New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.

Issue 832886 link

Starred by 2 users

Issue metadata

Status: Fixed
Owner:
OOO until 2019-01-24
Closed: Aug 24
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Android
Pri: 2
Type: Bug

Blocked on:
issue 575305
issue 837884
issue 858907

Blocking:
issue 879423



Sign in to add a comment

WebGL blocking/unblocking context_lost_tests fail on Android FYI Release (NVIDIA Shield TV) and (Nexus 9)

Project Member Reported by ynovikov@chromium.org, Apr 13 2018

Issue description

Comment 1 by kbr@chromium.org, Apr 13 2018

Blockedon: 575305
Labels: -Type-Bug-Regression Type-Bug
Owner: kbr@chromium.org
Status: Assigned (was: Untriaged)
Summary: WebGL blocking/unblocking context_lost_tests fail on Android FYI Release (NVIDIA Shield TV) and (Nexus 9) (was: Couple context_lost_tests fail on Android FYI Release (NVIDIA Shield TV) and (Nexus 9))
Not sure why these would only fail on those two Android devices. They should be running the GPU process in a similar configuration as other high-end devices – namely, a separate process.

Comment 2 by kbr@chromium.org, Apr 13 2018

Cc: kainino@chromium.org jdarpinian@chromium.org
Components: -Internals>GPU Blink>WebGL
Looks like we may need a try/catch around the closing of the about:gpucrash tab, after all.

Traceback (most recent call last):
  _RunGpuTest at content/test/gpu/gpu_tests/gpu_integration_test.py:132
    self.RunActualGpuTest(url, *args)
  RunActualGpuTest at content/test/gpu/gpu_tests/context_lost_integration_test.py:93
    getattr(self, test_name)(test_path)
  _ContextLost_WebGLBlockedAfterJSNavigation at content/test/gpu/gpu_tests/context_lost_integration_test.py:301
    gpucrash_tab.Close()
  traced_function at third_party/catapult/common/py_trace_event/py_trace_event/trace_event_impl/decorators.py:52
    return func(*args, **kwargs)
  Close at third_party/catapult/telemetry/telemetry/internal/browser/tab.py:104
    self._tab_list_backend.CloseTab(self.id)
  CloseTab at third_party/catapult/telemetry/telemetry/internal/backends/chrome/tab_list_backend.py:60
    response = self._browser_backend.devtools_client.CloseTab(tab_id, timeout)
  CloseTab at third_party/catapult/telemetry/telemetry/internal/backends/chrome_inspector/devtools_client_backend.py:333
    timeout=timeout)
  Request at third_party/catapult/telemetry/telemetry/internal/backends/chrome_inspector/devtools_http.py:88
    response = self._conn.getresponse()
  getresponse at .swarming_module/lib/python2.7/httplib.py:1121
    response.begin()
  begin at .swarming_module/lib/python2.7/httplib.py:438
    version, status, reason = self._read_status()
  _read_status at .swarming_module/lib/python2.7/httplib.py:402
    raise BadStatusLine(line)
DevToolsClientConnectionError: ''

Project Member

Comment 3 by bugdroid1@chromium.org, Apr 13 2018

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/6233e939ef5951618d8b1c941930781ce39f11d1

commit 6233e939ef5951618d8b1c941930781ce39f11d1
Author: Yuly Novikov <ynovikov@chromium.org>
Date: Fri Apr 13 23:03:53 2018

Mark couple context_lost_tests fail

ContextLost_WebGLBlockedAfterJSNavigation
ContextLost_WebGLUnblockedAfterUserInitiatedReload
on Android Nvidia (Nexus 9 and Shield TV)

BUG= 832886 
TBR=kbr@chromium.org

Cq-Include-Trybots: luci.chromium.try:android_optional_gpu_tests_rel;luci.chromium.try:linux_optional_gpu_tests_rel;luci.chromium.try:mac_optional_gpu_tests_rel;luci.chromium.try:win_optional_gpu_tests_rel
Change-Id: I0b945fe5c02824ae92524d7592d899a9f1bac504
Reviewed-on: https://chromium-review.googlesource.com/1012982
Commit-Queue: Yuly Novikov <ynovikov@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Cr-Commit-Position: refs/heads/master@{#550772}
[modify] https://crrev.com/6233e939ef5951618d8b1c941930781ce39f11d1/content/test/gpu/gpu_tests/context_lost_expectations.py

 Issue 832974  has been merged into this issue.
Project Member

Comment 5 by bugdroid1@chromium.org, Apr 17 2018

Labels: merge-merged-testbranch
The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/6233e939ef5951618d8b1c941930781ce39f11d1

commit 6233e939ef5951618d8b1c941930781ce39f11d1
Author: Yuly Novikov <ynovikov@chromium.org>
Date: Fri Apr 13 23:03:53 2018

Mark couple context_lost_tests fail

ContextLost_WebGLBlockedAfterJSNavigation
ContextLost_WebGLUnblockedAfterUserInitiatedReload
on Android Nvidia (Nexus 9 and Shield TV)

BUG= 832886 
TBR=kbr@chromium.org

Cq-Include-Trybots: luci.chromium.try:android_optional_gpu_tests_rel;luci.chromium.try:linux_optional_gpu_tests_rel;luci.chromium.try:mac_optional_gpu_tests_rel;luci.chromium.try:win_optional_gpu_tests_rel
Change-Id: I0b945fe5c02824ae92524d7592d899a9f1bac504
Reviewed-on: https://chromium-review.googlesource.com/1012982
Commit-Queue: Yuly Novikov <ynovikov@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Cr-Commit-Position: refs/heads/master@{#550772}
[modify] https://crrev.com/6233e939ef5951618d8b1c941930781ce39f11d1/content/test/gpu/gpu_tests/context_lost_expectations.py

Comment 6 by kbr@chromium.org, Apr 21 2018

 Issue 834896  has been merged into this issue.
Project Member

Comment 8 by bugdroid1@chromium.org, Apr 21 2018

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/ebd02a7a495999c710b82c638d0408bd01fcd587

commit ebd02a7a495999c710b82c638d0408bd01fcd587
Author: Kenneth Russell <kbr@chromium.org>
Date: Sat Apr 21 20:15:51 2018

Attempt to fix context_lost test flakes.

Add try/except around closing of about:gpucrash tab. This is still
causing exceptions in the DevTools backend.

Removed failure exceptions for Nexus 9 and Shield TV, since the same
failures were being seen on android_n5x_swarming_rel, and this should
address both. If not, failure expectations will be re-introduced.

Tbr: ynovikov@chromium.org
Bug:  832886 
Cq-Include-Trybots: luci.chromium.try:android_optional_gpu_tests_rel;luci.chromium.try:linux_optional_gpu_tests_rel;luci.chromium.try:mac_optional_gpu_tests_rel;luci.chromium.try:win_optional_gpu_tests_rel
Change-Id: I138925c7ab8fd46f47946683a939a6cfc5c75653
Reviewed-on: https://chromium-review.googlesource.com/1023335
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Commit-Queue: Kenneth Russell <kbr@chromium.org>
Cr-Commit-Position: refs/heads/master@{#552592}
[modify] https://crrev.com/ebd02a7a495999c710b82c638d0408bd01fcd587/content/test/gpu/gpu_tests/context_lost_expectations.py
[modify] https://crrev.com/ebd02a7a495999c710b82c638d0408bd01fcd587/content/test/gpu/gpu_tests/context_lost_integration_test.py

Project Member

Comment 9 by chromium...@appspot.gserviceaccount.com, Apr 23 2018

Labels: Sheriff-Chromium
Detected 3 new flakes for test/step "gpu_tests.context_lost_integration_test.ContextLostIntegrationTest.ContextLost_WebGLUnblockedAfterUserInitiatedReload". To see the actual flakes, please visit https://chromium-try-flakes.appspot.com/all_flake_occurrences?key=ahVzfmNocm9taXVtLXRyeS1mbGFrZXNygAELEgVGbGFrZSJ1Z3B1X3Rlc3RzLmNvbnRleHRfbG9zdF9pbnRlZ3JhdGlvbl90ZXN0LkNvbnRleHRMb3N0SW50ZWdyYXRpb25UZXN0LkNvbnRleHRMb3N0X1dlYkdMVW5ibG9ja2VkQWZ0ZXJVc2VySW5pdGlhdGVkUmVsb2FkDA. This message was posted automatically by the chromium-try-flakes app. Since flakiness is ongoing, the issue was moved back into Sheriff Bug Queue (unless already there).
Labels: -Sheriff-Chromium
Project Member

Comment 11 by chromium...@appspot.gserviceaccount.com, Apr 24 2018

Labels: Sheriff-Chromium
Detected 3 new flakes for test/step "gpu_tests.context_lost_integration_test.ContextLostIntegrationTest.ContextLost_WebGLUnblockedAfterUserInitiatedReload". To see the actual flakes, please visit https://chromium-try-flakes.appspot.com/all_flake_occurrences?key=ahVzfmNocm9taXVtLXRyeS1mbGFrZXNygAELEgVGbGFrZSJ1Z3B1X3Rlc3RzLmNvbnRleHRfbG9zdF9pbnRlZ3JhdGlvbl90ZXN0LkNvbnRleHRMb3N0SW50ZWdyYXRpb25UZXN0LkNvbnRleHRMb3N0X1dlYkdMVW5ibG9ja2VkQWZ0ZXJVc2VySW5pdGlhdGVkUmVsb2FkDA. This message was posted automatically by the chromium-try-flakes app. Since flakiness is ongoing, the issue was moved back into Sheriff Bug Queue (unless already there).
Project Member

Comment 12 by bugdroid1@chromium.org, Apr 24 2018

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/b6077ecacd144c279175b2db08eb9c6ccd6f4f8f

commit b6077ecacd144c279175b2db08eb9c6ccd6f4f8f
Author: Kenneth Russell <kbr@chromium.org>
Date: Tue Apr 24 22:42:02 2018

Re-suppress new context lost tests on Nexus 9 and Shield TV.

Suppress the following tests again:
  ContextLost_WebGLBlockedAfterJSNavigation
  ContextLost_WebGLUnblockedAfterUserInitiatedReload

They're still failing reliably on these two bots.

Also restart the browser after running these two tests; this was
forgotten, and appears necessary for any of these tests which tear
down the GPU process.

Tbr: ynovikov@chromium.org
Tbr: sugoi@chromium.org
No-Try: true
Bug:  832886 
Cq-Include-Trybots: luci.chromium.try:android_optional_gpu_tests_rel;luci.chromium.try:linux_optional_gpu_tests_rel;luci.chromium.try:mac_optional_gpu_tests_rel;luci.chromium.try:win_optional_gpu_tests_rel
Change-Id: I5918e7b731023b8efb8e1202f8c9b57e1089e1bd
Reviewed-on: https://chromium-review.googlesource.com/1025112
Commit-Queue: Yuly Novikov <ynovikov@chromium.org>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Reviewed-by: Alexis Hétu <sugoi@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Cr-Commit-Position: refs/heads/master@{#553330}
[modify] https://crrev.com/b6077ecacd144c279175b2db08eb9c6ccd6f4f8f/content/test/gpu/gpu_tests/context_lost_expectations.py
[modify] https://crrev.com/b6077ecacd144c279175b2db08eb9c6ccd6f4f8f/content/test/gpu/gpu_tests/context_lost_integration_test.py

Comment 13 by hbos@chromium.org, Apr 25 2018

Labels: -Sheriff-Chromium
Removing sheriff label since the tests were suppressed
Project Member

Comment 14 by chromium...@appspot.gserviceaccount.com, Apr 26 2018

Labels: Sheriff-Chromium
Detected 4 new flakes for test/step "gpu_tests.context_lost_integration_test.ContextLostIntegrationTest.ContextLost_WebGLUnblockedAfterUserInitiatedReload". To see the actual flakes, please visit https://chromium-try-flakes.appspot.com/all_flake_occurrences?key=ahVzfmNocm9taXVtLXRyeS1mbGFrZXNygAELEgVGbGFrZSJ1Z3B1X3Rlc3RzLmNvbnRleHRfbG9zdF9pbnRlZ3JhdGlvbl90ZXN0LkNvbnRleHRMb3N0SW50ZWdyYXRpb25UZXN0LkNvbnRleHRMb3N0X1dlYkdMVW5ibG9ja2VkQWZ0ZXJVc2VySW5pdGlhdGVkUmVsb2FkDA. This message was posted automatically by the chromium-try-flakes app. Since flakiness is ongoing, the issue was moved back into Sheriff Bug Queue (unless already there).
Looks like this is still flaking on Nexus 5 bots. I'm not sure why the change referred to by #12 didn't suppress that failure. Kenneth, can you take a look?
Change in #12 was only for Nexus9 and Shield TV.
I'll mark those tests flaky on other Android.
Project Member

Comment 17 by bugdroid1@chromium.org, Apr 26 2018

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/74e04301a05a4e3254192735802e1ffa0b160f91

commit 74e04301a05a4e3254192735802e1ffa0b160f91
Author: Yuly Novikov <ynovikov@chromium.org>
Date: Thu Apr 26 22:32:26 2018

Mark couple context_lost_tests Flaky on Android

Bug:  832886 
Cq-Include-Trybots: luci.chromium.try:android_optional_gpu_tests_rel;luci.chromium.try:linux_optional_gpu_tests_rel;luci.chromium.try:mac_optional_gpu_tests_rel;luci.chromium.try:win_optional_gpu_tests_rel
Change-Id: If566aa3fd7d00c6310b1919909f9a536a375627a
Tbr: kbr@chromium.org
No-try: true
Reviewed-on: https://chromium-review.googlesource.com/1030900
Commit-Queue: Yuly Novikov <ynovikov@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Cr-Commit-Position: refs/heads/master@{#554174}
[modify] https://crrev.com/74e04301a05a4e3254192735802e1ffa0b160f91/content/test/gpu/gpu_tests/context_lost_expectations.py

Labels: -Sheriff-Chromium

Comment 19 by kbr@chromium.org, Apr 28 2018

Blockedon: 837884

Comment 20 by kbr@chromium.org, Apr 28 2018

Labels: -Pri-1 Pri-2
The flakes of this test on Android are caused by  Issue 837884 . Since the flakes have been suppressed I'm downgrading this to P2, which is the priority used for the other bug.

Comment 21 by kbr@chromium.org, Jun 2 2018

Owner: vikassoni@chromium.org
vikassoni@ picked up  Issue 837884  and it turns out it was implicitly fixed by a code refactoring in another bug. Vikas will remove the failure expectations; thanks in advance.

Project Member

Comment 22 by bugdroid1@chromium.org, Jun 2 2018

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/f76b1477c25acbf5d5242164d84106f8b9a21693

commit f76b1477c25acbf5d5242164d84106f8b9a21693
Author: Vikas Soni <vikassoni@chromium.org>
Date: Sat Jun 02 01:50:17 2018

Remove Android exceptions for WebGL [un]blocking context_lost tests.

The assertion failure has been refactored away.

Removes these two:
  ContextLost_WebGLBlockedAfterJSNavigation
  ContextLost_WebGLUnblockedAfterUserInitiatedReload

Bug:  832886 
Cq-Include-Trybots: luci.chromium.try:android_optional_gpu_tests_rel;luci.chromium.try:linux_optional_gpu_tests_rel;luci.chromium.try:mac_optional_gpu_tests_rel;luci.chromium.try:win_optional_gpu_tests_rel
Change-Id: I95f885ef3e22babb3387079105fc9e3f49e21370
Reviewed-on: https://chromium-review.googlesource.com/1083951
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Commit-Queue: Kenneth Russell <kbr@chromium.org>
Cr-Commit-Position: refs/heads/master@{#563911}
[modify] https://crrev.com/f76b1477c25acbf5d5242164d84106f8b9a21693/content/test/gpu/gpu_tests/context_lost_expectations.py

Cc: vikassoni@chromium.org
Owner: kbr@chromium.org
Taking this back from Vikas. Still seeing flakes of these tests, for example:

https://ci.chromium.org/p/chromium/builders/luci.chromium.ci/Android%20FYI%20Release%20%28NVIDIA%20Shield%20TV%29/5302
https://ci.chromium.org/p/chromium/builders/luci.chromium.ci/Android%20FYI%20Release%20%28NVIDIA%20Shield%20TV%29/5292
https://ci.chromium.org/p/chromium/builders/luci.chromium.ci/Android%20FYI%20Release%20%28NVIDIA%20Shield%20TV%29/5274
https://ci.chromium.org/p/chromium/builders/luci.chromium.ci/Android%20FYI%20Release%20%28NVIDIA%20Shield%20TV%29/5273

The current failure mode is as follows:

  Traceback (most recent call last):
    _RunGpuTest at content/test/gpu/gpu_tests/gpu_integration_test.py:132
      self.RunActualGpuTest(url, *args)
    RunActualGpuTest at content/test/gpu/gpu_tests/context_lost_integration_test.py:102
      getattr(self, test_name)(test_path)
    _ContextLost_WebGLBlockedAfterJSNavigation at content/test/gpu/gpu_tests/context_lost_integration_test.py:299
      tab.Activate()
    traced_function at third_party/catapult/common/py_trace_event/py_trace_event/trace_event_impl/decorators.py:52
      return func(*args, **kwargs)
    Activate at third_party/catapult/telemetry/telemetry/internal/browser/tab.py:85
      self._tab_list_backend.ActivateTab(self.id)
    ActivateTab at third_party/catapult/telemetry/telemetry/internal/backends/chrome/tab_list_backend.py:87
      timeout=5)
    WaitFor at third_party/catapult/common/py_utils/py_utils/__init__.py:148
      (timeout, GetConditionString()))
  TimeoutException: Timed out while waiting 5s for lambda: tab_id == self._browser_backend.browser.foreground_tab.id,.


The mechanism used by the context_lost_tests to crash the GPU process is to create a new tab, navigate it to chrome://gpucrash, and then switch back to the original tab. The creation of the new tab and switch back to the original tab seems flaky. To address this will try adding a new entry point to the gpu_benchmarking_extension which provokes a GPU process crash, and use that from the tests.

Blockedon: 858907
Status: Fixed (was: Assigned)
After the fix in  Issue 858907  the flakiness of this test is gone. On these two bots:

https://ci.chromium.org/p/chromium/builders/luci.chromium.ci/Android%20FYI%20Release%20%28Nexus%209%29?limit=200
https://ci.chromium.org/p/chromium/builders/luci.chromium.ci/Android%20FYI%20Release%20%28NVIDIA%20Shield%20TV%29?limit=200

there's only one failure each of context_lost tests in the last 200 builds, and it was an infrastructure-level failure.

Closing as fixed.

Blocking: 879423

Sign in to add a comment