Bug 47388

Summary: [i965] Piglit fbo/fbo-sys-blit randomly fails under compositor
Product: Mesa Reporter: lu hua <huax.lu>
Component: Drivers/DRI/i965Assignee: Ian Romanick <idr>
Status: VERIFIED FIXED QA Contact:
Severity: major    
Priority: high CC: xunx.fang
Version: git   
Hardware: All   
OS: Linux (All)   
Whiteboard:
i915 platform: i915 features:
Attachments: Only invalidate if the WindowPixmap changes
Flush writes to FakeFront on creation

Description lu hua 2012-03-15 20:35:20 UTC
System Environment:
--------------------------
Arch:             i386
Platform:         Ivybridge
Libdrm:		(master)2.4.31-17-g99c73378a1b440bcf594742445dfe14ab1e89128
Mesa:		(master)99cd475cc9d3bd54140f84c24b55b9e05d7310a1
Xserver:  	(master)xorg-server-1.12.0-1-ga615b90cab7569fae9d123e4da1d3373c871d84b
Xf86_video_intel:(master)2.18.0-98-g27f8f121b3d7433c6c3470b048428ca088acd2c9
Libva:		(vaapi-ext)f8be4a5477e58c4a27170f0c87eeedac8de60aef
Libva_intel_driver:(vaapi-ext)304bf4dd752c5d0edd6bcdf87efaaa7db503a1c3
Kernel: 	(drm-intel-next-queued) fa37d39e4c6622d80bd8061d600701bcea1d6870

Bug detailed description:
-----------------------------
It happens on sandybridge, ivybridge and ironlake with mese master branch and mesa 8.0 branch.
It fails 9 in 20 cycles under gnome mode.
It passes under X mode.
Case 'fbo/fbo-sys-sub-blit' also randomly fails.

Output:
Probe at (0,0)
  Expected: 0.000000 1.000000 0.000000
  Observed: 0.000000 0.000000 0.000000
Probe at (127,0)
  Expected: 0.000000 1.000000 0.000000
  Observed: 0.000000 0.000000 0.000000
Probe at (0,127)
  Expected: 0.000000 1.000000 0.000000
  Observed: 0.000000 0.000000 0.000000
Probe at (127,127)
  Expected: 0.000000 1.000000 0.000000
  Observed: 0.000000 0.000000 0.000000
PIGLIT: {'result': 'fail' }



Reproduce steps:
----------------------------
1. start X
2. start gnome-session
3. ./bin/fbo-sys-sub-blit -auto
Comment 1 lu hua 2012-03-15 22:30:55 UTC
It does not happen on Pineview.
Comment 2 Chris Wilson 2012-03-16 13:20:38 UTC
Created attachment 58578 [details] [review]
Only invalidate if the WindowPixmap changes

But mesa can also squash this invalidation itself, the logic appears to be all there to spot the texture reuse.
Comment 3 Gordon Jin 2012-03-31 18:33:36 UTC
Hua, can you try the patch? And please retest with bug#44001 together.
Comment 4 Chris Wilson 2012-07-05 03:09:04 UTC
Created attachment 63854 [details] [review]
Flush writes to FakeFront on creation

Fails consistently now, without a compositor.
Comment 5 Chris Wilson 2012-07-05 04:40:53 UTC
*** Bug 47129 has been marked as a duplicate of this bug. ***
Comment 6 Gordon Jin 2012-10-08 03:47:47 UTC
Chris, do you need QA to test your patch?
Comment 7 Kenneth Graunke 2014-08-07 21:52:48 UTC
Seems to be working okay these days.  Feel free to reopen if it still fails for you.
Comment 8 lu hua 2014-08-12 06:26:14 UTC
Run 20 cycles, It works well. Close it.

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.