Mark all loop conditions.

Mark all loop conditions.

Previously we marked compare operations that occurred as for loop
conditions so as to avoid inlining the floating-point code for them.
Begin marking all expressions that occur as any loop condition
(because they are never dead code).

Committed: http://code.google.com/p/v8/source/detail?r=4227

[fschneider, 2010-03-23 13:19:38]

LGTM.

http://codereview.chromium.org/1074009/diff/4001/5004
File src/ia32/codegen-ia32.cc (right):

http://codereview.chromium.org/1074009/diff/4001/5004#newcode2475
src/ia32/codegen-ia32.cc:2475: bool is_loop_condition = (node->AsExpression() !=
NULL) &&
I think we should change CodeGenerator::Comparison to take an Expression* node
as argument.

[Søren Thygesen Gjesse, 2010-03-23 13:21:02]

LGTM

http://codereview.chromium.org/1074009/diff/4001/5003
File src/data-flow.cc (right):

http://codereview.chromium.org/1074009/diff/4001/5003#newcode223
src/data-flow.cc:223: cond_copy->set_is_loop_condition(false);
Strictly speaking this is a lie, and it will cause the number compare code to be
inlined (if already in a loop), which I am not sure we want. Should we keep this
as a loop condition?

[Kevin Millikin (Chromium), 2010-03-23 13:36:26]

http://codereview.chromium.org/1074009/diff/4001/5003
File src/data-flow.cc (right):

http://codereview.chromium.org/1074009/diff/4001/5003#newcode223
src/data-flow.cc:223: cond_copy->set_is_loop_condition(false);
On 2010/03/23 13:21:02, Søren Gjesse wrote:
> Strictly speaking this is a lie, and it will cause the number compare code to
be
> inlined (if already in a loop), which I am not sure we want. Should we keep
this
> as a loop condition?

I guess not.  I've removed it for now.