Reland of [interpreter] Add string type feedback to add

[interpreter] Add string type feedback to add

Adds string type feedback to Ignition's AddWithFeedback code stub, for now only
adding a special case for when both lhs and rhs are strings. This improves
octane's splay by >100%.

BUG=v8:5400
Committed: https://crrev.com/fb4ae2239d37adaf0321165034050316914de708
Committed: https://crrev.com/bf1a94f1b269914856a8c8763fd282367f066c67
Committed: https://crrev.com/b3c8b0ce2c9af0528837d8309625118d4096553b
Cr-Original-Original-Commit-Position: refs/heads/master@{#39987}
Cr-Original-Commit-Position: refs/heads/master@{#39996}
Cr-Commit-Position: refs/heads/master@{#40015}

[Leszek Swirski, 2016-10-03 13:55:22]

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

[commit-bot: I haz the power, 2016-10-03 13:55:30]

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, 2016-10-03 14:25:30]

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

[commit-bot: I haz the power, 2016-10-03 14:25:30]

Dry run: This issue passed the CQ dry run.

[Leszek Swirski, 2016-10-03 14:30:00]

leszeks@chromium.org changed reviewers:
+ mythria@chromium.org, rmcilroy@chromium.org

[rmcilroy, 2016-10-04 08:05:32]

https://codereview.chromium.org/2392533002/diff/1/src/code-stubs.cc
File src/code-stubs.cc (right):

https://codereview.chromium.org/2392533002/diff/1/src/code-stubs.cc#newcode1100
src/code-stubs.cc:1100: &call_add_stub);
We also need a check here. My suggestion would be to add checks here which check
the unknown side. The type feedback should be String if one side is a string,
and the other is any of a string, number or SMI. So for here you could add:

 assembler->GotoUnless(assembler->IsHeapNumberMap(rhs_map),
                           &check_rhs_string);

 ... 

 assembler->Bind(&check_rhs_string);
 {
   // We know that lhs is a SMI, if rhs is a string type feedback should be
kString, otherwise it should be kAny.
   Node* rhs_instance_type = assembler->LoadMapInstanceType(rhs_map);
   Node* rhs_is_string = assembler->Int32LessThan(rhs_instance_type,
       assembler->Int32Constant(FIRST_NONSTRING_TYPE)),
   var_type_feedback.Bind(
       assembler->Select(rhs_is_string,
           assembler->Int32Constant(BinaryOperationFeedback::kString),
           assembler->Int32Constant(BinaryOperationFeedback::kNumber));
   assembler->Goto(&call_add_stub);

And removing the "var_type_feedback.Bind(..." from call_add_stub.

https://codereview.chromium.org/2392533002/diff/1/src/code-stubs.cc#newcode1137
src/code-stubs.cc:1137: &call_add_stub);
Same comment here as above - here you know that lhs is a heap number, so again
should only have to check the rhs type (maybe you could have a common branch,
although you would need to reload the rhs_map).

https://codereview.chromium.org/2392533002/diff/1/src/code-stubs.cc#newcode1147
src/code-stubs.cc:1147: assembler->GotoIf(assembler->WordIsSmi(rhs),
&call_add_stub);
We should still count this as a string if the lhs is a string even if rhs is a
SMI (since the output will be string). Here we know that the lhs is a non-number
heap object, but we don't know anything about the rhs, so the code should be
something like:

assembler->Bind(&check_string);
 {
   // Check if lhs is a String.
   Node* lhs_instance_type = assembler->LoadMapInstanceType(lhs_map);
   assembler->GotoUnless(assembler->Int32LessThan(rhs_instance_type,
       assembler->Int32Constant(FIRST_NONSTRING_TYPE), &has_any_type);

   // Check if rhs is a Smi, HeapNumber or String.
   assembler->GotoIf(assembler->WordIsSmi(rhs), &has_string_type);
   Node* rhs_map = assembler->LoadMap(rhs);
   assembler->GotoIf(assembler->IsHeapNumberMap(rhs_map),
       &has_string_type);
   assembler->Branch(assembler->Int32LessThan(
                                rhs_instance_type,
                                assembler->Int32Constant(FIRST_NONSTRING_TYPE)
       &has_string_type,
       &has_any_type);

  assembler->Bind(&has_string_type);
  {
     var_type_feedback.Bind(
        assembler->Int32Constant(BinaryOperationFeedback::kString));
     assembler->Goto(&call_add_stub);
   }
 ...

https://codereview.chromium.org/2392533002/diff/1/src/code-stubs.cc#newcode1166
src/code-stubs.cc:1166: var_result.Bind(assembler->CallStub(callable, context,
lhs, rhs));
Let's keep the calling of the stub and updating of the type feedback in one
place (see comments above).

https://codereview.chromium.org/2392533002/diff/1/src/globals.h
File src/globals.h (right):

https://codereview.chromium.org/2392533002/diff/1/src/globals.h#newcode1223
src/globals.h:1223: kString = 0x4,
This needs to keep the or-combining from kNumber->kString->kAny, so kString
should be 0x7 and kAny should be 0xF (update comment above too).

[Leszek Swirski, 2016-10-04 08:30:40]

Replies without any code changes yet, the enum one needs clarification.

https://codereview.chromium.org/2392533002/diff/1/src/code-stubs.cc
File src/code-stubs.cc (right):

https://codereview.chromium.org/2392533002/diff/1/src/code-stubs.cc#newcode1100
src/code-stubs.cc:1100: &call_add_stub);
On 2016/10/04 08:05:31, rmcilroy wrote:
> We also need a check here. My suggestion would be to add checks here which
check
> the unknown side. The type feedback should be String if one side is a string,
> and the other is any of a string, number or SMI. So for here you could add:
> 
>  assembler->GotoUnless(assembler->IsHeapNumberMap(rhs_map),
>                            &check_rhs_string);
> 
>  ... 
> 
>  assembler->Bind(&check_rhs_string);
>  {
>    // We know that lhs is a SMI, if rhs is a string type feedback should be
> kString, otherwise it should be kAny.
>    Node* rhs_instance_type = assembler->LoadMapInstanceType(rhs_map);
>    Node* rhs_is_string = assembler->Int32LessThan(rhs_instance_type,
>        assembler->Int32Constant(FIRST_NONSTRING_TYPE)),
>    var_type_feedback.Bind(
>        assembler->Select(rhs_is_string,
>            assembler->Int32Constant(BinaryOperationFeedback::kString),
>            assembler->Int32Constant(BinaryOperationFeedback::kNumber));
>    assembler->Goto(&call_add_stub);
> 
> And removing the "var_type_feedback.Bind(..." from call_add_stub.

Ah, this patch was only adding string + string = string feedback, I can add
string + number = string feedback too if you want.

https://codereview.chromium.org/2392533002/diff/1/src/code-stubs.cc#newcode1137
src/code-stubs.cc:1137: &call_add_stub);
On 2016/10/04 08:05:31, rmcilroy wrote:
> Same comment here as above - here you know that lhs is a heap number, so again
> should only have to check the rhs type (maybe you could have a common branch,
> although you would need to reload the rhs_map).

As above.

https://codereview.chromium.org/2392533002/diff/1/src/code-stubs.cc#newcode1147
src/code-stubs.cc:1147: assembler->GotoIf(assembler->WordIsSmi(rhs),
&call_add_stub);
On 2016/10/04 08:05:31, rmcilroy wrote:
> We should still count this as a string if the lhs is a string even if rhs is a
> SMI (since the output will be string). Here we know that the lhs is a
non-number
> heap object, but we don't know anything about the rhs, so the code should be
> something like:
> 
> assembler->Bind(&check_string);
>  {
>    // Check if lhs is a String.
>    Node* lhs_instance_type = assembler->LoadMapInstanceType(lhs_map);
>    assembler->GotoUnless(assembler->Int32LessThan(rhs_instance_type,
>        assembler->Int32Constant(FIRST_NONSTRING_TYPE), &has_any_type);
> 
>    // Check if rhs is a Smi, HeapNumber or String.
>    assembler->GotoIf(assembler->WordIsSmi(rhs), &has_string_type);
>    Node* rhs_map = assembler->LoadMap(rhs);
>    assembler->GotoIf(assembler->IsHeapNumberMap(rhs_map),
>        &has_string_type);
>    assembler->Branch(assembler->Int32LessThan(
>                                 rhs_instance_type,
>                                 assembler->Int32Constant(FIRST_NONSTRING_TYPE)
>        &has_string_type,
>        &has_any_type);
> 
>   assembler->Bind(&has_string_type);
>   {
>      var_type_feedback.Bind(
>         assembler->Int32Constant(BinaryOperationFeedback::kString));
>      assembler->Goto(&call_add_stub);
>    }
>  ...

As above.

https://codereview.chromium.org/2392533002/diff/1/src/code-stubs.cc#newcode1166
src/code-stubs.cc:1166: var_result.Bind(assembler->CallStub(callable, context,
lhs, rhs));
On 2016/10/04 08:05:31, rmcilroy wrote:
> Let's keep the calling of the stub and updating of the type feedback in one
> place (see comments above).

Not sure about this, calling the stub is a small amount of code and I think it
might be better to inline it to skip a jump (no pun intended).

https://codereview.chromium.org/2392533002/diff/1/src/globals.h
File src/globals.h (right):

https://codereview.chromium.org/2392533002/diff/1/src/globals.h#newcode1223
src/globals.h:1223: kString = 0x4,
On 2016/10/04 08:05:32, rmcilroy wrote:
> This needs to keep the or-combining from kNumber->kString->kAny, so kString
> should be 0x7 and kAny should be 0xF (update comment above too).

Should it? or-combining kNumber and kString shouldn't output kString, I wouldn't
have thought?

[Leszek Swirski, 2016-10-04 08:50:00]

https://codereview.chromium.org/2392533002/diff/1/src/globals.h
File src/globals.h (right):

https://codereview.chromium.org/2392533002/diff/1/src/globals.h#newcode1223
src/globals.h:1223: kString = 0x4,
On 2016/10/04 08:30:40, Leszek Swirski wrote:
> On 2016/10/04 08:05:32, rmcilroy wrote:
> > This needs to keep the or-combining from kNumber->kString->kAny, so kString
> > should be 0x7 and kAny should be 0xF (update comment above too).
> 
> Should it? or-combining kNumber and kString shouldn't output kString, I
wouldn't
> have thought?

Never mind, Mythri explained f2f that number is a subtype of string, which
doesn't make sense in my heart but at least now makes sense in my brain.

[Leszek Swirski, 2016-10-04 10:16:58]

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

[commit-bot: I haz the power, 2016-10-04 10:17:06]

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

[Leszek Swirski, 2016-10-04 10:21:21]

PTAL.

After some more f2f discussion, it seems that I was right to follow my heart --
numbers don't seem to be treated by FCG or turbofan as a subtype of strings, and
so we can only give string binop feedback for string + string operations, and
kNumber | kString != kString.

FWIW, Benedikt claims that there is evidence that the string + number type
feedback in Crankshaft has little performance benefit.

[commit-bot: I haz the power, 2016-10-04 10:44:54]

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

[commit-bot: I haz the power, 2016-10-04 10:44:55]

Dry run: This issue passed the CQ dry run.

[mythria, 2016-10-05 08:28:00]

Thanks, lgtm.

https://codereview.chromium.org/2392533002/diff/20001/src/code-stubs.cc
File src/code-stubs.cc (right):

https://codereview.chromium.org/2392533002/diff/20001/src/code-stubs.cc#newco...
src/code-stubs.cc:1165: assembler->Goto(&call_add_stub);
I am not sure if it makes sense, but would it be possible to directly have code
to add two strings rather than going to calling the add stub. If the operation
is too complex it may increase the code size, but if it is simple enough we may
avoid some checks.

[rmcilroy, 2016-10-05 11:03:13]

LGTM thanks.

https://codereview.chromium.org/2392533002/diff/20001/src/code-stubs.cc
File src/code-stubs.cc (right):

https://codereview.chromium.org/2392533002/diff/20001/src/code-stubs.cc#newco...
src/code-stubs.cc:1153: &do_add_any);
Could you use IsStringInstanceType (as added in
https://codereview.chromium.org/2395453002/).

https://codereview.chromium.org/2392533002/diff/20001/src/code-stubs.cc#newco...
src/code-stubs.cc:1161: &do_add_any);
ditto

[Leszek Swirski, 2016-10-05 11:22:53]

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

[commit-bot: I haz the power, 2016-10-05 11:23:02]

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

[Leszek Swirski, 2016-10-05 11:23:02]

The CQ bit was unchecked by leszeks@chromium.org

[Leszek Swirski, 2016-10-05 11:23:02]

The CQ bit was checked by leszeks@chromium.org

[Leszek Swirski, 2016-10-05 11:23:03]

The patchset sent to the CQ was uploaded after l-g-t-m from
rmcilroy@chromium.org, mythria@chromium.org
Link to the patchset: https://codereview.chromium.org/2392533002/#ps40001
(title: "Use the fancy new IsStringInstanceType method")

[commit-bot: I haz the power, 2016-10-05 11:31:52]

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, 2016-10-05 11:52:32]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2016-10-05 11:52:50]

Description was changed from

==========
[interpreter] Add string type feedback to add

Adds string type feedback to Ignition's AddWithFeedback code stub, for now only
adding a special case for when both lhs and rhs are strings. This improves
octane's splay by >100%.

BUG=v8:5400
==========

to

==========
[interpreter] Add string type feedback to add

Adds string type feedback to Ignition's AddWithFeedback code stub, for now only
adding a special case for when both lhs and rhs are strings. This improves
octane's splay by >100%.

BUG=v8:5400

Committed: https://crrev.com/fb4ae2239d37adaf0321165034050316914de708
Cr-Commit-Position: refs/heads/master@{#39987}
==========

[commit-bot: I haz the power, 2016-10-05 11:52:51]

Patchset 3 (id:??) landed as
https://crrev.com/fb4ae2239d37adaf0321165034050316914de708
Cr-Commit-Position: refs/heads/master@{#39987}

[Michael Achenbach, 2016-10-05 13:04:22]

A revert of this CL (patchset #3 id:40001) has been created in
https://codereview.chromium.org/2395743004/ by machenbach@chromium.org.

The reason for reverting is: Fails unittests on win32 debug:
https://build.chromium.org/p/client.v8/builders/V8%20Win32%20-%20debug/builds....

[Michael Achenbach, 2016-10-05 13:05:21]

FYI: win dbg bots don't run by default on the CQ, but can be added manually on a
reland of this CL

[Leszek Swirski, 2016-10-05 13:10:55]

https://codereview.chromium.org/2392533002/diff/20001/src/code-stubs.cc
File src/code-stubs.cc (right):

https://codereview.chromium.org/2392533002/diff/20001/src/code-stubs.cc#newco...
src/code-stubs.cc:1153: &do_add_any);
On 2016/10/05 11:03:12, rmcilroy wrote:
> Could you use IsStringInstanceType (as added in
> https://codereview.chromium.org/2395453002/).

Done.

https://codereview.chromium.org/2392533002/diff/20001/src/code-stubs.cc#newco...
src/code-stubs.cc:1161: &do_add_any);
On 2016/10/05 11:03:12, rmcilroy wrote:
> ditto

Done.

https://codereview.chromium.org/2392533002/diff/20001/src/code-stubs.cc#newco...
src/code-stubs.cc:1165: assembler->Goto(&call_add_stub);
On 2016/10/05 08:28:00, mythria wrote:
> I am not sure if it makes sense, but would it be possible to directly have
code
> to add two strings rather than going to calling the add stub. If the operation
> is too complex it may increase the code size, but if it is simple enough we
may
> avoid some checks. 

Nice idea, I'll punt on this for this CL and will investigate for a subsequent
one.

[Leszek Swirski, 2016-10-05 13:11:26]

Description was changed from

==========
[interpreter] Add string type feedback to add

Adds string type feedback to Ignition's AddWithFeedback code stub, for now only
adding a special case for when both lhs and rhs are strings. This improves
octane's splay by >100%.

BUG=v8:5400

Committed: https://crrev.com/fb4ae2239d37adaf0321165034050316914de708
Cr-Commit-Position: refs/heads/master@{#39987}
==========

to

==========
[interpreter] Add string type feedback to add

Adds string type feedback to Ignition's AddWithFeedback code stub, for now only
adding a special case for when both lhs and rhs are strings. This improves
octane's splay by >100%.

BUG=v8:5400

Committed: https://crrev.com/fb4ae2239d37adaf0321165034050316914de708
Cr-Commit-Position: refs/heads/master@{#39987}
==========

[Leszek Swirski, 2016-10-05 13:12:06]

Description was changed from

==========
[interpreter] Add string type feedback to add

Adds string type feedback to Ignition's AddWithFeedback code stub, for now only
adding a special case for when both lhs and rhs are strings. This improves
octane's splay by >100%.

BUG=v8:5400

Committed: https://crrev.com/fb4ae2239d37adaf0321165034050316914de708
Cr-Commit-Position: refs/heads/master@{#39987}
==========

to

==========
Reland: [interpreter] Add string type feedback to add

Adds string type feedback to Ignition's AddWithFeedback code stub, for now only
adding a special case for when both lhs and rhs are strings. This improves
octane's splay by >100%.

BUG=v8:5400

Committed: https://crrev.com/fb4ae2239d37adaf0321165034050316914de708
Cr-Commit-Position: refs/heads/master@{#39987}
==========

[Michael Achenbach, 2016-10-05 13:21:12]

Strangely
https://build.chromium.org/p/client.v8/builders/V8%20Win32%20-%20debug/builds...
was green before the revert. Maybe it's flaky? Or wrong infra?

[Leszek Swirski, 2016-10-05 13:23:27]

On 2016/10/05 13:21:12, machenbach (slow) wrote:
> Strangely
>
https://build.chromium.org/p/client.v8/builders/V8%20Win32%20-%20debug/builds...
> was green before the revert. Maybe it's flaky? Or wrong infra?

Maybe flaky, I took a look at the broken tests and it doesn't seem that any
should have been broken by this patch. I'm running the win_dbg try bots again on
this.

[Leszek Swirski, 2016-10-05 13:23:45]

Description was changed from

==========
Reland: [interpreter] Add string type feedback to add

Adds string type feedback to Ignition's AddWithFeedback code stub, for now only
adding a special case for when both lhs and rhs are strings. This improves
octane's splay by >100%.

BUG=v8:5400

Committed: https://crrev.com/fb4ae2239d37adaf0321165034050316914de708
Cr-Commit-Position: refs/heads/master@{#39987}
==========

to

==========
[interpreter] Add string type feedback to add

Adds string type feedback to Ignition's AddWithFeedback code stub, for now only
adding a special case for when both lhs and rhs are strings. This improves
octane's splay by >100%.

BUG=v8:5400

Committed: https://crrev.com/fb4ae2239d37adaf0321165034050316914de708
Cr-Commit-Position: refs/heads/master@{#39987}
==========

[Leszek Swirski, 2016-10-05 14:31:04]

Well, three successes and one failure on v8_win_dbg. Has this bot been flaky
before like this?

[Michael Achenbach, 2016-10-05 14:37:48]

On 2016/10/05 14:31:04, Leszek Swirski wrote:
> Well, three successes and one failure on v8_win_dbg. Has this bot been flaky
> before like this?

Maybe we have a build dependency problem again. I also see this strange problem:
https://build.chromium.org/p/tryserver.v8/builders/v8_win_dbg/builds/409

Looks similar but is way older. Maybe just reland... If we get more problems,
lets land a landmine.

[Leszek Swirski, 2016-10-05 14:47:03]

The CQ bit was checked by leszeks@chromium.org

[commit-bot: I haz the power, 2016-10-05 14:47:07]

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

[Leszek Swirski, 2016-10-05 14:47:26]

sgtm

[commit-bot: I haz the power, 2016-10-05 14:49:08]

Description was changed from

==========
[interpreter] Add string type feedback to add

Adds string type feedback to Ignition's AddWithFeedback code stub, for now only
adding a special case for when both lhs and rhs are strings. This improves
octane's splay by >100%.

BUG=v8:5400

Committed: https://crrev.com/fb4ae2239d37adaf0321165034050316914de708
Cr-Commit-Position: refs/heads/master@{#39987}
==========

to

==========
[interpreter] Add string type feedback to add

Adds string type feedback to Ignition's AddWithFeedback code stub, for now only
adding a special case for when both lhs and rhs are strings. This improves
octane's splay by >100%.

BUG=v8:5400

Committed: https://crrev.com/fb4ae2239d37adaf0321165034050316914de708
Cr-Commit-Position: refs/heads/master@{#39987}
==========

[commit-bot: I haz the power, 2016-10-05 14:49:09]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2016-10-05 14:49:28]

Description was changed from

==========
[interpreter] Add string type feedback to add

Adds string type feedback to Ignition's AddWithFeedback code stub, for now only
adding a special case for when both lhs and rhs are strings. This improves
octane's splay by >100%.

BUG=v8:5400

Committed: https://crrev.com/fb4ae2239d37adaf0321165034050316914de708
Cr-Commit-Position: refs/heads/master@{#39987}
==========

to

==========
[interpreter] Add string type feedback to add

Adds string type feedback to Ignition's AddWithFeedback code stub, for now only
adding a special case for when both lhs and rhs are strings. This improves
octane's splay by >100%.

BUG=v8:5400

Committed: https://crrev.com/fb4ae2239d37adaf0321165034050316914de708
Committed: https://crrev.com/bf1a94f1b269914856a8c8763fd282367f066c67
Cr-Original-Commit-Position: refs/heads/master@{#39987}
Cr-Commit-Position: refs/heads/master@{#39996}
==========

[commit-bot: I haz the power, 2016-10-05 14:49:29]

Patchset 3 (id:??) landed as
https://crrev.com/bf1a94f1b269914856a8c8763fd282367f066c67
Cr-Commit-Position: refs/heads/master@{#39996}

[Leszek Swirski, 2016-10-05 15:26:25]

A revert of this CL (patchset #3 id:40001) has been created in
https://codereview.chromium.org/2393193002/ by leszeks@chromium.org.

The reason for reverting is: Broke the tree again, for no obvious reason :/.

[Michael Achenbach, 2016-10-05 19:44:41]

Description was changed from

==========
[interpreter] Add string type feedback to add

Adds string type feedback to Ignition's AddWithFeedback code stub, for now only
adding a special case for when both lhs and rhs are strings. This improves
octane's splay by >100%.

BUG=v8:5400

Committed: https://crrev.com/fb4ae2239d37adaf0321165034050316914de708
Committed: https://crrev.com/bf1a94f1b269914856a8c8763fd282367f066c67
Cr-Original-Commit-Position: refs/heads/master@{#39987}
Cr-Commit-Position: refs/heads/master@{#39996}
==========

to

==========
[interpreter] Add string type feedback to add

Adds string type feedback to Ignition's AddWithFeedback code stub, for now only
adding a special case for when both lhs and rhs are strings. This improves
octane's splay by >100%.

BUG=v8:5400

Committed: https://crrev.com/fb4ae2239d37adaf0321165034050316914de708
Committed: https://crrev.com/bf1a94f1b269914856a8c8763fd282367f066c67
Cr-Original-Commit-Position: refs/heads/master@{#39987}
Cr-Commit-Position: refs/heads/master@{#39996}
==========

[Michael Achenbach, 2016-10-05 19:45:16]

Landmine is in, and another suspect reverted. Relanding this.

[Michael Achenbach, 2016-10-05 19:45:26]

The CQ bit was checked by machenbach@chromium.org

[commit-bot: I haz the power, 2016-10-05 19:45:34]

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, 2016-10-05 19:48:36]

Description was changed from

==========
[interpreter] Add string type feedback to add

Adds string type feedback to Ignition's AddWithFeedback code stub, for now only
adding a special case for when both lhs and rhs are strings. This improves
octane's splay by >100%.

BUG=v8:5400

Committed: https://crrev.com/fb4ae2239d37adaf0321165034050316914de708
Committed: https://crrev.com/bf1a94f1b269914856a8c8763fd282367f066c67
Cr-Original-Commit-Position: refs/heads/master@{#39987}
Cr-Commit-Position: refs/heads/master@{#39996}
==========

to

==========
[interpreter] Add string type feedback to add

Adds string type feedback to Ignition's AddWithFeedback code stub, for now only
adding a special case for when both lhs and rhs are strings. This improves
octane's splay by >100%.

BUG=v8:5400

Committed: https://crrev.com/fb4ae2239d37adaf0321165034050316914de708
Committed: https://crrev.com/bf1a94f1b269914856a8c8763fd282367f066c67
Cr-Original-Commit-Position: refs/heads/master@{#39987}
Cr-Commit-Position: refs/heads/master@{#39996}
==========

[commit-bot: I haz the power, 2016-10-05 19:48:37]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2016-10-05 19:48:57]

Description was changed from

==========
[interpreter] Add string type feedback to add

Adds string type feedback to Ignition's AddWithFeedback code stub, for now only
adding a special case for when both lhs and rhs are strings. This improves
octane's splay by >100%.

BUG=v8:5400

Committed: https://crrev.com/fb4ae2239d37adaf0321165034050316914de708
Committed: https://crrev.com/bf1a94f1b269914856a8c8763fd282367f066c67
Cr-Original-Commit-Position: refs/heads/master@{#39987}
Cr-Commit-Position: refs/heads/master@{#39996}
==========

to

==========
[interpreter] Add string type feedback to add

Adds string type feedback to Ignition's AddWithFeedback code stub, for now only
adding a special case for when both lhs and rhs are strings. This improves
octane's splay by >100%.

BUG=v8:5400

Committed: https://crrev.com/fb4ae2239d37adaf0321165034050316914de708
Committed: https://crrev.com/bf1a94f1b269914856a8c8763fd282367f066c67
Committed: https://crrev.com/b3c8b0ce2c9af0528837d8309625118d4096553b
Cr-Original-Original-Commit-Position: refs/heads/master@{#39987}
Cr-Original-Commit-Position: refs/heads/master@{#39996}
Cr-Commit-Position: refs/heads/master@{#40015}
==========

[commit-bot: I haz the power, 2016-10-05 19:48:58]

Patchset 3 (id:??) landed as
https://crrev.com/b3c8b0ce2c9af0528837d8309625118d4096553b
Cr-Commit-Position: refs/heads/master@{#40015}