[Android] Use a "unique" remote debugging socket name on bind failure

[Android] Use a "unique" remote debugging socket name on bind failure

When socket bind failure happens on enabling remote web debugging,
retry using a socket name with PID suffix to ensure uniquiness.
This allows several channels of Chrome to have remote web debugging
enabled simultaneously. Also, this preserves backwards compatibility,
as in the case of the single Chrome instance on a device, remote
debugging socket name is unchanged.

It seems easier to add retry functionality to UnixDomainSocket,
as DevToolsHttpHandler creation is heavily asynchronous.

BUG=222338
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=203132

[mnaganov (inactive), 2013-05-29 11:01:15]

PTAL, I hope this solution is compliant to our prior discussions.

[pfeldman, 2013-05-29 11:48:25]

lgtm

[mnaganov (inactive), 2013-05-29 12:05:53]

Requesting an OWNERS' review:

bulach@: chrome/browser/android/
szym@: net/socket/

[bulach, 2013-05-29 12:34:54]

lgtm

[szym, 2013-05-29 18:49:04]

https://codereview.chromium.org/16093005/diff/1/net/socket/unix_domain_socket...
File net/socket/unix_domain_socket_posix_unittest.cc (right):

https://codereview.chromium.org/16093005/diff/1/net/socket/unix_domain_socket...
net/socket/unix_domain_socket_posix_unittest.cc:265: kSocketFilename, "",
socket_delegate_.get(), MakeAuthCallback());
Do not use kSocketFilename alone. Use MakeSocketPath() which appends
kSocketFilename to a temporary directory. Maybe extend MakeSocketPath to take a
suffix argument.

[mnaganov (inactive), 2013-05-29 20:11:10]

https://codereview.chromium.org/16093005/diff/1/net/socket/unix_domain_socket...
File net/socket/unix_domain_socket_posix_unittest.cc (right):

https://codereview.chromium.org/16093005/diff/1/net/socket/unix_domain_socket...
net/socket/unix_domain_socket_posix_unittest.cc:265: kSocketFilename, "",
socket_delegate_.get(), MakeAuthCallback());
On 2013/05/29 18:49:04, szym wrote:
> Do not use kSocketFilename alone. Use MakeSocketPath() which appends
> kSocketFilename to a temporary directory. Maybe extend MakeSocketPath to take
a
> suffix argument.

But that's for abstract namespace names only. Is full socket path also required
in this case?

[szym, 2013-05-29 20:13:53]

https://codereview.chromium.org/16093005/diff/1/net/socket/unix_domain_socket...
File net/socket/unix_domain_socket_posix_unittest.cc (right):

https://codereview.chromium.org/16093005/diff/1/net/socket/unix_domain_socket...
net/socket/unix_domain_socket_posix_unittest.cc:265: kSocketFilename, "",
socket_delegate_.get(), MakeAuthCallback());
On 2013/05/29 20:11:11, Mikhail Naganov (Chromium) wrote:
> On 2013/05/29 18:49:04, szym wrote:
> > Do not use kSocketFilename alone. Use MakeSocketPath() which appends
> > kSocketFilename to a temporary directory. Maybe extend MakeSocketPath to
take
> a
> > suffix argument.
> 
> But that's for abstract namespace names only. Is full socket path also
required
> in this case?

The concern is interference with other tests running in parallel. Each test
instance should use a unique name.

[mnaganov (inactive), 2013-05-29 20:37:11]

https://codereview.chromium.org/16093005/diff/1/net/socket/unix_domain_socket...
File net/socket/unix_domain_socket_posix_unittest.cc (right):

https://codereview.chromium.org/16093005/diff/1/net/socket/unix_domain_socket...
net/socket/unix_domain_socket_posix_unittest.cc:265: kSocketFilename, "",
socket_delegate_.get(), MakeAuthCallback());
On 2013/05/29 20:13:53, szym wrote:
> On 2013/05/29 20:11:11, Mikhail Naganov (Chromium) wrote:
> > On 2013/05/29 18:49:04, szym wrote:
> > > Do not use kSocketFilename alone. Use MakeSocketPath() which appends
> > > kSocketFilename to a temporary directory. Maybe extend MakeSocketPath to
> take
> > a
> > > suffix argument.
> > 
> > But that's for abstract namespace names only. Is full socket path also
> required
> > in this case?
> 
> The concern is interference with other tests running in parallel. Each test
> instance should use a unique name.

Makes sense. Fixed!

[szym, 2013-05-29 20:38:38]

lgtm

[commit-bot: I haz the power, 2013-05-30 06:03:25]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/mnaganov@chromium.org/16093005/16001

[commit-bot: I haz the power, 2013-05-30 12:38:41]

Change committed as 203132