Bug 67800 - Computer freezes after several hours (trinity)
Summary: Computer freezes after several hours (trinity)
Status: RESOLVED WORKSFORME
Alias: None
Product: DRI
Classification: Unclassified
Component: DRM/Radeon (show other bugs)
Version: DRI git
Hardware: x86-64 (AMD64) Linux (All)
: medium normal
Assignee: Default DRI bug account
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2013-08-05 19:47 UTC by Kertesz Laszlo
Modified: 2013-10-12 01:34 UTC (History)
1 user (show)

See Also:
i915 platform:
i915 features:


Attachments
Attached dmesg rright after boot (59.52 KB, text/plain)
2013-08-06 14:12 UTC, Kertesz Laszlo
no flags Details
netconsole capture until freeze (239.19 KB, text/plain)
2013-08-31 19:09 UTC, Kertesz Laszlo
no flags Details

Description Kertesz Laszlo 2013-08-05 19:47:20 UTC
I use Debian 64 bit and now i have the 3.11rc4 kernel installed, with mesa from git, xf86-video-ati from git, libdrm from git, all firmwares put in /lib/firmware.
The hardware is A8-5500 desktop APU with its built in 7560D card.
The system works very well, opengl, vdpau is functional.

I have xscreensaver set to turn off the monitor after 4 minutes, which it does. I can wake up after a few minutes by moving the mouse or pressing a key.

All is going well until i leave the computer for a longer time (it varies, usually a few hours) - the computer becomes totally unresponsive, doesnt react to keyboard/mouse and network wise is totally dead. The fans are on normally, no disk activity. I have to reset it from the reset button.
Nothing out of the ordinary in any logs, they just stop logging, presumably when the computer freezes.

Now if i disable dpms and leave only blanking, this doesnt happen (This is why i think it has to do with the kernel driver), but this mean the monitor is on all the tme which isnt good since the computer runs 24/7.

PS. It didnt happen previously with the fglrx driver in any circumstances.

Any ideas why this might happen?
Comment 1 Alex Deucher 2013-08-06 13:44:27 UTC
Please attach your dmesg output.
Comment 2 Kertesz Laszlo 2013-08-06 14:12:35 UTC
Created attachment 83722 [details]
Attached dmesg rright after boot
Comment 3 Kertesz Laszlo 2013-08-07 06:44:42 UTC
So. It seems that dpms isnt the (only) culprit after all. I left it disabled for the night and in the morning the computer was frozen again (the monitor was blanked, but still had signal).
Comment 4 Alex Deucher 2013-08-07 22:41:56 UTC
This may be a duplicate of bug 65958.  Does setting the env var RADEON_VA=0 in your /etc/environment fix the issue?
Comment 5 Kertesz Laszlo 2013-08-08 07:21:00 UTC
(In reply to comment #4)
> This may be a duplicate of bug 65958.  Does setting the env var RADEON_VA=0
> in your /etc/environment fix the issue?

(In reply to comment #4)
> This may be a duplicate of bug 65958.  Does setting the env var RADEON_VA=0
> in your /etc/environment fix the issue?

Unfortunately it doesnt solve it. I still get freezes after a few hours of idle.
Comment 6 Kertesz Laszlo 2013-08-09 09:43:05 UTC
I recompiled the kernel (from linus' git) with these patches applied:

http://lists.freedesktop.org/archives/dri-devel/2013-August/043145.html
I also disabled C6 state from BIOS.

I had no freezes since then. I still have to test some longer idle times (i had ~3 hours maximum so far), but it seems promising.
Comment 7 Kertesz Laszlo 2013-08-10 03:25:25 UTC
Last night i rebooted and re-activated C6 in the BIOS. Reactivated dpms (turn off display completely after 4 minutes).

It the morning (~7 hours of idle time) the computer works as expected. So it seems that those patches  solved this issue. I will test it further for anothey day.

As a sideline, sleep state C6 enabled in the BIOS (versus disabled) lowers the idle temperature of the APU by 2-4 C.
Comment 8 Kertesz Laszlo 2013-08-13 06:42:15 UTC
This bug seems to be gone now in this form after the aforementioned patches were installed.

But now i have random freezes here and there when i am actually working on the computer.

I had it happening during compilation (with 3/4 threads), using the browser (Firefox, Seamonkey) and playback (xv, mplayer).

I never had it happen if i used something opengl related such as games (TF2, DoD)
or xbmc. On the other hand xbmc exhibited severe memory leaks occasionally if i used vdpau.

Thermals, at least the ones indicated by sensors dont seem to be a problem - i compiled for hours stuff on it and didnt freeze. And on other occasions it froze after a few minutes. Core temperature was 44C on the last occasion.
Temperatures go up to 57C with 3-4 threads and nothing bad happens except thermal throttling (with fglrx i had up to 60C).
Comment 9 Kertesz Laszlo 2013-08-23 05:36:46 UTC
I may have found a crude workaround for the freezing issue.
It seems that if i completely disable screensaver (it was turning the monitor off with dpms after 4 minutes) and leave the monitor on permanently works. I now have 1 day uptime.
Comment 10 Kertesz Laszlo 2013-08-27 06:33:58 UTC
It froze after 1 days and 13 hours. Subsequent reboots all ended in freeze in 6-12 hours time.
Comment 11 Kertesz Laszlo 2013-08-31 19:09:31 UTC
Created attachment 84976 [details]
netconsole capture until freeze
Comment 12 Kertesz Laszlo 2013-08-31 19:17:09 UTC
I attached a netconsole capture, maybe it helps.

I set the following kernel command line options

radeon.dpm=1 ignore_loglevel drm.debug=14

drm.debug=15 had too much output, dozens of lines per second.

It doesnt have the boot part because i had to enable netconsole runtime as i use a bridge and it comes up after netconsole wants to begin logging (if i use eth0 it logs until br0 comes up) if i use the kernel command line to set up netconsole.

I have a 17" LCD Monitor connected to the CRT1 connector.
Comment 13 Kertesz Laszlo 2013-09-09 02:42:56 UTC
The freezes are still happening, both with kernel 3.11 stable or the later git versions, mesa 9.2 stable or with the 9.3 devel git.
Comment 14 Kertesz Laszlo 2013-09-28 23:25:06 UTC
Well interesting news. 
I played around with glamor acceleration, and to my surprise, my computer ran for 2 days and 16 hours or so without problems (did just about everything i do usually - browse, watch movies with vdpau, play games).
After that restarted X with exa rendering and the computer froze hard within minutes. Will test this further.
Comment 15 Kertesz Laszlo 2013-10-12 01:34:11 UTC
I think this bug can be closed. Using the kernels rc3-rc4 i never had any freezes when the computer was left idle.
With rc3 did happen a few times, but all of them happened when somehing was playing in flash player, so might be related to mesa or glamor.
Since rc3+, it never froze in any circumstances.


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.