Summary: | [GEN4] Graphics driver crashes occassionally | ||
---|---|---|---|
Product: | Mesa | Reporter: | Maarten Jacobs <maarten256> |
Component: | Drivers/DRI/i965 | Assignee: | Ian Romanick <idr> |
Status: | RESOLVED DUPLICATE | QA Contact: | Intel 3D Bugs Mailing List <intel-3d-bugs> |
Severity: | normal | ||
Priority: | medium | CC: | intel-gfx-bugs |
Version: | unspecified | ||
Hardware: | x86 (IA32) | ||
OS: | Linux (All) | ||
Whiteboard: | |||
i915 platform: | i915 features: | ||
Attachments: | crash dump (/sys/class/drm/card0/error) |
Description
Maarten Jacobs
2014-11-26 05:20:38 UTC
I managed to recreate the problem (more on that in a bit). The kern.log indicates again that a crash dump was saved in /sys/class/drm/card0/error (like before), however a listing of that location shows: $ ls -l /sys/class/drm/card0/ total 0 drwxr-xr-x 4 root root 0 Nov 26 00:15 card0-LVDS-1 drwxr-xr-x 3 root root 0 Nov 26 00:15 card0-VGA-1 -r--r--r-- 1 root root 4096 Nov 26 00:15 dev lrwxrwxrwx 1 root root 0 Nov 26 00:15 device -> ../../../0000:00:02.0 -rw------- 1 root root 0 Nov 26 00:15 error drwxr-xr-x 2 root root 0 Nov 26 00:15 power lrwxrwxrwx 1 root root 0 Nov 26 00:15 subsystem -> ../../../../../class/drm -rw-r--r-- 1 root root 4096 Nov 26 00:15 uevent So maybe it didn't save anything. The machine is still in it's "darkened" state. I recreated the issue by selecting a particular icon on the MS Outlook website... In Google Chrome. When I select the icon that is supposed to show me the contact list, the graphics driver goes into this state and the screen goes blank. I have seen the same issue on other occasions, so I know there are other interactions that can cause the same. However this "contact list" function seems to recreate the issue consistently - at least it did so twice in a row. ls lies; /sys/class/drm/card0/error is a virtual file and doesn't have a size until it is opened. Just compress it and attach, after reproducing the hang. Created attachment 110076 [details]
crash dump (/sys/class/drm/card0/error)
Comment on attachment 110076 [details]
crash dump (/sys/class/drm/card0/error)
I added the crash bump to this bug (I think). I executed:
# gzip -c /sys/class/drm/card0/error > crash_dump.gz
gzip: /sys/class/drm/card0/error: file size changed while zipping
I don't know if the error I got from gzip is relevant or not.
Let me know if this does the trick. If not please send instructions (obviously I'm out of my depth here).
For next time around: cat /sys/class/drm/card0/error | gzip -c > crash_dump.gz Everything more intelligent than cat will try to seek around in the file and get pissed about the size being 0. But it worked. !!!*** Carry this message to developers with them is simply impossible to be written off to foreigners. !!!*** Why is there no Ukrainian version of the site libra office ?? Although it has long been translated! Add it please. You have a very intricate system of reporting bugs (errors) Everything is very closed and difficult! Is it not possible to make a more friendly system? No support for the Ukrainian language ( https://translations.documentfoundation.org/uk/website/multiplanet-uk.po http://uk.libreoffice.org/ - no open(( I suspect this may be another duplicate of the bug 80568, fixed (worked-around) by this commit: commit c4fd0c9052dd391d6f2e9bb8e6da209dfc7ef35b Author: Kenneth Graunke <kenneth@whitecape.org> Date: Sat Jan 17 23:21:15 2015 -0800 i965: Work around mysterious Gen4 GPU hangs with minimal state changes. Gen4 hardware appears to GPU hang frequently when using Chromium, and also when running 'glmark2 -b ideas'. Most of the error states contain 3DPRIMITIVE commands in quick succession, with very few state packets between them - usually VERTEX_BUFFERS/ELEMENTS and CONSTANT_BUFFER. I trimmed an apitrace of the glmark2 hang down to two draw calls with a glUniformMatrix4fv call between the two. Either draw by itself works fine, but together, they hang the GPU. Removing the glUniform call makes the hangs disappear. In the hardware state, this translates to removing the CONSTANT_BUFFER packet between the two 3DPRIMITIVE packets. Flushing before emitting CONSTANT_BUFFER packets also appears to make the hangs disappear. I observed a slowdown in glxgears by doing it all the time, so I've chosen to only do it when BRW_NEW_BATCH and BRW_NEW_PSP are unset (i.e. we haven't done a CS_URB_STATE change or already flushed the whole pipeline). I'd much rather understand the problem, but at this point, I don't see how we'd ever be able to track it down further. We have no real tools, and the hardware people moved on years ago. I've analyzed 20+ error states and read every scrap of documentation I could find. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=80568 Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=85367 Signed-off-by: Kenneth Graunke <kenneth@whitecape.org> Acked-by: Matt Turner <mattst88@gmail.com> Cc: "10.4 10.3" <mesa-stable@lists.freedesktop.org> It's in git, and backports are in Mesa 10.4.x for x > 3. Please try upgrading to >10.4.3. If it's resolved by such an upgrade, please mark as a duplicate of bug 80568. |
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.