Bug 103267 - [BSW] dEQP-GLES31.functional.primitive_bounding_box.lines.global_state.vertex_tessellation_geometry_fragment.default_framebuffer_bbox_smaller
Summary: [BSW] dEQP-GLES31.functional.primitive_bounding_box.lines.global_state.vertex...
Status: RESOLVED FIXED
Alias: None
Product: Mesa
Classification: Unclassified
Component: Drivers/DRI/i965 (show other bugs)
Version: git
Hardware: Other All
: medium normal
Assignee: Iago Toral
QA Contact: Intel 3D Bugs Mailing List
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2017-10-13 22:07 UTC by Mark Janes
Modified: 2019-03-12 17:33 UTC (History)
2 users (show)

See Also:
i915 platform:
i915 features:


Attachments

Description Mark Janes 2017-10-13 22:07:46 UTC
This test asserts intermittently:

deqp-gles31:
src/intel/common/gen_urb_config.c:151:
          gen_get_urb_config: Assertion `total_needs <= urb_chunks' failed.

First seen on mesa=e915b8d267d
not reproduced on mesa=740a1618c34
Comment 1 Mark Janes 2017-10-14 00:59:15 UTC
Bisected to mesa 5e584a9db7f

Author:     Iago Toral Quiroga <itoral@igalia.com>
i965: skip reading unused slots at the begining of the URB for the FS
Comment 2 Iago Toral 2017-10-16 08:47:01 UTC
I have tested this in a Braswell machine with mesa from git (f1eb5e6399) which includes this commit and I see no crashes. I put the test to run in a loop 100 times and all of the executions are a pass.

Mark, is there any chance that this is already fixed in master?
Comment 3 Mark Janes 2017-10-16 22:45:55 UTC
no, it still fails all the time in CI.  It may have to do with running tests in parallel, or some other aspect of how we invoke the tests.
Comment 4 Denis 2019-03-11 17:09:35 UTC
hello. Mark, does CI have BSW machine in general? I didn't find it.
I built this test on local machine and ran it separately, and in "suite"

./glcts -n "dEQP-GLES31.functional.primitive_bounding_box.lines.global_state.vertex_tessellation_geometry_fragment.*"

./glcts -n "dEQP-GLES31.functional.primitive_bounding_box.lines.global_state.vertex_tessellation_geometry_fragment.default_framebuffer_bbox_smaller"


In both cases it passed. Could you please enable it in CI again and re-check?

My SW:
kernel 4.20.14-200.fc29.x86_64
OS Fedora 29
Mesa 19.1.0-devel (git-47fc359822)
GPU Intel(R) HD Graphics 400 (Braswell)

Also I found out, that this issue has the same error with https://bugs.freedesktop.org/show_bug.cgi?id=108787 this one. But, according to my results (issue with benchmark reproduces, issue with DEQP test - no), only error is the same.
Comment 5 Mark Janes 2019-03-12 16:15:42 UTC
BSW systems run tests only for daily builds, because they are so slow:

https://mesa-ci.01.org/mesa_master_daily/builds/

Looking in the logs, this test has been running reliably on BSW for some time.  Not sure when it was fixed.
Comment 6 Denis 2019-03-12 17:33:57 UTC
aha, I see, thank you.
I wanted to reproduce this issue on bisected commit, and try to find the fix, but for me it was also stable :( even if I ran this test in a loop for 100 times (I built mesa with debug enabled).


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.