Bug 154269

Summary: FTL should support CompareEq(String:, String:)
Product: WebKit Reporter: Filip Pizlo <fpizlo>
Component: JavaScriptCoreAssignee: Filip Pizlo <fpizlo>
Status: RESOLVED FIXED    
Severity: Normal CC: benjamin, commit-queue, ggaren, keith_miller, mark.lam, msaboff, rniwa, saam
Priority: P2    
Version: WebKit Nightly Build   
Hardware: All   
OS: All   
Bug Depends on:    
Bug Blocks: 154266    
Attachments:
Description Flags
voila
none
maybe the patch
none
the patch
benjamin: review+
rebased patch none

Filip Pizlo
Reported 2016-02-15 16:56:38 PST
Patch forthcoming.
Attachments
voila (6.45 KB, patch)
2016-02-16 13:31 PST, Filip Pizlo 		no flags
DetailsFormatted DiffDiff
maybe the patch (11.41 KB, patch)
2016-02-16 14:31 PST, Filip Pizlo 		no flags
DetailsFormatted DiffDiff
the patch (11.75 KB, patch)
2016-02-16 15:54 PST, Filip Pizlo 		benjamin: review+
DetailsFormatted DiffDiff
rebased patch (11.74 KB, patch)
2016-02-16 20:28 PST, Filip Pizlo 		no flags
DetailsFormatted DiffDiff
Show Obsolete (3) View All Add attachment proposed patch, testcase, etc.
Filip Pizlo
Comment 1 2016-02-16 13:31:14 PST
Created attachment 271478 [details] voila Haven't tried it yet
Filip Pizlo
Comment 2 2016-02-16 14:31:14 PST
Created attachment 271484 [details] maybe the patch Still running tests.
Filip Pizlo
Comment 3 2016-02-16 15:54:13 PST
Created attachment 271500 [details] the patch
Benjamin Poulain
Comment 4 2016-02-16 18:26:41 PST
Comment on attachment 271500 [details] the patch View in context: https://bugs.webkit.org/attachment.cgi?id=271500&action=review > Source/JavaScriptCore/ftl/FTLLowerDFGToLLVM.cpp:7739 > + m_out.branch(m_out.notNull(left), usually(leftReadyCase), rarely(slowCase)); > + > + m_out.appendTo(leftReadyCase, rightReadyCase); Do you also need those checks for StringIdent?
Filip Pizlo
Comment 5 2016-02-16 18:32:23 PST
Comment on attachment 271500 [details] the patch View in context: https://bugs.webkit.org/attachment.cgi?id=271500&action=review >> Source/JavaScriptCore/ftl/FTLLowerDFGToLLVM.cpp:7739 >> + m_out.appendTo(leftReadyCase, rightReadyCase); > > Do you also need those checks for StringIdent? StringIdent is handled elsewhere. I believe that speculates that left and right are resolved, as opposed to this, which handles that on the slow path. But no matter how it does it, it's unrelated.
Filip Pizlo
Comment 6 2016-02-16 20:28:01 PST
Created attachment 271523 [details] rebased patch
Filip Pizlo
Comment 7 2016-02-16 21:30:30 PST
Landed in http://trac.webkit.org/changeset/196685
Note You need to log in before you can comment on or make changes to this bug.