Reland r16751 with fix

Fixed a bug in CopyBytes() and new test cases for ARM macro assembler

TEST=cctest/test-macro-assembler-arm
R=bmeurer@chromium.org

Committed: https://code.google.com/p/v8/source/detail?r=16873

[Benedikt Meurer, 2013-09-20 10:31:20]

https://codereview.chromium.org/23447035/diff/1/src/arm/macro-assembler-arm.cc
File src/arm/macro-assembler-arm.cc (right):

https://codereview.chromium.org/23447035/diff/1/src/arm/macro-assembler-arm.c...
src/arm/macro-assembler-arm.cc:3224: b(ne, &align_loop_1);
Shouldn't this be an unconditional branch to align_loop_1? Because length cannot
become zero w/o src being properly aligned, right?

[Bangfu, 2013-09-20 10:47:12]

On 2013/09/20 10:31:20, Benedikt Meurer wrote:
> https://codereview.chromium.org/23447035/diff/1/src/arm/macro-assembler-arm.cc
> File src/arm/macro-assembler-arm.cc (right):
> 
>
https://codereview.chromium.org/23447035/diff/1/src/arm/macro-assembler-arm.c...
> src/arm/macro-assembler-arm.cc:3224: b(ne, &align_loop_1);
> Shouldn't this be an unconditional branch to align_loop_1? Because length
cannot
> become zero w/o src being properly aligned, right?

You are right, it would be better use unconditional branch since loop counter is
grater than 4. 
I am testing the changes now.

[Bangfu, 2013-09-20 13:24:05]

On 2013/09/20 10:47:12, Bangfu wrote:
> On 2013/09/20 10:31:20, Benedikt Meurer wrote:
> >
https://codereview.chromium.org/23447035/diff/1/src/arm/macro-assembler-arm.cc
> > File src/arm/macro-assembler-arm.cc (right):
> > 
> >
>
https://codereview.chromium.org/23447035/diff/1/src/arm/macro-assembler-arm.c...
> > src/arm/macro-assembler-arm.cc:3224: b(ne, &align_loop_1);
> > Shouldn't this be an unconditional branch to align_loop_1? Because length
> cannot
> > become zero w/o src being properly aligned, right?
> 
> You are right, it would be better use unconditional branch since loop counter
is
> grater than 4. 
> I am testing the changes now.

Done.

[Benedikt Meurer, 2013-09-20 13:37:00]

Ok, LGTM.

[Benedikt Meurer, 2013-09-23 08:07:14]

Committed patchset #2 manually as r16873 (presubmit successful).