Mojo: add mojo_message_pipe_perftests

Mojo: add mojo_message_pipe_perftests

This is analogous to ipc_perftests. It measures roundtrip
time with various message sizes.

TEST=message_pipe_perftest.cc
R=viettrungluu@chromium.org
BUG=none

Committed: https://crrev.com/5570dc0c4594f5466ef6846875130336ab02c486
Cr-Commit-Position: refs/heads/master@{#292024}

[Hajime Morrita, 2014-08-25 17:52:32]

I'll update the benchmark note to add result, but in summary, it is about

 - 200% slower for smaller message size
 -  20% slower for larger one

[viettrungluu, 2014-08-25 20:32:46]

LGTM w/nits & comments.

https://codereview.chromium.org/501983003/diff/1/mojo/system/message_pipe_per...
File mojo/system/message_pipe_perftest.cc (right):

https://codereview.chromium.org/501983003/diff/1/mojo/system/message_pipe_per...
mojo/system/message_pipe_perftest.cc:1: // Copyright 2013 The Chromium Authors.
All rights reserved.
nit: 2014

https://codereview.chromium.org/501983003/diff/1/mojo/system/message_pipe_per...
mojo/system/message_pipe_perftest.cc:20: #include "build/build_config.h"  //
TODO(vtl): Remove this.
I don't think you need this include.

https://codereview.chromium.org/501983003/diff/1/mojo/system/message_pipe_per...
mojo/system/message_pipe_perftest.cc:41: void SetupMeasurement(int
message_count, size_t message_size) {
nit: "Setup" -> "SetUp" (the verb is always "set up", as two words).

https://codereview.chromium.org/501983003/diff/1/mojo/system/message_pipe_per...
mojo/system/message_pipe_perftest.cc:156: // Sends a bunch of messages to the
child. Expects them "repeated" back. Waits
This description isn't entirely accurate. This test is really a latency test,
since it waits for the reply before sending the next one.

(Also, in contrast to ipc_perf_tests, this test necessitates a thread-hop on the
receive side, whereas ipc_perf_tests does everything on the I/O thread. This
isn't entirely unfair, the Mojo design doesn't currently really let you do
anything on the I/O thread, but it kind of pits the native IPC's best case
(handle everything on the I/O -- or test main -- thread) versus the Mojo normal
case. It'd be interesting to have a version of the native IPC where the
sender/receiver are on a different thread.)

https://codereview.chromium.org/501983003/diff/1/mojo/system/message_pipe_per...
mojo/system/message_pipe_perftest.cc:158:
TEST_F(MultiprocessMessagePipePerfTest, Performance) {
Possibly you should have a slightly more descriptive name than "Performance".
Maybe it should be "PingPong" or something like that.

https://codereview.chromium.org/501983003/diff/1/mojo/system/message_pipe_tes...
File mojo/system/message_pipe_test_utils.h (right):

https://codereview.chromium.org/501983003/diff/1/mojo/system/message_pipe_tes...
mojo/system/message_pipe_test_utils.h:7: 
#include "base/macros.h" for DISALLOW_COPY_AND_ASSIGN.

[Hajime Morrita, 2014-08-26 18:16:26]

Thanks for the review! Landing.

https://codereview.chromium.org/501983003/diff/1/mojo/system/message_pipe_per...
File mojo/system/message_pipe_perftest.cc (right):

https://codereview.chromium.org/501983003/diff/1/mojo/system/message_pipe_per...
mojo/system/message_pipe_perftest.cc:1: // Copyright 2013 The Chromium Authors.
All rights reserved.
On 2014/08/25 20:32:45, viettrungluu wrote:
> nit: 2014

Done.

https://codereview.chromium.org/501983003/diff/1/mojo/system/message_pipe_per...
mojo/system/message_pipe_perftest.cc:20: #include "build/build_config.h"  //
TODO(vtl): Remove this.
On 2014/08/25 20:32:45, viettrungluu wrote:
> I don't think you need this include.

Done.

https://codereview.chromium.org/501983003/diff/1/mojo/system/message_pipe_per...
mojo/system/message_pipe_perftest.cc:41: void SetupMeasurement(int
message_count, size_t message_size) {
On 2014/08/25 20:32:46, viettrungluu wrote:
> nit: "Setup" -> "SetUp" (the verb is always "set up", as two words).

Done.

https://codereview.chromium.org/501983003/diff/1/mojo/system/message_pipe_per...
mojo/system/message_pipe_perftest.cc:156: // Sends a bunch of messages to the
child. Expects them "repeated" back. Waits
On 2014/08/25 20:32:45, viettrungluu wrote:
> This description isn't entirely accurate. This test is really a latency test,
> since it waits for the reply before sending the next one.
> 
> (Also, in contrast to ipc_perf_tests, this test necessitates a thread-hop on
the
> receive side, whereas ipc_perf_tests does everything on the I/O thread. This
> isn't entirely unfair, the Mojo design doesn't currently really let you do
> anything on the I/O thread, but it kind of pits the native IPC's best case
> (handle everything on the I/O -- or test main -- thread) versus the Mojo
normal
> case. It'd be interesting to have a version of the native IPC where the
> sender/receiver are on a different thread.)

That's true. Updated the comment. I'll add ChannelProxy-based benchmark next.

https://codereview.chromium.org/501983003/diff/1/mojo/system/message_pipe_per...
mojo/system/message_pipe_perftest.cc:158:
TEST_F(MultiprocessMessagePipePerfTest, Performance) {
On 2014/08/25 20:32:45, viettrungluu wrote:
> Possibly you should have a slightly more descriptive name than "Performance".
> Maybe it should be "PingPong" or something like that.

Done.

[Hajime Morrita, 2014-08-26 18:16:47]

The CQ bit was checked by morrita@chromium.org

[commit-bot: I haz the power, 2014-08-26 18:18:41]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/morrita@chromium.org/501983003/20001

[commit-bot: I haz the power, 2014-08-26 19:30:00]

FYI, CQ is re-trying this CL (attempt #1).
The failing builders are:
  win_chromium_x64_rel_swarming on tryserver.chromium.win
(http://build.chromium.org/p/tryserver.chromium.win/builders/win_chromium_x64_...)

[commit-bot: I haz the power, 2014-08-26 19:39:43]

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

[commit-bot: I haz the power, 2014-08-26 19:39:46]

Try jobs failed on following builders:
  win_chromium_x64_rel_swarming on tryserver.chromium.win
(http://build.chromium.org/p/tryserver.chromium.win/builders/win_chromium_x64_...)

[Hajime Morrita, 2014-08-26 20:37:53]

The CQ bit was checked by morrita@chromium.org

[commit-bot: I haz the power, 2014-08-26 20:39:11]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/morrita@chromium.org/501983003/40001

[commit-bot: I haz the power, 2014-08-26 22:20:35]

FYI, CQ is re-trying this CL (attempt #1).
The failing builders are:
  ios_rel_device on tryserver.chromium.mac
(http://build.chromium.org/p/tryserver.chromium.mac/builders/ios_rel_device/bu...)
  mac_chromium_rel_swarming on tryserver.chromium.mac
(http://build.chromium.org/p/tryserver.chromium.mac/builders/mac_chromium_rel_...)
  win_chromium_compile_dbg on tryserver.chromium.win
(http://build.chromium.org/p/tryserver.chromium.win/builders/win_chromium_comp...)

[commit-bot: I haz the power, 2014-08-26 23:13:11]

Committed patchset #3 (40001) as a9865e8a1771cf6da2c2894d2c0202af0f327447

[commit-bot: I haz the power, 2014-09-10 02:47:09]

Patchset 3 (id:??) landed as
https://crrev.com/5570dc0c4594f5466ef6846875130336ab02c486
Cr-Commit-Position: refs/heads/master@{#292024}