Bug 30513

Summary: [GM965] artifacts when using metacity as compositing manager
Product: DRI Reporter: Ivan Bulatovic <combuster>
Component: DRM/IntelAssignee: Chris Wilson <chris>
Status: CLOSED FIXED QA Contact:
Severity: normal    
Priority: medium CC: jbarnes, krh
Version: XOrg git   
Hardware: x86-64 (AMD64)   
OS: Linux (All)   
Whiteboard:
i915 platform: i915 features:
Attachments:
Description Flags
persistent artifact
none
temporary artifact none

Description Ivan Bulatovic 2010-09-30 13:06:49 UTC
When using metacity as compositing manager, pop-ups or menu windows don't dissapear and they stay always on top with content intact. Only if I open the exact same menu again or a popup then it gets redrawn.

First occurence of this behaviour was when I've upgraded my box to xserver 1.9.

I can't reproduce it always but it happenes a lot (several times a day). 

OS: Arch Linux x86_64
xserver 1.9.0
xf86-video-intel 2.13 (happened with 2.12 also)
kernel 2.6.35.6/2.6.36-rc6
mesa 7.8.3
Kernel command line: root=/dev/sda1 ro video=LVDS-1:d

Maybe it's just me but I think it happenes more often on 2.6.36-rc6. I've encountered the exact same problem on Fedora 14 beta almost instantly as I've turned on compositing mode on metacity.

Bug report on archlinux bug tracker:

https://bugs.archlinux.org/task/20860?project=1&string=xorg-server

There is a screenshot also.
Comment 1 Chris Wilson 2010-09-30 16:43:40 UTC
Kristian, more damage problems? Is this something that may have been fixed post 1.9?
Comment 2 Kristian Høgsberg 2010-09-30 16:47:48 UTC
Can you try reverting 8d7b7a0d71e0b89321b3341b781bc8845386def6 in your X server?  This may be a duplicate of #30260, but I doubt it.  Does the artifact go away if your move a window over it or does it stay around?
Comment 3 Ivan Bulatovic 2010-09-30 17:31:12 UTC
Created attachment 39091 [details]
persistent artifact

No it stays around until I open the exact same menu item. Sometimes small rectangle as a part of an active window that I just minimized stays on the screen and when I move another window over it, it stays there but content of the overlaping window glues to it, but when I restore the window I've minimized and minize it back again the rectangle disappears.

I have also other artifacts but they dissappear very quickly, on mouse move etc.
Comment 4 Ivan Bulatovic 2010-09-30 17:40:21 UTC
Created attachment 39092 [details]
temporary artifact

I'll try reverting that commit as soon as I build X from git. Had it on my test partition and overwrote it :(
Comment 5 Ivan Bulatovic 2010-10-01 17:05:38 UTC
Yes, reverting 8d7b7a0d71e0b89321b3341b781bc8845386def6 did the trick. I've been poking around menus, popups, everything I did before to reproduce this bug and no artifacts at all.

TY Kristian, Chris.
Comment 6 Jonathan 2010-10-20 04:36:40 UTC
I would like to point out that this bug does not seem to be limited to intel.  I had the same problem with metacity compositing + ati and reverting 8d7b7a0d71e0b89321b3341b781bc8845386def6 solved it.
Comment 7 Ivan Bulatovic 2010-10-20 04:46:53 UTC
That is true. I've made a PKGBUILD for xserver that reverts this commit and many archlinux users reported that this fixes their problems with artifacts and compositing window managers and it's not limited to users with intel gpu's (ati and nvidia also).
Comment 8 Chris Wilson 2010-12-10 07:03:22 UTC
I believe this is now fixed with:

commit f36153e3ef6b13a87d016caab09cc9be274b0dd5
Author: Eric Anholt <eric@anholt.net>
Date:   Thu Oct 28 20:46:22 2010 -0700

    Replace usage of DamageRegionAppend with DamageDamageRegion to fix reportAft
    
    In all these cases, any rendering implied by this damage has already
    occurred, and we want to get the damage out to the client.  Some of
    the DamageRegionAppend calls were explicitly telling damage to flush
    the reportAfter damage out, but not all.
    
    Bug #30260. Fixes the compiz wallpaper plugin with client damage
    changed to reportAfter.
    
    Signed-off-by: Eric Anholt <eric@anholt.net>
    Reviewed-by: Keith Packard <keithp@keithp.com>
    Signed-off-by: Keith Packard <keithp@keithp.com>
Comment 9 Ivan Bulatovic 2010-12-29 17:17:57 UTC
It seems that for some commit f36153e3ef6b13a87d016caab09cc9be274b0dd5 doesn't solve issues with compositing wm (xcompmgr and urxvt in this case) while reverting 8d7b7a0d71e0b89321b3341b781bc8845386def6 helps.

https://bbs.archlinux.org/viewtopic.php?id=106017

For now only Intel users...

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.