Support Smi in CompareIDAndBranch

Support Smi in CompareIDAndBranch

BUG=
R=jkummerow@chromium.org

Committed: https://code.google.com/p/v8/source/detail?r=14842

[Toon Verwaest, 2013-05-28 08:16:47]

PTAL

[Jakob Kummerow, 2013-05-28 09:00:54]

https://chromiumcodereview.appspot.com/16134003/diff/1009/src/hydrogen-instru...
File src/hydrogen-instructions.cc (right):

https://chromiumcodereview.appspot.com/16134003/diff/1009/src/hydrogen-instru...
src/hydrogen-instructions.cc:2456: void
HCompareIDAndBranch::InferRepresentation(HInferRepresentation* h_infer) {
As discussed offline, please rewrite this method to make it clear that what it
computes is the most general representation out of the set
{observed_input_representation(0), observed_input_representation(1), left_rep,
right_rep}, ignoring any inputs that are Tagged.

Then, if the computed |rep| is Double, set the flag kDeoptimizeOnUndefined if
necessary.

https://chromiumcodereview.appspot.com/16134003/diff/1009/src/ia32/lithium-ia...
File src/ia32/lithium-ia32.cc (right):

https://chromiumcodereview.appspot.com/16134003/diff/1009/src/ia32/lithium-ia...
src/ia32/lithium-ia32.cc:1694:
ASSERT(instr->right()->representation().IsSmiOrInteger32());
I think
ASSERT(instr->right()->representation().Equals(instr->left()->representation()))
would be much more useful here. Bad things will happen if you compare a Smi with
an Integer32.

[Toon Verwaest, 2013-05-28 09:06:47]

Addressed comments, PTAL again.

[Jakob Kummerow, 2013-05-28 09:10:33]

lgtm

[Toon Verwaest, 2013-05-28 09:24:47]

Committed patchset #2 manually as r14842 (presubmit successful).