Widen the intake valve for TurboFan.

Widen the intake valve for TurboFan.

R=danno@chromium.org, mstarzinger@chromium.org
BUG=

Committed: https://code.google.com/p/v8/source/detail?r=24083

[Michael Starzinger, 2014-09-18 19:12:34]

https://codereview.chromium.org/582703002/diff/1/src/compiler.cc
File src/compiler.cc (right):

https://codereview.chromium.org/582703002/diff/1/src/compiler.cc#newcode401
src/compiler.cc:401: info()->function()->dont_optimize_reason() !=
kTryCatchStatement &&
Please move this blacklist together with the TODO's into Pipeline::GenerateCode,
so that both call-sites have the same set of blacklist items. The only thing
remaining in this predicate should be the FLAG_turbo_filter check.

https://codereview.chromium.org/582703002/diff/1/src/compiler.cc#newcode747
src/compiler.cc:747: if (!Parser::Parse(&info)) return MaybeHandle<Code>();
I don't particularly like this duplication, in fact I hate it. But since this
file is a mess, any suggestion at making this cleaner would probably involve
burning this file to the ground first. So I'll live with this duplication for
now I guess.

https://codereview.chromium.org/582703002/diff/1/src/scopes.cc
File src/scopes.cc (right):

https://codereview.chromium.org/582703002/diff/1/src/scopes.cc#newcode164
src/scopes.cc:164: asm_function_ = outer_scope != NULL &&
outer_scope->asm_module_;
This only applies of the scope_type is FUNCTION_SCOPE. Wouldn't false as a
default value suffice here anyways?

[titzer, 2014-09-19 11:38:22]

https://codereview.chromium.org/582703002/diff/1/src/compiler.cc
File src/compiler.cc (right):

https://codereview.chromium.org/582703002/diff/1/src/compiler.cc#newcode401
src/compiler.cc:401: info()->function()->dont_optimize_reason() !=
kTryCatchStatement &&
On 2014/09/18 19:12:34, Michael Starzinger wrote:
> Please move this blacklist together with the TODO's into
Pipeline::GenerateCode,
> so that both call-sites have the same set of blacklist items. The only thing
> remaining in this predicate should be the FLAG_turbo_filter check.

Done.

https://codereview.chromium.org/582703002/diff/1/src/compiler.cc#newcode747
src/compiler.cc:747: if (!Parser::Parse(&info)) return MaybeHandle<Code>();
On 2014/09/18 19:12:34, Michael Starzinger wrote:
> I don't particularly like this duplication, in fact I hate it. But since this
> file is a mess, any suggestion at making this cleaner would probably involve
> burning this file to the ground first. So I'll live with this duplication for
> now I guess.

Agreed. We need to factor out into a routine to parse, rewrite, scope analyze,
and commit the results to SharedFunctionInfo and ScopeInfo objects.

https://codereview.chromium.org/582703002/diff/1/src/scopes.cc
File src/scopes.cc (right):

https://codereview.chromium.org/582703002/diff/1/src/scopes.cc#newcode164
src/scopes.cc:164: asm_function_ = outer_scope != NULL &&
outer_scope->asm_module_;
On 2014/09/18 19:12:34, Michael Starzinger wrote:
> This only applies of the scope_type is FUNCTION_SCOPE. Wouldn't false as a
> default value suffice here anyways?

Eh, the scope creation tangle is hard to follow and I'm not if there is a path
that goes through here without going through one of the other places that
propagates the bit. :(

[Michael Starzinger, 2014-09-19 11:40:59]

LGTM.

https://codereview.chromium.org/582703002/diff/1/src/scopes.cc
File src/scopes.cc (right):

https://codereview.chromium.org/582703002/diff/1/src/scopes.cc#newcode164
src/scopes.cc:164: asm_function_ = outer_scope != NULL &&
outer_scope->asm_module_;
On 2014/09/19 11:38:21, titzer wrote:
> On 2014/09/18 19:12:34, Michael Starzinger wrote:
> > This only applies of the scope_type is FUNCTION_SCOPE. Wouldn't false as a
> > default value suffice here anyways?
> 
> Eh, the scope creation tangle is hard to follow and I'm not if there is a path
> that goes through here without going through one of the other places that
> propagates the bit. :(

Acknowledged.

[danno, 2014-09-19 12:15:30]

On 2014/09/19 11:40:59, Michael Starzinger wrote:
> LGTM.
> 
> https://codereview.chromium.org/582703002/diff/1/src/scopes.cc
> File src/scopes.cc (right):
> 
> https://codereview.chromium.org/582703002/diff/1/src/scopes.cc#newcode164
> src/scopes.cc:164: asm_function_ = outer_scope != NULL &&
> outer_scope->asm_module_;
> On 2014/09/19 11:38:21, titzer wrote:
> > On 2014/09/18 19:12:34, Michael Starzinger wrote:
> > > This only applies of the scope_type is FUNCTION_SCOPE. Wouldn't false as a
> > > default value suffice here anyways?
> > 
> > Eh, the scope creation tangle is hard to follow and I'm not if there is a
path
> > that goes through here without going through one of the other places that
> > propagates the bit. :(
> 
> Acknowledged.

LGTM, too. Land it.

[titzer, 2014-09-19 12:51:05]

Committed patchset #3 manually as 24083 (presubmit successful).