Bug 14163 - [G965 EXA] Display corruption on Intel 2.1 and 2.2
Summary: [G965 EXA] Display corruption on Intel 2.1 and 2.2
Status: RESOLVED FIXED
Alias: None
Product: xorg
Classification: Unclassified
Component: Driver/intel (show other bugs)
Version: unspecified
Hardware: x86 (IA32) FreeBSD
: medium major
Assignee: Eric Anholt
QA Contact: Xorg Project Team
URL:
Whiteboard:
Keywords:
Depends on:
Blocks: 15000
  Show dependency treegraph
 
Reported: 2008-01-20 19:28 UTC by Kevin Downey
Modified: 2008-04-18 01:25 UTC (History)
2 users (show)

See Also:
i915 platform:
i915 features:


Attachments
X log from offending system (95.84 KB, text/plain)
2008-01-21 23:12 UTC, Kevin Downey
no flags Details
xorg.conf from offending system (3.01 KB, text/plain)
2008-01-21 23:12 UTC, Kevin Downey
no flags Details

Description Kevin Downey 2008-01-20 19:28:46 UTC
Using EXA, fonts render as black blocks. I see this with the intel driver versions I have tried, 2.2 and 2.1. If I use XAA everything is fine.
Comment 1 Wang Zhenyu 2008-01-21 16:46:57 UTC
Too little info...pls upload your X log and xorg.conf at least.
Comment 2 Kevin Downey 2008-01-21 23:12:11 UTC
Created attachment 13850 [details]
X log from offending system
Comment 3 Kevin Downey 2008-01-21 23:12:49 UTC
Created attachment 13851 [details]
xorg.conf from offending system
Comment 4 Wang Zhenyu 2008-01-21 23:22:46 UTC
Eric, see similar thing on freebsd?
Comment 5 Kevin Downey 2008-01-24 20:51:15 UTC
I checked out the sources from git and installed what I guess is the head version of the intel driver. No change.

Poking around the list of intel bugs in the bugzilla I discovered the ExaNoComposite option. Enabling this option stops the font corruption.

I did notice that not only was there font corruption, but the drop shadows around windows turned in solid black bars around the window frame.
Comment 6 Wang Zhenyu 2008-02-02 23:34:12 UTC
My X builds on freebsd-7.0 rc1 works fine on t61 here.

Comment 7 Kevin Downey 2008-02-08 11:27:47 UTC
(In reply to comment #6)
> My X builds on freebsd-7.0 rc1 works fine on t61 here.
> 
well I have moved to the git head from intel and still have the same issues. I dunno how extensive the differences between intel's mobile chips (the t61 is a thinkpad right?) and their desktop chips are. 
Comment 8 Wang Zhenyu 2008-03-18 23:10:03 UTC
As Freebsd 7.0 has released, how about this one now?
Comment 9 Kevin Downey 2008-03-19 00:30:37 UTC
(In reply to comment #8)
> As Freebsd 7.0 has released, how about this one now?
> 

as of commit 4b9b7b007d729f94b01b0031d8ae478134b501da (xf86-video-intel) I am still seeing the same thing on 7.0-STABLE
Comment 10 Kevin Downey 2008-04-12 20:12:41 UTC
I am not sure which commit fixed this for me, I am running 03836067b77606c134c71b30c7078d09d77c95fa and EXA is working great. Thanks guys.
Comment 11 Wang Zhenyu 2008-04-13 19:38:30 UTC
So does it mean recent i965 render change also fixed the problem for you? could you try with xf86-video-intel-2.3-branch or http://xorg.freedesktop.org/archive/individual/driver/xf86-video-intel-2.2.99.903.tar.bz2?
Comment 12 Kevin Downey 2008-04-17 23:44:12 UTC
I tried xf86-video-intel-2.2.99.903 and it did not work. But it was different, weird rainbow sort of static, I get that with the current git head too. e27ce4238f748d024635835824c120eb298b84b2 is the last commit that works "perfectly" for me. I haven't rolled back to figure out what the first commit that works is.
Comment 13 Wang Zhenyu 2008-04-18 01:25:15 UTC
(In reply to comment #12)
> I tried xf86-video-intel-2.2.99.903 and it did not work. But it was different,
> weird rainbow sort of static, I get that with the current git head too.

You mean current 2.3-branch and master have same behavior to you?

> e27ce4238f748d024635835824c120eb298b84b2 is the last commit that works "perfectly" for me. 

Do you mean revert this commit then master works fine?

commit 23d1df22d177e54bfc46304053d8115047ff85d4
Author: Eric Anholt <eric@anholt.net>
Date:   Mon Apr 14 16:46:17 2008 -0700

    Abstract surface setup into a separate function.


> I haven't rolled back to figure out what the first commit that works is.
> 

Appreciate you can nail it down. It looks we have many weird behavior on desktop 965.


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.