[Interpreter] Add dispatch counters for each bytecode.

[Interpreter] Add dispatch counters for each bytecode.

Counters are enabled with FLAG_ignition_count_handler_dispatches.

Increment saturates instead of overflowing.

BUG=v8:4280
LOG=N

[Stefano Sanfilippo, 2016-03-21 13:40:03]

Description was changed from

==========
[Interpreter] Count dispatches to each bytecode handler.

Counters are enabled with --ignition_count_handler_dispatches.
--debug_code causes overflow checking code to be emitted for counters.
==========

to

==========
[Interpreter] Count dispatches to each bytecode handler.

Counters are enabled with FLAG_ignition_count_handler_dispatches.
FLAG_debug_code causes overflow checking code to be emitted for counters.
==========

[Stefano Sanfilippo, 2016-03-21 14:24:43]

Description was changed from

==========
[Interpreter] Count dispatches to each bytecode handler.

Counters are enabled with FLAG_ignition_count_handler_dispatches.
FLAG_debug_code causes overflow checking code to be emitted for counters.
==========

to

==========
[Interpreter] Count dispatches to each bytecode handler.

Counters are enabled with FLAG_ignition_count_handler_dispatches.
FLAG_debug_code causes overflow checking code to be emitted as well.
==========

[Stefano Sanfilippo, 2016-03-21 15:21:21]

Patchset #1 (id:1) has been deleted

[Stefano Sanfilippo, 2016-03-21 15:22:57]

ssanfilippo@chromium.org changed reviewers:
+ rmcilroy@chromium.org

[Stefano Sanfilippo, 2016-03-21 15:22:57]

PTAL.

[Stefano Sanfilippo, 2016-03-21 15:23:57]

Description was changed from

==========
[Interpreter] Count dispatches to each bytecode handler.

Counters are enabled with FLAG_ignition_count_handler_dispatches.
FLAG_debug_code causes overflow checking code to be emitted as well.
==========

to

==========
[Interpreter] Count dispatches to each bytecode handler.

Counters are enabled with FLAG_ignition_count_handler_dispatches.
FLAG_debug_code causes overflow checking code to be emitted as well.

BUG=v8:4280
LOG=N
==========

[Stefano Sanfilippo, 2016-03-21 15:33:15]

Description was changed from

==========
[Interpreter] Count dispatches to each bytecode handler.

Counters are enabled with FLAG_ignition_count_handler_dispatches.
FLAG_debug_code causes overflow checking code to be emitted as well.

BUG=v8:4280
LOG=N
==========

to

==========
[Interpreter] Add flag to count the number of dispatches for each bytecode.

Counters are enabled with FLAG_ignition_count_handler_dispatches.
FLAG_debug_code causes overflow checking code to be emitted as well.

BUG=v8:4280
LOG=N
==========

[Stefano Sanfilippo, 2016-03-21 15:40:53]

Description was changed from

==========
[Interpreter] Add flag to count the number of dispatches for each bytecode.

Counters are enabled with FLAG_ignition_count_handler_dispatches.
FLAG_debug_code causes overflow checking code to be emitted as well.

BUG=v8:4280
LOG=N
==========

to

==========
[Interpreter] Add flag to count the number of dispatches for each bytecode.

Counters are enabled with FLAG_ignition_count_handler_dispatches.
FLAG_debug_code causes overflow checks to be emitted as well.

BUG=v8:4280
LOG=N
==========

[rmcilroy, 2016-03-23 11:04:26]

https://codereview.chromium.org/1817033002/diff/20001/src/d8.cc
File src/d8.cc (right):

https://codereview.chromium.org/1817033002/diff/20001/src/d8.cc#newcode1353
src/d8.cc:1353: }
As discussed, maybe we could output json for this case too. Is there anywhere we
could put this where it wouldn't be d8 specific (i.e., could be used by Chrome
too?

https://codereview.chromium.org/1817033002/diff/40001/src/interpreter/interpr...
File src/interpreter/interpreter.h (right):

https://codereview.chromium.org/1817033002/diff/40001/src/interpreter/interpr...
src/interpreter/interpreter.h:152: uint32_t
handlers_dispatch_counters_[kDispatchTableSize];
Do we need the 3x size of the dispatch table, or do we only need counters for
each non-scaled bytecode variant (I think the code in IncrementDispatchCounter
only updates the initial 256 entries of this table (which is probably what we
want, rather than having a few sparse counts of wide bytecode variants).

[Stefano Sanfilippo, 2016-03-23 16:26:27]

https://codereview.chromium.org/1817033002/diff/20001/src/d8.cc
File src/d8.cc (right):

https://codereview.chromium.org/1817033002/diff/20001/src/d8.cc#newcode1353
src/d8.cc:1353: }
On 2016/03/23 11:04:26, rmcilroy wrote:
> As discussed, maybe we could output json for this case too. Is there anywhere
we
> could put this where it wouldn't be d8 specific (i.e., could be used by Chrome
> too?

Done JSON. As of where to move this, I'll start by extracting a function to do
the job, which does not depend on anything Shell specific, so we are free to
move it wherever it seems more appropriate.

https://codereview.chromium.org/1817033002/diff/40001/src/interpreter/interpr...
File src/interpreter/interpreter.h (right):

https://codereview.chromium.org/1817033002/diff/40001/src/interpreter/interpr...
src/interpreter/interpreter.h:152: uint32_t
handlers_dispatch_counters_[kDispatchTableSize];
On 2016/03/23 11:04:26, rmcilroy wrote:
> Do we need the 3x size of the dispatch table, or do we only need counters for
> each non-scaled bytecode variant (I think the code in IncrementDispatchCounter
> only updates the initial 256 entries of this table (which is probably what we
> want, rather than having a few sparse counts of wide bytecode variants).

I have created a new kCountersTableRowSize constant and moved it to the public
part, so that it can be used externally (e.g. in for loops).

[Stefano Sanfilippo, 2016-03-24 14:25:52]

Description was changed from

==========
[Interpreter] Add flag to count the number of dispatches for each bytecode.

Counters are enabled with FLAG_ignition_count_handler_dispatches.
FLAG_debug_code causes overflow checks to be emitted as well.

BUG=v8:4280
LOG=N
==========

to

==========
[Interpreter] Add dispatch counters for each bytecode.

Counters are enabled with FLAG_ignition_count_handler_dispatches.
FLAG_debug_code causes overflow checks to be emitted as well.

BUG=v8:4280
LOG=N
==========

[Stefano Sanfilippo, 2016-03-24 14:27:06]

Description was changed from

==========
[Interpreter] Add dispatch counters for each bytecode.

Counters are enabled with FLAG_ignition_count_handler_dispatches.
FLAG_debug_code causes overflow checks to be emitted as well.

BUG=v8:4280
LOG=N
==========

to

==========
[Interpreter] Add dispatch counters for each bytecode.

Counters are enabled with FLAG_ignition_count_handler_dispatches.

Increment saturates instead of overflowing.

BUG=v8:4280
LOG=N
==========

[rmcilroy, 2016-04-05 10:00:46]

https://codereview.chromium.org/1817033002/diff/120001/src/assembler.h
File src/assembler.h (right):

https://codereview.chromium.org/1817033002/diff/120001/src/assembler.h#newcod...
src/assembler.h:908: static ExternalReference
interpreter_handlers_dispatch_counters(
nit - just interpreter_dispatch_counters (throughout, and for the function and
field in interpreter.h too).

https://codereview.chromium.org/1817033002/diff/120001/src/d8.cc
File src/d8.cc (right):

https://codereview.chromium.org/1817033002/diff/120001/src/d8.cc#newcode1309
src/d8.cc:1309: void Shell::DumpHandlersDispatchCounters(FILE* stream, Isolate*
isolate) {
WriteInterpreterDispatchCounters

https://codereview.chromium.org/1817033002/diff/120001/src/d8.cc#newcode1314
src/d8.cc:1314: fputc('{', stream);
Can we just use an ostream instead of the c file output?

https://codereview.chromium.org/1817033002/diff/120001/src/d8.cc#newcode1362
src/d8.cc:1362: fopen(i::FLAG_ignition_handler_to_handler_output_file, "w");
nit - just do the open / close in WriteInterpreterDispatchCounters

https://codereview.chromium.org/1817033002/diff/120001/src/flag-definitions.h
File src/flag-definitions.h (right):

https://codereview.chromium.org/1817033002/diff/120001/src/flag-definitions.h...
src/flag-definitions.h:299: DEFINE_BOOL(ignition_count_handler_dispatches,
false,
trace_ignition_dispatches

https://codereview.chromium.org/1817033002/diff/120001/src/flag-definitions.h...
src/flag-definitions.h:300: "count dispatches to each handler")
traces the dispatches to bytecode handlers by the ignition interpreter

https://codereview.chromium.org/1817033002/diff/120001/src/flag-definitions.h...
src/flag-definitions.h:301:
DEFINE_STRING(ignition_handler_to_handler_output_file,
trace_ignition_dispatches_output_file

https://codereview.chromium.org/1817033002/diff/120001/src/flag-definitions.h...
src/flag-definitions.h:302: "v8.dispatch_counters_table.json",
v8.ignition_dispatches_table.json

https://codereview.chromium.org/1817033002/diff/120001/src/flag-definitions.h...
src/flag-definitions.h:303: "file to write the handler to handler dispatch
counters table to")
the file to which the bytecode handler dispatch table is written

https://codereview.chromium.org/1817033002/diff/120001/src/interpreter/interp...
File src/interpreter/interpreter-assembler.cc (right):

https://codereview.chromium.org/1817033002/diff/120001/src/interpreter/interp...
src/interpreter/interpreter-assembler.cc:617: void
InterpreterAssembler::IncrementDispatchCounter() {
nit - could you move this to be near TraceBytecode

https://codereview.chromium.org/1817033002/diff/120001/src/interpreter/interp...
File src/interpreter/interpreter-assembler.h (right):

https://codereview.chromium.org/1817033002/diff/120001/src/interpreter/interp...
src/interpreter/interpreter-assembler.h:170: void IncrementDispatchCounter();
TraceBytecodeDispatch

Also, please add a comment.

https://codereview.chromium.org/1817033002/diff/120001/src/interpreter/interp...
File src/interpreter/interpreter.cc (right):

https://codereview.chromium.org/1817033002/diff/120001/src/interpreter/interp...
src/interpreter/interpreter.cc:164: // or to print the assembly code generated
by TurboFan.
Please update this comment

https://codereview.chromium.org/1817033002/diff/120001/src/interpreter/interp...
File src/interpreter/interpreter.h (right):

https://codereview.chromium.org/1817033002/diff/120001/src/interpreter/interp...
src/interpreter/interpreter.h:67: static_cast<int>(Bytecode::kLast) + 1;
Please keep these private. I don't think you need kCountersTableRowSize either
(just use Bytecode::kLast + 1 where needed).

https://codereview.chromium.org/1817033002/diff/120001/src/interpreter/interp...
src/interpreter/interpreter.h:154: uint32_t
handlers_dispatch_counters_[kCountersTableRowSize];
Could we make this a smart pointer which is only allocated if the flag is set
(avoid wasting memory). Also, how about we just make this uintptr_t size, (with
appropriate updates in the interpreter-assembler generated code) that way we get
64bit counters on 64bit platforms.

[Stefano Sanfilippo, 2016-04-05 11:33:34]

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

[commit-bot: I haz the power, 2016-04-05 11:33:44]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1817033002/160001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1817033002/160001

[commit-bot: I haz the power, 2016-04-05 11:34:54]

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

[commit-bot: I haz the power, 2016-04-05 11:34:55]

Dry run: Try jobs failed on following builders:
  v8_linux64_asan_rel on tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_linux64_asan_rel/builds/...)
  v8_linux_arm64_rel on tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_linux_arm64_rel/builds/1...)
  v8_linux_dbg_ng on tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_linux_dbg_ng/builds/3789)
  v8_linux_mipsel_compile_rel on tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_linux_mipsel_compile_rel...)

[Stefano Sanfilippo, 2016-04-05 14:01:46]

https://codereview.chromium.org/1817033002/diff/120001/src/assembler.h
File src/assembler.h (right):

https://codereview.chromium.org/1817033002/diff/120001/src/assembler.h#newcod...
src/assembler.h:908: static ExternalReference
interpreter_handlers_dispatch_counters(
On 2016/04/05 10:00:45, rmcilroy (Slow) wrote:
> nit - just interpreter_dispatch_counters (throughout, and for the function and
> field in interpreter.h too).

Done.

https://codereview.chromium.org/1817033002/diff/120001/src/d8.cc
File src/d8.cc (right):

https://codereview.chromium.org/1817033002/diff/120001/src/d8.cc#newcode1309
src/d8.cc:1309: void Shell::DumpHandlersDispatchCounters(FILE* stream, Isolate*
isolate) {
On 2016/04/05 10:00:45, rmcilroy (Slow) wrote:
> WriteInterpreterDispatchCounters

Done.

https://codereview.chromium.org/1817033002/diff/120001/src/d8.cc#newcode1314
src/d8.cc:1314: fputc('{', stream);
On 2016/04/05 10:00:45, rmcilroy (Slow) wrote:
> Can we just use an ostream instead of the c file output?

Done.

https://codereview.chromium.org/1817033002/diff/120001/src/d8.cc#newcode1362
src/d8.cc:1362: fopen(i::FLAG_ignition_handler_to_handler_output_file, "w");
On 2016/04/05 10:00:45, rmcilroy (Slow) wrote:
> nit - just do the open / close in WriteInterpreterDispatchCounters

I'd rather open the stream externally and pass it to the function, since in the
followup we will be serializing both tables to the same file.

https://codereview.chromium.org/1817033002/diff/120001/src/flag-definitions.h
File src/flag-definitions.h (right):

https://codereview.chromium.org/1817033002/diff/120001/src/flag-definitions.h...
src/flag-definitions.h:299: DEFINE_BOOL(ignition_count_handler_dispatches,
false,
On 2016/04/05 10:00:45, rmcilroy (Slow) wrote:
> trace_ignition_dispatches

Done.

https://codereview.chromium.org/1817033002/diff/120001/src/flag-definitions.h...
src/flag-definitions.h:300: "count dispatches to each handler")
On 2016/04/05 10:00:45, rmcilroy (Slow) wrote:
> traces the dispatches to bytecode handlers by the ignition interpreter

Done.

https://codereview.chromium.org/1817033002/diff/120001/src/flag-definitions.h...
src/flag-definitions.h:301:
DEFINE_STRING(ignition_handler_to_handler_output_file,
On 2016/04/05 10:00:45, rmcilroy (Slow) wrote:
> trace_ignition_dispatches_output_file

Done.

https://codereview.chromium.org/1817033002/diff/120001/src/flag-definitions.h...
src/flag-definitions.h:302: "v8.dispatch_counters_table.json",
On 2016/04/05 10:00:45, rmcilroy (Slow) wrote:
> v8.ignition_dispatches_table.json

Done.

https://codereview.chromium.org/1817033002/diff/120001/src/flag-definitions.h...
src/flag-definitions.h:303: "file to write the handler to handler dispatch
counters table to")
On 2016/04/05 10:00:45, rmcilroy (Slow) wrote:
> the file to which the bytecode handler dispatch table is written

Done.

https://codereview.chromium.org/1817033002/diff/120001/src/interpreter/interp...
File src/interpreter/interpreter-assembler.cc (right):

https://codereview.chromium.org/1817033002/diff/120001/src/interpreter/interp...
src/interpreter/interpreter-assembler.cc:617: void
InterpreterAssembler::IncrementDispatchCounter() {
On 2016/04/05 10:00:46, rmcilroy (Slow) wrote:
> nit - could you move this to be near TraceBytecode

Done.

https://codereview.chromium.org/1817033002/diff/120001/src/interpreter/interp...
File src/interpreter/interpreter-assembler.h (right):

https://codereview.chromium.org/1817033002/diff/120001/src/interpreter/interp...
src/interpreter/interpreter-assembler.h:170: void IncrementDispatchCounter();
On 2016/04/05 10:00:46, rmcilroy (Slow) wrote:
> TraceBytecodeDispatch
> 
> Also, please add a comment.

Done.

https://codereview.chromium.org/1817033002/diff/120001/src/interpreter/interp...
File src/interpreter/interpreter.cc (right):

https://codereview.chromium.org/1817033002/diff/120001/src/interpreter/interp...
src/interpreter/interpreter.cc:164: // or to print the assembly code generated
by TurboFan.
On 2016/04/05 10:00:46, rmcilroy (Slow) wrote:
> Please update this comment

Done.

https://codereview.chromium.org/1817033002/diff/120001/src/interpreter/interp...
File src/interpreter/interpreter.h (right):

https://codereview.chromium.org/1817033002/diff/120001/src/interpreter/interp...
src/interpreter/interpreter.h:67: static_cast<int>(Bytecode::kLast) + 1;
On 2016/04/05 10:00:46, rmcilroy (Slow) wrote:
> Please keep these private. I don't think you need kCountersTableRowSize either
> (just use Bytecode::kLast + 1 where needed).

Moved to private.

I'd be in favour of keeping the kCountersTableRowSize constant, it will be
useful for the follow-up CL (handler to handler counters).

https://codereview.chromium.org/1817033002/diff/120001/src/interpreter/interp...
src/interpreter/interpreter.h:154: uint32_t
handlers_dispatch_counters_[kCountersTableRowSize];
On 2016/04/05 10:00:46, rmcilroy (Slow) wrote:
> Could we make this a smart pointer which is only allocated if the flag is set
> (avoid wasting memory). Also, how about we just make this uintptr_t size,
(with
> appropriate updates in the interpreter-assembler generated code) that way we
get
> 64bit counters on 64bit platforms.

I replaced the array with a std::vector which gets resize()'d only if the flag
is enabled. Done.

[Stefano Sanfilippo, 2016-04-05 14:02:13]

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

[commit-bot: I haz the power, 2016-04-05 14:02:21]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1817033002/160001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1817033002/160001

[commit-bot: I haz the power, 2016-04-05 14:03:50]

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

[commit-bot: I haz the power, 2016-04-05 14:03:50]

Dry run: Try jobs failed on following builders:
  v8_android_arm_compile_rel on tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_android_arm_compile_rel/...)
  v8_linux64_asan_rel on tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_linux64_asan_rel/builds/...)
  v8_linux64_rel_ng on tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_linux64_rel_ng/builds/3787)
  v8_linux_arm64_rel on tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_linux_arm64_rel/builds/1...)
  v8_linux_arm_rel on tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_linux_arm_rel/builds/16204)
  v8_linux_dbg_ng on tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_linux_dbg_ng/builds/3803)
  v8_linux_gcc_compile_rel on tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_linux_gcc_compile_rel/bu...)
  v8_linux_mips64el_compile_rel on tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_linux_mips64el_compile_r...)
  v8_linux_mipsel_compile_rel on tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_linux_mipsel_compile_rel...)
  v8_linux_nodcheck_rel on tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_linux_nodcheck_rel/build...)
  v8_linux_rel_ng on tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_linux_rel_ng/builds/3797)
  v8_presubmit on tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_presubmit/builds/13225)

[Stefano Sanfilippo, 2016-04-05 14:21:03]

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

[commit-bot: I haz the power, 2016-04-05 14:21:13]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1817033002/180001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1817033002/180001

[commit-bot: I haz the power, 2016-04-05 14:41:12]

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

[commit-bot: I haz the power, 2016-04-05 14:41:13]

Dry run: Try jobs failed on following builders:
  v8_win_nosnap_shared_compile_rel on tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_win_nosnap_shared_compil...)

[Stefano Sanfilippo, 2016-04-05 14:46:31]

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

[commit-bot: I haz the power, 2016-04-05 14:46:42]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1817033002/200001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1817033002/200001

[commit-bot: I haz the power, 2016-04-05 14:54:33]

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

[commit-bot: I haz the power, 2016-04-05 14:54:34]

Dry run: Try jobs failed on following builders:
  v8_win_nosnap_shared_compile_rel on tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_win_nosnap_shared_compil...)

[Stefano Sanfilippo, 2016-04-05 16:23:05]

Patchset #11 (id:220001) has been deleted

[Stefano Sanfilippo, 2016-04-05 16:24:35]

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

[commit-bot: I haz the power, 2016-04-05 16:24:42]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1817033002/240001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1817033002/240001

[Stefano Sanfilippo, 2016-04-05 17:53:58]

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

[commit-bot: I haz the power, 2016-04-05 17:54:02]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1817033002/280001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1817033002/280001

[commit-bot: I haz the power, 2016-04-05 18:25:50]

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

[commit-bot: I haz the power, 2016-04-05 18:25:51]

Dry run: Try jobs failed on following builders:
  v8_linux64_rel_ng on tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_linux64_rel_ng/builds/3813)
  v8_linux64_rel_ng_triggered on tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_linux64_rel_ng_triggered...)

[Stefano Sanfilippo, 2016-04-06 11:03:29]

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

[commit-bot: I haz the power, 2016-04-06 11:03:35]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1817033002/300001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1817033002/300001

[commit-bot: I haz the power, 2016-04-06 11:37:10]

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

[commit-bot: I haz the power, 2016-04-06 11:37:11]

Dry run: This issue passed the CQ dry run.

[Stefano Sanfilippo, 2016-04-18 12:54:24]

Closing, as we went for the approach in
https://crrev.com/1e3257d27f42ee1fd81aaf97536db171e7abd7fd