Trim underlying fixed array by one element from the left when doing shift.

Trim underlying fixed array by one element from the left when doing shift.

For now this trick is only done to objects in new space, see comments
for reasons.

Committed: http://code.google.com/p/v8/source/detail?r=4229

[antonm, 2010-03-23 10:26:11]

Mads,

may you have a look?

[Mads Ager (chromium), 2010-03-23 12:36:50]

LGTM

http://codereview.chromium.org/1076010/diff/1/2
File src/builtins.cc (right):

http://codereview.chromium.org/1076010/diff/1/2#newcode448
src/builtins.cc:448: // In large object space object's start must coincide with
chunk
object's -> the object's

http://codereview.chromium.org/1076010/diff/1/2#newcode450
src/builtins.cc:450: // In old space I'd rather not play rset tricks like
How about: In old space we do not use this trick to avoid dealing with
remembered sets.

http://codereview.chromium.org/1076010/diff/1/2#newcode473
src/builtins.cc:473: ASSERT(elms->address() + kPointerSize
Doesn't this fit on one line?

[antonm, 2010-03-23 13:27:10]

Thanks a lot for review, Mads.

http://codereview.chromium.org/1076010/diff/1/2
File src/builtins.cc (right):

http://codereview.chromium.org/1076010/diff/1/2#newcode448
src/builtins.cc:448: // In large object space object's start must coincide with
chunk
On 2010/03/23 12:36:50, Mads Ager wrote:
> object's -> the object's

Done.

http://codereview.chromium.org/1076010/diff/1/2#newcode450
src/builtins.cc:450: // In old space I'd rather not play rset tricks like
On 2010/03/23 12:36:50, Mads Ager wrote:
> How about: In old space we do not use this trick to avoid dealing with
> remembered sets.

Done.

http://codereview.chromium.org/1076010/diff/1/2#newcode473
src/builtins.cc:473: ASSERT(elms->address() + kPointerSize
On 2010/03/23 12:36:50, Mads Ager wrote:
> Doesn't this fit on one line?

Done.