Bug 231209

Summary: CodeBlock should not add/remove LoopHintExecutionCounters.
Product: WebKit Reporter: Mark Lam <mark.lam>
Component: JavaScriptCoreAssignee: Mark Lam <mark.lam>
Status: RESOLVED FIXED    
Severity: Normal CC: ews-watchlist, keith_miller, msaboff, saam, tzagallo, webkit-bug-importer
Priority: P2 Keywords: InRadar
Version: WebKit Nightly Build   
Hardware: Unspecified   
OS: Unspecified   
Attachments:
Description Flags
proposed patch.
none
proposed patch.
saam: review+
patch for landing. none

Mark Lam
Reported 2021-10-04 22:08:58 PDT
This is because cached unlinked baseline JIT code would retain a pointer to those counters. Hence, the UnlinkedCodeBlock should do the add /remove of the counters instead. rdar://83571235
Attachments
proposed patch. (6.49 KB, patch)
2021-10-04 22:21 PDT, Mark Lam 		no flags
DetailsFormatted DiffDiff
proposed patch. (6.49 KB, patch)
2021-10-04 22:23 PDT, Mark Lam 		saam: review+
DetailsFormatted DiffDiff
patch for landing. (6.63 KB, patch)
2021-10-04 23:06 PDT, Mark Lam 		no flags
DetailsFormatted DiffDiff
Show Obsolete (2) View All Add attachment proposed patch, testcase, etc.
Mark Lam
Comment 1 2021-10-04 22:21:34 PDT
Created attachment 440157 [details] proposed patch.
Mark Lam
Comment 2 2021-10-04 22:23:55 PDT
Created attachment 440158 [details] proposed patch.
Saam Barati
Comment 3 2021-10-04 22:45:07 PDT
Comment on attachment 440158 [details] proposed patch. View in context: https://bugs.webkit.org/attachment.cgi?id=440158&action=review r=me > Source/JavaScriptCore/bytecode/UnlinkedCodeBlock.h:144 > + void finalize(); Suggestion: “finalize” has some common meaning in cells, typically meaning doing some finalization work at the end of GC. Because of that, I’m not a fan of this name. And right now, it’s only doing one very specific thing. I suggest giving it a specific name to match the specific thing it’s doing.
Saam Barati
Comment 4 2021-10-04 22:45:13 PDT
Comment on attachment 440158 [details] proposed patch. View in context: https://bugs.webkit.org/attachment.cgi?id=440158&action=review r=me > Source/JavaScriptCore/bytecode/UnlinkedCodeBlock.h:144 > + void finalize(); Suggestion: “finalize” has some common meaning in cells, typically meaning doing some finalization work at the end of GC. Because of that, I’m not a fan of this name. And right now, it’s only doing one very specific thing. I suggest giving it a specific name to match the specific thing it’s doing.
Mark Lam
Comment 5 2021-10-04 22:55:36 PDT
Thanks for the review. (In reply to Saam Barati from comment #4) > > Source/JavaScriptCore/bytecode/UnlinkedCodeBlock.h:144 > > + void finalize(); > > Suggestion: “finalize” has some common meaning in cells, typically meaning > doing some finalization work at the end of GC. Because of that, I’m not a > fan of this name. And right now, it’s only doing one very specific thing. I > suggest giving it a specific name to match the specific thing it’s doing. I've renamed it to initializeLoopHintExecutionCounter(), and changed it to be called only when Options::returnEarlyFromInfiniteLoopsForFuzzing() is true.
Mark Lam
Comment 6 2021-10-04 23:06:27 PDT
Created attachment 440160 [details] patch for landing.
EWS
Comment 7 2021-10-05 12:20:36 PDT
Committed r283567 (242532@main): <https://commits.webkit.org/242532@main> All reviewed patches have been landed. Closing bug and clearing flags on attachment 440160 [details].
Note You need to log in before you can comment on or make changes to this bug.