Bug 232330

Summary: Regression (r284741?) : [ macOS Debug ] compositing/video/video-reflection.html is a flaky timeout
Product: WebKit Reporter: ayumi_kojima
Component: CompositingAssignee: Nobody <webkit-unassigned>
Status: NEW ---    
Severity: Normal CC: dino, simon.fraser, thorton, webkit-bot-watchers-bugzilla, webkit-bug-importer
Priority: P2 Keywords: InRadar
Version: WebKit Nightly Build   
Hardware: Unspecified   
OS: Unspecified   

Description ayumi_kojima 2021-10-26 13:48:56 PDT 
compositing/video/video-reflection.html

Is a flaky timeout on macOS wk2 Debug and Catalina Debug gpuprocess.

History: https://results.webkit.org/?suite=layout-tests&test=compositing%2Fvideo%2Fvideo-reflection.html

Result page: https://build.webkit.org/results/Apple-BigSur-Debug-AppleSilicon-WK2-Tests/r284866%20(3839)/results.html

Diff:

--- /Volumes/Data/worker/bigsur-debug-applesilicon-tests-wk2/build/layout-test-results/compositing/video/video-reflection-expected.txt
+++ /Volumes/Data/worker/bigsur-debug-applesilicon-tests-wk2/build/layout-test-results/compositing/video/video-reflection-actual.txt
@@ -1,3 +1,3 @@
+FAIL: Timed out waiting for notifyDone to be called
 
-EXPECTED (video.currentTime == '1') OK
Comment 1 Radar WebKit Bug Importer 2021-10-26 13:51:09 PDT 
<rdar://problem/84679805>
Comment 2 ayumi_kojima 2021-10-26 13:54:39 PDT 
Marked test expectations https://trac.webkit.org/changeset/284893/webkit
Comment 3 ayumi_kojima 2021-10-26 17:30:58 PDT 
I was able to reproduce the timeout locally on BigSur using run-webkit-tests --force --iterations 1000 --exit-after-n-crashes-or-timeouts 1 compositing/video/video-reflection.html --debug -f.

The test timed out with r284750 and passed with r284749.
Comment 4 Alexey Proskuryakov 2021-10-28 10:27:04 PDT 
Per a slack discussion, r284750 is unlikely to have been the culprit.

It's definitely something not too far before that, according to bot history.
Comment 5 ayumi_kojima 2021-10-28 12:08:18 PDT 
The test is flaky and hard to reproduce. Iterated 1000 times (no -f flag this time because the test didn't start running with -f).

It timed out with r284741 but passed with r284742. Tested a couple of revisions near r284741, but couldn't reproduce the timeout; hence couldn't identify the regression point. 

284739 pass
284740 pass
284741 timeout
284742 pass
284743 pass
284749 pass
284750 timeout