Recognize special comparisons via pattern matching on the hydrogen graph, 2nd attempt.

Recognize special comparisons via pattern matching on the hydrogen graph, 2nd attempt.

This time, we initially leave the HTypeof instruction in the Hydrogen graph,
even for the special cases. We later try to remove this instruction (and any
HConstant) in the canonicalization pass, if possible. Always removing the
HTypeof during the initial graph construction is wrong if e.g. it is used in an
HSimulate.

The removals can be generalized a bit, but this will happen in a separate CL.

TEST=mjsunit/optimized-typeof.js
Committed: http://code.google.com/p/v8/source/detail?r=9688

[Kevin Millikin (Chromium), 2011-10-19 07:24:41]

LGTM.

http://codereview.chromium.org/8334021/diff/1/src/hydrogen-instructions.cc
File src/hydrogen-instructions.cc (right):

http://codereview.chromium.org/8334021/diff/1/src/hydrogen-instructions.cc#ne...
src/hydrogen-instructions.cc:791: return HasNoUses()&& !IsBlockEntry() ? NULL :
this;
Missing a space before &&, also below.

http://codereview.chromium.org/8334021/diff/1/src/hydrogen.cc
File src/hydrogen.cc (right):

http://codereview.chromium.org/8334021/diff/1/src/hydrogen.cc#newcode5798
src/hydrogen.cc:5798: ast_context()->ReturnControl(instr, expr->id());
I like the "return".  This function is (sort of) assumed to be called in tail
position as part of translating an expression, so the return (sort of) enforces
that.

[Sven Panne, 2011-10-19 07:33:32]

http://codereview.chromium.org/8334021/diff/1/src/hydrogen-instructions.cc
File src/hydrogen-instructions.cc (right):

http://codereview.chromium.org/8334021/diff/1/src/hydrogen-instructions.cc#ne...
src/hydrogen-instructions.cc:791: return HasNoUses()&& !IsBlockEntry() ? NULL :
this;
On 2011/10/19 07:24:41, Kevin Millikin wrote:
> Missing a space before &&, also below.

Done.

http://codereview.chromium.org/8334021/diff/1/src/hydrogen.cc
File src/hydrogen.cc (right):

http://codereview.chromium.org/8334021/diff/1/src/hydrogen.cc#newcode5798
src/hydrogen.cc:5798: ast_context()->ReturnControl(instr, expr->id());
On 2011/10/19 07:24:41, Kevin Millikin wrote:
> I like the "return".  This function is (sort of) assumed to be called in tail
> position as part of translating an expression, so the return (sort of)
enforces
> that.

Done. Personally I like returning void, too, it is just like any other type
IMHO. :-) But I had discussions about this in other places in the past, so I
removed it during my various attempts to get this function right. Now it has
made its glorious comeback...