Bug 25559 - Lots of errors in dmesg and screen corruption with latest drm-radeon-testing
Summary: Lots of errors in dmesg and screen corruption with latest drm-radeon-testing
Status: RESOLVED FIXED
Alias: None
Product: xorg
Classification: Unclassified
Component: Driver/Radeon (show other bugs)
Version: git
Hardware: x86-64 (AMD64) Linux (All)
: high normal
Assignee: xf86-video-ati maintainers
QA Contact: Xorg Project Team
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2009-12-10 04:00 UTC by Bob Ham
Modified: 2009-12-18 02:24 UTC (History)
1 user (show)

See Also:
i915 platform:
i915 features:


Attachments
dmesg output (39.75 KB, text/plain)
2009-12-10 04:01 UTC, Bob Ham
no flags Details
verbose debug (991 bytes, patch)
2009-12-10 05:39 UTC, Jerome Glisse
no flags Details | Splinter Review
dmesg output with patch (60.05 KB, text/plain)
2009-12-10 06:07 UTC, Bob Ham
no flags Details
Xorg configuration file (1.56 KB, text/plain)
2009-12-10 06:47 UTC, Bob Ham
no flags Details
Allow to allocate bo from system ram (1.03 KB, patch)
2009-12-10 06:57 UTC, Jerome Glisse
no flags Details | Splinter Review
kernel output after a day's uptime (67.72 KB, text/plain)
2009-12-11 05:03 UTC, Bob Ham
no flags Details
corruption in Konversation (213.94 KB, image/png)
2009-12-13 04:49 UTC, Mikko C.
no flags Details

Description Bob Ham 2009-12-10 04:00:48 UTC
Subject says it all really.  Latest git commit is:

commit fb53f8621a3fab88776ae2450a1f3afc7920231b
Author: Jerome Glisse <jglisse@redhat.com>
Date:   Wed Dec 9 21:55:10 2009 +0100

    drm/ttm: Print debug information on memory manager when eviction fails
    
    This add helper function to print information on eviction placements
    and memory manager status when eviction fails to allocate memory
    space.
    
    Signed-off-by: Jerome Glisse <jglisse@redhat.com>
    Signed-off-by: Dave Airlie <airlied@redhat.com>
Comment 1 Bob Ham 2009-12-10 04:01:41 UTC
Created attachment 31927 [details]
dmesg output
Comment 2 Jerome Glisse 2009-12-10 05:39:59 UTC
Created attachment 31933 [details] [review]
verbose debug

Please run with the attached patch and attach new full dmesg
Comment 3 Bob Ham 2009-12-10 06:07:14 UTC
Created attachment 31934 [details]
dmesg output with patch

Output of dmesg with the debugging patch and one of the error messages
Comment 4 Bob Ham 2009-12-10 06:47:01 UTC
Created attachment 31939 [details]
Xorg configuration file
Comment 5 Jerome Glisse 2009-12-10 06:57:36 UTC
Created attachment 31940 [details] [review]
Allow to allocate bo from system ram

Please try with this patch, hopefully validation have more chance of succeeding than allocation.
Comment 6 Bob Ham 2009-12-11 05:03:24 UTC
Created attachment 31975 [details]
kernel output after a day's uptime

After a day, the major problems have lessened.  However, there are still the same kind of screen corruptions going on but at a lower frequency.  I'm still getting the following error messages from the kernel:

[drm:radeon_cs_ioctl] *ERROR* Failed to parse relocation !
[drm:radeon_cs_ioctl] *ERROR* Failed to parse relocation !
[drm:radeon_cs_ioctl] *ERROR* Failed to parse relocation !
Comment 7 Mikko C. 2009-12-13 04:06:44 UTC
(In reply to comment #5)
> Created an attachment (id=31940) [details]
> Allow to allocate bo from system ram
> 
> Please try with this patch, hopefully validation have more chance of succeeding
> than allocation.
> 

Patch fixes the issue for me so far: RS880 + 2.6.32 + drm-radeon-testing.
Comment 8 Mikko C. 2009-12-13 04:07:24 UTC
Just to add, I also still get:

[drm:radeon_cs_ioctl] *ERROR* Failed to parse relocation !
[drm:radeon_cs_ioctl] *ERROR* Failed to parse relocation !
[drm:radeon_cs_ioctl] *ERROR* Failed to parse relocation !
[drm:radeon_cs_ioctl] *ERROR* Failed to parse relocation !

But no evident corruption.
Comment 9 Mikko C. 2009-12-13 04:49:21 UTC
Created attachment 32032 [details]
corruption in Konversation

I spoke too soon.
After 5-10 mins the system gradually went foobar.
Here's an example in konversation, but the corruption was all over kde, chromium, thunderbird. And it's persistent corruption, changing window did not help.
Same errors in dmesg as above:
[drm:radeon_cs_ioctl] *ERROR* Failed to parse relocation !
Comment 10 Florian Scandella 2009-12-13 09:50:41 UTC
this happens to me too after some hours of usage, rv770, amd64, drm-radeon-testing, everything from git master.
Comment 11 Bob Ham 2009-12-14 04:42:13 UTC
(In reply to comment #9)
> Created an attachment (id=32032) [details]
> corruption in Konversation

Yeah, this is exactly the corruption I get at the same time as the errors.
Comment 12 Ed Tomlinson 2009-12-15 12:58:57 UTC
With a rs780 (790gx based hd3300) I get corruption similar to the attachment right from X startup.  Bisecting (manually) found  779720a3209849be202ac36a811e934865c50971 was causing the problem.  Reverting this fixed things here.

Hope this helps.
Comment 13 Mikko C. 2009-12-15 13:18:50 UTC
Most likely same as bug #25567
Comment 14 Ed Tomlinson 2009-12-15 18:14:50 UTC
After being up several hours with commit causing corruption at startup reverted I am beginning to see some corruption very much like the attachments... 
Comment 15 Alex Deucher 2009-12-15 18:47:57 UTC
please try the patch on bug 25567
Comment 16 Mikko C. 2009-12-16 00:02:30 UTC
(In reply to comment #15)
> please try the patch on bug 25567
> 

is this to apply on top of glisse's patch? http://bugs.freedesktop.org/attachment.cgi?id=31940
Comment 17 Mikko C. 2009-12-17 23:50:04 UTC
(In reply to comment #5)
> Created an attachment (id=31940) [details]
> Allow to allocate bo from system ram
> 
> Please try with this patch, hopefully validation have more chance of succeeding
> than allocation.
> 

This patch is not in drm-radeon-testing yet. Why?
And it does not apply to it either. Hence dmesg is still full of:
radeon 0000:01:05.0: object_init failed for (7983104, 0x00000004)                                                       
[drm:radeon_gem_object_create] *ERROR* Failed to allocate GEM object (7983104, 4, 4096)

On the other side, the *ERROR* Failed to parse relocation is fixed.
Comment 18 Mikko C. 2009-12-18 02:24:31 UTC
nevermind, i did not have all the patches from drm-radeon-testing.
It works fine now, no extra patches needed :)


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.