Remove separate construct stub for new.target users.

Remove separate construct stub for new.target users.

R=bmeurer@chromium.org

Committed: https://crrev.com/e50c861b099b3bd3e1174b5f2843567620cc6842
Cr-Commit-Position: refs/heads/master@{#29562}

[Michael Starzinger, 2015-07-08 14:41:50]

mstarzinger@chromium.org changed reviewers:
+ rossberg@chromium.org

[Michael Starzinger, 2015-07-08 14:42:24]

mstarzinger@chromium.org changed reviewers:
+ bmeurer@chromium.org
- rossberg@chromium.org

[Michael Starzinger, 2015-07-08 14:42:39]

https://codereview.chromium.org/1213623020/diff/1/src/ia32/builtins-ia32.cc
File src/ia32/builtins-ia32.cc (right):

https://codereview.chromium.org/1213623020/diff/1/src/ia32/builtins-ia32.cc#n...
src/ia32/builtins-ia32.cc:108: int offset = kPointerSize;
There actually seems to be a off-by-one here in the original implementation. I
wonder how this ever worked.

[arv (Not doing code reviews), 2015-07-08 15:14:28]

https://codereview.chromium.org/1213623020/diff/1/src/deoptimizer.cc
File src/deoptimizer.cc (right):

https://codereview.chromium.org/1213623020/diff/1/src/deoptimizer.cc#newcode1235
src/deoptimizer.cc:1235: value =
reinterpret_cast<intptr_t>(isolate_->heap()->undefined_value());
Maybe I'm missing something here. Why is this set to undefined. Shouldn't we use
the original constructor?

[Michael Starzinger, 2015-07-08 15:20:36]

https://codereview.chromium.org/1213623020/diff/1/src/deoptimizer.cc
File src/deoptimizer.cc (right):

https://codereview.chromium.org/1213623020/diff/1/src/deoptimizer.cc#newcode1235
src/deoptimizer.cc:1235: value =
reinterpret_cast<intptr_t>(isolate_->heap()->undefined_value());
On 2015/07/08 15:14:28, arv wrote:
> Maybe I'm missing something here. Why is this set to undefined. Shouldn't we
use
> the original constructor?

We do not yet support inlining of constructor calls where the callee uses the
new.target variable, so nobody will look at this field. Also, even if we would
support such inlining, this field would only be looked at if the deoptimization
actually happens in the prologue of the callee, before the new.target variable
is assigned.

But I am happy to leave a TODO behind in that regard if you feel we need one.

[arv (Not doing code reviews), 2015-07-08 15:31:02]

LGTM

https://codereview.chromium.org/1213623020/diff/1/src/deoptimizer.cc
File src/deoptimizer.cc (right):

https://codereview.chromium.org/1213623020/diff/1/src/deoptimizer.cc#newcode1235
src/deoptimizer.cc:1235: value =
reinterpret_cast<intptr_t>(isolate_->heap()->undefined_value());
On 2015/07/08 15:20:36, Michael Starzinger wrote:
> On 2015/07/08 15:14:28, arv wrote:
> > Maybe I'm missing something here. Why is this set to undefined. Shouldn't we
> use
> > the original constructor?
> 
> We do not yet support inlining of constructor calls where the callee uses the
> new.target variable, so nobody will look at this field. Also, even if we would
> support such inlining, this field would only be looked at if the
deoptimization
> actually happens in the prologue of the callee, before the new.target variable
> is assigned.
> 
> But I am happy to leave a TODO behind in that regard if you feel we need one.

No need.

[arv (Not doing code reviews), 2015-07-08 15:31:35]

Not LGTM... other ports needed of course.

[Michael Starzinger, 2015-07-08 15:32:57]

Just finished with the ports (except PPC and x87, /me is lazy).

[arv (Not doing code reviews), 2015-07-08 15:43:25]

LGTM

So much nicer.

We need to keep a close eye on the perf tests to make sure that this
simplification has no impact.

[Benedikt Meurer, 2015-07-08 19:13:42]

LGTM. I love it!

[Michael Starzinger, 2015-07-10 08:24:58]

The CQ bit was checked by mstarzinger@chromium.org

[commit-bot: I haz the power, 2015-07-10 08:25:07]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1213623020/20001

[commit-bot: I haz the power, 2015-07-10 08:50:29]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2015-07-10 08:50:40]

Patchset 2 (id:??) landed as
https://crrev.com/e50c861b099b3bd3e1174b5f2843567620cc6842
Cr-Commit-Position: refs/heads/master@{#29562}