Bug 184395

Summary: Add pointer profiling to the FTL and supporting code.
Product: WebKit Reporter: Mark Lam <mark.lam>
Component: JavaScriptCoreAssignee: Mark Lam <mark.lam>
Status: RESOLVED FIXED    
Severity: Normal CC: fpizlo, jfbastien, keith_miller, msaboff, rmorisset, saam, webkit-bug-importer
Priority: P2 Keywords: InRadar
Version: WebKit Nightly Build   
Hardware: Unspecified   
OS: Unspecified   
Bug Depends on:    
Bug Blocks: 184324    
Attachments:
Description Flags
proposed patch.
none
proposed patch. fpizlo: review+

Mark Lam
Reported 2018-04-07 22:37:21 PDT
Patch coming.
Attachments
proposed patch. (51.87 KB, patch)
2018-04-08 01:35 PDT, Mark Lam 		no flags
DetailsFormatted DiffDiff
proposed patch. (58.28 KB, patch)
2018-04-08 02:07 PDT, Mark Lam 		fpizlo: review+
DetailsFormatted DiffDiff
Show Obsolete (1) View All Add attachment proposed patch, testcase, etc.
Radar WebKit Bug Importer
Comment 1 2018-04-07 22:37:59 PDT
<rdar://problem/39264019>
Mark Lam
Comment 2 2018-04-08 01:35:19 PDT
Created attachment 337452 [details] proposed patch. Let's get some EWS testing.
Mark Lam
Comment 3 2018-04-08 02:07:06 PDT
Created attachment 337454 [details] proposed patch.
Michael Saboff
Comment 4 2018-04-09 10:25:28 PDT
Comment on attachment 337454 [details] proposed patch. r=me
Filip Pizlo
Comment 5 2018-04-09 10:33:10 PDT
Comment on attachment 337454 [details] proposed patch. View in context: https://bugs.webkit.org/attachment.cgi?id=337454&action=review > Source/JavaScriptCore/ftl/FTLOutput.h:408 > - LValue operation(FunctionType function) { return constIntPtr(bitwise_cast<void*>(function)); } > + LValue operation(FunctionType function) { return constIntPtr(tagCFunctionPtr<void*>(function, B3CCallPtrTag)); } Ideally we would support passing more specific diversity to B3::CCall. Can you file a bug and reference it here with a FIXME? I totally see how it's not necessary for making this feature work right now, but such a FIXME will help us remember that it's only this way because it was expedient.
Filip Pizlo
Comment 6 2018-04-09 10:33:24 PDT
(In reply to Michael Saboff from comment #4) > Comment on attachment 337454 [details] > proposed patch. > > r=me r=me too
Mark Lam
Comment 7 2018-04-09 10:39:13 PDT
(In reply to Filip Pizlo from comment #5) > Comment on attachment 337454 [details] > proposed patch. > > View in context: > https://bugs.webkit.org/attachment.cgi?id=337454&action=review > > > Source/JavaScriptCore/ftl/FTLOutput.h:408 > > - LValue operation(FunctionType function) { return constIntPtr(bitwise_cast<void*>(function)); } > > + LValue operation(FunctionType function) { return constIntPtr(tagCFunctionPtr<void*>(function, B3CCallPtrTag)); } > > Ideally we would support passing more specific diversity to B3::CCall. Can > you file a bug and reference it here with a FIXME? > > I totally see how it's not necessary for making this feature work right now, > but such a FIXME will help us remember that it's only this way because it > was expedient. Done. Ref: https://bugs.webkit.org/show_bug.cgi?id=184324
Mark Lam
Comment 8 2018-04-09 10:42:38 PDT
Thanks for the reviews. Landed in r230444: <http://trac.webkit.org/r230444>.
Note You need to log in before you can comment on or make changes to this bug.