Bug 218063
| Summary: | REGRESSION: [ Catalina ] 50+ webrtc test crashes on Catalina Debug | ||
|---|---|---|---|
| Product: | WebKit | Reporter: | Truitt Savell <tsavell> |
| Component: | WebRTC | Assignee: | Nobody <webkit-unassigned> |
| Status: | RESOLVED CONFIGURATION CHANGED | ||
| Severity: | Normal | CC: | ap, eric.carlson, peng.liu6, webkit-bot-watchers-bugzilla, webkit-bug-importer, youennf |
| Priority: | P2 | Keywords: | InRadar |
| Version: | WebKit Nightly Build | ||
| Hardware: | Unspecified | ||
| OS: | Unspecified | ||
Truitt Savell
After a commit around r268758 50+ webrtc tests began crashing on Catalina WK2 Debug and around 20 on WK1 Debug. These tests are flaky and we will see runs with 50+ crashes and runs with 1 failure.
history of runs:
https://build.webkit.org/builders/Apple-Catalina-Debug-WK2-Tests?numbuilds=50
Example:
https://build.webkit.org/builders/Apple-Catalina-Debug-WK2-Tests/builds/7339
This is reproducible on the bots, I am looking to find a test list that can reproduce these to be more specific.
Results:
https://build.webkit.org/results/Apple-Catalina-Debug-WK2-Tests/r268822%20(7339)/results.html
| Attachments | ||
|---|---|---|
| Add attachment proposed patch, testcase, etc. |
Radar WebKit Bug Importer
<rdar://problem/70551805>
Alexey Proskuryakov
# Fatal error in: Source/webrtc/rtc_base/operations_chain.h, line 136
# last system error: 0
# Check failed: (&sequence_checker_)->IsCurrent()
It would seem like r268758 only changes GPU Process code, which I think is not supposed to be on in tests. But there are no other potential culprits anywhere near.
Truitt Savell
Speculatively reverted https://trac.webkit.org/changeset/268938/webkit to fix this. Bisection seems to have shown this as the regression point though the commit seems very unrelated
Ryan Haddad
(In reply to Truitt Savell from comment #3)
> Speculatively reverted https://trac.webkit.org/changeset/268938/webkit to
> fix this. Bisection seems to have shown this as the regression point though
> the commit seems very unrelated
The issue persists, so this commit was not the cause of the crashes.
youenn fablet
Marking as configuration changed, please reopen if crash is still happenning.