Introduce DisallowDeoptimization scope.

Introduce DisallowDeoptimization scope.

R=yangguo@chromium.org

Committed: https://code.google.com/p/v8/source/detail?r=20998

[Benedikt Meurer, 2014-04-28 06:34:02]

PTAL

[Yang, 2014-04-28 06:38:38]

LGTM.

https://codereview.chromium.org/254763007/diff/20001/src/runtime.cc
File src/runtime.cc (right):

https://codereview.chromium.org/254763007/diff/20001/src/runtime.cc#newcode14587
src/runtime.cc:14587: if (!JSObject::TryMigrateInstance(js_object)) return
Smi::FromInt(0);
Isn't JSObject::TryMigrateInstance not supposed to cause deopts altogether? In
that case, moving the disallow scope inside it would be nicer. If not, ignore
this comment.

[Benedikt Meurer, 2014-04-28 06:44:35]

https://codereview.chromium.org/254763007/diff/20001/src/runtime.cc
File src/runtime.cc (right):

https://codereview.chromium.org/254763007/diff/20001/src/runtime.cc#newcode14587
src/runtime.cc:14587: if (!JSObject::TryMigrateInstance(js_object)) return
Smi::FromInt(0);
On 2014/04/28 06:38:38, Yang wrote:
> Isn't JSObject::TryMigrateInstance not supposed to cause deopts altogether? In
> that case, moving the disallow scope inside it would be nicer. If not, ignore
> this comment.

Done.

[Yang, 2014-04-28 06:45:31]

On 2014/04/28 06:44:35, Benedikt Meurer wrote:
> https://codereview.chromium.org/254763007/diff/20001/src/runtime.cc
> File src/runtime.cc (right):
> 
>
https://codereview.chromium.org/254763007/diff/20001/src/runtime.cc#newcode14587
> src/runtime.cc:14587: if (!JSObject::TryMigrateInstance(js_object)) return
> Smi::FromInt(0);
> On 2014/04/28 06:38:38, Yang wrote:
> > Isn't JSObject::TryMigrateInstance not supposed to cause deopts altogether?
In
> > that case, moving the disallow scope inside it would be nicer. If not,
ignore
> > this comment.
> 
> Done.

lgtm.

[Benedikt Meurer, 2014-04-28 06:47:11]

Committed patchset #2 manually as r20998 (presubmit successful).