Bug 209837 – [ Mojave wk1 Release ] fast/canvas/webgl/texImage2D-mse-flipY-true.html is timing out.

• Status: RESOLVED
• Resolution: FIXED
• Component: WebGL

[Jason Lawrence]

fast/canvas/webgl/texImage2D-mse-flipY-true.html

Description:
This test is flaky timing out on Mojave wk1 Release. The current batch of flaky timeouts, which appeared in the visible history after the test was passing consistently, appears to have started on 3/23/2020.

History:
https://results.webkit.org/?suite=layout-tests&test=fast%2Fcanvas%2Fwebgl%2FtexImage2D-mse-flipY-true.html&flavor=wk1&platform=mac&version_name=Mojave&limit=50000&style=release

Diff:
Not available.

[Radar WebKit Bug Importer]

<rdar://problem/61128322>

[Jason Lawrence]

I have marked this test as timing out while this issue is investigated.
https://trac.webkit.org/changeset/259319/webkit

[Jason Lawrence]

I was able to reproduce this issue with r259310 by running the command below.

run-webkit-tests --iterations 1000 --exit-after-n-failures 3 --child-processes 100 fast/canvas/webgl/texImage2D-mse-flipY-true.html

[145/1000] fast/canvas/webgl/texImage2D-mse-flipY-true.html failed unexpectedly (text diff)
[231/1000] fast/canvas/webgl/texImage2D-mse-flipY-true.html failed unexpectedly (test timed out, test was not run)
[231/1000] fast/canvas/webgl/texImage2D-mse-flipY-true.html failed unexpectedly (test timed out, test was not run)

[Kenneth Russell]

Is there an approximate regression range? Aside from the recent fix for related tests in Issue 205734, the last WebGL validation change was in Issue 209098 and that was on 3/27/2020.

[Jer Noble]

This test has been flakily timing out on this bot since at least 1/23/2020.  So if there was a regression, it wasn't super recent.

[Jer Noble]

Also, it has been generating between 1 and 10 flakes per day, starting on 3/24/2020, but has stopped flaking altogether as of 4/7, except for a single flake on 4/28.

[Kenneth Russell]

Good to hear that the flakes stopped recently. The fix in Bug 205734 should have made this test reliably, but that was done before these flakes started.

How should this bug be handled at this point? Would it be appropriate to remove the Timeout suppression and close it as WontFix?

[Kenneth Russell]

Created attachment 398894 [details]
Patch

[Kenneth Russell]

Looking at the layout test results, a timeout hasn't happened in over a month, so let's remove the suppression and close this.

[EWS]

Committed r261445: <https://trac.webkit.org/changeset/261445>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 398894 [details].