[turbofan] Support variable size argument popping in TF-generated functions

[turbofan] Support variable size argument removal in TF-generated functions

This is preparation for using TF to create builtins that handle variable number of
arguments and have to remove these arguments dynamically from the stack upon
return.

The gist of the changes:
- Added a second argument to the Return node which specifies the number of stack
  slots to pop upon return in addition to those specified by the Linkage of the
  compiled function.
- Removed Tail -> Non-Tail fallback in the instruction selector. Since TF now should
  handles all tail-call cases except where the return value type differs, this fallback
  was not really useful and in fact caused unexpected behavior with variable
  sized argument popping, since it wasn't possible to materialize a Return node
  with the right pop count from the TailCall without additional context.
- Modified existing Return generation to pass a constant zero as the additional
  pop argument since the variable pop functionality

LOG=N

Committed: https://crrev.com/fe552636be3afd99210c991545e91fd705e4ee46
Cr-Commit-Position: refs/heads/master@{#40699}

[danno, 2016-10-24 08:23:45]

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

[commit-bot: I haz the power, 2016-10-24 08:23:46]

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, 2016-10-24 08:28:18]

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

[commit-bot: I haz the power, 2016-10-24 08:28:18]

Dry run: Try jobs failed on following builders:
  v8_linux_mips64el_compile_rel on master.tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_linux_mips64el_compile_r...)

[danno, 2016-10-24 08:29:24]

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

[commit-bot: I haz the power, 2016-10-24 08:29:27]

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, 2016-10-24 08:41:19]

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

[commit-bot: I haz the power, 2016-10-24 08:41:20]

Dry run: Try jobs failed on following builders:
  v8_linux64_gyp_rel_ng on master.tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_linux64_gyp_rel_ng/build...)

[danno, 2016-10-24 08:43:11]

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

[commit-bot: I haz the power, 2016-10-24 08:43:22]

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

[danno, 2016-10-24 08:44:54]

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

[commit-bot: I haz the power, 2016-10-24 08:45:00]

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, 2016-10-24 08:49:09]

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

[commit-bot: I haz the power, 2016-10-24 08:49:10]

Dry run: Try jobs failed on following builders:
  v8_linux_mips64el_compile_rel on master.tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_linux_mips64el_compile_r...)

[danno, 2016-10-24 08:58:42]

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

[commit-bot: I haz the power, 2016-10-24 08:58:50]

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

[danno, 2016-10-24 09:09:14]

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

[commit-bot: I haz the power, 2016-10-24 09:09:19]

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, 2016-10-24 09:20:50]

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

[commit-bot: I haz the power, 2016-10-24 09:20:51]

Dry run: Try jobs failed on following builders:
  v8_linux_mips64el_compile_rel on master.tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_linux_mips64el_compile_r...)

[danno, 2016-10-24 09:25:14]

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

[commit-bot: I haz the power, 2016-10-24 09:25:19]

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

[danno, 2016-10-24 09:27:27]

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

[commit-bot: I haz the power, 2016-10-24 09:27:30]

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, 2016-10-24 09:53:09]

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

[commit-bot: I haz the power, 2016-10-24 09:53:10]

Dry run: Try jobs failed on following builders:
  v8_linux_mipsel_compile_rel on master.tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_linux_mipsel_compile_rel...)

[danno, 2016-10-24 15:42:57]

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

[commit-bot: I haz the power, 2016-10-24 15:43:01]

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, 2016-10-24 16:27:36]

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

[commit-bot: I haz the power, 2016-10-24 16:27:37]

Dry run: This issue passed the CQ dry run.

[danno, 2016-10-25 06:42:23]

Description was changed from

==========
[turbofan] Support variable size argument popping in TF-generated functions
==========

to

==========
[turbofan] Support variable size argument removal in TF-generated functions

This is preparation for using TF to create builtins that handle variable number
of
arguments and have to remove these arguments dynamically from the stack upon
return.

The gist of the changes:
- Added a second argument to the Return node which specifies the number of stack
  slots to pop upon return in addition to those specified by the Linkage of the 
  compiled function.
- Removed Tail -> Non-Tail fallback in the instruction selector. Since TF now
should
  handles all tail-call cases except where the return value type differs, this
fallback
  was not really useful and in fact caused unexpected behavior with variable 
  sized argument popping, since it wasn't possible to materialize a Return node
  with the right pop count from the TailCall without additional context.
- Modified existing Return generation to pass a constant zero as the additional 
  pop argument since the variable pop functionality 

LOG=N
==========

[danno, 2016-10-25 06:42:23]

danno@chromium.org changed reviewers:
+ bmeurer@chromium.org, epertoso@chromium.org, mstarzinger@chromium.org

[danno, 2016-10-25 06:43:32]

ptal

https://codereview.chromium.org/2446543002/diff/140001/src/code-stubs.cc
File src/code-stubs.cc (right):

https://codereview.chromium.org/2446543002/diff/140001/src/code-stubs.cc#newc...
src/code-stubs.cc:2904: result.Bind(assembler->UndefinedConstant());
It's not a good idea to tail call with a different return signature, which can
happen easily by tail-calling abort from a stub that doesn't return a value.
This started triggering CHECKs when tinkering with this CL.

https://codereview.chromium.org/2446543002/diff/140001/src/compiler/instructi...
File src/compiler/instruction-selector.cc (left):

https://codereview.chromium.org/2446543002/diff/140001/src/compiler/instructi...
src/compiler/instruction-selector.cc:1952: } else {
Given that return instructions now need to know how much extra they need to pop
from the stack, turning TailCalls into regular calls is no longer possible,
since it would require materializing that pop value from nothing... all other
reductions know where the get/inherit the value.

Besides, implicitly converting tail calls into something else seems unexpected
and we never should (or do) rely on it.

[Michael Starzinger, 2016-10-28 14:12:15]

Looking good from my end. Just a couple of nits.

https://codereview.chromium.org/2446543002/diff/220001/src/compiler/ast-graph...
File src/compiler/ast-graph-builder.cc (right):

https://codereview.chromium.org/2446543002/diff/220001/src/compiler/ast-graph...
src/compiler/ast-graph-builder.cc:3780: Node* pop_node =
graph()->NewNode(common()->Int32Constant(0));
nit: Please use jsgraph()->Int32Constant(0) here. Direct calls to
graph()->NewNode() are dangerous.

https://codereview.chromium.org/2446543002/diff/220001/src/compiler/bytecode-...
File src/compiler/bytecode-graph-builder.cc (right):

https://codereview.chromium.org/2446543002/diff/220001/src/compiler/bytecode-...
src/compiler/bytecode-graph-builder.cc:1730: Node* pop_node =
graph()->NewNode(common()->Int32Constant(0));
nit: Likewise, please use jsgraph()->Int32Constant(0) here.

https://codereview.chromium.org/2446543002/diff/220001/src/compiler/instructi...
File src/compiler/instruction-selector.cc (right):

https://codereview.chromium.org/2446543002/diff/220001/src/compiler/instructi...
src/compiler/instruction-selector.cc:1901: }
nit: Indentation of the rest of the function off, please re-indent.

https://codereview.chromium.org/2446543002/diff/220001/src/compiler/instructi...
src/compiler/instruction-selector.cc:1963: const int ret_count =
linkage()->GetIncomingDescriptor()->ReturnCount() == 0
nit: s/ret_count/input_count/ because the semantics of that value changed.

https://codereview.chromium.org/2446543002/diff/220001/src/compiler/instructi...
src/compiler/instruction-selector.cc:1969: value_locations[0] =
pop_count->opcode() == IrOpcode::kInt32Constant
nit: Can we use g.CanBeImmediate here?

https://codereview.chromium.org/2446543002/diff/220001/src/compiler/simplifie...
File src/compiler/simplified-lowering.cc (right):

https://codereview.chromium.org/2446543002/diff/220001/src/compiler/simplifie...
src/compiler/simplified-lowering.cc:2506: case IrOpcode::kReturn:
nit: Please move this up to before the block where "all uniform" opcodes are
handled.

https://codereview.chromium.org/2446543002/diff/220001/src/compiler/tail-call...
File src/compiler/tail-call-optimization.cc (right):

https://codereview.chromium.org/2446543002/diff/220001/src/compiler/tail-call...
src/compiler/tail-call-optimization.cc:26: Node* const control =
NodeProperties::GetControlInput(node);
nit: Can we add a DCHECK here that ensures we are not expected to drop elements
from the stack? Something along the lines of "DCHECK_EQ(0,
Int32Matcher(x).Value())" or similar.

[danno, 2016-10-31 09:23:33]

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

[commit-bot: I haz the power, 2016-10-31 09:23:44]

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, 2016-10-31 09:51:00]

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

[commit-bot: I haz the power, 2016-10-31 09:51:01]

Dry run: This issue passed the CQ dry run.

[danno, 2016-10-31 10:06:40]

review feedback addressed, please take another look.

https://codereview.chromium.org/2446543002/diff/220001/src/compiler/ast-graph...
File src/compiler/ast-graph-builder.cc (right):

https://codereview.chromium.org/2446543002/diff/220001/src/compiler/ast-graph...
src/compiler/ast-graph-builder.cc:3780: Node* pop_node =
graph()->NewNode(common()->Int32Constant(0));
On 2016/10/28 14:12:14, Michael Starzinger wrote:
> nit: Please use jsgraph()->Int32Constant(0) here. Direct calls to
> graph()->NewNode() are dangerous.

Done.

https://codereview.chromium.org/2446543002/diff/220001/src/compiler/bytecode-...
File src/compiler/bytecode-graph-builder.cc (right):

https://codereview.chromium.org/2446543002/diff/220001/src/compiler/bytecode-...
src/compiler/bytecode-graph-builder.cc:1730: Node* pop_node =
graph()->NewNode(common()->Int32Constant(0));
On 2016/10/28 14:12:14, Michael Starzinger wrote:
> nit: Likewise, please use jsgraph()->Int32Constant(0) here.

Done.

https://codereview.chromium.org/2446543002/diff/220001/src/compiler/instructi...
File src/compiler/instruction-selector.cc (right):

https://codereview.chromium.org/2446543002/diff/220001/src/compiler/instructi...
src/compiler/instruction-selector.cc:1901: }
On 2016/10/28 14:12:14, Michael Starzinger wrote:
> nit: Indentation of the rest of the function off, please re-indent.

Done.

https://codereview.chromium.org/2446543002/diff/220001/src/compiler/instructi...
src/compiler/instruction-selector.cc:1963: const int ret_count =
linkage()->GetIncomingDescriptor()->ReturnCount() == 0
On 2016/10/28 14:12:14, Michael Starzinger wrote:
> nit: s/ret_count/input_count/ because the semantics of that value changed.

Done.

https://codereview.chromium.org/2446543002/diff/220001/src/compiler/instructi...
src/compiler/instruction-selector.cc:1969: value_locations[0] =
pop_count->opcode() == IrOpcode::kInt32Constant
On 2016/10/28 14:12:14, Michael Starzinger wrote:
> nit: Can we use g.CanBeImmediate here?

Unfortunately, no. IFAICT, CanBeImmediate is platform-specific. It certainly
would be possible to turn it into a virtual method that's available in the
InstructionSelector superclass and override it in the subclasses, but should I
do that for this CL?

https://codereview.chromium.org/2446543002/diff/220001/src/compiler/simplifie...
File src/compiler/simplified-lowering.cc (right):

https://codereview.chromium.org/2446543002/diff/220001/src/compiler/simplifie...
src/compiler/simplified-lowering.cc:2506: case IrOpcode::kReturn:
On 2016/10/28 14:12:14, Michael Starzinger wrote:
> nit: Please move this up to before the block where "all uniform" opcodes are
> handled.

Done.

https://codereview.chromium.org/2446543002/diff/220001/src/compiler/tail-call...
File src/compiler/tail-call-optimization.cc (right):

https://codereview.chromium.org/2446543002/diff/220001/src/compiler/tail-call...
src/compiler/tail-call-optimization.cc:26: Node* const control =
NodeProperties::GetControlInput(node);
On 2016/10/28 14:12:14, Michael Starzinger wrote:
> nit: Can we add a DCHECK here that ensures we are not expected to drop
elements
> from the stack? Something along the lines of "DCHECK_EQ(0,
> Int32Matcher(x).Value())" or similar.

Done.

[Michael Starzinger, 2016-10-31 10:23:19]

Thanks! LGTM from my end.

https://codereview.chromium.org/2446543002/diff/220001/src/compiler/instructi...
File src/compiler/instruction-selector.cc (right):

https://codereview.chromium.org/2446543002/diff/220001/src/compiler/instructi...
src/compiler/instruction-selector.cc:1969: value_locations[0] =
pop_count->opcode() == IrOpcode::kInt32Constant
On 2016/10/31 10:06:39, danno wrote:
> On 2016/10/28 14:12:14, Michael Starzinger wrote:
> > nit: Can we use g.CanBeImmediate here?
> 
> Unfortunately, no. IFAICT, CanBeImmediate is platform-specific. It certainly
> would be possible to turn it into a virtual method that's available in the
> InstructionSelector superclass and override it in the subclasses, but should I
> do that for this CL?

Acknowledged. Nah, I thought it would be "an easy" change. I am fine with
keeping it as it is.

[danno, 2016-10-31 16:51:30]

The CQ bit was checked by danno@chromium.org

[commit-bot: I haz the power, 2016-10-31 16:51:34]

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, 2016-10-31 16:53:42]

Description was changed from

==========
[turbofan] Support variable size argument removal in TF-generated functions

This is preparation for using TF to create builtins that handle variable number
of
arguments and have to remove these arguments dynamically from the stack upon
return.

The gist of the changes:
- Added a second argument to the Return node which specifies the number of stack
  slots to pop upon return in addition to those specified by the Linkage of the 
  compiled function.
- Removed Tail -> Non-Tail fallback in the instruction selector. Since TF now
should
  handles all tail-call cases except where the return value type differs, this
fallback
  was not really useful and in fact caused unexpected behavior with variable 
  sized argument popping, since it wasn't possible to materialize a Return node
  with the right pop count from the TailCall without additional context.
- Modified existing Return generation to pass a constant zero as the additional 
  pop argument since the variable pop functionality 

LOG=N
==========

to

==========
[turbofan] Support variable size argument removal in TF-generated functions

This is preparation for using TF to create builtins that handle variable number
of
arguments and have to remove these arguments dynamically from the stack upon
return.

The gist of the changes:
- Added a second argument to the Return node which specifies the number of stack
  slots to pop upon return in addition to those specified by the Linkage of the 
  compiled function.
- Removed Tail -> Non-Tail fallback in the instruction selector. Since TF now
should
  handles all tail-call cases except where the return value type differs, this
fallback
  was not really useful and in fact caused unexpected behavior with variable 
  sized argument popping, since it wasn't possible to materialize a Return node
  with the right pop count from the TailCall without additional context.
- Modified existing Return generation to pass a constant zero as the additional 
  pop argument since the variable pop functionality 

LOG=N
==========

[commit-bot: I haz the power, 2016-10-31 16:53:44]

Committed patchset #13 (id:240001)

[Michael Achenbach, 2016-11-02 07:21:22]

A revert of this CL (patchset #13 id:240001) has been created in
https://codereview.chromium.org/2473643002/ by machenbach@chromium.org.

The reason for reverting is: Seems to break arm64 sim debug and blocks roll:
https://build.chromium.org/p/client.v8.ports/builders/V8%20Linux%20-%20arm64%....

[danno, 2016-11-02 12:41:04]

Description was changed from

==========
[turbofan] Support variable size argument removal in TF-generated functions

This is preparation for using TF to create builtins that handle variable number
of
arguments and have to remove these arguments dynamically from the stack upon
return.

The gist of the changes:
- Added a second argument to the Return node which specifies the number of stack
  slots to pop upon return in addition to those specified by the Linkage of the 
  compiled function.
- Removed Tail -> Non-Tail fallback in the instruction selector. Since TF now
should
  handles all tail-call cases except where the return value type differs, this
fallback
  was not really useful and in fact caused unexpected behavior with variable 
  sized argument popping, since it wasn't possible to materialize a Return node
  with the right pop count from the TailCall without additional context.
- Modified existing Return generation to pass a constant zero as the additional 
  pop argument since the variable pop functionality 

LOG=N
==========

to

==========
[turbofan] Support variable size argument removal in TF-generated functions

This is preparation for using TF to create builtins that handle variable number
of
arguments and have to remove these arguments dynamically from the stack upon
return.

The gist of the changes:
- Added a second argument to the Return node which specifies the number of stack
  slots to pop upon return in addition to those specified by the Linkage of the 
  compiled function.
- Removed Tail -> Non-Tail fallback in the instruction selector. Since TF now
should
  handles all tail-call cases except where the return value type differs, this
fallback
  was not really useful and in fact caused unexpected behavior with variable 
  sized argument popping, since it wasn't possible to materialize a Return node
  with the right pop count from the TailCall without additional context.
- Modified existing Return generation to pass a constant zero as the additional 
  pop argument since the variable pop functionality 

LOG=N
==========

[danno, 2016-11-02 12:41:52]

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

[commit-bot: I haz the power, 2016-11-02 12:41:58]

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, 2016-11-02 13:07:46]

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

[commit-bot: I haz the power, 2016-11-02 13:07:47]

Dry run: This issue passed the CQ dry run.

[danno, 2016-11-02 13:13:01]

Fixed arm64 and tests, relanding

[danno, 2016-11-02 13:13:08]

The CQ bit was checked by danno@chromium.org

[danno, 2016-11-02 13:13:09]

The patchset sent to the CQ was uploaded after l-g-t-m from
mstarzinger@chromium.org
Link to the patchset: https://codereview.chromium.org/2446543002/#ps260001
(title: "Fix tests and arm64")

[commit-bot: I haz the power, 2016-11-02 13:13:17]

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, 2016-11-02 13:15:47]

Description was changed from

==========
[turbofan] Support variable size argument removal in TF-generated functions

This is preparation for using TF to create builtins that handle variable number
of
arguments and have to remove these arguments dynamically from the stack upon
return.

The gist of the changes:
- Added a second argument to the Return node which specifies the number of stack
  slots to pop upon return in addition to those specified by the Linkage of the 
  compiled function.
- Removed Tail -> Non-Tail fallback in the instruction selector. Since TF now
should
  handles all tail-call cases except where the return value type differs, this
fallback
  was not really useful and in fact caused unexpected behavior with variable 
  sized argument popping, since it wasn't possible to materialize a Return node
  with the right pop count from the TailCall without additional context.
- Modified existing Return generation to pass a constant zero as the additional 
  pop argument since the variable pop functionality 

LOG=N
==========

to

==========
[turbofan] Support variable size argument removal in TF-generated functions

This is preparation for using TF to create builtins that handle variable number
of
arguments and have to remove these arguments dynamically from the stack upon
return.

The gist of the changes:
- Added a second argument to the Return node which specifies the number of stack
  slots to pop upon return in addition to those specified by the Linkage of the 
  compiled function.
- Removed Tail -> Non-Tail fallback in the instruction selector. Since TF now
should
  handles all tail-call cases except where the return value type differs, this
fallback
  was not really useful and in fact caused unexpected behavior with variable 
  sized argument popping, since it wasn't possible to materialize a Return node
  with the right pop count from the TailCall without additional context.
- Modified existing Return generation to pass a constant zero as the additional 
  pop argument since the variable pop functionality 

LOG=N
==========

[commit-bot: I haz the power, 2016-11-02 13:15:49]

Committed patchset #14 (id:260001)

[commit-bot: I haz the power, 2016-11-17 22:18:24]

Description was changed from

==========
[turbofan] Support variable size argument removal in TF-generated functions

This is preparation for using TF to create builtins that handle variable number
of
arguments and have to remove these arguments dynamically from the stack upon
return.

The gist of the changes:
- Added a second argument to the Return node which specifies the number of stack
  slots to pop upon return in addition to those specified by the Linkage of the 
  compiled function.
- Removed Tail -> Non-Tail fallback in the instruction selector. Since TF now
should
  handles all tail-call cases except where the return value type differs, this
fallback
  was not really useful and in fact caused unexpected behavior with variable 
  sized argument popping, since it wasn't possible to materialize a Return node
  with the right pop count from the TailCall without additional context.
- Modified existing Return generation to pass a constant zero as the additional 
  pop argument since the variable pop functionality 

LOG=N
==========

to

==========
[turbofan] Support variable size argument removal in TF-generated functions

This is preparation for using TF to create builtins that handle variable number
of
arguments and have to remove these arguments dynamically from the stack upon
return.

The gist of the changes:
- Added a second argument to the Return node which specifies the number of stack
  slots to pop upon return in addition to those specified by the Linkage of the
  compiled function.
- Removed Tail -> Non-Tail fallback in the instruction selector. Since TF now
should
  handles all tail-call cases except where the return value type differs, this
fallback
  was not really useful and in fact caused unexpected behavior with variable
  sized argument popping, since it wasn't possible to materialize a Return node
  with the right pop count from the TailCall without additional context.
- Modified existing Return generation to pass a constant zero as the additional
  pop argument since the variable pop functionality

LOG=N

Committed: https://crrev.com/5319b50c853c4213d825aa7cf620fde5d827f7eb
Cr-Commit-Position: refs/heads/master@{#40678}
==========

[commit-bot: I haz the power, 2016-11-17 22:18:26]

Patchset 14 (id:??) landed as
https://crrev.com/5319b50c853c4213d825aa7cf620fde5d827f7eb
Cr-Commit-Position: refs/heads/master@{#40678}

[commit-bot: I haz the power, 2016-11-17 22:19:18]

Description was changed from

==========
[turbofan] Support variable size argument removal in TF-generated functions

This is preparation for using TF to create builtins that handle variable number
of
arguments and have to remove these arguments dynamically from the stack upon
return.

The gist of the changes:
- Added a second argument to the Return node which specifies the number of stack
  slots to pop upon return in addition to those specified by the Linkage of the
  compiled function.
- Removed Tail -> Non-Tail fallback in the instruction selector. Since TF now
should
  handles all tail-call cases except where the return value type differs, this
fallback
  was not really useful and in fact caused unexpected behavior with variable
  sized argument popping, since it wasn't possible to materialize a Return node
  with the right pop count from the TailCall without additional context.
- Modified existing Return generation to pass a constant zero as the additional
  pop argument since the variable pop functionality

LOG=N

Committed: https://crrev.com/5319b50c853c4213d825aa7cf620fde5d827f7eb
Cr-Commit-Position: refs/heads/master@{#40678}
==========

to

==========
[turbofan] Support variable size argument removal in TF-generated functions

This is preparation for using TF to create builtins that handle variable number
of
arguments and have to remove these arguments dynamically from the stack upon
return.

The gist of the changes:
- Added a second argument to the Return node which specifies the number of stack
  slots to pop upon return in addition to those specified by the Linkage of the
  compiled function.
- Removed Tail -> Non-Tail fallback in the instruction selector. Since TF now
should
  handles all tail-call cases except where the return value type differs, this
fallback
  was not really useful and in fact caused unexpected behavior with variable
  sized argument popping, since it wasn't possible to materialize a Return node
  with the right pop count from the TailCall without additional context.
- Modified existing Return generation to pass a constant zero as the additional
  pop argument since the variable pop functionality

LOG=N

Committed: https://crrev.com/fe552636be3afd99210c991545e91fd705e4ee46
Cr-Commit-Position: refs/heads/master@{#40699}
==========

[commit-bot: I haz the power, 2016-11-17 22:19:20]

Patchset 14 (id:??) landed as
https://crrev.com/fe552636be3afd99210c991545e91fd705e4ee46
Cr-Commit-Position: refs/heads/master@{#40699}