[compiler] Profiler reconstructs inlined frames for deopts.

[compiler] Profiler reconstructs inlined frames for deopts.

This makes the profiler reconstruct inlined function frames at eager
deopt points from the deoptimization data. The main goal of this is to
remove the last side-channel where Crankshaft communicates directly to
the profiler. This is the last preparatory step towards deprecating the
side-channel in question.

R=yangguo@chromium.org

Committed: https://crrev.com/29001f44fc8be9308c1255ec73b07e425ae206aa
Cr-Commit-Position: refs/heads/master@{#36229}

[Michael Starzinger, 2016-05-12 13:39:54]

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

[Michael Starzinger, 2016-05-12 13:39:55]

Benedikt: PTAL at Crankshaft.
Yang: PTAL at everything else.

[Benedikt Meurer, 2016-05-12 17:43:35]

LGTM on crankyshaft.

[Yang, 2016-05-13 06:48:47]

On 2016/05/12 17:43:35, Benedikt Meurer wrote:
> LGTM on crankyshaft.

  _      _____ _______ __  __ 
 | |    / ____|__   __|  \/  |
 | |   | |  __   | |  | \  / |
 | |   | | |_ |  | |  | |\/| |
 | |___| |__| |  | |  | |  | |
 |lgtm__\_____|  |_|  |_|  |_|

[Yang, 2016-05-13 06:49:10]

On 2016/05/13 06:48:47, Yang wrote:
> On 2016/05/12 17:43:35, Benedikt Meurer wrote:
> > LGTM on crankyshaft.
> 
>   _      _____ _______ __  __ 
>  | |    / ____|__   __|  \/  |
>  | |   | |  __   | |  | \  / |
>  | |   | | |_ |  | |  | |\/| |
>  | |___| |__| |  | |  | |  | |
>  |lgtm _\_____|  |_|  |_|  |_|

[Yang, 2016-05-13 06:49:46]

rietveld fails to parse my lgtm :/

[Yang, 2016-05-13 06:52:42]

On 2016/05/13 06:49:46, Yang wrote:
> rietveld fails to parse my lgtm :/

One last try:

  _       _____  _______  __  __ 
 | |     / ____||__   __||  \/  |
 | |    | |  __    | |   | \  / |
 | |    | | |_ |   | |   | |\/| |
 | |___ | |__| |   | |   | |  | |
 | lgtm  \_____|   |_|   |_|  |_|

[Yang, 2016-05-13 06:53:22]

https://codereview.chromium.org/1973993002/diff/1/src/profiler/cpu-profiler.cc
File src/profiler/cpu-profiler.cc (right):

https://codereview.chromium.org/1973993002/diff/1/src/profiler/cpu-profiler.c...
src/profiler/cpu-profiler.cc:455: if (code->kind() != Code::OPTIMIZED_FUNCTION)
return;
You could just check for abstract_code->kind(), and save the first line.

https://codereview.chromium.org/1973993002/diff/1/src/profiler/cpu-profiler.c...
src/profiler/cpu-profiler.cc:458: for (RelocIterator rit(code); !rit.done();
rit.next()) {
You could pass (1 << DEOPT_ID) as reloc info mask to the RelocIterator
constructor and just assert that what you get are only deopt ids.

[Michael Starzinger, 2016-05-13 08:08:19]

Thanks! Addressed comments. Landing.

https://codereview.chromium.org/1973993002/diff/1/src/profiler/cpu-profiler.cc
File src/profiler/cpu-profiler.cc (right):

https://codereview.chromium.org/1973993002/diff/1/src/profiler/cpu-profiler.c...
src/profiler/cpu-profiler.cc:455: if (code->kind() != Code::OPTIMIZED_FUNCTION)
return;
On 2016/05/13 06:53:22, Yang wrote:
> You could just check for abstract_code->kind(), and save the first line.

Done.

https://codereview.chromium.org/1973993002/diff/1/src/profiler/cpu-profiler.c...
src/profiler/cpu-profiler.cc:458: for (RelocIterator rit(code); !rit.done();
rit.next()) {
On 2016/05/13 06:53:22, Yang wrote:
> You could pass (1 << DEOPT_ID) as reloc info mask to the RelocIterator
> constructor and just assert that what you get are only deopt ids.

Done.

[Michael Starzinger, 2016-05-13 08:43:08]

The CQ bit was checked by mstarzinger@chromium.org

[Michael Starzinger, 2016-05-13 08:43:10]

The patchset sent to the CQ was uploaded after l-g-t-m from
bmeurer@chromium.org, yangguo@chromium.org
Link to the patchset: https://codereview.chromium.org/1973993002/#ps40001
(title: "Fix build.")

[commit-bot: I haz the power, 2016-05-13 08:43:12]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1973993002/40001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1973993002/40001

[commit-bot: I haz the power, 2016-05-13 08:49:54]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2016-05-13 08:51:32]

Description was changed from

==========
[compiler] Profiler reconstructs inlined frames for deopts.

This makes the profiler reconstruct inlined function frames at eager
deopt points from the deoptimization data. The main goal of this is to
remove the last side-channel where Crankshaft communicates directly to
the profiler. This is the last preparatory step towards deprecating the
side-channel in question.

R=yangguo@chromium.org
==========

to

==========
[compiler] Profiler reconstructs inlined frames for deopts.

This makes the profiler reconstruct inlined function frames at eager
deopt points from the deoptimization data. The main goal of this is to
remove the last side-channel where Crankshaft communicates directly to
the profiler. This is the last preparatory step towards deprecating the
side-channel in question.

R=yangguo@chromium.org

Committed: https://crrev.com/29001f44fc8be9308c1255ec73b07e425ae206aa
Cr-Commit-Position: refs/heads/master@{#36229}
==========

[commit-bot: I haz the power, 2016-05-13 08:51:34]

Patchset 3 (id:??) landed as
https://crrev.com/29001f44fc8be9308c1255ec73b07e425ae206aa
Cr-Commit-Position: refs/heads/master@{#36229}