Stop asynchronously deleting the network manager in P2PPortAllocator.

Stop asynchronously deleting the network manager in P2PPortAllocator.

WebRTC now calls the PortAllocator destructor from the correct thread,
so this is no longer necessary.

Committed: https://crrev.com/874f972df6d281501f8e77cf40186a2b58ab3b88
Cr-Commit-Position: refs/heads/master@{#397785}

[Taylor_Brandstetter, 2016-05-13 18:21:02]

deadbeef@chromium.org changed reviewers:
+ tommi@chromium.org

[Taylor_Brandstetter, 2016-05-13 18:22:10]

deadbeef@chromium.org changed reviewers:
+ sergeyu@chromium.org

[Taylor_Brandstetter, 2016-05-13 18:22:10]

sergeyu@chromium.org: Please review changes in p2p

[tommi (sloooow) - chröme, 2016-05-13 19:01:28]

lgtm

[Sergey Ulanov, 2016-05-13 19:43:33]

lgtm

[Sergey Ulanov, 2016-05-13 19:44:21]

https://codereview.chromium.org/1976263002/diff/1/content/renderer/p2p/port_a...
File content/renderer/p2p/port_allocator.cc (right):

https://codereview.chromium.org/1976263002/diff/1/content/renderer/p2p/port_a...
content/renderer/p2p/port_allocator.cc:52: P2PPortAllocator::~P2PPortAllocator()
{}
Maybe add ThreadChecker in P2PPortAllocator to make sure it's created and
destroyed on the same thread.

[Taylor_Brandstetter, 2016-05-13 19:57:45]

On 2016/05/13 19:44:21, Sergey Ulanov wrote:
>
https://codereview.chromium.org/1976263002/diff/1/content/renderer/p2p/port_a...
> File content/renderer/p2p/port_allocator.cc (right):
> 
>
https://codereview.chromium.org/1976263002/diff/1/content/renderer/p2p/port_a...
> content/renderer/p2p/port_allocator.cc:52:
P2PPortAllocator::~P2PPortAllocator()
> {}
> Maybe add ThreadChecker in P2PPortAllocator to make sure it's created and
> destroyed on the same thread.

Currently it's created on a different thread, though, in
peer_connection_dependency_factory.cc.

tommi, what do you think is the best way to fix this? Should we go back to
having a PortAllocatorFactory, this time for the sole purpose of constructing a
PortAllocator on the correct thread? Or cam we just do something similar to
CreateIpcNetworkManagerOnWorkerThread (which has the drawback of requiring an
extra thread hop)?

[tommi, 2016-05-13 20:18:17]

On 2016/05/13 19:57:45, Taylor_Brandstetter wrote:
> On 2016/05/13 19:44:21, Sergey Ulanov wrote:
> >
>
https://codereview.chromium.org/1976263002/diff/1/content/renderer/p2p/port_a...
> > File content/renderer/p2p/port_allocator.cc (right):
> > 
> >
>
https://codereview.chromium.org/1976263002/diff/1/content/renderer/p2p/port_a...
> > content/renderer/p2p/port_allocator.cc:52:
> P2PPortAllocator::~P2PPortAllocator()
> > {}
> > Maybe add ThreadChecker in P2PPortAllocator to make sure it's created and
> > destroyed on the same thread.
> 
> Currently it's created on a different thread, though, in
> peer_connection_dependency_factory.cc.
> 
> tommi, what do you think is the best way to fix this? Should we go back to
> having a PortAllocatorFactory, this time for the sole purpose of constructing
a
> PortAllocator on the correct thread? Or cam we just do something similar to
> CreateIpcNetworkManagerOnWorkerThread (which has the drawback of requiring an
> extra thread hop)?

I think the CreateIpcNetworkManagerOnWorkerThread is OK for now.  The worker
thread concept will change at some point and it will be good to be able to
assume single-threadedness in this class as we adopt to whatever we'll end up
using.

[Taylor_Brandstetter, 2016-06-03 18:26:30]

The CQ bit was checked by deadbeef@chromium.org

[commit-bot: I haz the power, 2016-06-03 18:26:59]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1976263002/1

[commit-bot: I haz the power, 2016-06-03 20:07:39]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2016-06-03 20:08:54]

Description was changed from

==========
Stop asynchronously deleting the network manager in P2PPortAllocator.

WebRTC now calls the PortAllocator destructor from the correct thread,
so this is no longer necessary.
==========

to

==========
Stop asynchronously deleting the network manager in P2PPortAllocator.

WebRTC now calls the PortAllocator destructor from the correct thread,
so this is no longer necessary.

Committed: https://crrev.com/874f972df6d281501f8e77cf40186a2b58ab3b88
Cr-Commit-Position: refs/heads/master@{#397785}
==========

[commit-bot: I haz the power, 2016-06-03 20:08:55]

Patchset 1 (id:??) landed as
https://crrev.com/874f972df6d281501f8e77cf40186a2b58ab3b88
Cr-Commit-Position: refs/heads/master@{#397785}