Porting r9605 to x64 (elements kind conversion in generated code).

Porting r9605 to x64 (elements kind conversion in generated code).

Committed: http://code.google.com/p/v8/source/detail?r=9655

[Yang, 2011-10-14 08:32:09]

Please take a look.

This also includes small changes to the ia32 version, cutting some corners.

[Jakob Kummerow, 2011-10-14 09:25:38]

LGTM with comments addressed.

http://codereview.chromium.org/8271007/diff/1001/src/x64/code-stubs-x64.cc
File src/x64/code-stubs-x64.cc (right):

http://codereview.chromium.org/8271007/diff/1001/src/x64/code-stubs-x64.cc#ne...
src/x64/code-stubs-x64.cc:6006: __ jmp(&gc_required);
lolwut?

http://codereview.chromium.org/8271007/diff/1001/src/x64/code-stubs-x64.cc#ne...
src/x64/code-stubs-x64.cc:6016: // r11: destination FixedArray
Please move this up to after the AllocateInNewSpace line (where r11 gets set).

http://codereview.chromium.org/8271007/diff/1001/src/x64/code-stubs-x64.cc#ne...
src/x64/code-stubs-x64.cc:6040: __ movq(FieldOperand(r11, r9, times_8,
FixedDoubleArray::kHeaderSize), r12);
FixedArray::kHeaderSize (I know they're the same, but still). Also, I think
times_pointer_size would be clearer than times_8.

http://codereview.chromium.org/8271007/diff/1001/src/x64/code-stubs-x64.cc#ne...
src/x64/code-stubs-x64.cc:6052: __ movq(FieldOperand(r11, r9, times_8,
FixedDoubleArray::kHeaderSize), rdi);
times_pointer_size, FixedArray::kHeaderSize

http://codereview.chromium.org/8271007/diff/1001/src/x64/code-stubs-x64.cc#ne...
src/x64/code-stubs-x64.cc:6070: HeapObject::kMapOffset,
JSObject::kElementsOffset!

[Yang, 2011-10-14 16:04:27]

Please take another look. Issue v8:1767 seems to be unrelated to this CL since
it reproduces on bleeding_edge. I'll take a look at that problem independently
from this.

http://codereview.chromium.org/8271007/diff/1001/src/x64/code-stubs-x64.cc
File src/x64/code-stubs-x64.cc (right):

http://codereview.chromium.org/8271007/diff/1001/src/x64/code-stubs-x64.cc#ne...
src/x64/code-stubs-x64.cc:6006: __ jmp(&gc_required);
On 2011/10/14 09:25:38, Jakob wrote:
> lolwut?

Forgot to take this out after some debugging.

http://codereview.chromium.org/8271007/diff/1001/src/x64/code-stubs-x64.cc#ne...
src/x64/code-stubs-x64.cc:6016: // r11: destination FixedArray
On 2011/10/14 09:25:38, Jakob wrote:
> Please move this up to after the AllocateInNewSpace line (where r11 gets set).

Done.

http://codereview.chromium.org/8271007/diff/1001/src/x64/code-stubs-x64.cc#ne...
src/x64/code-stubs-x64.cc:6040: __ movq(FieldOperand(r11, r9, times_8,
FixedDoubleArray::kHeaderSize), r12);
On 2011/10/14 09:25:38, Jakob wrote:
> FixedArray::kHeaderSize (I know they're the same, but still). Also, I think
> times_pointer_size would be clearer than times_8.

Done.

http://codereview.chromium.org/8271007/diff/1001/src/x64/code-stubs-x64.cc#ne...
src/x64/code-stubs-x64.cc:6052: __ movq(FieldOperand(r11, r9, times_8,
FixedDoubleArray::kHeaderSize), rdi);
On 2011/10/14 09:25:38, Jakob wrote:
> times_pointer_size, FixedArray::kHeaderSize

Done.

http://codereview.chromium.org/8271007/diff/1001/src/x64/code-stubs-x64.cc#ne...
src/x64/code-stubs-x64.cc:6070: HeapObject::kMapOffset,
On 2011/10/14 09:25:38, Jakob wrote:
> JSObject::kElementsOffset!

Done.

[Jakob Kummerow, 2011-10-17 07:29:50]

LGTM