Bug 51530

Summary: Accelerated rendering corruption for certain window sizes
Product: xorg Reporter: denisgolovan
Component: Driver/RadeonAssignee: xf86-video-ati maintainers <xorg-driver-ati>
Status: RESOLVED INVALID QA Contact: Xorg Project Team <xorg-team>
Severity: normal    
Priority: medium    
Version: unspecified   
Hardware: Other   
OS: All   
Whiteboard:
i915 platform: i915 features:
Attachments:
Description Flags
xorg.conf
none
dmesg none

Description denisgolovan 2012-06-28 11:32:13 UTC 
Hi

I am getting corrupted output when some windows try to use OpenGL acceleration (like Chrome, Opera, etc). It looks like it is related to specific window sizes. 
Some width/height combinations work correctly.

Following errors appear in syslog.

Jun 28 21:23:15 comp1 radeon 0000:01:00.0: evergreen_surface_check_2d:285 cb height 1088 invalid must be aligned with 128
Jun 28 21:23:15 comp1 radeon 0000:01:00.0: evergreen_cs_track_validate_cb:427 cb[0] invalid (0x000000e7 0x00007b3f 0x000108a0 0x01088468)
Jun 28 21:23:15 comp1 radeon 0000:01:00.0: evergreen_packet3_check:2096 invalid cmd stream 509
Jun 28 21:23:15 comp1 [drm:radeon_cs_ib_chunk] *ERROR* Invalid command stream !

P.S. I am on latest radeon git driver (179b035835bfa99283d44ba2d5ee17e686196535), ATI Radeon 5850 dual-head, Gentoo x86, xorg-server 1.8.2
Comment 1 Alex Deucher 2012-06-28 12:13:25 UTC 
I think this should be fixed with the latest libdrm and ddx and kernel.
Comment 2 denisgolovan 2012-06-28 13:43:40 UTC 
Could you be more specific?

I am using latest libdrm (git) and 3.4.4 kernel.
The only commit related to radeon made recently in 3.4.4 is 789ed2afd385af95798c6ec419ad41f8801975a4, made by you.

Unfortunately, it does not solve the issue.
Comment 3 Michel Dänzer 2012-06-28 23:53:40 UTC 
What version of Mesa are you using? 3D acceleration is mostly handled by the Mesa driver.
Comment 4 denisgolovan 2012-06-29 10:38:13 UTC 
(In reply to comment #3)
> What version of Mesa are you using? 3D acceleration is mostly handled by the
> Mesa driver.

Mesa 8.0.3. But it is innocent :)

I've bisected down to 615033f2b5e3817e335e9d022fc9fdcf8ac8b11a (r600-evergreen: use common surface allocator for tiling v11). That's the first commit that gives problems.

Unfortunately I could not revert it on top of current HEAD (5ebebfdb4fadada157d1369f35a8610c0fa72e02). It gives conflicts.

Please investigate.
Comment 5 Alex Deucher 2012-06-29 10:43:37 UTC 
Have you enabled 2D tiling in your xorg.conf?  mesa 8.0.x is too old for 2D tiling.  Please attach your xorg log and dmesg output.
Comment 6 denisgolovan 2012-06-29 12:30:20 UTC 
Created attachment 63617 [details]
xorg.conf
Comment 7 denisgolovan 2012-06-29 12:34:10 UTC 
Created attachment 63618 [details]
dmesg
Comment 8 denisgolovan 2012-06-29 12:35:55 UTC 
Yes. 2D tiling is enabled.
Comment 9 Alex Deucher 2012-06-29 13:04:54 UTC 
You need mesa from git to use 2D tiling.
Comment 10 denisgolovan 2012-06-30 02:48:56 UTC 
Corfirmed.
Using mesa from git fixes problems with ColorTiling 2D.
Thanks for explanations.

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.