with latest ddx/mesa/drm i get strange rendering color problems only with cairo engine used by firefox for example. if i go to a site with offending colors (like padoca.wordpress.com), the grey/cyan background is rendered almost black instead of the pattern, and i get A LOT of this in syslog:
Jul 6 14:49:52 kerberos kernel: [ 6933.106204] [drm:r300_cs_track_check] *ERROR* [drm] No buffer for color buffer 0 !
Jul 6 14:49:52 kerberos kernel: [ 6933.106209] [drm:radeon_cs_ioctl] *ERROR* Invalid command stream !
and by a LOT i really mean a LOT :D
xorg is xorg-server 18.104.22.168+git20090705.fd4eed69-0~padoka1 (groo@kerberos).
Created attachment 27422 [details] [review]
Does it still happen with current xf86-video-ati Git?
If so, does this patch fix it?
(In reply to comment #1)
> Created an attachment (id=27422) [details]
> Possible fix
> Does it still happen with current xf86-video-ati Git?
> If so, does this patch fix it?
half fixed, messages and syslog dont dump any errors anymore, but i still get the wrong colors, see snap1 for firefox and snap2 for arora that uses the qt render (i enabled it by default) and doesnt suffer from the aforementioned bug.
Created attachment 27426 [details]
snap1 firefox render error
Created attachment 27427 [details]
arora correct colors
work around the problem?
(In reply to comment #5)
> Option "EXANoComposite"
> work around the problem?
yes, it does, but then i have a lot of render problems with kde4 (icons not showing in kde systray for ex).
*** Bug 23625 has been marked as a duplicate of this bug. ***
(In reply to comment #6)
> yes, it does, but then i have a lot of render problems with kde4 (icons not
> showing in kde systray for ex).
I think that was a regression in pixman 0.15.x which should be fixed in 0.16.
So the problem only happens with KMS, right? Does it not happen when booting without KMS even using the same driver binary?
I've just updated pixman to latest git, and recompile xorg-server but no change, same corrupted background
(In reply to comment #9)
> I've just updated pixman to latest git, and recompile xorg-server but no
> change, same corrupted background
That paragraph was only referring to the problems with ExaNoComposite.
Oh my, the patch had the dst/src pictures reversed in the PrepareComposite call...
Fixed now in commit 794ae743c305331741e7cf45dd71c755fe325ed5 .
i can confirm it fixed the bug, thank you very much :)
*** Bug 23383 has been marked as a duplicate of this bug. ***