Use +kCallerSPOffset instead of -kMarkerOffset to compute the args pointer

Use +kCallerSPOffset instead of -kMarkerOffset to compute the args pointer

R=mstarzinger@chromium.org

Committed: http://code.google.com/p/v8/source/detail?r=15238

[haitao.feng, 2013-06-20 05:29:23]

It makes sense to use +kCallerSPOffset instead of -kMarkerOffset to compute the
args pointer. kCallerSPOffset is what we have lost from (output_frame_size -
output_frame_offset).

This change is needed for X32 as  +kCallerSPOffset is not equal to
-kMarkerOffset.

[Michael Starzinger, 2013-06-20 11:39:27]

LGTM with a nit.

https://codereview.chromium.org/17451014/diff/1/src/deoptimizer.cc
File src/deoptimizer.cc (right):

https://codereview.chromium.org/17451014/diff/1/src/deoptimizer.cc#newcode1514
src/deoptimizer.cc:1514: value = frame_ptr - (output_frame_size -
output_frame_offset) +
nit: Can we re-order this calculation slightly? This would make it in line with
the calculation of args.arguments and also easier to parse IMHO. I was thinking
about the following order:

value = frame_ptr + StandardFrameConstants::kCallerSPOffset -
    (output_frame_size - output_frame_offset) + kPointerSize;

[haitao.feng, 2013-06-20 12:26:19]

Michael, thanks for review.

https://codereview.chromium.org/17451014/diff/1/src/deoptimizer.cc
File src/deoptimizer.cc (right):

https://codereview.chromium.org/17451014/diff/1/src/deoptimizer.cc#newcode1514
src/deoptimizer.cc:1514: value = frame_ptr - (output_frame_size -
output_frame_offset) +
On 2013/06/20 11:39:27, Michael Starzinger wrote:
> nit: Can we re-order this calculation slightly? This would make it in line
with
> the calculation of args.arguments and also easier to parse IMHO. I was
thinking
> about the following order:
> 
> value = frame_ptr + StandardFrameConstants::kCallerSPOffset -
>     (output_frame_size - output_frame_offset) + kPointerSize;

Done.

[Michael Starzinger, 2013-06-20 12:51:41]

Committed patchset #2 manually as r15238 (presubmit successful).