Bug 109684 - [SNBGT1] KHR-GL33.transform_feedback.query_geometry_separate_test fails
Summary: [SNBGT1] KHR-GL33.transform_feedback.query_geometry_separate_test fails
Status: NEEDINFO
Alias: None
Product: Mesa
Classification: Unclassified
Component: Drivers/DRI/i965 (show other bugs)
Version: git
Hardware: Other All
: medium normal
Assignee: Intel 3D Bugs Mailing List
QA Contact: Intel 3D Bugs Mailing List
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2019-02-20 02:43 UTC by Mark Janes
Modified: 2019-02-21 10:56 UTC (History)
0 users

See Also:
i915 platform:
i915 features:


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Mark Janes 2019-02-20 02:43:04 UTC
In Mesa i965 CI, this test fails on SNBGT1, but passes on other skus.  Typically, tests pass or fail consistently on all hardware configurations of a particular gen.

Rather than track the GT1 results separately, we will disable this test on SNB.
Comment 1 Denis 2019-02-20 11:50:41 UTC
hi Mark. It is stable failure in CI? I checked "daily" and "master" branches, and see that this test fails sometimes on SNB. But I don't see, what exact HW it was.

And it passes for me, I ran it about 10 times in a row.

I have:
Ubuntu 18.04
kernel 4.15.0-34-generic
Mesa 19.1.0-devel (git-f509213675) (latest from master)

VK-CTS from this commit (should be latest):

> commit 18fef9d688debbaa63234084a0e9baab0f11e60f (HEAD -> master, origin/master, origin/HEAD)

How I configured test:
> cmake -DDEQP_TARGET=x11_egl

Test result:

./glcts -n KHR-GL33.transform_feedback.query_geometry_separate_test
Writing test log into TestResults.qpa
dEQP Core git-18fef9d688debbaa63234084a0e9baab0f11e60f (0x18fef9d6) starting..
  target implementation = 'X11 EGL'

Test case 'KHR-GL33.transform_feedback.query_geometry_separate_test'..
  Pass (Pass)

DONE!

Test run totals:
  Passed:        1/1 (100.0%)
  Failed:        0/1 (0.0%)
  Not supported: 0/1 (0.0%)
  Warnings:      0/1 (0.0%)
Comment 2 Denis 2019-02-20 11:53:57 UTC
oh, forgot to mention that I have exactly GT1 gpu:

Device: Mesa DRI Intel(R) Sandybridge Desktop  (0x102)

Intel® Core™ i5-2500 Processor
HD Graphics 2000
Comment 3 Mark Janes 2019-02-20 16:45:12 UTC
You can determine the sku of the test machine by looking at the name of the tester in the components table:

  https://mesa-ci.01.org/mesa_master/builds/15303/group/63a9f0ea7bb98050796b649e85481845

"snbgt1-04" is the fourth sandy bridge GT1 in our CI.

The devices that we have are mobile snb gt1: 0x106

My theory may be wrong, if you can't reproduce locally.  I'll run the test by itself.
Comment 4 Mark Janes 2019-02-20 18:58:24 UTC
More information:  The test passes reliably when run by itself.



./glcts -n KHR-GL33.transform_feedback.query_geometry_separate_test
Writing test log into TestResults.qpa
dEQP Core git-cc4ad11ac991c67820dbc88e539f46ae787a3dd0 (0xcc4ad11a) starting..
  target implementation = 'X11 EGL'

Test case 'KHR-GL33.transform_feedback.query_geometry_separate_test'..
  Pass (Pass)


------------------------------------------------------------

It fails when it is run with the rest of the transform feedback tests:

./glcts -n KHR-GL33.transform_feedback.*
Writing test log into TestResults.qpa
dEQP Core git-cc4ad11ac991c67820dbc88e539f46ae787a3dd0 (0xcc4ad11a) starting..
  target implementation = 'X11 EGL'

Test case 'KHR-GL33.transform_feedback.api_errors_test'..
Mesa: User error: GL_INVALID_ENUM in glCreateShader(GL_TESS_CONTROL_SHADER)
  Fail (Fail)
<snip>
Test case 'KHR-GL33.transform_feedback.capture_geometry_separate_test'..
  Fail (Fail)
<snip>
Test run totals:
  Passed:        15/20 (75.0%)
  Failed:        5/20 (25.0%)


Denis, can you reproduce this on your system?
Comment 5 Mark Janes 2019-02-20 19:00:06 UTC
oops, I snipped the wrong section of output.  The test does fail when run with the others:

Test case 'KHR-GL33.transform_feedback.query_geometry_separate_test'..
  Fail (Fail)
Comment 6 Denis 2019-02-21 10:56:10 UTC
Yes, good catch. You are right, only during running suite - test fails.
I will ask our devs to investigate this issue, thank you.


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.