Bug 226135

Summary:

Remove WTF::holdLock() / WTF::holdLockIf()

Product:

WebKit

Reporter:

Chris Dumez <cdumez>

Component:

Web Template Framework

Assignee:

Chris Dumez <cdumez>

Status:

RESOLVED FIXED

Severity:

Normal

CC:

achristensen, benjamin, cmarcelo, darin, ddkilzer, esprehn+autocc, ews-watchlist, ggaren, kangil.han, keith_miller, kkinnunen, mark.lam, msaboff, saam, sam, tzagallo, webkit-bug-importer

Priority:

Keywords:

InRadar

Version:

WebKit Nightly Build

Hardware:

Unspecified

OS:

Unspecified

Attachments:

Description	Flags
Patch	none

Description Chris Dumez 2021-05-21 22:02:09 PDT

Remove WTF::holdLock() / WTF::holdLockIf() because they are not compatible with Clang Thread Safety Analysis. All call sites have been converted to use the Locker constructor directly.

Comment 1 Chris Dumez 2021-05-21 22:04:09 PDT

Created attachment 429401 [details]
Patch

Comment 2 EWS 2021-05-21 23:39:52 PDT

Committed r277914 (238047@main): <https://commits.webkit.org/238047@main>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 429401 [details].

Comment 3 Radar WebKit Bug Importer 2021-05-21 23:40:16 PDT

<rdar://problem/78344289>

Comment 4 Keith Miller 2021-05-22 06:26:03 PDT

Out of curiosity, why is it incompatible? You'd think and r-value reference would be the same as as the r-value constructor?

Comment 5 Chris Dumez 2021-05-22 08:45:33 PDT

(In reply to Keith Miller from comment #4)
> Out of curiosity, why is it incompatible? You'd think and r-value reference
> would be the same as as the r-value constructor?

Not sure which r-value constructor you're referring to. The Locker<CheckedLock> specialization in CheckedLock.h does not have a move or copy constructor. AFAIK, this isn't supposed by clang thread safety analysis (https://clang.llvm.org/docs/ThreadSafetyAnalysis.html).