Bug 223826

Summary: REGRESSION (r275038?): Perf test IndexedDB/stress/large-binary-keys.html failing with logging about memory pressure events
Product: WebKit Reporter: Ryan Haddad <ryanhaddad>
Component: New BugsAssignee: Chris Dumez <cdumez>
Status: RESOLVED FIXED    
Severity: Normal CC: achristensen, benjamin, cdumez, cmarcelo, darin, ews-watchlist, ggaren, webkit-bot-watchers-bugzilla, webkit-bug-importer
Priority: P2 Keywords: InRadar
Version: WebKit Nightly Build   
Hardware: Unspecified   
OS: Unspecified   
See Also: https://bugs.webkit.org/show_bug.cgi?id=223717
Attachments:
Description Flags
Patch none

Description Ryan Haddad 2021-03-26 16:16:07 PDT 
REGRESSION: Catalina-Release-WK2-Perf tests failing with logging about memory pressure events:

Running IndexedDB/stress/large-binary-keys.html (98 of 207)
error: IndexedDB/stress/large-binary-keys.html
Received memory pressure event 16 vm pressure 0
Received memory pressure event 32 vm pressure 0
FAILED
Finished: 208.580084 s

https://build.webkit.org/#/builders/20/builds/416
Comment 1 Radar WebKit Bug Importer 2021-03-26 16:16:23 PDT 
<rdar://problem/75906134>
Comment 2 Ryan Haddad 2021-03-26 16:17:53 PDT 
https://trac.webkit.org/changeset/275038/webkit looks like it could have been related
Comment 3 Chris Dumez 2021-03-26 16:29:20 PDT 
Created attachment 424418 [details]
Patch
Comment 4 Darin Adler 2021-03-26 17:40:39 PDT 
Comment on attachment 424418 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=424418&action=review

> Source/WTF/ChangeLog:10
> +        Use RELEASE_LOG() instead of WTFLogAlways() to log memory pressure events, so that they don't
> +        show on stdout in the performance tests and make them fail.

Does this still achieve what we want in production builds?
Comment 5 Chris Dumez 2021-03-26 17:43:20 PDT 
(In reply to Darin Adler from comment #4)
> Comment on attachment 424418 [details]
> Patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=424418&action=review
> 
> > Source/WTF/ChangeLog:10
> > +        Use RELEASE_LOG() instead of WTFLogAlways() to log memory pressure events, so that they don't
> > +        show on stdout in the performance tests and make them fail.
> 
> Does this still achieve what we want in production builds?

We will see the logging is os_log which I believe is suitable for such logging.
Comment 6 EWS 2021-03-26 18:00:40 PDT 
Committed r275125: <https://commits.webkit.org/r275125>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 424418 [details].