Generate DataViewInitialize built-in in hydrogen.

Generate DataViewInitialize built-in in hydrogen.

R=bmeurer@chromium.org, mvstanton@chromium.org

Committed: https://code.google.com/p/v8/source/detail?r=17831

[Dmitry Lomov (no reviews), 2013-11-15 14:37:09]

This CL is a close friend of https://codereview.chromium.org/59023003/

[Benedikt Meurer, 2013-11-18 08:02:31]

https://codereview.chromium.org/66843011/diff/1/src/hydrogen.cc
File src/hydrogen.cc (right):

https://codereview.chromium.org/66843011/diff/1/src/hydrogen.cc#newcode7953
src/hydrogen.cc:7953: byte_offset);
byte_offset should be a SMI, so you should pass Representation::Smi() to
ForJSObjectOffset() to avoid the write barrier.

https://codereview.chromium.org/66843011/diff/1/src/hydrogen.cc#newcode7955
src/hydrogen.cc:7955:
HObjectAccess::ForJSObjectOffset(JSDataView::kByteLengthOffset),
byte_length is a SMI, so you should pass Representation::Smi() to
ForJSObjectOffset() to avoid the write barrier.

[Dmitry Lomov (no reviews), 2013-11-18 08:46:46]

https://codereview.chromium.org/66843011/diff/1/src/hydrogen.cc
File src/hydrogen.cc (right):

https://codereview.chromium.org/66843011/diff/1/src/hydrogen.cc#newcode7953
src/hydrogen.cc:7953: byte_offset);
On 2013/11/18 08:02:31, Benedikt Meurer wrote:
> byte_offset should be a SMI, so you should pass Representation::Smi() to
> ForJSObjectOffset() to avoid the write barrier.

Byte offset is not neccessarily a smi here. Ditto byte length below.

[Benedikt Meurer, 2013-11-18 08:56:01]

Ok.

LGTM

[mvstanton, 2013-11-18 09:40:31]

lgtm. I like this idea of generating optimized versions of runtime functions,
may be helpful for literals too. thx!

[Dmitry Lomov (no reviews), 2013-11-18 13:57:58]

Committed patchset #1 manually as r17831 (presubmit successful).