Bug 144205 – [JSC] Implement Math.clz32(), remove Number.clz()

• Component: New Bugs
• Status: RESOLVED FIXED

[Benjamin Poulain]

[JSC] Implement Math.clz32(), remove Number.clz()

[Benjamin Poulain]

Created attachment 251664 [details]
Patch

[Michael Saboff]

Comment on attachment 251664 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=251664&action=review

r=me with one question.

> Source/JavaScriptCore/assembler/MacroAssemblerX86Common.h:194
> +        xor32(TrustedImm32(0x1f), dst);

Why is the xor needed?  Since you protect for the zero case and bsr is spec'ed to return an unsigned value from 0..31, you shouldn't need this xor.

[Benjamin Poulain]

(In reply to comment #2)
> > Source/JavaScriptCore/assembler/MacroAssemblerX86Common.h:194
> > +        xor32(TrustedImm32(0x1f), dst);
> 
> Why is the xor needed?  Since you protect for the zero case and bsr is
> spec'ed to return an unsigned value from 0..31, you shouldn't need this xor.

BSR returns the position of the most significant bit, the opposite of what we need.
The xor32 is there to inverse the result.

[Benjamin Poulain]

Comment on attachment 251664 [details]
Patch

Clearing flags on attachment: 251664

Committed r183358: <http://trac.webkit.org/changeset/183358>

[Benjamin Poulain]

All reviewed patches have been landed.  Closing bug.

[Michael Saboff]

(In reply to comment #3)
> (In reply to comment #2)
> > > Source/JavaScriptCore/assembler/MacroAssemblerX86Common.h:194
> > > +        xor32(TrustedImm32(0x1f), dst);
> > 
> > Why is the xor needed?  Since you protect for the zero case and bsr is
> > spec'ed to return an unsigned value from 0..31, you shouldn't need this xor.
> 
> BSR returns the position of the most significant bit, the opposite of what
> we need.
> The xor32 is there to inverse the result.

Got it.