Bug 33943 - Image disturbances after update to xf86-video-ati-6.14.0
Summary: Image disturbances after update to xf86-video-ati-6.14.0
Status: RESOLVED FIXED
Alias: None
Product: xorg
Classification: Unclassified
Component: Driver/Radeon (show other bugs)
Version: 7.6 (2010.12)
Hardware: x86-64 (AMD64) Linux (All)
: medium major
Assignee: xf86-video-ati maintainers
QA Contact: Xorg Project Team
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2011-02-05 13:06 UTC by Jonathan Weber
Modified: 2012-01-05 16:37 UTC (History)
0 users

See Also:
i915 platform:
i915 features:


Attachments
Image disturbances (KDE Kickoff) (15.63 KB, image/jpeg)
2011-02-05 13:06 UTC, Jonathan Weber
no flags Details
Image disturbances (web page dict.leo.org) (45.05 KB, image/jpeg)
2011-02-05 13:09 UTC, Jonathan Weber
no flags Details
xorg configuration in xorg.conf.d/ (786 bytes, text/plain)
2011-02-05 13:17 UTC, Jonathan Weber
no flags Details
Xorg.0.log (51.55 KB, text/plain)
2011-02-05 13:20 UTC, Jonathan Weber
no flags Details
Xorg.0.log with driver version 6.14.2 (43.24 KB, text/plain)
2011-05-31 14:25 UTC, Jonathan Weber
no flags Details

Description Jonathan Weber 2011-02-05 13:06:48 UTC
Created attachment 42971 [details]
Image disturbances (KDE Kickoff)

Update the xf86-video-ati from version 6.13.2 to 6.14.0 results in image disturbances on the KDE desktop.

Steps to Reproduce (on my computer):
PC configuration: Radeon HD5830, AMD X6, dual screen both over DVI
1) Update xf86-video-ati from 6.13.2 to 6.14.0 on 64bit Arch Linux
2) Start KDE (kdm)
3) User login

Actual Results:
Image disturbance in KDE, see attachment

Expected Results:
Nice KDE desktop

Build Date & Platform:
Fre 04 Feb 2011 14:05:30 CET (xf86-video-ati 6.14.0)

Additional Builds and Platforms:
Doesn't occur on build xf86-video-ati 6.13.2-2 (Arch Linux version) build date Sam 09 Oct 2010 15:52:37 CEST

Additional Information:
+ Add Option "EXAPixmaps" "Off" to xorg.conf resolves the image disturbances, but KDE slows down and other screen refresh disturbance occure
- Clean pixmap caches (kpc directory in /tmp/var and ~/.kde4) in KDE doesn't solve the problem
- Turn on/off KDE composite doesn't solve the problem
+ Reinstall xf86-video-ati 6.13.2-2 resolves the image disturbances
Comment 1 Jonathan Weber 2011-02-05 13:09:06 UTC
Created attachment 42972 [details]
Image disturbances (web page dict.leo.org)
Comment 2 Jonathan Weber 2011-02-05 13:17:47 UTC
Created attachment 42973 [details]
xorg configuration in xorg.conf.d/
Comment 3 Jonathan Weber 2011-02-05 13:20:06 UTC
Created attachment 42974 [details]
Xorg.0.log
Comment 4 Jonathan Weber 2011-02-06 08:14:40 UTC
Some information about the older xf86-video-ati driver in Arch Linux:
Arch Linux xf86-video-ati version 6.13.2-2 is the 6.13.2 version (http://xorg.freedesktop.org/releases/individual/driver/xf86-video-ati
-6.13.2.tar.bz2) , it is just a safety rebuild for new libdrm 

Arch Linux package file in git:
http://projects.archlinux.org/svntogit/packages.git/commit/xf86-video-ati/trunk/PKGBUILD?id=e924af0a311e2716db9fdfb80464b7dad97eb01a
Comment 5 Michel Dänzer 2011-02-08 01:14:09 UTC
Can you try to isolate which driver change introduced the problem with git bisect?
Comment 6 Alex 2011-02-08 09:49:22 UTC
I have updated yesterday to xf86-video-ati 6.14.0.

Steps to repeat the problem:
startx
opening xterm, aterm or eterm
output: http://i.imgur.com/midmJ.jpg
xorg cannot be used after and must be killed

However, after downgrading to xf86-video-ati-6.13.2-2 the problem goes away.

[2011-02-08 12:02] Running 'pacman -U xf86-video-ati-6.13.2-2-i686.pkg.tar.xz'
[2011-02-08 12:02] upgraded xf86-video-ati (6.14.0-1 -> 6.13.2-2)

My WM:
openbox
My video card:
01:00.0 VGA compatible controller: ATI Technologies Inc RV350 AS [Radeon 9550]

uname -a
Linux p4p 2.6.37-ARCH #1 SMP PREEMPT Sat Jan 29 19:40:04 UTC 2011 i686 Intel(R) Pentium(R) 4 CPU 2.66GHz GenuineIntel GNU/Linux
Comment 7 Alex Deucher 2011-02-08 10:25:13 UTC
Does:
Option "ColorTiling" "False"
in the device section of your xorg.conf fix the issue?

I think the following bugs are probably the same issue:
bug 33929
bug 33952
bug 33943
bug 33963
Comment 8 Alex 2011-02-08 10:52:21 UTC
(In reply to comment #7)
> Does:
> Option "ColorTiling" "False"
> in the device section of your xorg.conf fix the issue?

Yes! I can now open any xterms with no problems.

> 
> I think the following bugs are probably the same issue:
> bug 33929
> bug 33952
> bug 33943
> bug 33963
Comment 9 Michel Dänzer 2011-02-08 11:00:21 UTC
Hmm, the attached log file claimed tiling was disabled...

*** This bug has been marked as a duplicate of bug 33929 ***
Comment 10 Alex Deucher 2011-02-08 11:12:35 UTC
(In reply to comment #9)
> Hmm, the attached log file claimed tiling was disabled...

I think there are two issues on this bug.  Jonathan has an evergreen card which did not support acceleration until 6.14.0, so his issue is probably related to no accel (6.13.x) vs. accel (6.14.x).  Alex's issue is a dup of bug 33929.
Comment 11 Michel Dänzer 2011-02-08 11:16:18 UTC
Sorry, failed to notice Alex wasn't the original reporter.
Comment 12 Jonathan Weber 2011-02-09 03:42:01 UTC
What should I try to find the source of this error? Did you need any other log files?
Comment 13 Jonathan Weber 2011-05-14 05:06:05 UTC
Hallo again

I sticked around with the closed source driver for a while (Arch Linux). But now I switched back to the open source driver (new installation, probelem still exist).
I tried the Kubuntu 11.04 live DVD and I had the same problem.
So I think there is a driver or a hardware problem.

Is there such a thing like a text bench for EXA or how can I try to track down the problem?

Hardware configuration:
-HIS HD 5830 iCooler V 1GB (single screen for the moment)
-Asus M4A89GTD Pro/USB3 with AMD 890GX/SB850 chipset (890GX/Radeon HD4290 VGA controller not visible with lspci)
-AMD Phenom II X6 1055T

Any help is appreciated.
Jonathan
Comment 14 Jonathan Weber 2011-05-31 14:21:04 UTC
With the latest radeon git version (one commite after xf86-video-ati-6.14.2).

I get

[   102.453] (II) RADEON(0): VRAM usage limit set to 222991K

in Xorg.0.log but my card has 1GB of VRAM in the older Xorg.0.log (2011-02-05) file it reports

[  2666.329] (II) RADEON(0): VRAM usage limit set to 921888K

Is this normal or can it be an initialisation problem?
Can it be a conflict with the internal GPU in the northbridge (should be disabled, not visible with lspci)?

Maybe next week I can test this HIS HD 5830 card in a sandy bridge system (with kubuntu 11.04 live CD).

Greeting
Jonathan
Comment 15 Jonathan Weber 2011-05-31 14:25:47 UTC
Created attachment 47408 [details]
Xorg.0.log with driver version 6.14.2

Strange VRAM report in Xorg.0.log file.
Driver version 6.14.2
Comment 16 Jonathan Weber 2011-05-31 14:43:20 UTC
(In reply to comment #14)
> With the latest radeon git version (one commite after xf86-video-ati-6.14.2).
> 
> I get
> 
> [   102.453] (II) RADEON(0): VRAM usage limit set to 222991K
> 
> in Xorg.0.log but my card has 1GB of VRAM in the older Xorg.0.log (2011-02-05)
> file it reports
> 
> [  2666.329] (II) RADEON(0): VRAM usage limit set to 921888K
> 
> Is this normal or can it be an initialisation problem?
> Can it be a conflict with the internal GPU in the northbridge (should be
> disabled, not visible with lspci)?
> 
> Maybe next week I can test this HIS HD 5830 card in a sandy bridge system (with
> kubuntu 11.04 live CD).
> 
> Greeting
> Jonathan

ok dmesg report:
[    5.991245] [drm] radeon: 1024M of VRAM memory ready
Comment 17 Jonathan Weber 2012-01-05 16:37:49 UTC
With the git driver from 6. jan 2012 the disturbance are gone. Maybe this is relatede to the solved bug #42913.

But now the mouse cursor is gone :/ in KDE (in xfce the cursor is visible)  I think this is an other problem. Some people on http://forums.opensuse.org/english/get-technical-help-here/install-boot-login/469968-mouse-cursor-disappears-when-hovering.html have the same problem.


So I will close this bug, many thanks for continuously improving the open source driver
Jonathan


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.