Bug 207857 - [ Mojave wk1 ] fast/canvas/webgl/texImage2D-mse-flipY-false.html is flaky timing out
Summary: [ Mojave wk1 ] fast/canvas/webgl/texImage2D-mse-flipY-false.html is flaky tim...
Status: RESOLVED DUPLICATE of bug 205734
Alias: None
Product: WebKit
Classification: Unclassified
Component: WebGL (show other bugs)
Version: WebKit Nightly Build
Hardware: Unspecified Unspecified
: P2 Normal
Assignee: Kenneth Russell
URL:
Keywords: InRadar
Depends on: 205483 205734
Blocks:
  Show dependency treegraph
 
Reported: 2020-02-17 13:35 PST by Jacob Uphoff
Modified: 2020-03-05 11:37 PST (History)
10 users (show)

See Also:


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Jacob Uphoff 2020-02-17 13:35:55 PST
fast/canvas/webgl/texImage2D-mse-flipY-false.html

This test recently started to timeout and the first timeout was on 255997 however there seems to be a few commits where no tests were run.

This test is timing out on Mojave release wk1 pretty consistently now.

History:

https://results.webkit.org/?suite=layout-tests&test=fast%2Fcanvas%2Fwebgl%2FtexImage2D-mse-flipY-false.html

Diff:

--- /Volumes/Data/slave/mojave-release-tests-wk2/build/layout-test-results/fast/canvas/webgl/texImage2D-mse-flipY-false-expected.txt
+++ /Volumes/Data/slave/mojave-release-tests-wk2/build/layout-test-results/fast/canvas/webgl/texImage2D-mse-flipY-false-actual.txt
@@ -1,6 +1,5 @@
-Setting video src.
-Video can play.
-Checking the canvas pixels.
-PASS: Bottom edge is red.
-PASS: Top edge is blue.
+#PID UNRESPONSIVE - WebKitTestRunner (pid 90890)
+FAIL: Timed out waiting for notifyDone to be called
 
+#EOF
+#EOF
Comment 1 Jacob Uphoff 2020-02-17 13:44:58 PST
Set expectations here: https://trac.webkit.org/changeset/256767/webkit
Comment 2 Radar WebKit Bug Importer 2020-02-17 13:45:30 PST
<rdar://problem/59524346>
Comment 3 Alexey Proskuryakov 2020-02-17 16:56:03 PST
bot132:~ buildbot$ uptime
16:55  up 34 days,  8:57, 2 users, load averages: 5.82 5.84 4.84

We need to periodically reboot Mojave bots, perhaps that would help here?
Comment 4 Jason Lawrence 2020-02-18 17:35:59 PST
I am able to reproduce this issue with r256844 using the command below.
run-webkit-tests fast/canvas/webgl/texImage2D-mse-flipY-false.html --iterations 3000 --exit-after-n-failures 5 -f --force -1

[5/3000] fast/canvas/webgl/texImage2D-mse-flipY-false.html failed unexpectedly (text diff)
[6/3000] fast/canvas/webgl/texImage2D-mse-flipY-false.html failed unexpectedly (text diff)
[7/3000] fast/canvas/webgl/texImage2D-mse-flipY-false.html failed unexpectedly (text diff)

Retrying 1 unexpected failure ...
Running 1 DumpRenderTree.
2997 tests ran as expected, 3 didn't:
Comment 5 Kenneth Russell 2020-02-20 11:34:25 PST
I'll investigate this.
Comment 6 Kenneth Russell 2020-02-21 17:47:03 PST
Whatever is going on in this test is almost surely the same problem as Bug 205734.
Comment 7 Alexey Proskuryakov 2020-02-22 17:05:30 PST
> [5/3000] fast/canvas/webgl/texImage2D-mse-flipY-false.html failed
> unexpectedly (text diff)
> [6/3000] fast/canvas/webgl/texImage2D-mse-flipY-false.html failed
> unexpectedly (text diff)
> [7/3000] fast/canvas/webgl/texImage2D-mse-flipY-false.html failed
> unexpectedly (text diff)

Is this always failing on some of the first iterations like here? That would imply that per-process initialization has some kind of a race I think. Running with --run-singly may shed more light, as that makes DumpRenderTree be relaunched each time.
Comment 8 Kenneth Russell 2020-02-22 17:25:59 PST
Please see Bug 205734 - something's going wrong in the video stack. Investigation of a related test is underway there - any assistance in debugging definitely appreciated.
Comment 9 Kenneth Russell 2020-03-05 11:37:00 PST
I believe this flake has the same root cause as Bug 205734. In my local testing, after the fix for that bug, the test is completely reliable. There weren't any TestExpectations entries for this test (aside from iOS, which has suppressed all 4 of these tests since 2017) so I think it's safe to duplicate it into the other bug. Please reopen if it still seems flaky.

*** This bug has been marked as a duplicate of bug 205734 ***