Bug 67241 - [IVB/HSW bisected] 3DMMES v2.0 render error ”Green Screen” with gnome-session enabling SNA
Summary: [IVB/HSW bisected] 3DMMES v2.0 render error ”Green Screen” with gnome-session...
Status: VERIFIED FIXED
Alias: None
Product: xorg
Classification: Unclassified
Component: Driver/intel (show other bugs)
Version: unspecified
Hardware: All Linux (All)
: high major
Assignee: Chris Wilson
QA Contact: Intel GFX Bugs mailing list
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2013-07-24 05:59 UTC by meng
Modified: 2013-07-26 09:23 UTC (History)
0 users

See Also:
i915 platform:
i915 features:


Attachments
dmesg.log (11.40 KB, text/plain)
2013-07-24 05:59 UTC, meng
no flags Details
Xorg.0.log (18.73 KB, text/plain)
2013-07-24 06:00 UTC, meng
no flags Details
scrot for the problem (36.42 KB, image/png)
2013-07-24 06:00 UTC, meng
no flags Details
HSW-ULT_urbanterror_Xorg.0.log (19.13 KB, text/plain)
2013-07-26 05:40 UTC, meng
no flags Details

Description meng 2013-07-24 05:59:02 UTC
System Environment:       
----------------------------------------------
Platform:  IVB/HSW
Mesa:	         (master)5a7bdd4b4173958c53109517b7c95f1039623e7e
Xf86_video_intel:(master)74dd10d252df2d826b6db5da7e53bd7c5d76dec0	
Kernel:	(drm-intel-nightly) 1a7fc2d3e919e151a2e87dd6c98f20eccf95e041

Bug detailed description:
----------------------------------------------
3DMMES v2.0 render error ”Green Screen” with gnome-session enabling SNA on IVB/HSW. It would be Ok on Raw X(xinit) or disable SNA. 
BTW, the render error make 3DMMES v2.0 performance “improved” such as:
3DMMES v2.0 taiji 147 -> 178 fps.
Please see Xorg.0.log and dmesg.

It’s xf86_video_intel regression, bisecting show the first bad commit is that:
commit e5036fb15983a1d3547eaab381c18b3edf4310b4
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date:   Fri Jul 19 11:51:40 2013 +0100

    sna/dri: Reject invalid DRI2Drawables if STRICT_BLIT is defined


Reproduce steps:
---------------------------------------------
1. enable SNA
2. xinit&
3. gnome-session
4. run 3DMMES v2.0
Comment 1 meng 2013-07-24 05:59:31 UTC
Created attachment 82897 [details]
dmesg.log
Comment 2 meng 2013-07-24 06:00:10 UTC
Created attachment 82898 [details]
Xorg.0.log
Comment 3 meng 2013-07-24 06:00:50 UTC
Created attachment 82899 [details]
scrot for the problem
Comment 4 Chris Wilson 2013-07-24 09:08:10 UTC
It should be informing you about the race behind errors such as #64940 and #66606. ./configure --enable-debug=full will give the reason why it rejected the DRI2 blit.
Comment 5 meng 2013-07-25 01:02:10 UTC
(In reply to comment #4)
> It should be informing you about the race behind errors such as #64940 and
> #66606. ./configure --enable-debug=full will give the reason why it rejected
> the DRI2 blit.

Hi Chris, how can I add "./configure --enable-debug=full"? Does it need other testing?
Comment 6 meng 2013-07-26 05:40:56 UTC
Created attachment 83015 [details]
HSW-ULT_urbanterror_Xorg.0.log

HSW ULT would blank screen, when run 3D OpenGL case such as urbanterror,Smokin' Guns.
Comment 7 Chris Wilson 2013-07-26 06:44:47 UTC
The root cause is the same missing DRI2Invalidate (https://bugs.freedesktop.org/attachment.cgi?id=81263) worked around by

commit 6f5fd772c7ca656b86394a0f036d4e0cf5b33d8e
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date:   Thu Jul 25 08:29:55 2013 +0100

    sna/dri: Discard the strict checking for stale bo before performing a blit
Comment 8 meng 2013-07-26 09:23:43 UTC
Verified it with above commit.


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.