[type-profile] Handle returns correctly.

[type-profile] Handle returns correctly.

Handle fall-off returns and returns inside try blocks.

Store the type-profile feedback slot on the FunctionLiteral
rather than on every return statement.

Next steps: 
* Store entries in nexus that can be identified as 'return' (rather than parameter or assignment)
* Collect types for parameters and assignments
* Distinguish multiple parameters and assignments correctly

R=mstarzinger@chromium.org
BUG=v8:5935
Review-Url: https://codereview.chromium.org/2764113002
Cr-Commit-Position: refs/heads/master@{#44014}
Committed: https://chromium.googlesource.com/v8/v8/+/86c2db5e33ccbff8acfa6c076a86b64c0b291675

[Franzi, 2017-03-21 18:14:41]

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

[commit-bot: I haz the power, 2017-03-21 18:14:44]

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

[Franzi, 2017-03-21 18:22:49]

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

[commit-bot: I haz the power, 2017-03-21 18:22:55]

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

[Franzi, 2017-03-21 18:24:02]

Description was changed from

==========
[type-profile] Handle returns correctly.

Handle fall-off returns and returns inside try blocks.

Store the type-profile feedback slot on the FunctionLiteral
rather than on every return statement.

R=mstarzinger@chromium.org
BUG=v8:5935
==========

to

==========
[type-profile] Handle returns correctly.

Handle fall-off returns and returns inside try blocks.

Store the type-profile feedback slot on the FunctionLiteral
rather than on every return statement.

R=mstarzinger@chromium.org
BUG=v8:5935
==========

[Franzi, 2017-03-21 18:24:03]

franzih@chromium.org changed reviewers:
+ yangguo@chromium.org

[Franzi, 2017-03-21 18:28:06]

Description was changed from

==========
[type-profile] Handle returns correctly.

Handle fall-off returns and returns inside try blocks.

Store the type-profile feedback slot on the FunctionLiteral
rather than on every return statement.

R=mstarzinger@chromium.org
BUG=v8:5935
==========

to

==========
[type-profile] Handle returns correctly.

Handle fall-off returns and returns inside try blocks.

Store the type-profile feedback slot on the FunctionLiteral
rather than on every return statement.

Next steps: 
* Store entries in nexus that can be identified as 'return' (rather than
parameter or assignment)
* Collect types for parameters and assignments
* Distinguish multiple parameters and assignments correctly

R=mstarzinger@chromium.org
BUG=v8:5935
==========

[commit-bot: I haz the power, 2017-03-21 19:42:13]

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

[commit-bot: I haz the power, 2017-03-21 19:42:14]

Dry run: This issue passed the CQ dry run.

[Franzi, 2017-03-21 19:51:09]

Ptal. Thanks.

[Michael Starzinger, 2017-03-22 09:34:10]

LGTM. I like it. Thanks!

[Yang, 2017-03-22 09:43:06]

LGTM.

https://codereview.chromium.org/2764113002/diff/20001/src/runtime/runtime-obj...
File src/runtime/runtime-object.cc (right):

https://codereview.chromium.org/2764113002/diff/20001/src/runtime/runtime-obj...
src/runtime/runtime-object.cc:700: CONVERT_SMI_ARG_CHECKED(index, 3);
Do I understand correctly that this index is actually constant now, since we
only have a single type profile slot per feedback vector?

[Franzi, 2017-03-22 09:48:39]

On 2017/03/22 09:43:06, Yang wrote:
> LGTM.
> 
>
https://codereview.chromium.org/2764113002/diff/20001/src/runtime/runtime-obj...
> File src/runtime/runtime-object.cc (right):
> 
>
https://codereview.chromium.org/2764113002/diff/20001/src/runtime/runtime-obj...
> src/runtime/runtime-object.cc:700: CONVERT_SMI_ARG_CHECKED(index, 3);
> Do I understand correctly that this index is actually constant now, since we
> only have a single type profile slot per feedback vector?

Only one slot per function. But not the same index for all functions. I don't
know if we're guaranteed that the index is constant. Doesn't it change when the
function is optimized?

[Yang, 2017-03-22 09:50:56]

On 2017/03/22 09:48:39, Franzi wrote:
> On 2017/03/22 09:43:06, Yang wrote:
> > LGTM.
> > 
> >
>
https://codereview.chromium.org/2764113002/diff/20001/src/runtime/runtime-obj...
> > File src/runtime/runtime-object.cc (right):
> > 
> >
>
https://codereview.chromium.org/2764113002/diff/20001/src/runtime/runtime-obj...
> > src/runtime/runtime-object.cc:700: CONVERT_SMI_ARG_CHECKED(index, 3);
> > Do I understand correctly that this index is actually constant now, since we
> > only have a single type profile slot per feedback vector?
> 
> Only one slot per function. But not the same index for all functions. I don't
> know if we're guaranteed that the index is constant. Doesn't it change when
the
> function is optimized?

I don't think the feedback vector changes once it has been created once, because
it has to match the feedback metadata.

We probably could make sure the index is constant if we reserve it early enough
during renumber. We could then also find it without having to iterate through
the vector?

[Franzi, 2017-03-22 10:10:09]

On 2017/03/22 09:50:56, Yang wrote:
> On 2017/03/22 09:48:39, Franzi wrote:
> > On 2017/03/22 09:43:06, Yang wrote:
> > > LGTM.
> > > 
> > >
> >
>
https://codereview.chromium.org/2764113002/diff/20001/src/runtime/runtime-obj...
> > > File src/runtime/runtime-object.cc (right):
> > > 
> > >
> >
>
https://codereview.chromium.org/2764113002/diff/20001/src/runtime/runtime-obj...
> > > src/runtime/runtime-object.cc:700: CONVERT_SMI_ARG_CHECKED(index, 3);
> > > Do I understand correctly that this index is actually constant now, since
we
> > > only have a single type profile slot per feedback vector?
> > 
> > Only one slot per function. But not the same index for all functions. I
don't
> > know if we're guaranteed that the index is constant. Doesn't it change when
> the
> > function is optimized?
> 
> I don't think the feedback vector changes once it has been created once,
because
> it has to match the feedback metadata.
> 
> We probably could make sure the index is constant if we reserve it early
enough
> during renumber. We could then also find it without having to iterate through
> the vector?

OK, I'll look into that in a separate CL.

[Franzi, 2017-03-22 10:19:54]

The CQ bit was checked by franzih@chromium.org

[commit-bot: I haz the power, 2017-03-22 10:20:05]

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, 2017-03-22 10:23:17]

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

[commit-bot: I haz the power, 2017-03-22 10:23:18]

Try jobs failed on following builders:
  v8_presubmit on master.tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_presubmit/builds/37303)

[Franzi, 2017-03-22 10:38:30]

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

[commit-bot: I haz the power, 2017-03-22 10:38:37]

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

[Franzi, 2017-03-22 10:40:38]

The CQ bit was unchecked by franzih@chromium.org

[Franzi, 2017-03-22 10:40:38]

The CQ bit was checked by franzih@chromium.org

[Franzi, 2017-03-22 10:40:39]

The patchset sent to the CQ was uploaded after l-g-t-m from
mstarzinger@chromium.org, yangguo@chromium.org
Link to the patchset: https://codereview.chromium.org/2764113002/#ps40001
(title: "Rebase.")

[commit-bot: I haz the power, 2017-03-22 10:40:48]

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, 2017-03-22 11:51:08]

CQ is committing da patch.

Bot data: {"patchset_id": 40001, "attempt_start_ts": 1490179238729940,
"parent_rev": "d82aca1928a8633213f73d9e31163e57c23c8e11", "commit_rev":
"86c2db5e33ccbff8acfa6c076a86b64c0b291675"}

[commit-bot: I haz the power, 2017-03-22 11:51:13]

Description was changed from

==========
[type-profile] Handle returns correctly.

Handle fall-off returns and returns inside try blocks.

Store the type-profile feedback slot on the FunctionLiteral
rather than on every return statement.

Next steps: 
* Store entries in nexus that can be identified as 'return' (rather than
parameter or assignment)
* Collect types for parameters and assignments
* Distinguish multiple parameters and assignments correctly

R=mstarzinger@chromium.org
BUG=v8:5935
==========

to

==========
[type-profile] Handle returns correctly.

Handle fall-off returns and returns inside try blocks.

Store the type-profile feedback slot on the FunctionLiteral
rather than on every return statement.

Next steps: 
* Store entries in nexus that can be identified as 'return' (rather than
parameter or assignment)
* Collect types for parameters and assignments
* Distinguish multiple parameters and assignments correctly

R=mstarzinger@chromium.org
BUG=v8:5935

Review-Url: https://codereview.chromium.org/2764113002
Cr-Commit-Position: refs/heads/master@{#44014}
Committed:
https://chromium.googlesource.com/v8/v8/+/86c2db5e33ccbff8acfa6c076a86b64c0b2...
==========

[commit-bot: I haz the power, 2017-03-22 11:51:14]

Committed patchset #3 (id:40001) as
https://chromium.googlesource.com/v8/v8/+/86c2db5e33ccbff8acfa6c076a86b64c0b2...