[Webkit-unassigned] [Bug 84324] DFG can have the spill format identical to the register format

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=84324





--- Comment #6 from Filip Pizlo <fpizlo at apple.com>  2012-06-05 20:55:44 PST ---
(From update of attachment 145924)
View in context: https://bugs.webkit.org/attachment.cgi?id=145924&action=review

I think that the code is probably right, in the sense that isKnownNotFooBar() will only return true if the CFA would have also concluded that the code will OSR exit.  But maybe we can put some tighter assertions about this into the code?

Also, the isKnownNotFooBar() and isKnownFooBar() methods are kind of a hack that is left over from the old DFG days.  It would be nice to get rid of them...

> Source/JavaScriptCore/dfg/DFGSpeculativeJIT64.cpp:1042
> +    if (isKnownNotInteger(nodeIndex)) {
> +        terminateSpeculativeExecution(Uncountable, JSValueRegs(), NoNode);
> +        returnFormat = DataFormatInteger;
> +        return allocate();
> +    }

Same comment as before...

> Source/JavaScriptCore/dfg/DFGSpeculativeJIT64.cpp:1175
> +    if (isKnownNotNumber(nodeIndex)) {
> +        terminateSpeculativeExecution(Uncountable, JSValueRegs(), NoNode);
> +        return fprAllocate();
> +    }

Ditto

> Source/JavaScriptCore/dfg/DFGSpeculativeJIT64.cpp:1329
> +    if (isKnownNotCell(nodeIndex)) {
> +        terminateSpeculativeExecution(Uncountable, JSValueRegs(), NoNode);
> +        return allocate();
> +    }

Ditto

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.