[turbofan] Unify various bailout hacks for super call.

[turbofan] Unify various bailout hacks for super call.

This removes various boilouts for super constructor calls from the
TurboFan pipeline and unifies them. It also disables and optimization
which breaks references to uninitialized const this variables.

R=bmeurer@chromium.org

Committed: https://crrev.com/cc149578cf1ee5c59fbc8cb0ff4b1455b11717ba
Cr-Commit-Position: refs/heads/master@{#29516}

[Benedikt Meurer, 2015-07-07 10:41:40]

lgtm

[Michael Starzinger, 2015-07-07 13:24:15]

The CQ bit was checked by mstarzinger@chromium.org

[Michael Starzinger, 2015-07-07 13:24:16]

The patchset sent to the CQ was uploaded after l-g-t-m from bmeurer@chromium.org
Link to the patchset: https://codereview.chromium.org/1222843004/#ps20001
(title: "Fix throwing of reference error.")

[commit-bot: I haz the power, 2015-07-07 13:24:22]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1222843004/20001

[commit-bot: I haz the power, 2015-07-07 13:25:58]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2015-07-07 13:33:50]

Patchset 2 (id:??) landed as
https://crrev.com/cc149578cf1ee5c59fbc8cb0ff4b1455b11717ba
Cr-Commit-Position: refs/heads/master@{#29516}

[arv (Not doing code reviews), 2015-07-07 15:48:08]

arv@chromium.org changed reviewers:
+ arv@chromium.org

[arv (Not doing code reviews), 2015-07-07 15:48:09]

https://codereview.chromium.org/1222843004/diff/20001/src/compiler/ast-graph-...
File src/compiler/ast-graph-builder.cc (right):

https://codereview.chromium.org/1222843004/diff/20001/src/compiler/ast-graph-...
src/compiler/ast-graph-builder.cc:2839: Node* value =
BuildThrowUnsupportedSuperError(expr->id());
Shouldn't this be UNREACHABLE? We are supposed to handle all of these at a
higher level.

https://codereview.chromium.org/1222843004/diff/20001/src/compiler/ast-graph-...
src/compiler/ast-graph-builder.cc:2845: Node* value =
BuildThrowUnsupportedSuperError(expr->id());
Same here. I don't see how these errors can ever be observed?

[Michael Starzinger, 2015-07-07 15:59:58]

https://codereview.chromium.org/1222843004/diff/20001/src/compiler/ast-graph-...
File src/compiler/ast-graph-builder.cc (right):

https://codereview.chromium.org/1222843004/diff/20001/src/compiler/ast-graph-...
src/compiler/ast-graph-builder.cc:2839: Node* value =
BuildThrowUnsupportedSuperError(expr->id());
On 2015/07/07 15:48:09, arv wrote:
> Shouldn't this be UNREACHABLE? We are supposed to handle all of these at a
> higher level.

Yeah, I though so too. Please admire patch set #1 and the associated failures of
the try job.

https://codereview.chromium.org/1222843004/diff/20001/src/compiler/ast-graph-...
src/compiler/ast-graph-builder.cc:2845: Node* value =
BuildThrowUnsupportedSuperError(expr->id());
On 2015/07/07 15:48:09, arv wrote:
> Same here. I don't see how these errors can ever be observed?

Likewise.

[arv (Not doing code reviews), 2015-07-07 16:04:43]

LGTM

https://codereview.chromium.org/1222843004/diff/20001/src/compiler/ast-graph-...
File src/compiler/ast-graph-builder.cc (right):

https://codereview.chromium.org/1222843004/diff/20001/src/compiler/ast-graph-...
src/compiler/ast-graph-builder.cc:2839: Node* value =
BuildThrowUnsupportedSuperError(expr->id());
On 2015/07/07 15:59:58, Michael Starzinger wrote:
> On 2015/07/07 15:48:09, arv wrote:
> > Shouldn't this be UNREACHABLE? We are supposed to handle all of these at a
> > higher level.
> 
> Yeah, I though so too. Please admire patch set #1 and the associated failures
of
> the try job.

I see.

This one comes from

  delete (super.prop)

and it looks like we do need to keek the throw here.

https://codereview.chromium.org/1222843004/diff/20001/src/compiler/ast-graph-...
src/compiler/ast-graph-builder.cc:2845: Node* value =
BuildThrowUnsupportedSuperError(expr->id());
On 2015/07/07 15:59:58, Michael Starzinger wrote:
> On 2015/07/07 15:48:09, arv wrote:
> > Same here. I don't see how these errors can ever be observed?
> 
> Likewise.

This one on the other hand I don't think we can get to.

I'll make the change locally and see what might break.