load ics for js api accessors

load ics for js api accessors

R=mstarzinger@chromium.org
BUG=

Committed: https://code.google.com/p/v8/source/detail?r=16543

[dcarney, 2013-08-28 11:23:10]

pls take a look and if you're okay with the approach, i'll extend to keyed loads
and add remaining architectures

[Michael Starzinger, 2013-09-04 08:33:40]

Yes, I like the approach. Looking good.

[dcarney, 2013-09-04 13:54:10]

okay this is ready for review

https://codereview.chromium.org/23699002/diff/7001/src/ic.cc
File src/ic.cc (right):

https://codereview.chromium.org/23699002/diff/7001/src/ic.cc#newcode1575
src/ic.cc:1575: }
do we want to add computekeyedloadviagetter in follow up??

[Michael Starzinger, 2013-09-04 16:32:31]

LGTM if comments are addressed. 

Also test cases where the getter and setter functions have signatures would be
helpful!

https://codereview.chromium.org/23699002/diff/7001/src/ia32/stub-cache-ia32.cc
File src/ia32/stub-cache-ia32.cc (right):

https://codereview.chromium.org/23699002/diff/7001/src/ia32/stub-cache-ia32.c...
src/ia32/stub-cache-ia32.cc:1366: // assign stack space for the call arguments
nit: Capitalize and punctuate comments.

https://codereview.chromium.org/23699002/diff/7001/src/ia32/stub-cache-ia32.c...
src/ia32/stub-cache-ia32.cc:1375:
call_optimization.expected_receiver_type().is_null() ?
We should just always use receiver() here, because either the expected receiver
type is null or the first compatible object is in receiver(). Using "reg" here
is never what we intend it to be. This applies to all architectures.

[dcarney, 2013-09-05 08:34:35]

Committed patchset #4 manually as r16543 (presubmit successful).