Fix some MultiplexRouter behavior for sync messages

Fix some MultiplexRouter behavior for sync messages

Two small fixes:
 - ProcessTasks was unintentionally scoping the
   interface id for a sync message task, so
   an unprocessed task would then get pushed
   back onto the queue for kInvalidInterfaceId
 - Unprocessed task pointer was being pushed back
   onto the sync message queue after moving the
   task ptr back into the task queue, so the sync
   message queue entry would always be null.

Also minor cleanup to ProcessFirstSyncMessageForEndpoint
to wipe out empty entries in sync_message_tasks_ when
possible.

R=yzshen@chromium.org
BUG=

Committed: https://crrev.com/b22a087001308e421c1c4b6fa7cf33d0b9b2d4e7
Cr-Commit-Position: refs/heads/master@{#397719}

[Ken Rockot(use gerrit already), 2016-06-03 07:22:36]

Had an a-ha moment while staring at the code for a few minutes :)

Fixing the id scoping revealed the second bug. This CL resolves the crashes my
SyncChannel patch was hitting, though there are still renderer deadlocks I need
to sort out even after fixing my associated interface setup as we discussed.

Not sure my "cleanup" in in ProcessFirstSyncMessageForEndpoint is really
necessary but thought it couldn't hurt. WDYT?

[yzshen1, 2016-06-03 15:30:18]

LGTM


Thanks! And sorry for the stupid bugs... The test cases for sync methods are all
single-threaded. So that path is never hit. Need more tests. :/

[Ken Rockot(use gerrit already), 2016-06-03 15:32:02]

The CQ bit was checked by rockot@chromium.org

[commit-bot: I haz the power, 2016-06-03 15:32:17]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/2035893003/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/2035893003/1

[commit-bot: I haz the power, 2016-06-03 16:09:17]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2016-06-03 16:11:06]

Description was changed from

==========
Fix some MultiplexRouter behavior for sync messages

Two small fixes:
 - ProcessTasks was unintentionally scoping the
   interface id for a sync message task, so
   an unprocessed task would then get pushed
   back onto the queue for kInvalidInterfaceId
 - Unprocessed task pointer was being pushed back
   onto the sync message queue after moving the
   task ptr back into the task queue, so the sync
   message queue entry would always be null.

Also minor cleanup to ProcessFirstSyncMessageForEndpoint
to wipe out empty entries in sync_message_tasks_ when
possible.

R=yzshen@chromium.org
BUG=
==========

to

==========
Fix some MultiplexRouter behavior for sync messages

Two small fixes:
 - ProcessTasks was unintentionally scoping the
   interface id for a sync message task, so
   an unprocessed task would then get pushed
   back onto the queue for kInvalidInterfaceId
 - Unprocessed task pointer was being pushed back
   onto the sync message queue after moving the
   task ptr back into the task queue, so the sync
   message queue entry would always be null.

Also minor cleanup to ProcessFirstSyncMessageForEndpoint
to wipe out empty entries in sync_message_tasks_ when
possible.

R=yzshen@chromium.org
BUG=

Committed: https://crrev.com/b22a087001308e421c1c4b6fa7cf33d0b9b2d4e7
Cr-Commit-Position: refs/heads/master@{#397719}
==========

[commit-bot: I haz the power, 2016-06-03 16:11:07]

Patchset 1 (id:??) landed as
https://crrev.com/b22a087001308e421c1c4b6fa7cf33d0b9b2d4e7
Cr-Commit-Position: refs/heads/master@{#397719}