[webkit-reviews] review denied: [Bug 72857] Optimize JITArithmetic32_64.cpp:: op_add & op_sub : [Attachment 116030] the patch

[bugzilla-daemon at webkit.org]

Filip Pizlo <fpizlo at apple.com> has denied zeng huiqing
<huiqing.zeng at intel.com>'s request for review:
Bug 72857: Optimize JITArithmetic32_64.cpp:: op_add & op_sub
https://bugs.webkit.org/show_bug.cgi?id=72857

Attachment 116030: the patch
https://bugs.webkit.org/attachment.cgi?id=116030&action=review

------- Additional Comments from Filip Pizlo <fpizlo at apple.com>
View in context: https://bugs.webkit.org/attachment.cgi?id=116030&action=review


> Source/JavaScriptCore/jit/JITArithmetic32_64.cpp:632
> -    addSlowCase(branchAdd32(Overflow, regT2, regT0));
> -    emitStoreInt32(dst, regT0, (op1 == dst || op2 == dst));
> +    Jump overflow = branchAdd32(Overflow, regT2, regT0, regT3);
> +    emitStoreInt32(dst, regT3, (op1 == dst || op2 == dst));
> +    end.append(jump());

This change means that overflow is no longer counted by the RareCaseProfile,
which will lead to potentially disastrous performance regressions in DFG.