Bug 58428 - [NVA8] X crashes when running opengl application after reclocking nouveau driver to use highest speed
Summary: [NVA8] X crashes when running opengl application after reclocking nouveau dri...
Status: RESOLVED WONTFIX
Alias: None
Product: xorg
Classification: Unclassified
Component: Driver/nouveau (show other bugs)
Version: unspecified
Hardware: x86-64 (AMD64) Linux (All)
: medium normal
Assignee: Nouveau Project
QA Contact: Xorg Project Team
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2012-12-17 19:16 UTC by Amadeusz
Modified: 2013-10-08 17:03 UTC (History)
1 user (show)

See Also:
i915 platform:
i915 features:


Attachments
X log from vanilla kernel (36.52 KB, text/plain)
2012-12-17 19:17 UTC, Amadeusz
no flags Details
dmesg | grep nouveau from vanilla kernel (8.66 KB, text/plain)
2012-12-17 19:17 UTC, Amadeusz
no flags Details
X log from hardened kernel (37.73 KB, text/plain)
2012-12-17 19:18 UTC, Amadeusz
no flags Details
dmesg | grep nouveau from hardened kernel (45.67 KB, text/plain)
2012-12-17 19:19 UTC, Amadeusz
no flags Details
X log from vanilla kernel (39.17 KB, text/plain)
2012-12-17 19:37 UTC, Amadeusz
no flags Details
dmesg | grep nouveau from vanilla kernel (11.83 KB, text/plain)
2012-12-17 19:38 UTC, Amadeusz
no flags Details

Description Amadeusz 2012-12-17 19:16:16 UTC
Linux lain 3.7.0-hardened #1 SMP PREEMPT Mon Dec 17 03:03:30 CET 2012 x86_64 Intel(R) Core(TM) i3 CPU M 350 @ 2.27GHz GenuineIntel GNU/Linux

(Tested both on hardened and vanilla kernels)

[    2.719980] nouveau  [     DRM] 3 available performance level(s)
[    2.719984] nouveau  [     DRM] 0: core 135MHz shader 270MHz memory 135MHz voltage 850mV
[    2.719987] nouveau  [     DRM] 1: core 405MHz shader 810MHz memory 405MHz voltage 850mV
[    2.719989] nouveau  [     DRM] 3: core 625MHz shader 1530MHz memory 790MHz voltage 1030mV
[    2.719992] nouveau  [     DRM] c: core 405MHz shader 810MHz memory 405MHz voltage 850mV

echo 3 > /sys/class/drm/card0/device/performance_level

run opengl app (I run xonotic-glx)
at first you can move mouse a bit but it just hangs a bit later, after a while X crashes

(It reclocks works fine when reclocking between 0 and 1)
Comment 1 Amadeusz 2012-12-17 19:17:00 UTC
Created attachment 71678 [details]
X log from vanilla kernel
Comment 2 Amadeusz 2012-12-17 19:17:42 UTC
Created attachment 71679 [details]
dmesg | grep nouveau from vanilla kernel
Comment 3 Amadeusz 2012-12-17 19:18:40 UTC
Created attachment 71680 [details]
X log from hardened kernel
Comment 4 Amadeusz 2012-12-17 19:19:14 UTC
Created attachment 71681 [details]
dmesg | grep nouveau from hardened kernel
Comment 5 Amadeusz 2012-12-17 19:37:28 UTC
Created attachment 71684 [details]
X log from vanilla kernel

Just noticed that for whatever reason my bootloader decided that if it can't load kernel it will just load another one... vanilla for real this time
Comment 6 Amadeusz 2012-12-17 19:38:13 UTC
Created attachment 71685 [details]
dmesg | grep nouveau from vanilla kernel
Comment 7 Ilia Mirkin 2013-09-06 22:22:34 UTC
Does this happen if you don't do any reclocking? If so, let's get reclocking out of the picture.

If not, you could try a newer kernel, but I think in general, reclocking isn't really supported (you probably noticed that you had to add a special parameter to get it to work, that's not a coincidence). I think some work is currently being done for the NVA3+ cards, you could try to find out status in #nouveau on freenode if you're interested.
Comment 8 Ilia Mirkin 2013-10-08 17:03:58 UTC
Reclocking is presently not supported. We know it kills many different cards for many different reasons. Closing as WONTFIX. (Not that it'll never get fixed, just not worth keeping this bug open for a presently-unsupported feature.) When/if we get good reclocking support, and you still have issues, feel free to reopen.

[It appears that you're seeing mem faults, and reclocking memory is one of the buggiest parts of the current reclocking logic. Makes sense.]


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.