Bug 302674
| Summary: | fast/webgpu/nocrash/fuzz-279542.html appears to cause other tests to timeout | ||
|---|---|---|---|
| Product: | WebKit | Reporter: | Robert Jenner <jenner> |
| Component: | WebGPU | Assignee: | Nobody <webkit-unassigned> |
| Status: | RESOLVED FIXED | ||
| Severity: | Normal | CC: | mwyrzykowski, tzagallo, webkit-bot-watchers-bugzilla, webkit-bug-importer |
| Priority: | P2 | Keywords: | InRadar |
| Version: | WebKit Nightly Build | ||
| Hardware: | Unspecified | ||
| OS: | Unspecified | ||
Robert Jenner
It has been difficult to determine the scope. However, we have observed a high amount of tests timing out on Tahoe-Apple-Silicon-WK2-Debug queue.
An example test run would be this:
https://build.webkit.org/#/builders/1701/builds/180
I have also sometimes been able to reproduce a timeout locally when running `fast/webgpu/nocrash/fuzz-279542.html` in front of another test that I wasn't able to reproduce a timeout standalone in.
| Attachments | ||
|---|---|---|
| Add attachment proposed patch, testcase, etc. |
Radar WebKit Bug Importer
<rdar://problem/164922577>
EWS
Test gardening commit 303147@main (a93e131e5248): <https://commits.webkit.org/303147@main>
Reviewed commits have been landed. Closing PR #54080 and removing active labels.
Robert Jenner
(In reply to EWS from comment #2)
> Test gardening commit 303147@main (a93e131e5248):
> <https://commits.webkit.org/303147@main>
>
> Reviewed commits have been landed. Closing PR #54080 and removing active
> labels.
I have opted to skip this test entirely at the commit above. We are going to see if that results in less test timeouts.
Robert Jenner
Skipping that one test did nothing. We are now going to try skipping a few others to see if we can get the timeouts to stop.
EWS
Test gardening commit 303189@main (6e5dcb91113f): <https://commits.webkit.org/303189@main>
Reviewed commits have been landed. Closing PR #54114 and removing active labels.
Mike Wyrzykowski
Will these still run in post-commit?
Robert Jenner
(In reply to Mike Wyrzykowski from comment #6)
> Will these still run in post-commit?
Yes, but not on Tahoe Debug arm64. Until we can spend the time figuring out what's going on. However, I will say that by skipping these tests we now get test runs to completion on Tahoe Debug arm64. Whereas before, some of these tests were causing a jump in timeouts to over 50 on a test run, which would cause them to exit early. But again, only for Tahoe Debug arm64 wk2.
Mike Wyrzykowski
As long as we have Tahoe Release arm64 that should be sufficient, thank you