Record the lazy deoptimization environmnent only at LLazyBailout-instructions...

Record the lazy deoptimization environmnent only at LLazyBailout-instructions.

This saves one pointer per LIR instructions. The change relies on the fact
that there is a LazyBailout instruction at all the places where we need
to register a lazy deoptimization environment.


Committed: http://code.google.com/p/v8/source/detail?r=6441

[fschneider, 2011-01-24 12:48:20]

Please have a look - I'm not yet sure if this is the right approach.

[Søren Thygesen Gjesse, 2011-01-24 13:02:43]

LGTM

http://codereview.chromium.org/6348016/diff/1/src/ia32/lithium-codegen-ia32.cc
File src/ia32/lithium-codegen-ia32.cc (right):

http://codereview.chromium.org/6348016/diff/1/src/ia32/lithium-codegen-ia32.c...
src/ia32/lithium-codegen-ia32.cc:419:
ASSERT(instructions_->at(current_instruction_ + 1)->IsGap());
As far as I remember there used to be places where there could be two gaps
before a lazy bailout, but that might have been changed.

[Kevin Millikin (Chromium), 2011-01-24 13:31:38]

One small question below:

http://codereview.chromium.org/6348016/diff/1/src/ia32/lithium-ia32.cc
File src/ia32/lithium-ia32.cc (right):

http://codereview.chromium.org/6348016/diff/1/src/ia32/lithium-ia32.cc#newcod...
src/ia32/lithium-ia32.cc:1861: if (pending_deoptimization_ast_id_ ==
instr->ast_id()) {
This should really be:

if (pending_deoptimization_ast_id_ != AstNode::kNoNumber) {
  ASSERT(pending_deoptimization_ast_id_ == instr->ast_id());
  ...

http://codereview.chromium.org/6348016/diff/1/src/ia32/lithium-ia32.cc#newcod...
src/ia32/lithium-ia32.cc:1863: LInstruction* result =
AssignEnvironment(lazy_bailout);
I'm worried that the uses in this environment appear to come after the gap the
precedes it, but the environment before the gap is the one that is recorded.

Is that a problem?