Add test cases for fast smi loops.

Add test cases for fast smi loops.


Committed: http://code.google.com/p/v8/source/detail?r=4174

[William Hesse, 2010-03-18 08:52:50]

LGTM.

http://codereview.chromium.org/1014007/diff/1/3
File src/ia32/codegen-ia32.cc (right):

http://codereview.chromium.org/1014007/diff/1/3#newcode3731
src/ia32/codegen-ia32.cc:3731: // and (b) the loop induction variable is not
assignend inside the
The added sentence is not capitalized, and has no verb.
How about, instead,

// We know that the loop index is a smi if it is not modified in the loop body
and it is checked against a constant limit in the loop condition.  In this case,
we reset the static type information of the loop index to smi before compiling
the body, the update expression, and the bottom check of the loop condition.

http://codereview.chromium.org/1014007/diff/1/3#newcode3768
src/ia32/codegen-ia32.cc:3768: // expression if we are in a fast smi loop.
loop condition?

[fschneider, 2010-03-18 09:08:20]

http://codereview.chromium.org/1014007/diff/1/3
File src/ia32/codegen-ia32.cc (right):

http://codereview.chromium.org/1014007/diff/1/3#newcode3731
src/ia32/codegen-ia32.cc:3731: // and (b) the loop induction variable is not
assignend inside the
On 2010/03/18 08:52:50, William Hesse wrote:
> The added sentence is not capitalized, and has no verb.
> How about, instead,
> 
> // We know that the loop index is a smi if it is not modified in the loop body
> and it is checked against a constant limit in the loop condition.  In this
case,
> we reset the static type information of the loop index to smi before compiling
> the body, the update expression, and the bottom check of the loop condition.

Done.

http://codereview.chromium.org/1014007/diff/1/3#newcode3768
src/ia32/codegen-ia32.cc:3768: // expression if we are in a fast smi loop.
On 2010/03/18 08:52:50, William Hesse wrote:
> loop condition?

Done.