Bug 98996

Summary: DFG should inline code blocks that use new_array_buffer
Product: WebKit Reporter: Filip Pizlo <fpizlo>
Component: JavaScriptCoreAssignee: Filip Pizlo <fpizlo>
Status: RESOLVED FIXED    
Severity: Normal CC: barraclough, ggaren, mark.lam, mhahnenberg, msaboff, oliver
Priority: P2    
Version: 528+ (Nightly build)   
Hardware: All   
OS: All   
Bug Depends on:    
Bug Blocks: 98608    
Attachments:
Description Flags
the patch ggaren: review+, buildbot: commit-queue-

Description Filip Pizlo 2012-10-10 23:42:35 PDT 
It doesn't do so right now because of a really silly piece of missing functionality.
Comment 1 Filip Pizlo 2012-10-11 00:23:26 PDT 
Created attachment 168153 [details]
the patch
Comment 2 Build Bot 2012-10-11 00:48:24 PDT 
Comment on attachment 168153 [details]
the patch

Attachment 168153 [details] did not pass mac-ews (mac):
Output: http://queues.webkit.org/results/14252494

New failing tests:
inspector/profiler/memory-instrumentation-canvas.html
Comment 3 Geoffrey Garen 2012-10-11 09:04:54 PDT 
Comment on attachment 168153 [details]
the patch

View in context: https://bugs.webkit.org/attachment.cgi?id=168153&action=review

Are those test failures real?

This patch does not appear to do the inlining yet...

> Source/JavaScriptCore/ChangeLog:10
> +        hash-cons them, either.

"const"
Comment 4 Filip Pizlo 2012-10-11 09:19:23 PDT 
(In reply to comment #3)
> (From update of attachment 168153 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=168153&action=review
> 
> Are those test failures real?

I don't think so. :-/  I will investigate though.

> 
> This patch does not appear to do the inlining yet...

It just enables inlining of code blocks that use new_array_buffer.  Array allocations themselves are not yet inlined; that's https://bugs.webkit.org/show_bug.cgi?id=98608.

> 
> > Source/JavaScriptCore/ChangeLog:10
> > +        hash-cons them, either.
> 
> "const"

http://en.wikipedia.org/wiki/Hash_consing
Comment 5 Filip Pizlo 2012-10-11 12:36:20 PDT 
Landed in http://trac.webkit.org/changeset/131087
Comment 6 Filip Pizlo 2012-10-11 12:36:35 PDT 
(In reply to comment #4)
> (In reply to comment #3)
> > (From update of attachment 168153 [details] [details])
> > View in context: https://bugs.webkit.org/attachment.cgi?id=168153&action=review
> > 
> > Are those test failures real?
> 
> I don't think so. :-/  I will investigate though.

I don't see this locally.  Assuming it's a flaky test.

> 
> > 
> > This patch does not appear to do the inlining yet...
> 
> It just enables inlining of code blocks that use new_array_buffer.  Array allocations themselves are not yet inlined; that's https://bugs.webkit.org/show_bug.cgi?id=98608.
> 
> > 
> > > Source/JavaScriptCore/ChangeLog:10
> > > +        hash-cons them, either.
> > 
> > "const"
> 
> http://en.wikipedia.org/wiki/Hash_consing