Fixup ChromeOS enrollment and renderer p2p sockets

Fixup ChromeOS enrollment and renderer p2p sockets

Both of these codepaths mint new SSL socket pools from
existing URLRequestContexts, which is already buggy
and error prone (improperly handling channel ID and
TLS session caches). However, ensure they also copy
over the Certificate Transparency state, so that the
socket does not crash when it tries to verify the
Certificate Transparency information.

BUG=623463, 623619
Committed: https://crrev.com/0364ad9c02e4d0374ef6daebcd24e79cb10d7fd4
Cr-Commit-Position: refs/heads/master@{#402285}

[Ryan Sleevi, 2016-06-27 17:51:00]

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

[Ryan Sleevi, 2016-06-27 17:51:01]

Sergey: Separate from this review, would you be willing to create a bug so that
we can better handle socket separation for these features? I'm not sure what the
features are actually doing / how to test (it passed all waterfall tests), but
we should make sure that creating SSLSockets from the URLRequesContext doesn't
mutate / share any state.

[Sergey Ulanov, 2016-06-27 19:15:54]

LGTM

socket_host_tcp.cc is used for by WebRTC. It can use TLS to connect to TURN and
STUN servers. I've filed crbug.com/623660 to add more tests.
xmpp_client_socket_factory.cc is used for by Sync and CloudPrint, but my
understanding was that Sync stopped using XMPP by default, so I'm not sure why
it failed in that scenario.

How does SSLSocket creation mutates URLRequestContext?

On 2016/06/27 17:51:01, Ryan Sleevi wrote:
> Sergey: Separate from this review, would you be willing to create a bug so
that
> we can better handle socket separation for these features? I'm not sure what
the
> features are actually doing / how to test (it passed all waterfall tests), but
> we should make sure that creating SSLSockets from the URLRequesContext doesn't
> mutate / share any state.

[Ryan Sleevi, 2016-06-27 19:20:05]

On 2016/06/27 19:15:54, Sergey Ulanov wrote:
> How does SSLSocket creation mutates URLRequestContext?

It's a bit of a messy API.

The CertVerifier may cache things affected by inputs.
The ChannelID store would be mutated - and though these two places were
intentually nulling it, this may have been part of the cause for mismatched
channel IDs when accompanied with Sync / CloudPrint cookies.
The CTVerifier can end up mutating the profile (in tracking what
SignedCertificateTimestamps its seen)

Hopefully this will be the last of the immediate low-hanging fruit, and the
//net folks can discuss how to make this an easier to use API in the future.

[Ryan Sleevi, 2016-06-27 19:20:10]

The CQ bit was checked by rsleevi@chromium.org

[commit-bot: I haz the power, 2016-06-27 19:21:01]

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, 2016-06-27 19:34:28]

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

[commit-bot: I haz the power, 2016-06-27 19:34:29]

Try jobs failed on following builders:
  cast_shell_android on master.tryserver.chromium.android (JOB_FAILED,
https://build.chromium.org/p/tryserver.chromium.android/builders/cast_shell_a...)
  mac_chromium_compile_dbg_ng on master.tryserver.chromium.mac (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.mac/builders/mac_chromium_comp...)

[Ryan Sleevi, 2016-06-27 19:40:55]

The CQ bit was checked by rsleevi@chromium.org

[Ryan Sleevi, 2016-06-27 19:40:55]

The patchset sent to the CQ was uploaded after l-g-t-m from sergeyu@chromium.org
Link to the patchset: https://codereview.chromium.org/2104493002/#ps20001
(title: "Copypasta")

[commit-bot: I haz the power, 2016-06-27 19:41:24]

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, 2016-06-27 21:17:45]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2016-06-27 21:19:19]

Description was changed from

==========
Fixup ChromeOS enrollment and renderer p2p sockets

Both of these codepaths mint new SSL socket pools from
existing URLRequestContexts, which is already buggy
and error prone (improperly handling channel ID and
TLS session caches). However, ensure they also copy
over the Certificate Transparency state, so that the
socket does not crash when it tries to verify the
Certificate Transparency information.

BUG=623463, 623619
==========

to

==========
Fixup ChromeOS enrollment and renderer p2p sockets

Both of these codepaths mint new SSL socket pools from
existing URLRequestContexts, which is already buggy
and error prone (improperly handling channel ID and
TLS session caches). However, ensure they also copy
over the Certificate Transparency state, so that the
socket does not crash when it tries to verify the
Certificate Transparency information.

BUG=623463, 623619

Committed: https://crrev.com/0364ad9c02e4d0374ef6daebcd24e79cb10d7fd4
Cr-Commit-Position: refs/heads/master@{#402285}
==========

[commit-bot: I haz the power, 2016-06-27 21:19:20]

Patchset 2 (id:??) landed as
https://crrev.com/0364ad9c02e4d0374ef6daebcd24e79cb10d7fd4
Cr-Commit-Position: refs/heads/master@{#402285}