X64 Implementation: Implement Generate_Function(Call,Apply)

X64 Implementation: Implement Generate_Function(Call,Apply)

Committed: http://code.google.com/p/v8/source/detail?r=2287

[Lasse Reichstein, 2009-06-29 06:52:34]

LGTM

http://codereview.chromium.org/147205/diff/1/2
File src/x64/builtins-x64.cc (right):

http://codereview.chromium.org/147205/diff/1/2#newcode185
Line 185: // +1 ~ return address.
Not understood. The +1 is added because of the return address?
Make a constant and refer to the stack structure.

http://codereview.chromium.org/147205/diff/1/2#newcode316
Line 316: __ subq(rcx, Operand(kScratchRegister, 0));  // We need this number
below.
Make a comment that the value in "rcx" is reused later.

http://codereview.chromium.org/147205/diff/1/2#newcode316
Line 316: __ subq(rcx, Operand(kScratchRegister, 0));  // We need this number
below.
Reword comment. Not clear what "number" refers to, or what "need" means.
After the subq, rcx contains the location of rsp relative to the stack guard
limit. If the limit's above the stack pointer, we are have hit the limit and
must take action.

http://codereview.chromium.org/147205/diff/1/2#newcode320
Line 320: // Because builtins always remove the receiver from the stack, we
It's a runtime function, not a builtin (well, not the way V8 usually uses
"builtin")

http://codereview.chromium.org/147205/diff/1/2#newcode356
Line 356: __ movq(rdi, Operand(rbp, 4 * kPointerSize));
Could you add a telling name for rbp + 4 * kPointerSize, if one does not exist
already.

http://codereview.chromium.org/147205/diff/1/2#newcode361
Line 361: __ movq(rbx, Operand(rbp, 3 * kPointerSize));
And for rbp + 3 * kPointerSize

[William Hesse, 2009-06-29 08:06:48]

Includes changelist from issue 151004 because 151004 was on a machine that
cannot commit. http://codereview.chromium.org/151004

http://codereview.chromium.org/147205/diff/1/2
File src/x64/builtins-x64.cc (right):

http://codereview.chromium.org/147205/diff/1/2#newcode185
Line 185: // +1 ~ return address.
On 2009/06/29 06:52:35, Lasse Reichstein wrote:
> Not understood. The +1 is added because of the return address?
> Make a constant and refer to the stack structure.

Done.

http://codereview.chromium.org/147205/diff/1/2#newcode316
Line 316: __ subq(rcx, Operand(kScratchRegister, 0));  // We need this number
below.
On 2009/06/29 06:52:35, Lasse Reichstein wrote:
> Make a comment that the value in "rcx" is reused later.

Done.

http://codereview.chromium.org/147205/diff/1/2#newcode316
Line 316: __ subq(rcx, Operand(kScratchRegister, 0));  // We need this number
below.
On 2009/06/29 06:52:35, Lasse Reichstein wrote:
> Reword comment. Not clear what "number" refers to, or what "need" means.
> After the subq, rcx contains the location of rsp relative to the stack guard
> limit. If the limit's above the stack pointer, we are have hit the limit and
> must take action.

Done.

http://codereview.chromium.org/147205/diff/1/2#newcode320
Line 320: // Because builtins always remove the receiver from the stack, we
On 2009/06/29 06:52:35, Lasse Reichstein wrote:
> It's a runtime function, not a builtin (well, not the way V8 usually uses
> "builtin")

Done.

http://codereview.chromium.org/147205/diff/1/2#newcode356
Line 356: __ movq(rdi, Operand(rbp, 4 * kPointerSize));
On 2009/06/29 06:52:35, Lasse Reichstein wrote:
> Could you add a telling name for rbp + 4 * kPointerSize, if one does not exist
> already.

Done.

http://codereview.chromium.org/147205/diff/1/2#newcode361
Line 361: __ movq(rbx, Operand(rbp, 3 * kPointerSize));
On 2009/06/29 06:52:35, Lasse Reichstein wrote:
> And for rbp + 3 * kPointerSize

Done.