[for-in] Ensure that we learn from deopts within for-in loop bodies.

[for-in] Ensure that we learn from deopts within for-in loop bodies.

If we deoptimize from TurboFan or Crankshaft into the body of a for-in
loop and that for-in mode then switches to slow mode (i.e. has to call
%ForInFilter), we have to record that feedback, because otherwise we
might actually OSR into that loop assuming that it's fast mode still,
or even worse recompile the function later when we call it again w/o
having rerun the for-in loop in fullcodegen from the beginning (where
was previously the only place we could learn).

R=mstarzinger@chromium.org
BUG=v8:3650
LOG=n

Committed: https://crrev.com/0637f5f64c3bc5d8fc930e65730b7d86cce7a74c
Cr-Commit-Position: refs/heads/master@{#33612}

[Benedikt Meurer, 2016-01-29 11:58:42]

Hey Michi,

Less deopt loops are best deopt loops.
Please take a look.

Thanks,
Benedikt

[Michael Starzinger, 2016-01-29 12:27:46]

LGTM.

[Benedikt Meurer, 2016-01-29 12:28:06]

The CQ bit was checked by bmeurer@chromium.org

[commit-bot: I haz the power, 2016-01-29 12:28:18]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1638303008/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1638303008/1

[commit-bot: I haz the power, 2016-01-29 12:30:05]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2016-01-29 12:30:26]

Description was changed from

==========
[for-in] Ensure that we learn from deopts within for-in loop bodies.

If we deoptimize from TurboFan or Crankshaft into the body of a for-in
loop and that for-in mode then switches to slow mode (i.e. has to call
%ForInFilter), we have to record that feedback, because otherwise we
might actually OSR into that loop assuming that it's fast mode still,
or even worse recompile the function later when we call it again w/o
having rerun the for-in loop in fullcodegen from the beginning (where
was previously the only place we could learn).

R=mstarzinger@chromium.org
BUG=v8:3650
LOG=n
==========

to

==========
[for-in] Ensure that we learn from deopts within for-in loop bodies.

If we deoptimize from TurboFan or Crankshaft into the body of a for-in
loop and that for-in mode then switches to slow mode (i.e. has to call
%ForInFilter), we have to record that feedback, because otherwise we
might actually OSR into that loop assuming that it's fast mode still,
or even worse recompile the function later when we call it again w/o
having rerun the for-in loop in fullcodegen from the beginning (where
was previously the only place we could learn).

R=mstarzinger@chromium.org
BUG=v8:3650
LOG=n

Committed: https://crrev.com/0637f5f64c3bc5d8fc930e65730b7d86cce7a74c
Cr-Commit-Position: refs/heads/master@{#33612}
==========

[commit-bot: I haz the power, 2016-01-29 12:30:27]

Patchset 1 (id:??) landed as
https://crrev.com/0637f5f64c3bc5d8fc930e65730b7d86cce7a74c
Cr-Commit-Position: refs/heads/master@{#33612}