[Nouveau] [Bug 76732] Kworker using 100% CPU

bugzilla-daemon at freedesktop.org bugzilla-daemon at freedesktop.org
Fri Mar 28 09:50:59 PDT 2014


https://bugs.freedesktop.org/show_bug.cgi?id=76732

--- Comment #12 from Ilia Mirkin <imirkin at alum.mit.edu> ---
(In reply to comment #11)
> Maybe now I can clarify a little bit.
> 
> 
> If nouveau loads at boot, the problem is always there, even with
> drm_kms_helper.poll=0.
> 
> 
> If nouveau is blacklisted at boot i have two possible situations:
> 1)
> - i have gdm on the screen
> - through ssh load drm_kms_helper with poll=0
> - load nouveau
> - the screen goes black, or sometimes it displays some rubbish
> - in dmesg whis appears:
> [   66.938644] pci_pm_runtime_suspend():
> nouveau_pmops_runtime_suspend+0x0/0xd0 [nouveau] returns -22
> - i press ctrl+alt+F1 and nothing happens but this appears in dmesg:
> [  175.868791] detected fb_set_par error, error code: -22
> -> CPU load is normal

Yeah, nouveau suspends and is gone forever. You can disable runtime pm with
nouveau.runpm=0. A fix for this should be in 3.14-final, I think.

> 
> 2)
> - switch to tty1
> - through ssh load drm_kms_helper with poll=0
> - load nouveau
> - screen changes resolution and still works, also changing tty works
> - CPU load at 100%
> 
> 
> All the 3 situations are reproducible all the time

In one of the cpu load at 100% situations, could you do a

perf record -g -a

(as root), and attach the resulting perf.data file after like 30s of runtime,
along with your kallsyms (which should include nouveau and drm* symbols...
grabbing /proc/kallsyms from a running kernel should work).

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/nouveau/attachments/20140328/06772ecc/attachment.html>


More information about the Nouveau mailing list