Bug 125979 – Clarified stack maintenance code

• Status: RESOLVED
• Resolution: FIXED
• Component: New Bugs

[Geoffrey Garen]

Clarified stack maintainence code

[Geoffrey Garen]

Created attachment 219616 [details]
Patch

[Mark Lam]

Comment on attachment 219616 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=219616&action=review

Nice. r=me too with unused code removed.

> Source/JavaScriptCore/llint/LLIntEntrypoint.cpp:137
> +int stackPointerOffsetFor(CodeBlock* codeBlock)
> +{
> +    return virtualRegisterForLocal(frameRegisterCountFor(codeBlock) - 1).offset();
> +}
> +

Is this used anywhere?  If not, it’s better to remove it.

> Source/JavaScriptCore/llint/LLIntEntrypoint.h:45
> +int stackPointerOffsetFor(CodeBlock*);

Ditto ... not used.

[Mark Lam]

I see that DFGOSREntry.cpp and FTLOSREntry.cpp still uses JSStack::grow() with the expectation that it takes a "past the end" pointer, but that is inconsequential considering they should be checking against the C stack limits instead of the JSStack (via grow()).  These appear to be part of several stack checking issues which still remains broken in the jsCStack branch.  I'll take care of these in https://bugs.webkit.org/show_bug.cgi?id=125928 when I fix the other stack issues I reported there later.

[Geoffrey Garen]

> > Source/JavaScriptCore/llint/LLIntEntrypoint.cpp:137
> > +int stackPointerOffsetFor(CodeBlock* codeBlock)
> > +{
> > +    return virtualRegisterForLocal(frameRegisterCountFor(codeBlock) - 1).offset();
> > +}
> > +
> 
> Is this used anywhere?  If not, it’s better to remove it.

Oops! Will remove.

[Geoffrey Garen]

<http://trac.webkit.org/changeset/160835>