Bug 203873

Summary: WTF::RunLoop should not depend on isMainThread() idiom.
Product: WebKit Reporter: Mark Lam <mark.lam>
Component: JavaScriptCoreAssignee: Mark Lam <mark.lam>
Status: RESOLVED FIXED    
Severity: Normal CC: benjamin, cdumez, cmarcelo, dbates, ews-watchlist, fpizlo, hi, joepeck, keith_miller, msaboff, rmorisset, rniwa, saam, simon.fraser, tzagallo, webkit-bug-importer, ysuzuki
Priority: P2 Keywords: InRadar
Version: WebKit Nightly Build   
Hardware: Unspecified   
OS: Unspecified   
See Also: https://bugs.webkit.org/show_bug.cgi?id=202716
https://bugs.webkit.org/show_bug.cgi?id=203875
Attachments:
Description Flags
proposed patch.
none
proposed patch. saam: review+

Mark Lam
Reported 2019-11-05 17:49:02 PST
The isMainThread() idiom is only meaningful for WebCore. It is less meaningful for JSC since a VM instance can be entered from multiple threads, as long as only one thread enters it at any time. Hence, the concept of a main thread doesn't make sense at the JSC level. Since r251036, we started using a WTF::String to represent the RunLoop mode. This caused problems for JSC clients when USE(CF) since it necessitated the use of StringWrapperCFAllocator to track the life cycle of the CFStringRef generated from the WTF::String. To fix this problem, we should restore the original behavior of using CFStringRefs as the RunLoop mode token. <rdar://problem/56524251>
Attachments
proposed patch. (11.22 KB, patch)
2019-11-05 17:58 PST, Mark Lam 		no flags
DetailsFormatted DiffDiff
proposed patch. (11.20 KB, patch)
2019-11-05 18:08 PST, Mark Lam 		saam: review+
DetailsFormatted DiffDiff
Show Obsolete (1) View All Add attachment proposed patch, testcase, etc.
Mark Lam
Comment 1 2019-11-05 17:58:23 PST
Created attachment 382873 [details] proposed patch.
Mark Lam
Comment 2 2019-11-05 18:04:55 PST
Comment on attachment 382873 [details] proposed patch. Need to fix non abstraction to also work for non USE(CF) ports.
Mark Lam
Comment 3 2019-11-05 18:08:46 PST
Created attachment 382877 [details] proposed patch.
Saam Barati
Comment 4 2019-11-05 18:14:58 PST
Comment on attachment 382877 [details] proposed patch. View in context: https://bugs.webkit.org/attachment.cgi?id=382877&action=review > Source/WTF/ChangeLog:7 > + Reviewed by NOBODY (OOPS!). can you also open a bug about the Date function which uses createCFString?
Mark Lam
Comment 5 2019-11-05 18:21:15 PST
Comment on attachment 382877 [details] proposed patch. View in context: https://bugs.webkit.org/attachment.cgi?id=382877&action=review >> Source/WTF/ChangeLog:7 >> + Reviewed by NOBODY (OOPS!). > > can you also open a bug about the Date function which uses createCFString? Will do.
Mark Lam
Comment 6 2019-11-05 18:27:19 PST
(In reply to Mark Lam from comment #5) > Comment on attachment 382877 [details] > proposed patch. > > View in context: > https://bugs.webkit.org/attachment.cgi?id=382877&action=review > > >> Source/WTF/ChangeLog:7 > >> + Reviewed by NOBODY (OOPS!). > > > > can you also open a bug about the Date function which uses createCFString? > > Will do. See https://bugs.webkit.org/show_bug.cgi?id=203875
Ryosuke Niwa
Comment 7 2019-11-05 18:59:02 PST
Looks good to me. Sorry for the regression.
Devin Rousso
Comment 8 2019-11-05 19:24:02 PST
Comment on attachment 382877 [details] proposed patch. View in context: https://bugs.webkit.org/attachment.cgi?id=382877&action=review r=me as well, nice fix! > Source/WTF/wtf/RunLoop.h:56 > +#if USE(CF) > +using RunLoopMode = CFStringRef; > +#define DefaultRunLoopMode kCFRunLoopDefaultMode > +#else > +using RunLoopMode = unsigned; > +#define DefaultRunLoopMode 0 > +#endif Should we move these to be inside `RunLoop`? ``` class RunLoop : public FunctionDispatcher { ... #if USE(CF) using Mode = CFStringRef; using DefaultMode = kCFRunLoopDefaultMode; #else using Mode = unsigned; using DefaultMode = 0; #endif ... WTF_EXPORT_PRIVATE CycleResult static cycle(Mode = DefaultMode); ... } ```
Mark Lam
Comment 9 2019-11-05 22:30:48 PST
Comment on attachment 382877 [details] proposed patch. View in context: https://bugs.webkit.org/attachment.cgi?id=382877&action=review >> Source/WTF/wtf/RunLoop.h:56 >> +#endif > > Should we move these to be inside `RunLoop`? > > ``` > class RunLoop : public FunctionDispatcher { > ... > > #if USE(CF) > using Mode = CFStringRef; > using DefaultMode = kCFRunLoopDefaultMode; > #else > using Mode = unsigned; > using DefaultMode = 0; > #endif > > ... > > WTF_EXPORT_PRIVATE CycleResult static cycle(Mode = DefaultMode); > > ... > } > ``` This unfortunately is not possible because you can't do "using DefaultMode = kCFRunLoopDefaultMode;". We can only use "using" for types, and kCFRunLoopDefaultMode is not a type. We also can't use "static constexpr DefaultMode = kCFRunLoopDefaultMode;" because kCFRunLoopDefaultMode is not a constexpr expression. That's why I settled on the current form that uses the #define for DefaultRunLoopMode. I'll stick with this unless we run into a namespace collision issue.
Mark Lam
Comment 10 2019-11-05 22:37:37 PST
Thanks for the reviews. Landed in r252124: <http://trac.webkit.org/r252124>.
Note You need to log in before you can comment on or make changes to this bug.