Bug 19923

Summary: OpenOffice toolbar button corruption [EXA, r500]
Product: xorg Reporter: Øyvind Stegard <oyvind>
Component: Driver/RadeonAssignee: xf86-video-ati maintainers <xorg-driver-ati>
Status: RESOLVED FIXED QA Contact: Xorg Project Team <xorg-team>
Severity: normal    
Priority: medium CC: pedretti.fabio
Version: 7.4 (2008.09)   
Hardware: x86 (IA32)   
OS: Linux (All)   
Whiteboard:
i915 platform: i915 features:
Attachments:
Description Flags
Xorg log
none
Screenshot which shows the corruption. none

Description Øyvind Stegard 2009-02-03 01:32:06 UTC
Created attachment 22513 [details]
Xorg log

After a recent update of radeon from GIT, all the toolbar icons in OpenOffice have a "tile"-like corruption. I cannot remember seeing this problem before, and I haven't seen it in other applications, yet. Please see the attached screenshot.

I'm using xf86-video-ati with latest GIT commit being:
6fac3cefd1f46161c1e276ba40e72da2823aa9f6 (Return NULL for encoder if no active device is assigned).

Attaching Xorg.0.log as well ..
Comment 1 Øyvind Stegard 2009-02-03 01:33:01 UTC
Created attachment 22514 [details]
Screenshot which shows the corruption.
Comment 2 Øyvind Stegard 2009-02-03 01:33:58 UTC
I'm not using Compiz, only plain non-composited environment with Metacity.
Comment 3 Øyvind Stegard 2009-02-03 01:37:15 UTC
The problem is very persistent, doesn't go away on maximize/minimize, and is always there. But like stated, I've only seen it in OpenOffice thus far. And it's the first time I've seen corruption in OpenOffice since I installed Ubuntu 8.10, so I'm pretty certain it is related to me updating the radeon-driver.
Comment 4 Øyvind Stegard 2009-02-03 01:45:03 UTC
I reverted the radeon-driver back to commit 3a6e958c59b4b44fe620c8abe760ed1abf0007b1 (rs780: include RS780 in the InitMemory to leave alone), and the problem disappeared. I did this because of an EXA-related commit with more acceleration just after it (EXA: Accelerate Composite of RepeatPad/Reflect pictures when possible.).
Comment 5 Øyvind Stegard 2009-02-03 01:52:09 UTC
.. and went back up to commit 6fac3cefd1f46161c1e276ba40e72da2823aa9f6 (Return NULL for encoder if no active
device is assigned), and the problem re-appeared.
Comment 6 Michel Dänzer 2009-02-03 02:31:53 UTC
I'm seeing this as well, but I'm not sure how this can break, as OpenOffice doesn't seem to use RepeatPad or RepeatReflect... no time to dig deeper right now unfortunately.
Comment 7 Fabio Pedretti 2009-02-09 02:46:58 UTC
I am confirming the problem. A git bisect revealed:

fa8e5a4fc236f8f15f462cb0d6164b194a65a118 is first bad commit
commit fa8e5a4fc236f8f15f462cb0d6164b194a65a118
Author: Michel Dänzer <daenzer@vmware.com>
Date:   Fri Jan 30 16:16:26 2009 +0100

    EXA: Accelerate Composite of RepeatPad/Reflect pictures when possible.

:040000 040000 4b4ec3b8ee7347995c3e61bd09c01f106f956cf3 ebb471ff05268df7ad89e4098be159704f450d23 M	src
Comment 8 Alex Deucher 2009-02-12 09:52:40 UTC
fix pushed:
c6e1bf75c399807716b99486ab3301aeba60fa79

We weren't handling the non-repeat case.  Looks like pPict->repeatType does get set properly if pPict->repeat is false.
Comment 9 Alex Deucher 2009-02-12 09:57:00 UTC
s/does get/does not get/
Comment 10 Fabio Pedretti 2009-02-13 01:26:49 UTC
confirmed fixed

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.