Your comments addressed, some additional changes made.  Please check
again.



http://codereview.chromium.org/113997/diff/1012/9
File src/x64/assembler-x64.cc (right):

http://codereview.chromium.org/113997/diff/1012/9#newcode126
Line 126: __ movq(rdx, rcx);
On 2009/06/02 08:57:20, Lasse Reichstein wrote:
> Here you should combine ecx and edx into eax to return it as a 64-bit
value. The
> F0-function typedef returns a 64-bit value in rax, not in edx:eax as
in 32-bit
> mode.
> I.e.,
> __ movl(rax, rdx);  // i.e., 32-bit zero-extending move
> __ or_(rax, 1 << CPUID);
> __ shl(ecx, 32);
> __ or_(rax, ecx);


Done.

http://codereview.chromium.org/113997/diff/1012/9#newcode476
Line 476: emit_rex_64(rax);
On 2009/06/02 08:57:20, Lasse Reichstein wrote:
> Just emit 0x40, or make a plain emit_rex_64(). The rax register use is
> confusing.

Done, emit_rex_64().  This is 0x48, not 0x40.  The W bit is set.

http://codereview.chromium.org/113997/diff/1012/10
File src/x64/assembler-x64.h (right):

http://codereview.chromium.org/113997/diff/1012/10#newcode403
Line 403: // TODO(): Move to macro assembler.
On 2009/06/02 08:57:20, Lasse Reichstein wrote:
> Labels are part of the assembler proper. No need to move it to the
> macro-assembler.

Done.

http://codereview.chromium.org/113997

--~--~---------~--~----~------------~-------~--~----~
v8-dev mailing list
[email protected]
http://groups.google.com/group/v8-dev
-~----------~----~----~----~------~----~------~--~---

Reply via email to