[i965] Open GL Renders flat surfaces incorrectly

Bug #206287 reported by Alvin Gao
36
Affects Status Importance Assigned to Milestone
Mesa
Fix Released
Medium
mesa (Ubuntu)
Fix Released
Undecided
Unassigned
Hardy
Fix Released
Undecided
Unassigned

Bug Description

Flat surfaces in games/OpenGL apps render oddly, making it very hard to use them.

Other than that, I really have no further information.

Hardy Beta x86
Graphics Card: Display controller: Intel Corporation Mobile GM965/GL960 Integrated Graphics Controller (rev 0c)

SCREENSHOTS:
http://farm4.static.flickr.com/3184/2299007958_6a34ce5bb1_o.jpg

Tags: gm965 intel

Related branches

Revision history for this message
In , Gordon Jin (gordon-jin) wrote :

At least open arena works on my GM965.

Revision history for this message
In , Mario-galanti (mario-galanti) wrote :

Hi,

I can confirm that the bug is present if I use the 7.0.3-rc2 version of the mesa library; I use Debian testing and I have an i965 video card. If I try to play Vegastrike or Armagetron, I can see exactly the same kind of graphics corruption that is shown in the screenshot:

http://bugs.debian.org/cgi-bin/bugreport.cgi?msg=5;filename=Mesaproblem.png;att=1;bug=470084

I have not tried other 3D games or apps, so I cannot say whether there are other programs affected by this bug on my system.

Also for me, the 7.0.2-4 (as named by Debian) version of the library produced no artefacts.

Kind regards,

Mario

Revision history for this message
In , Gordon Jin (gordon-jin) wrote :

*** Bug 14969 has been marked as a duplicate of this bug. ***

Revision history for this message
Alvin Gao (picardpwnskirk) wrote : Open GL Renders "Funny"

Flat surfaces in games/OpenGL apps render oddly, making it very hard to use them.

Other than that, I really have no further information.

Hardy Beta x86
Graphics Card: Display controller: Intel Corporation Mobile GM965/GL960 Integrated Graphics Controller (rev 0c)

SCREENSHOTS:
http://farm4.static.flickr.com/3184/2299007958_6a34ce5bb1_o.jpg

Revision history for this message
ewanm89 (ewanm89-googlemail) wrote :

Could you please attach xorg.conf and xorg.0.log?

Revision history for this message
Alvin Gao (picardpwnskirk) wrote :

Xorg.conf

Revision history for this message
Alvin Gao (picardpwnskirk) wrote :

xorg.0.log

Revision history for this message
Gunni (fgunni) wrote :

Just for confirmation: I have the same effect here with game sauerbraten on hardy with GM965 chipset.

Revision history for this message
unggnu (unggnu) wrote :

Do you have INTEL_BATCH=1 enabled/added?

Changed in xserver-xorg-video-intel:
status: New → Incomplete
Revision history for this message
Alvin Gao (picardpwnskirk) wrote :

I enabled INTEL_BATCH=1, but nothing happened. I experienced the same problems that I had without this enabled.

Revision history for this message
Bryce Harrington (bryce) wrote : Re: Open GL Renders flat surfaces incorrectly

Sort of sounds like it might be bug 96991 - does the issue go away if you turn off compiz? If so, then this is just a dupe of that bug, and will be fixed when TTM becomes available (Intrepid or Intrepid+1 probably).

Also, INTEL_BATCH=1 is not recommended - according to upstream, in theory it can cause instabilities for certain corner cases, so if you don't notice a benefit from it, don't use it.

Revision history for this message
Alvin Gao (picardpwnskirk) wrote :

I never turn on compiz, in fact, I removed the packages from my install.

This was working correctly in the previous Hardy Alphas, and in Gutsy.

Revision history for this message
Bryce Harrington (bryce) wrote :

Ah, then perhaps it was the "greedy" migration heuristic bug that triggered it. Do you remember which of the Alphas that it last worked, and which one it started to misbehave? Or do you recall the driver version where the problem started?

Revision history for this message
Alvin Gao (picardpwnskirk) wrote :

It was around Alpha 4 that it stopped working, I think.

My skills at memorization still need refining.

Revision history for this message
unggnu (unggnu) wrote :

I guess it could be tested through adding the line
Option "MigrationHeuristic" "always"
to your xorg.conf device section and restarting X.
This should enable the previous default mode.

Revision history for this message
Alvin Gao (picardpwnskirk) wrote :

That did not change anything, so I guess it wasn't that problem.

Revision history for this message
Bryce Harrington (bryce) wrote :

No, I think with the patch we did, we force "greedy" regardless of what's set in xorg.conf. Perhaps what we should do is collect pci id's of people experiencing this problem and not set "greedy" in those cases. Not sure.

Revision history for this message
KillerKiwi (killerkiwi2005) wrote :

confirmed same issue Mobile GM965/GL960 Integrated Graphics Controller

Games are unplayable (SDL I think), compiz has no effect, worked in gutsy

Revision history for this message
Alvin Gao (picardpwnskirk) wrote :

I think this bug is confirmed...

Anyone want to change the status, or provide a workaround?

Revision history for this message
unggnu (unggnu) wrote :

Has anyone proofed with the older Intel driver/ oldHardy Live CD if it is really the fault of Greedy?

Revision history for this message
KillerKiwi (killerkiwi2005) wrote :

which alpha was greedy turned on at? I'll test it.

Revision history for this message
unggnu (unggnu) wrote :

Greedy was enabled for i965 chipset with Intel driver version 2:2.2.0.90-2ubuntu4 on 14 Feb 2008 so it should be Hardy Alpha 4. The problem is that not only greedy was activated since then but a new Intel upstream release was uploaded afaik.

Revision history for this message
Christophe Dumez (hydr0g3n) wrote :

Maybe you could post a package with Greedy disabled here so that we can test and tell you if it works better?

Revision history for this message
Timo Aaltonen (tjaalton) wrote :

note that the 3D-bits are in mesa, not the intel driver. That said, to get the default EXA behaviour, use 'Option "MigrationHeuristic" "always"'.

Revision history for this message
KillerKiwi (killerkiwi2005) wrote :

'Option "MigrationHeuristic" "always"'.

does not fix the issue..... if any thing it seemed slightly worse (but that might be my imagination)

Revision history for this message
In , Otakar-trunecek (otakar-trunecek) wrote :

Same bug in archlinux
http://bugs.archlinux.org/task/9866
Similar glitches in opengl games with mesa 7.0.3rc2 and intel graphics driver.

Revision history for this message
In , Tristan (tristan-willy) wrote :

Created an attachment (id=15574)
Re-enable -ve rhw workaround for all i965 chips.

Revision history for this message
In , Tristan (tristan-willy) wrote :

I noticed the same corruption in git (mesa_7_0_branch) and decided to track it down yesterday. The offending commit is da476ff02dd357d9909a2df9891ee645f5f776a0. It appears a workaround that was always on was changed to IGD chips only. It appears that my G965 requires this workaround, so here's a patch to include all i965 chips when executing the workaround. Patch included.

Regarding the patch, should the -ve rhw workaround even need a test? The DRI driver is for all 965 chips after all.

Revision history for this message
In , Tristan (tristan-willy) wrote :
Revision history for this message
In , Haihao-xiang (haihao-xiang) wrote :

(In reply to comment #6)
> I noticed the same corruption in git (mesa_7_0_branch) and decided to track it
> down yesterday. The offending commit is
> da476ff02dd357d9909a2df9891ee645f5f776a0. It appears a workaround that was
> always on was changed to IGD chips only. It appears that my G965 requires this
> workaround, so here's a patch to include all i965 chips when executing the
> workaround. Patch included.
>
> Regarding the patch, should the -ve rhw workaround even need a test? The DRI
> driver is for all 965 chips after all.
>

Thanks . I make a mistake in that commit. The -ve rhw workaround should be applied to 965, not IGD

Revision history for this message
In , Haihao-xiang (haihao-xiang) wrote :

*** Bug 14794 has been marked as a duplicate of this bug. ***

Revision history for this message
kakukri (kakukri) wrote :

Looking at the screenshot it looks like the same problem as in bug #199823. The good news is that it is supposed to be fixed upstream.

Revision history for this message
Christophe Dumez (hydr0g3n) wrote :

The patch is available there: https://bugs.freedesktop.org/show_bug.cgi?id=14940

Couldn't you repackage libgl1-mesa-glx and apply this patch (please :)?

Revision history for this message
KillerKiwi (killerkiwi2005) wrote :

pretty please......

Revision history for this message
Christophe Dumez (hydr0g3n) wrote :

Just so you know. I repackaged mesa after applying the patch from the other bug report. My games are working fine again now :) I think it would really be nice if you packaged a patched version (or maybe a git snapshot) for Hardy. A lot of users have this video chipset...

Revision history for this message
Christophe Dumez (hydr0g3n) wrote :

By the way, the bug can be marked as a duplicate (this one is the same: #199823)

Revision history for this message
Christophe Dumez (hydr0g3n) wrote :
Revision history for this message
Alvin Gao (picardpwnskirk) wrote :

Uhh, n00b question.

How do I apply this patch?

Alternative: Is there any way I can get a package with the patch applied?

Thanks

Revision history for this message
Christophe Dumez (hydr0g3n) wrote :

You can use mine (http://www.dchris.eu/libmesa). I'm using them and it works.

Changed in mesa:
status: Unknown → Fix Released
Revision history for this message
Alvin Gao (picardpwnskirk) wrote :

Dude....

I love you.

(not in any sick or perverted way)

Revision history for this message
Launchpad Janitor (janitor) wrote :

This bug was fixed in the package mesa - 7.0.3~rc2-1ubuntu3

---------------
mesa (7.0.3~rc2-1ubuntu3) hardy; urgency=low

  * debian/patches/107_fix_-ve_rhw_regression.patch: Fixes regression
    introduced upstream where a -ve rhw workaround is being applied only
    for IGD systems, when it should be applied only for *non*-IGD systems.
    This resulted in rendering issues in 3D games and drawing programs.
    (LP: #199823, #206287)

 -- Bryce Harrington <email address hidden> Fri, 04 Apr 2008 19:46:40 -0700

Changed in mesa:
status: Incomplete → Fix Released
Revision history for this message
Christophe Dumez (hydr0g3n) wrote :

Thanks, but I read somewhere that mesa 7.0.3 final was released :)

Revision history for this message
In , Gordon Jin (gordon-jin) wrote :

*** Bug 15351 has been marked as a duplicate of this bug. ***

Changed in mesa:
importance: Unknown → Medium
Changed in mesa:
importance: Medium → Unknown
Changed in mesa:
importance: Unknown → Medium
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Duplicates of this bug

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.