Bisect points to crrev.com/9435ed8fa7d997aa54609dd3080f367eb20c8ee4.

Robert, would you be able to take a look please? This is causing a large number of Chrome OS video unittests to fail. Should we revert for now?

To revert, it seems that we'd also need to revert:
crrev.com/abed4f3887d920d5ac0300034a2c7637fa7cda8
crrev.com/2460e71463578a452799e945b0e96c4f3cfdadcf.


To reproduce, please run video_decode_accelerator_unittest on any CrOS device (you may need to ensure you have the proper test file on the device, test files are located in src/media/test/data/) and that the argument --test_video_data points to the location of the test file, e.g.:

./video_decode_accelerator_unittest --test_video_data=/tmp/test-25fps.h264:320:240:250:258:::1  --ozone-platform=gbm

(Full documentation of the arguments at: https://cs.chromium.org/chromium/src/media/gpu/video_decode_accelerator_unittest.cc?l=110)

Thank you!

CL is posted for review.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/7a0f354be912e8aa9f0e6eef2d55905416f32922

commit 7a0f354be912e8aa9f0e6eef2d55905416f32922
Author: Robert Kroeger <rjkroege@chromium.org>
Date: Fri Feb 09 23:37:29 2018

Initialize ozone/drm correctly in video tests

In https://chromium-review.googlesource.com/861749, I changed
ozone/drm intialization semantics to make ozone/drm more flexible
about setup from WS or GpuProcessHost. I missed updating one call site
in media. Added it here.

BUG= 808962 

Cq-Include-Trybots: master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel
Change-Id: I22982b2d3019638083fb21858881a29681ee4559
Reviewed-on: https://chromium-review.googlesource.com/912405
Reviewed-by: Pawel Osciak <posciak@chromium.org>
Commit-Queue: Robert Kroeger <rjkroege@chromium.org>
Cr-Commit-Position: refs/heads/master@{#535883}
[modify] https://crrev.com/7a0f354be912e8aa9f0e6eef2d55905416f32922/media/gpu/rendering_helper.cc

video_VideoDecodeAccelerator, video_VDASanity and video_VDAPerf  are still failing on ToT.
https://cros-goldeneye.corp.google.com/chromeos/healthmonitoring/testDetails?testName=video_VideoDecodeAccelerator&suite=&daysBack=14&board=&architecture=&boardFamily=&buildConfig=&reason=&version=&milestone=66&dut=

Is this fix available in ToT?

Which boards? Which tests? I tested on a chell and after #5, the tests described in #2 passed.

This is happening on all boards for video_VideoDecodeAccelerator, video_VDASanity and video_VDAPerf. Please chck below stainless link:
https://stainless.corp.google.com/search?view=matrix&row=model&col=build&first_date=2018-02-14&last_date=2018-02-20&test=video_VideoDecodeAccelerator%7Cvideo_VDASanity%7Cvideo_VDAPerf&build=R66-*&exclude_cts=false&exclude_not_run=false&exclude_non_release=false&exclude_au=true&exclude_acts=true&exclude_retried=true&exclude_non_production=true

posciak@ can you help me out here?

I just built ToT video_decode_accelerator_unittest, copied it to my chell and:

./video_decode_accelerator_unittest --test_video_data=/tmp/test-25fps.h264:320:240:250:258:::1  --ozone-platform=gbm

ran successfully (23 of 23 tests completed in 110 seconds)

I can't figure out which from the stainless link how to replicate this. I see a lot of redness (and get a lot of infrastructure errors) but couldn't find the targets that I need to run on device.

Just in case if it helps,
Below link has lab target device(host name) list which are failing. 

https://stainless.corp.google.com/search?view=matrix&row=model&col=hostname&first_date=2018-02-14&last_date=2018-02-20&test=video_VideoDecodeAccelerator%7Cvideo_VDASanity%7Cvideo_VDAPerf&build=%5ER66%5C-10422%5C.0%5C.0%24&status=FAIL&status=ERROR&status=ABORT&exclude_cts=false&exclude_not_run=false&exclude_non_release=false&exclude_au=true&exclude_acts=true&exclude_retried=true&exclude_non_production=true

avkodipelli@chromium.org doesn't help at all. Can you answer my question above?

I've no idea about that, Assigning back to posciak@.

Reopening so it doesn't accidentally get overlooked. 

10422.0.0 linked above may not have all the fixes in, 10424.0.0 should be the first version this was fixed in.

Test passing from R66-10424.0.0.
https://stainless.corp.google.com/search?view=matrix&row=model&col=build&first_date=2018-02-21&last_date=2018-02-23&test=video_VideoDecodeAccelerator%7Cvideo_VDASanity%7Cvideo_VDAPerf&build=R66-*&exclude_cts=false&exclude_not_run=false&exclude_non_release=false&exclude_au=true&exclude_acts=true&exclude_retried=true&exclude_non_production=true