Bug 76558 - [radeon] GPU lockup: radeon: wait for empty RBBM fifo failed ! when playing sauerbraten
Summary: [radeon] GPU lockup: radeon: wait for empty RBBM fifo failed ! when playing s...
Status: RESOLVED MOVED
Alias: None
Product: DRI
Classification: Unclassified
Component: DRM/Radeon (show other bugs)
Version: XOrg git
Hardware: x86 (IA32) Linux (All)
: high critical
Assignee: Default DRI bug account
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2014-03-24 16:16 UTC by Fabio Pedretti
Modified: 2019-11-19 08:47 UTC (History)
0 users

See Also:
i915 platform:
i915 features:


Attachments
dmesg after the lockup with Ubuntu kernel 3.11.0-18.32 (based on 3.11.10.4) (66.96 KB, text/plain)
2014-03-24 16:16 UTC, Fabio Pedretti
no flags Details
dmesg after the lockup with Ubuntu kernel 3.13.0-24.47 (based on 3.13.9) (78.61 KB, text/plain)
2014-05-21 10:29 UTC, Fabio Pedretti
no flags Details

Description Fabio Pedretti 2014-03-24 16:16:54 UTC
Created attachment 96295 [details]
dmesg after the lockup with Ubuntu kernel 3.11.0-18.32 (based on 3.11.10.4)

I am using kernel 3.11.0-18.32 (default Ubuntu 13.10 kernel) with mesa from git on a RV530. The system is usually very stable, however when running sauerbreaten with wake6 map (sauerbraten-wake6 package on Debian/Ubuntu) this way:
sauerbraten -lwake6

after some minutes I get the following lockup:

[ 8425.176130] radeon 0000:01:00.0: GPU lockup CP stall for more than 10000msec
[ 8425.176144] radeon 0000:01:00.0: GPU lockup (waiting for 0x0000000000002845 last fence id 0x0000000000002844)
[ 8425.349835] radeon: wait for empty RBBM fifo failed ! Bad things might happen.
[ 8425.523469] Failed to wait GUI idle while programming pipes. Bad things might happen.
[ 8425.524520] radeon 0000:01:00.0: Saved 59 dwords of commands on ring 0.
[ 8425.530923] radeon 0000:01:00.0: (rs600_asic_reset:401) RBBM_STATUS=0x9401C100
[ 8426.029068] radeon 0000:01:00.0: (rs600_asic_reset:421) RBBM_STATUS=0x9401C100
[ 8426.526183] radeon 0000:01:00.0: (rs600_asic_reset:429) RBBM_STATUS=0x9400C100
[ 8427.023299] radeon 0000:01:00.0: (rs600_asic_reset:437) RBBM_STATUS=0x9400C100
[ 8427.023370] radeon 0000:01:00.0: failed to reset GPU
[ 8427.024701] radeon 0000:01:00.0: GPU reset failed
[ 8427.028124] radeon 0000:01:00.0: couldn't schedule ib
[ 8427.028133] [drm:radeon_cs_ib_chunk] *ERROR* Failed to schedule IB !
[ 8427.057855] radeon 0000:01:00.0: couldn't schedule ib
...

Full dmesg is attached. After the lockup I can switch to/from console, but I have to reboot because the screen is corrupted/flashing.
Comment 1 Fabio Pedretti 2014-04-09 13:36:43 UTC
After upgrading to Ubuntu 14.04 I cannot reproduce this issue. Note that I was using current git of mesa/ati/libdrm with both 13.10 and 14.04. What changed was the kernel and sauerbraten get updated. I also tried with the 13.10 kernel, but the issue is no more reproducible.

It could be that the updated sauerbraten has a different path that doesn't trigger this issue. Closing since I cannot reproduce it anymore anyway.
Comment 2 Fabio Pedretti 2014-04-19 09:45:36 UTC
This is still happening with updated sauerbraten. However I cannot find an easy way to reproduce it.

I suggest using the old sauerbraten to reproduce it.
Comment 3 Fabio Pedretti 2014-05-21 10:29:09 UTC
Created attachment 99497 [details]
dmesg after the lockup with Ubuntu kernel 3.13.0-24.47 (based on 3.13.9)

Here is an updated dmesg with kernel 3.13.9 (ubuntu 3.13.0-24.47), only relevant difference is "on ring 0" is added to this line:
[11757.716140] radeon 0000:01:00.0: GPU lockup (waiting for 0x000000000003dd41 last fence id 0x000000000003dd40 on ring 0)
Comment 4 Martin Peres 2019-11-19 08:47:08 UTC
-- GitLab Migration Automatic Message --

This bug has been migrated to freedesktop.org's GitLab instance and has been closed from further activity.

You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.freedesktop.org/drm/amd/issues/480.


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.