[turbofan] Fixes to validator

[turbofan] Fixes to validator

If we have 2 phis with the exact same operand list, and the first phi is
used before the second one, via the operand incoming to the block
that defines the phi, and the second one's operand is defined (via a
parallel move) after the use, then the original operand will be assigned
to the first phi. This will lead to a spurious validation error.

To fix this, we look at the original pending assessment.

Committed: https://crrev.com/7bb766978ba887028c83e37a13e55388c9f53354
Cr-Commit-Position: refs/heads/master@{#35601}

[Mircea Trofin, 2016-04-19 01:25:34]

Description was changed from

==========
fixups to validator

BUG=
==========

to

==========
If we have 2 phis with the exact same operand list, and the first phi is
used before the second one, via the operand incoming to the block
that defines the phi, and the second one's operand is defined (via a 
parallel move) after the use, then the original operand will be assigned 
to the first phi. This will lead to a spurious validation error.

To fix this, we look at the original pending assessment.

BUG=
==========

[Mircea Trofin, 2016-04-19 01:25:34]

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

[Benedikt Meurer, 2016-04-19 04:17:15]

Description was changed from

==========
If we have 2 phis with the exact same operand list, and the first phi is
used before the second one, via the operand incoming to the block
that defines the phi, and the second one's operand is defined (via a 
parallel move) after the use, then the original operand will be assigned 
to the first phi. This will lead to a spurious validation error.

To fix this, we look at the original pending assessment.

BUG=
==========

to

==========
[turbofan] Fixes to validator

If we have 2 phis with the exact same operand list, and the first phi is
used before the second one, via the operand incoming to the block
that defines the phi, and the second one's operand is defined (via a 
parallel move) after the use, then the original operand will be assigned 
to the first phi. This will lead to a spurious validation error.

To fix this, we look at the original pending assessment.
==========

[Benedikt Meurer, 2016-04-19 04:17:35]

lgtm

[Mircea Trofin, 2016-04-19 04:22:58]

The CQ bit was checked by mtrofin@chromium.org

[commit-bot: I haz the power, 2016-04-19 04:23:08]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1895013003/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1895013003/20001

[commit-bot: I haz the power, 2016-04-19 04:25:08]

Description was changed from

==========
[turbofan] Fixes to validator

If we have 2 phis with the exact same operand list, and the first phi is
used before the second one, via the operand incoming to the block
that defines the phi, and the second one's operand is defined (via a 
parallel move) after the use, then the original operand will be assigned 
to the first phi. This will lead to a spurious validation error.

To fix this, we look at the original pending assessment.
==========

to

==========
[turbofan] Fixes to validator

If we have 2 phis with the exact same operand list, and the first phi is
used before the second one, via the operand incoming to the block
that defines the phi, and the second one's operand is defined (via a 
parallel move) after the use, then the original operand will be assigned 
to the first phi. This will lead to a spurious validation error.

To fix this, we look at the original pending assessment.
==========

[commit-bot: I haz the power, 2016-04-19 04:25:08]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2016-04-19 04:26:44]

Description was changed from

==========
[turbofan] Fixes to validator

If we have 2 phis with the exact same operand list, and the first phi is
used before the second one, via the operand incoming to the block
that defines the phi, and the second one's operand is defined (via a 
parallel move) after the use, then the original operand will be assigned 
to the first phi. This will lead to a spurious validation error.

To fix this, we look at the original pending assessment.
==========

to

==========
[turbofan] Fixes to validator

If we have 2 phis with the exact same operand list, and the first phi is
used before the second one, via the operand incoming to the block
that defines the phi, and the second one's operand is defined (via a
parallel move) after the use, then the original operand will be assigned
to the first phi. This will lead to a spurious validation error.

To fix this, we look at the original pending assessment.

Committed: https://crrev.com/7bb766978ba887028c83e37a13e55388c9f53354
Cr-Commit-Position: refs/heads/master@{#35601}
==========

[commit-bot: I haz the power, 2016-04-19 04:26:45]

Patchset 2 (id:??) landed as
https://crrev.com/7bb766978ba887028c83e37a13e55388c9f53354
Cr-Commit-Position: refs/heads/master@{#35601}