Bug 384022

Summary: intel: Fonts sometimes get corrupted on VT switch with EXA ( 965GM)
Product: [openSUSE] openSUSE 11.0 Reporter: Andrew Jorgensen <ajorgensen>
Component: X.OrgAssignee: Stefan Dirsch <sndirsch>
Status: RESOLVED FIXED QA Contact: E-mail List <xorg-maintainer-bugs>
Severity: Normal    
Priority: P5 - None    
Version: Beta 1   
Target Milestone: ---   
Hardware: Other   
OS: Other   
Whiteboard:
Found By: --- Services Priority:
Business Priority: Blocker: ---
Marketing QA Status: --- IT Deployment: ---
Attachments: font corruption

Description Andrew Jorgensen 2008-04-26 00:18:39 UTC
Sometimes if I do a VT switch or if I suspend and resume (which does the same) the fonts appear corrupted when I return to X.  The fonts appear extremely blocky.
Comment 1 Andrew Jorgensen 2008-04-26 00:22:44 UTC
Created attachment 210633 [details]
font corruption
Comment 2 Andrew Jorgensen 2008-04-26 00:24:44 UTC
This happened on first boot as well.
Comment 3 Stefan Dirsch 2008-04-26 01:20:42 UTC
Thanks for the reminder. Can you still reproduce this issue when adding

  Option "AccelMethod" "XAA"

to Section "Device" of your /etc/X11/xorg.conf?
Comment 4 Forgotten User ZhJd0F0L3x 2008-04-26 09:56:42 UTC
(In reply to comment #3 from Stefan Dirsch)

>   Option "AccelMethod" "XAA"

This certainly helps for my after-suspend rendering problems.
Comment 5 Stefan Dirsch 2008-04-26 11:57:47 UTC
Thanks. It would be interesting wich EXA acceleration breaks it. Could you test the available "EXANo*" options to figure this out? See "man exa" for details.
Comment 6 Andrew Jorgensen 2008-04-29 15:24:05 UTC
Yes, the option given does avert the issue.  I have not tried the EXANo options yet.
Comment 7 Stefan Dirsch 2008-04-29 15:31:48 UTC
Thanks for the promise to test with these options. ;-)
Comment 8 Stefan Dirsch 2008-04-30 09:04:55 UTC
Due to performance reasons I decided to make XAA the default for intel driver again. Therefore I'm closing this one now as fixed. Still it would be interesting to know, which EXANo* can workaround this issue.
Comment 9 Andrew Jorgensen 2008-04-30 22:01:07 UTC
It's EXANoComposite=Yes that works around the issue.
Comment 10 Stefan Dirsch 2008-04-30 23:37:40 UTC
Thanks for the update. This is also the option, which is always mentioned, when it comes to bad performance with EXA and how to workaround it.
Comment 11 Andrew Jorgensen 2008-05-01 13:50:57 UTC
I noticed that Ubuntu uses the "greedy" EXA MigrationHeuristic (from their X logs, they must have it hard-coded in the intel driver) so I tried that in oS 11.0 and it also works around this issue.  It may also work around the performance issues.  Please give it a try by setting

Option "MigrationHeuristic" "greedy"
Comment 12 Stefan Dirsch 2008-05-01 14:20:51 UTC
This is what also the nv driver manual page recommends. But I'm afraid there is a reason, why "greedy" is *not* the default.