[turbofan] In effect control linearizer, only delay effect phi update for loops.

[turbofan] In effect control linearizer, only delay effect phi update for loops.

Delaying for merges caused branch cloning using dummy effect phi inputs,
potentially splitting the effect chain at start.

We still have to delay the creation for loops because we need to break
cycles.

BUG=chromium:628403
Committed: https://crrev.com/514951f651bce8c245f60f8df16a02905401e4c8
Cr-Commit-Position: refs/heads/master@{#37808}

[Jarin, 2016-07-16 07:28:24]

The CQ bit was checked by jarin@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2016-07-16 07:28:32]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[Jarin, 2016-07-16 07:49:38]

Description was changed from

==========
[turbofan] In effect control linearizer, only delay effect phi update for loops.

Delaying for merges caused branch cloning using dummy effect phi inputs,
splitting the effect chain at start.

We still have to delay the creation for loops because we need to break
cycles.

BUG=chromium:628403
==========

to

==========
[turbofan] In effect control linearizer, only delay effect phi update for loops.

Delaying for merges caused branch cloning using dummy effect phi inputs,
potentially splitting the effect chain at start.

We still have to delay the creation for loops because we need to break
cycles.

BUG=chromium:628403
==========

[Jarin, 2016-07-16 07:50:46]

The CQ bit was checked by jarin@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2016-07-16 07:50:50]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[Jarin, 2016-07-16 07:51:32]

jarin@chromium.org changed reviewers:
+ bmeurer@chromium.org

[Jarin, 2016-07-16 07:51:33]

Could you take a look, please?

[Jarin, 2016-07-16 07:53:28]

The CQ bit was checked by jarin@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2016-07-16 07:53:30]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-07-16 08:16:51]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-07-16 08:16:52]

Dry run: This issue passed the CQ dry run.

[Benedikt Meurer, 2016-07-16 12:35:30]

Awesome, thanks for fixing (kudos to Jakob for the simple repro).

LGTM, gonna land this now.

https://codereview.chromium.org/2159603002/diff/40001/src/compiler/effect-con...
File src/compiler/effect-control-linearizer.cc (right):

https://codereview.chromium.org/2159603002/diff/40001/src/compiler/effect-con...
src/compiler/effect-control-linearizer.cc:381:
inputs_buffer.resize(block->PredecessorCount(), jsgraph()->Dead());
Nice!

[Benedikt Meurer, 2016-07-16 12:35:37]

The CQ bit was checked by bmeurer@chromium.org

[commit-bot: I haz the power, 2016-07-16 12:35:48]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-07-16 12:37:52]

Description was changed from

==========
[turbofan] In effect control linearizer, only delay effect phi update for loops.

Delaying for merges caused branch cloning using dummy effect phi inputs,
potentially splitting the effect chain at start.

We still have to delay the creation for loops because we need to break
cycles.

BUG=chromium:628403
==========

to

==========
[turbofan] In effect control linearizer, only delay effect phi update for loops.

Delaying for merges caused branch cloning using dummy effect phi inputs,
potentially splitting the effect chain at start.

We still have to delay the creation for loops because we need to break
cycles.

BUG=chromium:628403
==========

[commit-bot: I haz the power, 2016-07-16 12:37:53]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2016-07-16 12:39:22]

Description was changed from

==========
[turbofan] In effect control linearizer, only delay effect phi update for loops.

Delaying for merges caused branch cloning using dummy effect phi inputs,
potentially splitting the effect chain at start.

We still have to delay the creation for loops because we need to break
cycles.

BUG=chromium:628403
==========

to

==========
[turbofan] In effect control linearizer, only delay effect phi update for loops.

Delaying for merges caused branch cloning using dummy effect phi inputs,
potentially splitting the effect chain at start.

We still have to delay the creation for loops because we need to break
cycles.

BUG=chromium:628403

Committed: https://crrev.com/514951f651bce8c245f60f8df16a02905401e4c8
Cr-Commit-Position: refs/heads/master@{#37808}
==========

[commit-bot: I haz the power, 2016-07-16 12:39:23]

Patchset 3 (id:??) landed as
https://crrev.com/514951f651bce8c245f60f8df16a02905401e4c8
Cr-Commit-Position: refs/heads/master@{#37808}