this function up.
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Backmerge drm-next - I need to backmerge drm-intel-fixes patches
touching the error capture code to be able to merge Ben's cleanup
patches.
Conflicts:
drivers/gpu/drm/i915/i915_gpu_error.c
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Many times in the past we have concluded that the cause of the GPU hang
has been that the hw status page was stale, usually because the GPU and
CPU disagreed over the address of the page. Having stumbled across yet
another issue that seems to be related to the HWSP, it is time to
include that information in the GPU error dump.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Currently we report through our error state only the rings that have
been initialised (as detected by ring->obj). This check is done after
the GPU reset and ring re-initialisation, which means that the software
state may not be the same as when we captured the hardware error and we
may not print out any of the vital information for debugging the hang.
This (and the implied object leak) is a regression from
commit 3d57e5bd12
Author: Ben Widawsky <ben@bwidawsk.net>
Date: Mon Oct 14 10:01:36 2013 -0700
drm/i915: Do a fuller init after reset
Note that we are already starting to get bug reports with incomplete
error states from 3.13, which also hampers debugging userspace driver
issues.
v2: Prevent a NULL dereference on 830gm/845g after a GPU reset where
the scratch obj may be NULL.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Ben Widawsky <ben@bwidawsk.net>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
References: https://bugs.freedesktop.org/show_bug.cgi?id=74094
Cc: stable@vger.kernel.org # please don't delay since it's a
vital support/debug feature for the intel gfx stack in general
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
[danvet: Add a bit of fluff to make it clear we need this expedited in
stable.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Because whatever.*
* This should contain a fairly long list of issues and still
unresolved resgressions, but I didn't really get a vote.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
As with processes which run on the CPU, the goal of multiple VMs is to
provide process isolation. Specific to GEN, there is also the ability to
map more objects per process (2GB each instead of 2Gb-2k total).
For the most part, all the pipes have been laid, and all we need to do
is remove asserts and actually start changing address spaces with the
context switch. Since prior to this we've converted the setting of the
page tables to a streamed version, this is quite easy.
One important thing to point out (since it'd been hotly contested) is
that with this patch, every context created will have it's own address
space (provided the HW can do it).
v2: Disable BDW on rebase
NOTE: I tried to make this commit as small as possible. I needed one
place where I could "turn everything on" and that is here. It could be
split into finer commits, but I didn't really see much point.
Cc: Eric Anholt <eric@anholt.net>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Using the current state of the page directory registers, we can
determine which of our address spaces was active when the hang occurred.
This allows us to scan through all the address spaces to identify the
"active" one during error capture.
v2: Rebased for BDW error detection. BDW error detection is similar
except instead of PP_DIR_BASE, we can use the PDP registers.
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
[danvet: Add FIXME about global gtt misuse.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The existing check was insufficient to determine whether we can use the
GTT mapping to read out the object during error capture.
The previous condition was, if the object has a GGTT mapping, and the
reloc is in the GTT range... the can happen with opjects mapped into
multiple vms (one of which being the GTT).
There are two solutions to this problem:
1. This patch, which avoid reading the io mapping
2. Use the GGTT offset with the io mapping.
Since error capture is about recording the most accurate possible error
state, and the error was caused by the object not in the GGTT - I opted
for the former.
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Every ring seems to have a BB_ADDR registers, so include them all in the
error state.
v2: Also include the _UDW on BDW
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Ben Widawsky <ben@bwidawsk.net>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The BB_ADDR register is documented to be 32bits at least since SNB.
Prior to that the high 32bits were listed as MBZ, so using a 64bit read
doesn't seem worth anything. Also the simulator doesn't like the 64bit
read. So just switch to using a 32bit read instead.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Ben Widawsky <ben@bwidawsk.net>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The bbstate contains useful bits of debugging information such as
whether the batch is being read from GTT or PPGTT, or whether it is
allowed to execute privileged instructions.
v2: Only record BB_STATE for gen4+
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Untangling me-too reports that actually aren't is really messy. And we
need to make sure the blame is put where it should be right from the
start ;-)
v2: Improve the wording from Ben's suggestions.
Cc: Ben Widawsky <ben@bwidawsk.net>
Acked-by: Ben Widawsky <ben@bwidawsk.net>
[danvet: Frob the message as suggested by Paulo on irc.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The conflict in intel_drv.h tripped me up a bit since a patch in dinq
moves all the functions around, but another one in drm-next removes a
single function. So I'ev figured backing this into a backmerge would
be good.
i915_dma.c is just adjacent lines changed, nothing nefarious there.
Conflicts:
drivers/gpu/drm/i915/i915_dma.c
drivers/gpu/drm/i915/intel_drv.h
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
We can get the PCI vendor and device IDs via dev->pdev. So we can drop
the duplicated information.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
When we switched to always using a timeout in conjunction with
wait_seqno, we lost the ability to detect missed interrupts. Since, we
have had issues with interrupts on a number of generations, and they are
required to be delivered in a timely fashion for a smooth UX, it is
important that we do log errors found in the wild and prevent the
display stalling for upwards of 1s every time the seqno interrupt is
missed.
Rather than continue to fix up the timeouts to work around the interface
impedence in wait_event_*(), open code the combination of
wait_event[_interruptible][_timeout], and use the exposed timer to
poll for seqno should we detect a lost interrupt.
v2: In order to satisfy the debug requirement of logging missed
interrupts with the real world requirments of making machines work even
if interrupts are hosed, we revert to polling after detecting a missed
interrupt.
v3: Throw in a debugfs interface to simulate broken hw not reporting
interrupts.
v4: s/EGAIN/EAGAIN/ (Imre)
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Imre Deak <imre.deak@intel.com>
[danvet: Don't use the struct typedef in new code.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Add the missing cache-level to the describe_obj() function for debug and
error reporting.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
No buffer overflows here, but better safe than sorry.
v2:
- Fixup the sizeof conversion, I've missed the pointer deref (Jani).
- Drop the redundant GFP_ZERO, kcalloc alreads memsets (Jani).
- Use kmalloc_array for the execbuf fastpath to avoid the memset
(Chris). I've opted to leave all other conversions as-is since they
aren't in a fastpath and dealing with cleared memory instead of
random garbage is just generally nicer.
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
[danvet: Drop the contentious kmalloc_array hunk in execbuf.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
drm-intel-next-2013-09-21:
- clock state handling rework from Ville
- l3 parity handling fixes for hsw from Ben
- some more watermark improvements from Ville
- ban badly behaved context from Mika
- a few vlv improvements from Jesse
- VGA power domain handling from Ville
drm-intel-next-2013-09-06:
- Basic mipi dsi support from Jani. Not yet converted over to drm_bridge
since that was too fresh, but the porting is in progress already.
- More vma patches from Ben, this time the code to convert the execbuffer
code. Now that the shrinker recursion bug is tracked down we can move
ahead here again. Yay!
- Optimize hw context switching to not generate needless interrupts (Chris
Wilson). Also some shuffling for the oustanding request allocation.
- Opregion support for SWSCI, although not yet fully wired up (we need a
bit of runtime D3 support for that apparently, due to Windows design
deficiencies), from Jani Nikula.
- A few smaller changes all over.
[airlied: merge conflict fix in i9xx_set_pipeconf]
* tag 'drm-intel-next-2013-09-21-merged' of git://people.freedesktop.org/~danvet/drm-intel: (119 commits)
drm/i915: assume all GM45 Acer laptops use inverted backlight PWM
drm/i915: cleanup a min_t() cast
drm/i915: Pull intel_init_power_well() out of intel_modeset_init_hw()
drm/i915: Add POWER_DOMAIN_VGA
drm/i915: Refactor power well refcount inc/dec operations
drm/i915: Add intel_display_power_{get, put} to request power for specific domains
drm/i915: Change i915_request power well handling
drm/i915: POSTING_READ IPS_CTL before waiting for the vblank
drm/i915: don't disable ERR_INT on the IRQ handler
drm/i915/vlv: disable rc6p and rc6pp residency reporting on BYT
drm/i915/vlv: honor i915_enable_rc6 boot param on VLV
drm/i915: s/HAS_L3_GPU_CACHE/HAS_L3_DPF
drm/i915: Do remaps for all contexts
drm/i915: Keep a list of all contexts
drm/i915: Make l3 remapping use the ring
drm/i915: Add second slice l3 remapping
drm/i915: Fix HSW parity test
drm/i915: dump crtc timings from the pipe config
drm/i915: register backlight device also when backlight class is a module
drm/i915: write D_COMP using the mailbox
...
Conflicts:
drivers/gpu/drm/i915/intel_display.c
In
commit edc3d8848d
Author: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Date: Thu May 23 13:55:35 2013 +0300
drm/i915: avoid big kmallocs on reading error state
we introduce a two-pass mechanism for splitting long strings being
formatted into the error-state. The first pass finds the length, and the
second pass emits the right portion of the string into the accumulation
buffer. Unfortunately we use the same va_list for both passes, resulting
in the second pass reading garbage off the end of the argument list. As
the two passes are only used for boundaries between read() calls, the
corruption is only rarely seen.
This fixes the root cause behind
commit baf27f9b17
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date: Sat Jun 29 23:26:50 2013 +0100
drm/i915: Break up the large vsnprintf() in print_error_buffers()
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: stable@vger.kernel.org
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Score and action reveals what all the rings were doing
and why hang was declared. Add idle state so that
we can distinguish between waiting and idle ring.
v2: - add idle as a hangcheck action
- consensed hangcheck status to single line (Chris)
- mark active explicitly when we are making progress (Chris)
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
We now have more devices using ring->private than not, and they all want
the same structure. Worse, I would like to use a scratch page from
outside of intel_ringbuffer.c and so for convenience would like to reuse
ring->private. Embed the object into the struct intel_ringbuffer so that
we can keep the code clean.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Ideally we could use for_each_ring with the ring flags as I've done a
couple times
(http://lists.freedesktop.org/archives/intel-gfx/2013-June/029450.html).
Until Daniel merges that patch though, we can just use this.
Cc: Mika Kuoppala <mika.kuoppala@intel.com>
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
formerly: "drm/i915: Create VMAs (part 4) - Error capture"
Since the active/inactive lists are per VM, we need to modify the error
capture code to be aware of this, and also extend it to capture the
buffers from all the VMs. For now all the code assumes only 1 VM, but it
will become more generic over the next few patches.
NOTE: If the number of VMs in a real world system grows significantly
we'll have to focus on only capturing the guilty VM, or else it's likely
there won't be enough space for error capture.
v2: Squashed in the "part 6" which had dependencies on the mm_list
change. Since I've moved the mm_list change to an earlier point in the
series, we were able to accomplish it here and now.
v3: Rebased over new error capture
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
formerly: "drm/i915: Create VMAs (part 5) - move mm_list"
The mm_list is used for the active/inactive LRUs. Since those LRUs are
per address space, the link should be per VMx .
Because we'll only ever have 1 VMA before this point, it's not incorrect
to defer this change until this point in the patch series, and doing it
here makes the change much easier to understand.
Shamelessly manipulated out of Daniel:
"active/inactive stuff is used by eviction when we run out of address
space, so needs to be per-vma and per-address space. Bound/unbound otoh
is used by the shrinker which only cares about the amount of memory used
and not one bit about in which address space this memory is all used in.
Of course to actual kick out an object we need to unbind it from every
address space, but for that we have the per-object list of vmas."
v2: only bump GGTT LRU in i915_gem_object_set_to_gtt_domain (Chris)
v3: Moved earlier in the series
v4: Add dropped message from v3
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
[danvet: Frob patch to apply and use vma->node.size directly as
discused with Ben. Also drop a needles BUG_ON before move_to_inactive,
the function itself has the same check.]
[danvet 2nd: Rebase on top of the lost "drm/i915: Cleanup more of VMA
in destroy", specifically unlink the vma from the mm_list in
vma_unbind (to keep it symmetric with bind_to_vm) instead of
vma_destroy.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
MLC_LLC was never validated for Sandybridge and was superseded by a new
level of cacheing for the GPU in Ivybridge. Update our names to be
consistent with usage, and in the process stop setting the unwanted bit
on Sandybridge.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
[danvet: s/BUG/WARN_ON(1) bikeshed.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Shamelessly manipulated out of Daniel :-)
"When moving the lists around explain that the active/inactive stuff is
used by eviction when we run out of address space, so needs to be
per-vma and per-address space. Bound/unbound otoh is used by the
shrinker which only cares about the amount of memory used and not one
bit about in which address space this memory is all used in. Of course
to actual kick out an object we need to unbind it from every address
space, but for that we have the per-object list of vmas."
v2: Leave the bound list as a global one. (Chris, indirectly)
v3: Rebased with no i915_gtt_vm. In most places I added a new *vm local,
since it will eventually be replaces by a vm argument.
Put comment back inline, since it no longer makes sense to do otherwise.
v4: Rebased on hangcheck/error state movement
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Reviewed-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Move error state generation and stringification to it's
own compilation unit. Sysfs also uses this so it can't be
under CONFIG_DEBUG_FS
This fixes a regression introduced in
commit ef86ddced7
Author: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Date: Thu Jun 6 17:38:54 2013 +0300
drm/i915: add error_state sysfs entry
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=66814
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Reported-by: kbuild test robot <fengguang.wu@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
