Fix for failing asserts in HBoundsCheck code generation on x64: use proper cmp operation width inst…

Fix for failing asserts in HBoundsCheck code generation on x64: use proper cmp operation width instead of asserting that Integer32 values should be zero extended. Similar to chromium:345820.

BUG=349465
LOG=N
R=verwaest@chromium.org

Committed: https://code.google.com/p/v8/source/detail?r=19694

[Igor Sheludko, 2014-03-06 14:28:05]

PTAL

[Toon Verwaest, 2014-03-06 15:39:18]

lgtm with nit

https://codereview.chromium.org/188703002/diff/1/test/mjsunit/regress/regress...
File test/mjsunit/regress/regress-crbug-349465-1.js (right):

https://codereview.chromium.org/188703002/diff/1/test/mjsunit/regress/regress...
test/mjsunit/regress/regress-crbug-349465-1.js:7: __v_1 = new Int32Array(1024);
Could you please sanitize these tests a bit? Try to reduce as much as possible,
and remove all the __ stuff. That'll just make it easier to understand the test
in the future.

[Igor Sheludko, 2014-03-06 16:22:54]

Committed patchset #2 manually as r19694 (tree was closed).

[Igor Sheludko, 2014-03-06 17:21:20]

https://codereview.chromium.org/188703002/diff/1/test/mjsunit/regress/regress...
File test/mjsunit/regress/regress-crbug-349465-1.js (right):

https://codereview.chromium.org/188703002/diff/1/test/mjsunit/regress/regress...
test/mjsunit/regress/regress-crbug-349465-1.js:7: __v_1 = new Int32Array(1024);
On 2014/03/06 15:39:19, Toon Verwaest wrote:
> Could you please sanitize these tests a bit? Try to reduce as much as
possible,
> and remove all the __ stuff. That'll just make it easier to understand the
test
> in the future.

Done.