Mojo C++ bindings: fix MultiplexRouter and ChannelAssociatedGroupController.

Mojo C++ bindings: fix MultiplexRouter and ChannelAssociatedGroupController.

- There were a few places where a local ScopedInterfaceEndpointHandle was
  created for an endpoint but the handle_created field wasn't updated.

- Previously there was a race: when processing a message, the router's lock was
  released to create local ScopedInterfaceEndpointHandles for the transferred
  interface IDs. (Let's say it was going to register endpoint x.) At the same
  time, another thread could acquire the lock to process subsequent messages. It
  might find that a message targeted endpoint x but x was not registered. The
  previous code inserted endpoint x in this case. MultiplexRouter also marked
  the newly-inserted endpoint as CLOSED, which avoided blocking processing
  subsequent messages, but was wrong in this case.

  With this CL, endpoints are registered as soon as incoming messages arrives.
  It fixes the race and also simplifies IncomingMessageWrapper.

BUG=682334
Review-Url: https://codereview.chromium.org/2674483002
Cr-Commit-Position: refs/heads/master@{#447701}
Committed: https://chromium.googlesource.com/chromium/src/+/0a597131aec73063576b26e6caabde99b09ec890

[yzshen1, 2017-02-01 22:24:55]

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

[commit-bot: I haz the power, 2017-02-01 22:25:23]

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

[yzshen1, 2017-02-01 22:26:55]

Description was changed from

==========
Mojo C++ bindings: fix MultiplexRouter and ChannelAssociatedGroupController.

- There were a few places where a local ScopedInterfaceEndpointHandle was
  created for an endpoint but the handle_created field wasn't updated.

- Previously there was a race: when processing a message, the router's lock is
  released to create local ScopedInterfaceEndpointHandles for the transferred
  interface IDs. (Let's say it is going to register endpoint x.) At the same
  time, another thread could acquire the lock to process subsequent messages. It
  may find that a message targets endpoint x but x is not registered. The
  previous code inserted endpoint x in this case. MultiplexRouter also marked
  the newly-inserted endpoint as CLOSED, which avoided blocking processing
  subsequent messages, but was wrong in this case.

  With this CL, endpoints are registered as soon as incoming messages arrives.
  It fixes the race and also simplifies IncomingMessageWrapper.

BUG=682334
==========

to

==========
Mojo C++ bindings: fix MultiplexRouter and ChannelAssociatedGroupController.

- There were a few places where a local ScopedInterfaceEndpointHandle was
  created for an endpoint but the handle_created field wasn't updated.

- Previously there was a race: when processing a message, the router's lock was
  released to create local ScopedInterfaceEndpointHandles for the transferred
  interface IDs. (Let's say it was going to register endpoint x.) At the same
  time, another thread could acquire the lock to process subsequent messages. It
  might find that a message targeted endpoint x but x was not registered. The
  previous code inserted endpoint x in this case. MultiplexRouter also marked
  the newly-inserted endpoint as CLOSED, which avoided blocking processing
  subsequent messages, but was wrong in this case.

  With this CL, endpoints are registered as soon as incoming messages arrives.
  It fixes the race and also simplifies IncomingMessageWrapper.

BUG=682334
==========

[yzshen1, 2017-02-01 22:26:55]

yzshen@chromium.org changed reviewers:
+ rockot@chromium.org

[yzshen1, 2017-02-01 23:07:05]

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

[commit-bot: I haz the power, 2017-02-01 23:07:14]

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

[yzshen1, 2017-02-01 23:07:19]

Hi, Ken.

Would you please take a look? Thanks!

[commit-bot: I haz the power, 2017-02-02 00:28:13]

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

[commit-bot: I haz the power, 2017-02-02 00:28:14]

Dry run: This issue passed the CQ dry run.

[Ken Rockot(use gerrit already), 2017-02-02 01:50:19]

lgtm

[yzshen1, 2017-02-02 03:25:48]

The CQ bit was checked by yzshen@chromium.org

[commit-bot: I haz the power, 2017-02-02 03:26:30]

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, 2017-02-02 05:13:53]

CQ is committing da patch.

Bot data: {"patchset_id": 20001, "attempt_start_ts": 1486005948398130,
"parent_rev": "cfc91f159f1635c66d0e2f82ba4039c024019e97", "commit_rev":
"0a597131aec73063576b26e6caabde99b09ec890"}

[commit-bot: I haz the power, 2017-02-02 05:14:23]

Description was changed from

==========
Mojo C++ bindings: fix MultiplexRouter and ChannelAssociatedGroupController.

- There were a few places where a local ScopedInterfaceEndpointHandle was
  created for an endpoint but the handle_created field wasn't updated.

- Previously there was a race: when processing a message, the router's lock was
  released to create local ScopedInterfaceEndpointHandles for the transferred
  interface IDs. (Let's say it was going to register endpoint x.) At the same
  time, another thread could acquire the lock to process subsequent messages. It
  might find that a message targeted endpoint x but x was not registered. The
  previous code inserted endpoint x in this case. MultiplexRouter also marked
  the newly-inserted endpoint as CLOSED, which avoided blocking processing
  subsequent messages, but was wrong in this case.

  With this CL, endpoints are registered as soon as incoming messages arrives.
  It fixes the race and also simplifies IncomingMessageWrapper.

BUG=682334
==========

to

==========
Mojo C++ bindings: fix MultiplexRouter and ChannelAssociatedGroupController.

- There were a few places where a local ScopedInterfaceEndpointHandle was
  created for an endpoint but the handle_created field wasn't updated.

- Previously there was a race: when processing a message, the router's lock was
  released to create local ScopedInterfaceEndpointHandles for the transferred
  interface IDs. (Let's say it was going to register endpoint x.) At the same
  time, another thread could acquire the lock to process subsequent messages. It
  might find that a message targeted endpoint x but x was not registered. The
  previous code inserted endpoint x in this case. MultiplexRouter also marked
  the newly-inserted endpoint as CLOSED, which avoided blocking processing
  subsequent messages, but was wrong in this case.

  With this CL, endpoints are registered as soon as incoming messages arrives.
  It fixes the race and also simplifies IncomingMessageWrapper.

BUG=682334

Review-Url: https://codereview.chromium.org/2674483002
Cr-Commit-Position: refs/heads/master@{#447701}
Committed:
https://chromium.googlesource.com/chromium/src/+/0a597131aec73063576b26e6caab...
==========

[commit-bot: I haz the power, 2017-02-02 05:14:24]

Committed patchset #2 (id:20001) as
https://chromium.googlesource.com/chromium/src/+/0a597131aec73063576b26e6caab...