[Webkit-unassigned] [Bug 99154] Refactor MacroAssembler interfaces to differentiate the pointer operands from the 64-bit integer operands

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=99154


Gavin Barraclough <barraclough at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
 Attachment #168383|                            |review-
               Flag|                            |




--- Comment #3 from Gavin Barraclough <barraclough at apple.com>  2012-10-12 17:41:36 PST ---
(From update of attachment 168383)
View in context: https://bugs.webkit.org/attachment.cgi?id=168383&action=review

(r-, comments to follow)

> Source/JavaScriptCore/assembler/AbstractMacroAssembler.h:278
> +        explicit TrustedImm64(const void* value)

Do we really need this constructor? – might be nice to try to remove it – if you have a void*, you should be creating a TrustedImmPtr.

> Source/JavaScriptCore/assembler/AbstractMacroAssembler.h:286
> +

Extra newline.

> Source/JavaScriptCore/assembler/AbstractMacroAssembler.h:299
> +

Extra newline.

> Source/JavaScriptCore/jit/JITOpcodes.cpp:1333
> +    move(TrustedImm64(reinterpret_cast<uintptr_t>(globalThis)), regT0);

I think this code is implicitly making an assumption about JSValue boxing.  I think this would be better:
    move(TrustedImm64(JSValue::encode(JSValue(globalThis))), regT0);

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