Bug 13634 - [i915 texture cubemap is wrong
Summary: [i915 texture cubemap is wrong
Status: VERIFIED FIXED
Alias: None
Product: Mesa
Classification: Unclassified
Component: Drivers/DRI/i915 (show other bugs)
Version: unspecified
Hardware: Other Linux (All)
: medium normal
Assignee: Eric Anholt
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2007-12-12 19:50 UTC by Shuang He
Modified: 2008-03-18 22:30 UTC (History)
2 users (show)

See Also:
i915 platform:
i915 features:


Attachments
test case (3.70 KB, text/plain)
2007-12-12 19:55 UTC, Shuang He
Details
xorg conf (3.65 KB, text/plain)
2007-12-12 19:56 UTC, Shuang He
Details
xorg log (49.52 KB, text/plain)
2008-03-18 20:04 UTC, Shuang He
Details

Description Shuang He 2007-12-12 19:50:56 UTC
System Environment:
--------------------------
--Platform: FC6
--Xf86_video_intel:6f92200db39a790c95c1fac64baf0bd41a70e481 
--Mesa:e2ca788ae700aae75bf8d024c1374c38cc5574f9 
--DRM:f1a99ddc14ebca303f20b6c23bd289fc887243ae
--Kernel:2.6.22

Bug detailed description:
--------------------------
This issue doesn't happen when INTEL_NO_TTM is set
And, q965 doesn't have this issue.

when applied cubemap texture, a part of the textured rectangle is incorrect. seems this incorrect part is using another cube face's color

Reproduce steps:
----------------
start X
compile and run the attached test case


Current result:
----------------
a part of the textured rectangle is incorrect. seems this incorrect part is using    another cube face's color

Expected result:
----------------
texture cubemap should work correctly
Comment 1 Shuang He 2007-12-12 19:55:40 UTC
Created attachment 13080 [details]
test case

this test case create a cube map texture, using different color for each face.
then render a rectangle with positive x face texture, but it turns out part of the rectangle is using another face's color
Comment 2 Shuang He 2007-12-12 19:56:07 UTC
Created attachment 13081 [details]
xorg conf
Comment 3 Eric Anholt 2008-03-18 13:42:01 UTC
a better demo of the problem is comparing mesa cubemap in software versus with mesa.  both tests fail equally in ttm and non-ttm modes.
Comment 4 Eric Anholt 2008-03-18 19:52:47 UTC
Fixed many bugs in 945 cube mipmap layout, which is what I suspect this bug was about (failure to include Xorg.0.log).  Test case and cubemap now work on my GM945.
Comment 5 Shuang He 2008-03-18 20:04:39 UTC
Created attachment 15282 [details]
xorg log
Comment 6 Shuang He 2008-03-18 20:06:52 UTC
(In reply to comment #4)
> Fixed many bugs in 945 cube mipmap layout, which is what I suspect this bug was
> about (failure to include Xorg.0.log).  Test case and cubemap now work on my
> GM945.
> 

with git tip mesa, it's still not working well, only the bottom lines are rendered.
and it's working well in software rendering.
Comment 7 Eric Anholt 2008-03-18 20:18:54 UTC
I lost the commit race. The changes are actually pushed now.
Comment 8 Shuang He 2008-03-18 22:30:07 UTC
(In reply to comment #7)
> I lost the commit race. The changes are actually pushed now.
> 

x86 build is blocked by bug #15119.
So I tested x86-64, it's fixed now.
Comment 9 Shuang He 2008-03-18 22:30:27 UTC
verified, thanks, Eric


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.