Issue metadata
Sign in to add a comment
|
http/tests/devtools/tracing/timeline-paint/layer-tree.html failing on WebKit Win7 (dbg) |
||||||||||||||||||||||
Issue descriptionStarted failing at https://build.chromium.org/p/chromium.webkit/builders/WebKit%20Win7%20%28dbg%29/builds/10777 17:01:53.591 1172 worker/0 http/tests/devtools/tracing/timeline-paint/layer-tree.html output stderr lines: 17:01:53.591 1172 [4248:3336:0831/170130.094:14164750:ERROR:direct_composition_surface_win.cc(1095)] Failing to detect HDR, couldn't retrieve D3D11 device from ANGLE. 17:01:53.591 1172 [4248:3336:0831/170130.128:14164781:INFO:media_foundation_video_encode_accelerator_win.cc(335)] Windows versions earlier than 8 are not supported. 17:01:53.604 3508 [9/10] http/tests/devtools/tracing/timeline-paint/layer-tree.html failed unexpectedly (text diff) 17:01:53.599 1172 worker/0 http/tests/devtools/tracing/timeline-paint/layer-tree.html failed: 17:01:53.600 1172 worker/0 text diff 17:01:53.600 1172 worker/0 killing secondary driver 17:01:53.600 1172 worker/0 exiting 17:01:53.600 1172 worker/0 cleaning up 17:01:53.600 1172 worker/0 killing primary driver 17:01:53.689 1172 "taskkill.exe /f /t /pid 7144" took 0.09s 17:01:53.690 1172 killed pid 7144 17:01:53.690 1172 worker/0 killing secondary driver
,
Sep 1 2017
In that case, why doesn't the existing [ Pass Failure ] expectation keep the test green?
,
Sep 1 2017
I have a CL at https://chromium-review.googlesource.com/c/chromium/src/+/647607 to mark the test as slow. Do you think that's worth a shot?
,
Sep 1 2017
So, it looks like the timeout is actually scheduled from within the test -- it must be using the setTimeout in https://cs.chromium.org/chromium/src/third_party/WebKit/LayoutTests/http/tests/inspector/inspector-test.js?type=cs&q=%22internal+watchdog%22&sq=package:chromium&l=1166 That explains why a timeout is caught in JS and interpreted as a failure by the test runner. Marking it as a slow test probably won't do anything, since it is forcably aborted after 20 seconds. The TestExpectations line for this one looks like crbug.com/757955 [ Win7 Debug ] http/tests/devtools/tracing/timeline-paint/layer-tree.html [ Pass Timeout ] with [ Pass Timeout ] rather than [ Pass Fail ] -- If we want to disable it as flaky, then we should probably just change it to [ Pass Fail Timeout ].
,
Sep 1 2017
FWIW, this failure seems to go back for a while - for example I also see a similar failure in the following builds: https://build.chromium.org/p/chromium.webkit/builders/WebKit%20Win7%20%28dbg%29/builds/10749
,
Sep 1 2017
+chenwilliam@ [touched these tests most recently] +pfeldman@ [introduced that "_watchDogTimer in inspector-test.js]
,
Sep 1 2017
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/b7bb9ad1050138687a6c70d7a4ef3aa3c295b616 commit b7bb9ad1050138687a6c70d7a4ef3aa3c295b616 Author: Ian Clelland <iclelland@google.com> Date: Fri Sep 01 19:36:18 2017 Marking timeline-paint/layer-tree.html as failing This test has an internal JS-based timeout, which is causing a failure on the WebKit Win7 (dbg) bot before the test harness itself can time out. Marking the test as [ Pass Failure Timeout ]. NOTRY=True Bug: 761333 , 757955 Change-Id: I9e324e52129681ab88abfadd417bdbd2dcba8928 TBR: chenwilliam@chromium.org Reviewed-on: https://chromium-review.googlesource.com/647777 Reviewed-by: Ian Clelland <iclelland@chromium.org> Commit-Queue: Ian Clelland <iclelland@chromium.org> Cr-Commit-Position: refs/heads/master@{#499259} [modify] https://crrev.com/b7bb9ad1050138687a6c70d7a4ef3aa3c295b616/third_party/WebKit/LayoutTests/TestExpectations
,
Sep 1 2017
Assigning to remove from sheriff queue.
,
Dec 20 2017
No longer flaky |
|||||||||||||||||||||||
►
Sign in to add a comment |
|||||||||||||||||||||||
Comment 1 by iclell...@chromium.org
, Sep 1 2017