[Interpreter] Enable tracing of bytecode handler dispatches.

[Interpreter] Enable tracing of bytecode handler dispatches.

When FLAG_trace_ignition_dispatches is enabled, a dispatch counter is
kept for each pair of source-destination bytecode handlers.

Each counter saturates at max uintptr_t value.

Counters are dumped as a JSON-encoded object of objects, such that
each key on the top level object is a source bytecode name, and each key
on the corresponding value is a destination bytecode name, with the
associated counter as value. The output file name can be controlled
with the FLAG_trace_ignition_dispatches_output_file flag.

The JSON file may be written by calling
Interpreter::WriteDispatchCounters(), which is done for d8 in
Shell::OnExit, if FLAG_trace_ignition_dispatches is enabled.

BUG=v8:4899
LOG=N

Committed: https://crrev.com/1e3257d27f42ee1fd81aaf97536db171e7abd7fd
Cr-Commit-Position: refs/heads/master@{#35380}

[Stefano Sanfilippo, 2016-03-23 17:42:34]

Patchset #1 (id:1) has been deleted

[Stefano Sanfilippo, 2016-03-23 17:42:42]

Patchset #1 (id:20001) has been deleted

[Stefano Sanfilippo, 2016-03-23 17:42:49]

Patchset #1 (id:40001) has been deleted

[Stefano Sanfilippo, 2016-03-23 18:26:06]

Patchset #1 (id:60001) has been deleted

[Stefano Sanfilippo, 2016-03-23 18:30:48]

Patchset #1 (id:80001) has been deleted

[Stefano Sanfilippo, 2016-03-24 13:47:08]

Patchset #1 (id:100001) has been deleted

[Stefano Sanfilippo, 2016-03-24 13:50:25]

Patchset #1 (id:120001) has been deleted

[Stefano Sanfilippo, 2016-03-24 14:12:21]

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

[Stefano Sanfilippo, 2016-03-24 14:12:21]

Here is the follow-up to https://codereview.chromium.org/1817033002/ (dispatch
counters). PTAL.

[Stefano Sanfilippo, 2016-03-24 14:25:26]

Description was changed from

==========
[Interpreter] Count dispatches from each handler to each other.

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

to

==========
[Interpreter] Add dispatch counters from each handler to each other.

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

[Stefano Sanfilippo, 2016-04-05 17:54:02]

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

[commit-bot: I haz the power, 2016-04-05 17:54:13]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1828633003/320001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1828633003/320001

[commit-bot: I haz the power, 2016-04-05 18:08:30]

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

[commit-bot: I haz the power, 2016-04-05 18:08:31]

Dry run: Try jobs failed on following builders:
  v8_win_compile_dbg on tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_win_compile_dbg/builds/1...)

[Stefano Sanfilippo, 2016-04-06 09:26:52]

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

[commit-bot: I haz the power, 2016-04-06 09:27:03]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1828633003/340001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1828633003/340001

[commit-bot: I haz the power, 2016-04-06 10:07:24]

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

[commit-bot: I haz the power, 2016-04-06 10:07:25]

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/3850)
  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 11:11:52]

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

[commit-bot: I haz the power, 2016-04-06 11:12:05]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1828633003/360001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1828633003/360001

[commit-bot: I haz the power, 2016-04-06 11:52:36]

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

[commit-bot: I haz the power, 2016-04-06 11:52:37]

Dry run: This issue passed the CQ dry run.

[Stefano Sanfilippo, 2016-04-07 09:36:39]

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

[commit-bot: I haz the power, 2016-04-07 09:36:45]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1828633003/380001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1828633003/380001

[commit-bot: I haz the power, 2016-04-07 10:01:34]

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

[commit-bot: I haz the power, 2016-04-07 10:01:35]

Dry run: This issue passed the CQ dry run.

[Stefano Sanfilippo, 2016-04-07 10:52:55]

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

[commit-bot: I haz the power, 2016-04-07 10:53:02]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1828633003/400001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1828633003/400001

[commit-bot: I haz the power, 2016-04-07 11:17:27]

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

[commit-bot: I haz the power, 2016-04-07 11:17:28]

Dry run: This issue passed the CQ dry run.

[Stefano Sanfilippo, 2016-04-07 13:59:10]

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

[commit-bot: I haz the power, 2016-04-07 13:59:13]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1828633003/420001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1828633003/420001

[commit-bot: I haz the power, 2016-04-07 14:22:54]

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

[commit-bot: I haz the power, 2016-04-07 14:22:55]

Dry run: This issue passed the CQ dry run.

[Stefano Sanfilippo, 2016-04-07 17:41:18]

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

[commit-bot: I haz the power, 2016-04-07 17:41:29]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1828633003/440001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1828633003/440001

[commit-bot: I haz the power, 2016-04-07 18:07:26]

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

[commit-bot: I haz the power, 2016-04-07 18:07:27]

Dry run: This issue passed the CQ dry run.

[rmcilroy, 2016-04-08 11:24:44]

https://codereview.chromium.org/1828633003/diff/460001/src/assembler.h
File src/assembler.h (right):

https://codereview.chromium.org/1828633003/diff/460001/src/assembler.h#newcod...
src/assembler.h:912: static ExternalReference
interpreter_handler_to_handler_dispatch_counters(
interpreter_dispatch_counters should be fine (throughout)

https://codereview.chromium.org/1828633003/diff/460001/src/d8.cc
File src/d8.cc (right):

https://codereview.chromium.org/1828633003/diff/460001/src/d8.cc#newcode1289
src/d8.cc:1289: int kCountersTableRowSize =
static const

https://codereview.chromium.org/1828633003/diff/460001/src/d8.cc#newcode1297
src/d8.cc:1297: // Do not use an object, we need the order of entries to be
preserved
Not sure what this comment means? Also, could we have the output be a dictionary
of named target bytecode (only for those target bytecodes with a dispatch count
> 0) rather than an ordered list. Relying on the ordering of the source
bytecodes to show the order of the unnamed target bytecode list seems brittle.

https://codereview.chromium.org/1828633003/diff/460001/src/interpreter/interp...
File src/interpreter/interpreter-assembler.cc (right):

https://codereview.chromium.org/1828633003/diff/460001/src/interpreter/interp...
src/interpreter/interpreter-assembler.cc:670: void
InterpreterAssembler::TraceHandlerToHandlerBytecodeDispatch(
Just TraceBytecodeDispatch

https://codereview.chromium.org/1828633003/diff/460001/src/interpreter/interp...
src/interpreter/interpreter-assembler.cc:671: Node* target_index) {
/s/target_index/target_bytecode

https://codereview.chromium.org/1828633003/diff/460001/src/interpreter/interp...
src/interpreter/interpreter-assembler.cc:675: Node* handler_row_base =
IntPtrConstant(
source_bytecode_table_index

https://codereview.chromium.org/1828633003/diff/460001/src/interpreter/interp...
src/interpreter/interpreter-assembler.cc:684: void
InterpreterAssembler::SaturatedCounterIncrement(Node* counters_table,
Just do this inline above

https://codereview.chromium.org/1828633003/diff/460001/src/interpreter/interp...
src/interpreter/interpreter-assembler.cc:699: : MachineRepresentation::kWord32,
PointerRepresentation()

https://codereview.chromium.org/1828633003/diff/460001/src/interpreter/interp...
File src/interpreter/interpreter.cc (right):

https://codereview.chromium.org/1828633003/diff/460001/src/interpreter/interp...
src/interpreter/interpreter.cc:36: int kCountersTableRowSize =
static_cast<int>(Bytecode::kLast) + 1;
nit - move into if below. Also make static const.

https://codereview.chromium.org/1828633003/diff/460001/src/interpreter/interp...
File src/interpreter/interpreter.h (right):

https://codereview.chromium.org/1828633003/diff/460001/src/interpreter/interp...
src/interpreter/interpreter.h:62: static bool BytecodeHasHandler(Bytecode
bytecode, OperandScale operand_scale);
Remove

https://codereview.chromium.org/1828633003/diff/460001/src/interpreter/interp...
src/interpreter/interpreter.h:152: v8::base::SmartArrayPointer<uintptr_t>
handler_to_handler_dispatch_counters_;
/s/handler_to_handler_dispatch_counters/bytecode_dispatch_count_table

https://codereview.chromium.org/1828633003/diff/460001/tools/ignition_plot_ha...
File tools/ignition_plot_handler_counters.py (right):

https://codereview.chromium.org/1828633003/diff/460001/tools/ignition_plot_ha...
tools/ignition_plot_handler_counters.py:7: 
Name should be something like plot_ignition_bytecode_dispatches.py. Also, should
this be in a directory of ignition specific tracing tools (along with your other
tool when you land it)?

https://codereview.chromium.org/1828633003/diff/460001/tools/ignition_plot_ha...
tools/ignition_plot_handler_counters.py:11: #mpl.use('Agg')  ## Uncomment this
line if running headless
Remove?

[Stefano Sanfilippo, 2016-04-08 14:09:11]

Patchset #20 (id:520001) has been deleted

[Stefano Sanfilippo, 2016-04-08 14:28:30]

Patchset #21 (id:560001) has been deleted

[Stefano Sanfilippo, 2016-04-08 14:42:44]

https://codereview.chromium.org/1828633003/diff/460001/src/assembler.h
File src/assembler.h (right):

https://codereview.chromium.org/1828633003/diff/460001/src/assembler.h#newcod...
src/assembler.h:912: static ExternalReference
interpreter_handler_to_handler_dispatch_counters(
On 2016/04/08 11:24:43, rmcilroy (Slow) wrote:
> interpreter_dispatch_counters should be fine (throughout)

Done.

https://codereview.chromium.org/1828633003/diff/460001/src/d8.cc
File src/d8.cc (right):

https://codereview.chromium.org/1828633003/diff/460001/src/d8.cc#newcode1289
src/d8.cc:1289: int kCountersTableRowSize =
On 2016/04/08 11:24:43, rmcilroy (Slow) wrote:
> static const

Done.

https://codereview.chromium.org/1828633003/diff/460001/src/d8.cc#newcode1297
src/d8.cc:1297: // Do not use an object, we need the order of entries to be
preserved
On 2016/04/08 11:24:43, rmcilroy (Slow) wrote:
> Not sure what this comment means? Also, could we have the output be a
dictionary
> of named target bytecode (only for those target bytecodes with a dispatch
count
> > 0) rather than an ordered list. Relying on the ordering of the source
> bytecodes to show the order of the unnamed target bytecode list seems brittle.

Replaced with dictionary of dictionaries.

https://codereview.chromium.org/1828633003/diff/460001/src/interpreter/interp...
File src/interpreter/interpreter-assembler.cc (right):

https://codereview.chromium.org/1828633003/diff/460001/src/interpreter/interp...
src/interpreter/interpreter-assembler.cc:670: void
InterpreterAssembler::TraceHandlerToHandlerBytecodeDispatch(
On 2016/04/08 11:24:44, rmcilroy (Slow) wrote:
> Just TraceBytecodeDispatch

Done.

https://codereview.chromium.org/1828633003/diff/460001/src/interpreter/interp...
src/interpreter/interpreter-assembler.cc:671: Node* target_index) {
On 2016/04/08 11:24:44, rmcilroy (Slow) wrote:
> /s/target_index/target_bytecode

Done.

https://codereview.chromium.org/1828633003/diff/460001/src/interpreter/interp...
src/interpreter/interpreter-assembler.cc:675: Node* handler_row_base =
IntPtrConstant(
On 2016/04/08 11:24:44, rmcilroy (Slow) wrote:
> source_bytecode_table_index

Done.

https://codereview.chromium.org/1828633003/diff/460001/src/interpreter/interp...
src/interpreter/interpreter-assembler.cc:684: void
InterpreterAssembler::SaturatedCounterIncrement(Node* counters_table,
On 2016/04/08 11:24:44, rmcilroy (Slow) wrote:
> Just do this inline above

Done.

https://codereview.chromium.org/1828633003/diff/460001/src/interpreter/interp...
src/interpreter/interpreter-assembler.cc:699: : MachineRepresentation::kWord32,
On 2016/04/08 11:24:43, rmcilroy (Slow) wrote:
> PointerRepresentation()

Done.

https://codereview.chromium.org/1828633003/diff/460001/src/interpreter/interp...
File src/interpreter/interpreter.cc (right):

https://codereview.chromium.org/1828633003/diff/460001/src/interpreter/interp...
src/interpreter/interpreter.cc:36: int kCountersTableRowSize =
static_cast<int>(Bytecode::kLast) + 1;
On 2016/04/08 11:24:44, rmcilroy (Slow) wrote:
> nit - move into if below. Also make static const.

Done.

https://codereview.chromium.org/1828633003/diff/460001/src/interpreter/interp...
File src/interpreter/interpreter.h (right):

https://codereview.chromium.org/1828633003/diff/460001/src/interpreter/interp...
src/interpreter/interpreter.h:62: static bool BytecodeHasHandler(Bytecode
bytecode, OperandScale operand_scale);
On 2016/04/08 11:24:44, rmcilroy (Slow) wrote:
> Remove

Done.

https://codereview.chromium.org/1828633003/diff/460001/src/interpreter/interp...
src/interpreter/interpreter.h:152: v8::base::SmartArrayPointer<uintptr_t>
handler_to_handler_dispatch_counters_;
On 2016/04/08 11:24:44, rmcilroy (Slow) wrote:
> /s/handler_to_handler_dispatch_counters/bytecode_dispatch_count_table

Done.

https://codereview.chromium.org/1828633003/diff/460001/tools/ignition_plot_ha...
File tools/ignition_plot_handler_counters.py (right):

https://codereview.chromium.org/1828633003/diff/460001/tools/ignition_plot_ha...
tools/ignition_plot_handler_counters.py:7: 
On 2016/04/08 11:24:44, rmcilroy (Slow) wrote:
> Name should be something like plot_ignition_bytecode_dispatches.py. Also,
should
> this be in a directory of ignition specific tracing tools (along with your
other
> tool when you land it)?

Done.

https://codereview.chromium.org/1828633003/diff/460001/tools/ignition_plot_ha...
tools/ignition_plot_handler_counters.py:11: #mpl.use('Agg')  ## Uncomment this
line if running headless
On 2016/04/08 11:24:44, rmcilroy (Slow) wrote:
> Remove?

It might be of some use if this script is run on a server or as part of some
automated process. Whether we keep it or not, it's the same for me, though.

About the weird positioning, this is because it must come before the other
matplotlib imports.

[Stefano Sanfilippo, 2016-04-08 16:24:36]

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

[commit-bot: I haz the power, 2016-04-08 16:24:39]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1828633003/580001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1828633003/580001

[rmcilroy, 2016-04-08 16:40:31]

LGTM once comments are addressed.

https://codereview.chromium.org/1828633003/diff/580001/src/d8.cc
File src/d8.cc (right):

https://codereview.chromium.org/1828633003/diff/580001/src/d8.cc#newcode1284
src/d8.cc:1284: uintptr_t* handler_to_handler_dispatch_counters =
update variable name.

https://codereview.chromium.org/1828633003/diff/580001/src/d8.cc#newcode1292
src/d8.cc:1292: stream << '{';
nit - brief comment explaining this emits the output as JSON  dictionary.

https://codereview.chromium.org/1828633003/diff/580001/src/d8.cc#newcode1302
src/d8.cc:1302: bool emitted_first_non_zero_cell_of_the_row = false;
This name is a bit long winded, how about just emitted_first

https://codereview.chromium.org/1828633003/diff/580001/src/interpreter/interp...
File src/interpreter/interpreter.cc (right):

https://codereview.chromium.org/1828633003/diff/580001/src/interpreter/interp...
src/interpreter/interpreter.cc:37: static const int kCountersTableRowSize =
kBytecodeCount would be a more appropriate name (given it is both the row and
column count). You could add this as a constant in Bytecodes.h if you like and
reuse it everywhere we currently use static_cast<int>(Bytecode::kLast) + 1.

[Stefano Sanfilippo, 2016-04-08 17:01:05]

All issues addressed. PT a last look.

https://codereview.chromium.org/1828633003/diff/580001/src/d8.cc
File src/d8.cc (right):

https://codereview.chromium.org/1828633003/diff/580001/src/d8.cc#newcode1284
src/d8.cc:1284: uintptr_t* handler_to_handler_dispatch_counters =
On 2016/04/08 16:40:31, rmcilroy wrote:
> update variable name.

Done.

https://codereview.chromium.org/1828633003/diff/580001/src/d8.cc#newcode1292
src/d8.cc:1292: stream << '{';
On 2016/04/08 16:40:31, rmcilroy wrote:
> nit - brief comment explaining this emits the output as JSON  dictionary.

Done.

https://codereview.chromium.org/1828633003/diff/580001/src/d8.cc#newcode1302
src/d8.cc:1302: bool emitted_first_non_zero_cell_of_the_row = false;
On 2016/04/08 16:40:31, rmcilroy wrote:
> This name is a bit long winded, how about just emitted_first

Done.

https://codereview.chromium.org/1828633003/diff/580001/src/interpreter/interp...
File src/interpreter/interpreter.cc (right):

https://codereview.chromium.org/1828633003/diff/580001/src/interpreter/interp...
src/interpreter/interpreter.cc:37: static const int kCountersTableRowSize =
On 2016/04/08 16:40:31, rmcilroy wrote:
> kBytecodeCount would be a more appropriate name (given it is both the row and
> column count). You could add this as a constant in Bytecodes.h if you like and
> reuse it everywhere we currently use static_cast<int>(Bytecode::kLast) + 1.

I second this proposal, however I fill it qualifies for a separate CL. For the
moment I am renaming the local variable, though.

[Stefano Sanfilippo, 2016-04-08 17:03:24]

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

[commit-bot: I haz the power, 2016-04-08 17:03:27]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1828633003/600001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1828633003/600001

[rmcilroy, 2016-04-08 17:15:47]

https://codereview.chromium.org/1828633003/diff/580001/src/interpreter/interp...
File src/interpreter/interpreter.cc (right):

https://codereview.chromium.org/1828633003/diff/580001/src/interpreter/interp...
src/interpreter/interpreter.cc:37: static const int kCountersTableRowSize =
On 2016/04/08 17:01:05, Stefano Sanfilippo wrote:
> On 2016/04/08 16:40:31, rmcilroy wrote:
> > kBytecodeCount would be a more appropriate name (given it is both the row
and
> > column count). You could add this as a constant in Bytecodes.h if you like
and
> > reuse it everywhere we currently use static_cast<int>(Bytecode::kLast) + 1.
> 
> I second this proposal, however I fill it qualifies for a separate CL. For the
> moment I am renaming the local variable, though.

Sure.

https://codereview.chromium.org/1828633003/diff/600001/src/d8.cc
File src/d8.cc (right):

https://codereview.chromium.org/1828633003/diff/600001/src/d8.cc#newcode1281
src/d8.cc:1281: void Shell::WriteInterpreterDispatchCounters(Isolate* isolate) {
Actually thinking about it, let's move this function to the interpreter class so
we can call it in Chrome in the future.

https://codereview.chromium.org/1828633003/diff/600001/src/d8.cc#newcode1290
src/d8.cc:1290: static_cast<int>(i::interpreter::Bytecode::kLast) + 1;
Update this constant name too.

[Stefano Sanfilippo, 2016-04-08 18:09:57]

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

[Stefano Sanfilippo, 2016-04-08 18:10:02]

https://codereview.chromium.org/1828633003/diff/600001/src/d8.cc
File src/d8.cc (right):

https://codereview.chromium.org/1828633003/diff/600001/src/d8.cc#newcode1281
src/d8.cc:1281: void Shell::WriteInterpreterDispatchCounters(Isolate* isolate) {
On 2016/04/08 17:15:47, rmcilroy wrote:
> Actually thinking about it, let's move this function to the interpreter class
so
> we can call it in Chrome in the future.

Done.

https://codereview.chromium.org/1828633003/diff/600001/src/d8.cc#newcode1290
src/d8.cc:1290: static_cast<int>(i::interpreter::Bytecode::kLast) + 1;
On 2016/04/08 17:15:47, rmcilroy wrote:
> Update this constant name too.

Done.

[commit-bot: I haz the power, 2016-04-08 18:10:07]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1828633003/620001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1828633003/620001

[commit-bot: I haz the power, 2016-04-08 18:34:24]

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

[commit-bot: I haz the power, 2016-04-08 18:34:25]

Dry run: This issue passed the CQ dry run.

[Stefano Sanfilippo, 2016-04-09 11:29:17]

Patchset #23 (id:620001) has been deleted

[Stefano Sanfilippo, 2016-04-09 16:34:57]

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

[commit-bot: I haz the power, 2016-04-09 16:35:08]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1828633003/640001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1828633003/640001

[commit-bot: I haz the power, 2016-04-09 17:06:11]

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

[commit-bot: I haz the power, 2016-04-09 17:06:13]

Dry run: This issue passed the CQ dry run.

[rmcilroy, 2016-04-11 09:20:33]

LGTM, but please update the title and description to be more informative.

[Stefano Sanfilippo, 2016-04-11 09:58:45]

Description was changed from

==========
[Interpreter] Add dispatch counters from each handler to each other.

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

to

==========
[Interpreter] Add flag to count dispatches from each handler to each other.

When FLAG_trace_ignition_dispatches is enabled, a dispatch counter is kept
for each pair of source-destination bytecode handlers.

Each counter saturates at max uintptr_t value.

Counters are dumped as a JSON-encoded object of objects, such that each
key on the top level object is a source bytecode name, and each key on
the corresponding value is a destination bytecode name, with the
associated counter as value. The output file name can be controlled with
the FLAG_trace_ignition_dispatches_output_file flag.

The JSON file may be written by calling
Interpreter::WriteDispatchCounters(), which is done for d8 in
Shell::OnExit, if the FLAG_trace_ignition_dispatches is enabled.

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

[Stefano Sanfilippo, 2016-04-11 10:00:32]

Description was changed from

==========
[Interpreter] Add flag to count dispatches from each handler to each other.

When FLAG_trace_ignition_dispatches is enabled, a dispatch counter is kept
for each pair of source-destination bytecode handlers.

Each counter saturates at max uintptr_t value.

Counters are dumped as a JSON-encoded object of objects, such that each
key on the top level object is a source bytecode name, and each key on
the corresponding value is a destination bytecode name, with the
associated counter as value. The output file name can be controlled with
the FLAG_trace_ignition_dispatches_output_file flag.

The JSON file may be written by calling
Interpreter::WriteDispatchCounters(), which is done for d8 in
Shell::OnExit, if the FLAG_trace_ignition_dispatches is enabled.

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

to

==========
[Interpreter] Add flag to count dispatches from each handler to each other.

When FLAG_trace_ignition_dispatches is enabled, a dispatch counter is
kept for each pair of source-destination bytecode handlers.

Each counter saturates at max uintptr_t value.

Counters are dumped as a JSON-encoded object of objects, such that
each key on the top level object is a source bytecode name, and each key
on the corresponding value is a destination bytecode name, with the
associated counter as value. The output file name can be controlled
with the FLAG_trace_ignition_dispatches_output_file flag.

The JSON file may be written by calling
Interpreter::WriteDispatchCounters(), which is done for d8 in
Shell::OnExit, if the FLAG_trace_ignition_dispatches is enabled.

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

[Stefano Sanfilippo, 2016-04-11 10:01:43]

On 2016/04/11 09:20:33, rmcilroy wrote:
> LGTM, but please update the title and description to be more informative.

Sorry, this was the old description when the CL was meant to be a follow-up to
the other.

Description updated, could you please give it a last check?

--S

[Stefano Sanfilippo, 2016-04-11 10:07:07]

Description was changed from

==========
[Interpreter] Add flag to count dispatches from each handler to each other.

When FLAG_trace_ignition_dispatches is enabled, a dispatch counter is
kept for each pair of source-destination bytecode handlers.

Each counter saturates at max uintptr_t value.

Counters are dumped as a JSON-encoded object of objects, such that
each key on the top level object is a source bytecode name, and each key
on the corresponding value is a destination bytecode name, with the
associated counter as value. The output file name can be controlled
with the FLAG_trace_ignition_dispatches_output_file flag.

The JSON file may be written by calling
Interpreter::WriteDispatchCounters(), which is done for d8 in
Shell::OnExit, if the FLAG_trace_ignition_dispatches is enabled.

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

to

==========
[Interpreter] Enable tracing of bytecode handler dispatches.

When FLAG_trace_ignition_dispatches is enabled, a dispatch counter is
kept for each pair of source-destination bytecode handlers.

Each counter saturates at max uintptr_t value.

Counters are dumped as a JSON-encoded object of objects, such that
each key on the top level object is a source bytecode name, and each key
on the corresponding value is a destination bytecode name, with the
associated counter as value. The output file name can be controlled
with the FLAG_trace_ignition_dispatches_output_file flag.

The JSON file may be written by calling
Interpreter::WriteDispatchCounters(), which is done for d8 in
Shell::OnExit, if the FLAG_trace_ignition_dispatches is enabled.

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

[Stefano Sanfilippo, 2016-04-11 10:21:05]

Description was changed from

==========
[Interpreter] Enable tracing of bytecode handler dispatches.

When FLAG_trace_ignition_dispatches is enabled, a dispatch counter is
kept for each pair of source-destination bytecode handlers.

Each counter saturates at max uintptr_t value.

Counters are dumped as a JSON-encoded object of objects, such that
each key on the top level object is a source bytecode name, and each key
on the corresponding value is a destination bytecode name, with the
associated counter as value. The output file name can be controlled
with the FLAG_trace_ignition_dispatches_output_file flag.

The JSON file may be written by calling
Interpreter::WriteDispatchCounters(), which is done for d8 in
Shell::OnExit, if the FLAG_trace_ignition_dispatches is enabled.

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

to

==========
[Interpreter] Enable tracing of bytecode handler dispatches.

When FLAG_trace_ignition_dispatches is enabled, a dispatch counter is
kept for each pair of source-destination bytecode handlers.

Each counter saturates at max uintptr_t value.

Counters are dumped as a JSON-encoded object of objects, such that
each key on the top level object is a source bytecode name, and each key
on the corresponding value is a destination bytecode name, with the
associated counter as value. The output file name can be controlled
with the FLAG_trace_ignition_dispatches_output_file flag.

The JSON file may be written by calling
Interpreter::WriteDispatchCounters(), which is done for d8 in
Shell::OnExit, if the FLAG_trace_ignition_dispatches is enabled.

BUG=v8:4899
LOG=N
==========

[Stefano Sanfilippo, 2016-04-11 10:27:56]

Description was changed from

==========
[Interpreter] Enable tracing of bytecode handler dispatches.

When FLAG_trace_ignition_dispatches is enabled, a dispatch counter is
kept for each pair of source-destination bytecode handlers.

Each counter saturates at max uintptr_t value.

Counters are dumped as a JSON-encoded object of objects, such that
each key on the top level object is a source bytecode name, and each key
on the corresponding value is a destination bytecode name, with the
associated counter as value. The output file name can be controlled
with the FLAG_trace_ignition_dispatches_output_file flag.

The JSON file may be written by calling
Interpreter::WriteDispatchCounters(), which is done for d8 in
Shell::OnExit, if the FLAG_trace_ignition_dispatches is enabled.

BUG=v8:4899
LOG=N
==========

to

==========
[Interpreter] Enable tracing of bytecode handler dispatches.

When FLAG_trace_ignition_dispatches is enabled, a dispatch counter is
kept for each pair of source-destination bytecode handlers.

Each counter saturates at max uintptr_t value.

Counters are dumped as a JSON-encoded object of objects, such that
each key on the top level object is a source bytecode name, and each key
on the corresponding value is a destination bytecode name, with the
associated counter as value. The output file name can be controlled
with the FLAG_trace_ignition_dispatches_output_file flag.

The JSON file may be written by calling
Interpreter::WriteDispatchCounters(), which is done for d8 in
Shell::OnExit, if FLAG_trace_ignition_dispatches is enabled.

BUG=v8:4899
LOG=N
==========

[Stefano Sanfilippo, 2016-04-11 10:33:04]

The CQ bit was checked by ssanfilippo@chromium.org

[commit-bot: I haz the power, 2016-04-11 10:33:12]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1828633003/640001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1828633003/640001

[commit-bot: I haz the power, 2016-04-11 11:57:40]

Description was changed from

==========
[Interpreter] Enable tracing of bytecode handler dispatches.

When FLAG_trace_ignition_dispatches is enabled, a dispatch counter is
kept for each pair of source-destination bytecode handlers.

Each counter saturates at max uintptr_t value.

Counters are dumped as a JSON-encoded object of objects, such that
each key on the top level object is a source bytecode name, and each key
on the corresponding value is a destination bytecode name, with the
associated counter as value. The output file name can be controlled
with the FLAG_trace_ignition_dispatches_output_file flag.

The JSON file may be written by calling
Interpreter::WriteDispatchCounters(), which is done for d8 in
Shell::OnExit, if FLAG_trace_ignition_dispatches is enabled.

BUG=v8:4899
LOG=N
==========

to

==========
[Interpreter] Enable tracing of bytecode handler dispatches.

When FLAG_trace_ignition_dispatches is enabled, a dispatch counter is
kept for each pair of source-destination bytecode handlers.

Each counter saturates at max uintptr_t value.

Counters are dumped as a JSON-encoded object of objects, such that
each key on the top level object is a source bytecode name, and each key
on the corresponding value is a destination bytecode name, with the
associated counter as value. The output file name can be controlled
with the FLAG_trace_ignition_dispatches_output_file flag.

The JSON file may be written by calling
Interpreter::WriteDispatchCounters(), which is done for d8 in
Shell::OnExit, if FLAG_trace_ignition_dispatches is enabled.

BUG=v8:4899
LOG=N
==========

[commit-bot: I haz the power, 2016-04-11 11:57:42]

Committed patchset #23 (id:640001)

[commit-bot: I haz the power, 2016-04-11 11:58:11]

Description was changed from

==========
[Interpreter] Enable tracing of bytecode handler dispatches.

When FLAG_trace_ignition_dispatches is enabled, a dispatch counter is
kept for each pair of source-destination bytecode handlers.

Each counter saturates at max uintptr_t value.

Counters are dumped as a JSON-encoded object of objects, such that
each key on the top level object is a source bytecode name, and each key
on the corresponding value is a destination bytecode name, with the
associated counter as value. The output file name can be controlled
with the FLAG_trace_ignition_dispatches_output_file flag.

The JSON file may be written by calling
Interpreter::WriteDispatchCounters(), which is done for d8 in
Shell::OnExit, if FLAG_trace_ignition_dispatches is enabled.

BUG=v8:4899
LOG=N
==========

to

==========
[Interpreter] Enable tracing of bytecode handler dispatches.

When FLAG_trace_ignition_dispatches is enabled, a dispatch counter is
kept for each pair of source-destination bytecode handlers.

Each counter saturates at max uintptr_t value.

Counters are dumped as a JSON-encoded object of objects, such that
each key on the top level object is a source bytecode name, and each key
on the corresponding value is a destination bytecode name, with the
associated counter as value. The output file name can be controlled
with the FLAG_trace_ignition_dispatches_output_file flag.

The JSON file may be written by calling
Interpreter::WriteDispatchCounters(), which is done for d8 in
Shell::OnExit, if FLAG_trace_ignition_dispatches is enabled.

BUG=v8:4899
LOG=N

Committed: https://crrev.com/1e3257d27f42ee1fd81aaf97536db171e7abd7fd
Cr-Commit-Position: refs/heads/master@{#35380}
==========

[commit-bot: I haz the power, 2016-04-11 11:58:12]

Patchset 23 (id:??) landed as
https://crrev.com/1e3257d27f42ee1fd81aaf97536db171e7abd7fd
Cr-Commit-Position: refs/heads/master@{#35380}