Bug 141915

Summary: Scopes should always be created with a previously-created symbol table rather than creating one on the fly
Product: WebKit Reporter: Filip Pizlo <fpizlo>
Component: JavaScriptCoreAssignee: Filip Pizlo <fpizlo>
Status: RESOLVED FIXED    
Severity: Normal CC: barraclough, benjamin, commit-queue, ggaren, mark.lam, mhahnenb, mmirman, msaboff, nrotem, oliver, ossy, saam, sam
Priority: P2    
Version: 528+ (Nightly build)   
Hardware: All   
OS: All   
Bug Depends on: 141921, 141953    
Bug Blocks: 141174    
Attachments:
Description Flags
the patch
mark.lam: review+
patch for landing
none
fix 32-bit none

Description Filip Pizlo 2015-02-23 10:33:03 PST 
Patch forthcoming.
Comment 1 Filip Pizlo 2015-02-23 10:55:25 PST 
Ossy: this will have some new callOperation() overloads in JIT.h.  Expect breakage.
Comment 2 Filip Pizlo 2015-02-23 11:27:09 PST 
(In reply to comment #1)
> Ossy: this will have some new callOperation() overloads in JIT.h.  Expect
> breakage.

Actually, we might get lucky.  The new overloads seem to bottom out in existing setupArgumentsXYZ overloads.
Comment 3 Filip Pizlo 2015-02-23 11:41:51 PST 
Created attachment 247131 [details]
the patch
Comment 4 WebKit Commit Bot 2015-02-23 11:42:51 PST 
Attachment 247131 [details] did not pass style-queue:


ERROR: Source/JavaScriptCore/runtime/JSNameScope.h:52:  The parameter name "symbolTable" adds no information, so it should be removed.  [readability/parameter_name] [5]
Total errors found: 1 in 20 files


If any of these errors are false positives, please file a bug against check-webkit-style.
Comment 5 Mark Lam 2015-02-23 12:00:19 PST 
Comment on attachment 247131 [details]
the patch

View in context: https://bugs.webkit.org/attachment.cgi?id=247131&action=review

r=me

> Source/JavaScriptCore/jit/JITOperations.cpp:71
> +    // FIXME: This won't work if this operation is called from the DFG or FTL.
> +    // This should be changed to pass in the new scope.

I see that this came from pre-existing code, but can you add a debug ASSERT to ensure that the current exec/frame is not a DFG/FTL frame so that we'll catch this if someone unwittingly uses this in the DFG/FTL without seeing this FIXME?

> Source/JavaScriptCore/runtime/JSSymbolTableObject.h:64
> +    

nit: unneeded white space.
Comment 6 Filip Pizlo 2015-02-23 12:05:09 PST 
(In reply to comment #5)
> Comment on attachment 247131 [details]
> the patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=247131&action=review
> 
> r=me
> 
> > Source/JavaScriptCore/jit/JITOperations.cpp:71
> > +    // FIXME: This won't work if this operation is called from the DFG or FTL.
> > +    // This should be changed to pass in the new scope.
> 
> I see that this came from pre-existing code, but can you add a debug ASSERT
> to ensure that the current exec/frame is not a DFG/FTL frame so that we'll
> catch this if someone unwittingly uses this in the DFG/FTL without seeing
> this FIXME?

Sure.

> 
> > Source/JavaScriptCore/runtime/JSSymbolTableObject.h:64
> > +    
> 
> nit: unneeded white space.

OK.
Comment 7 Filip Pizlo 2015-02-23 12:16:05 PST 
Created attachment 247134 [details]
patch for landing
Comment 8 Filip Pizlo 2015-02-23 12:53:17 PST 
Created attachment 247137 [details]
fix 32-bit
Comment 9 Filip Pizlo 2015-02-23 13:54:55 PST 
Landed in http://trac.webkit.org/changeset/180514
Comment 10 Csaba Osztrogonác 2015-02-23 14:03:30 PST 
(In reply to comment #1)
> Ossy: this will have some new callOperation() overloads in JIT.h.  Expect
> breakage.

Thanks for the notice, fix landed in http://trac.webkit.org/changeset/180515