Use ChannelMojo between the remoting daemon and network processes.

Use ChannelMojo between the remoting daemon and network processes.

This refactors LaunchProcessWithToken to take an explicit list of
handles for the child to inherit (copied from the equivalent process
launch code in //base).

BUG=604282
Committed: https://crrev.com/41c99b466b9bbba9dcb002e8cb00f1d7da22dff4
Committed: https://crrev.com/e5d78a2c2a7e226ebfd531f0c42ba39759f00043
Cr-Original-Commit-Position: refs/heads/master@{#427238}
Cr-Commit-Position: refs/heads/master@{#427843}

[Sam McNally, 2016-10-18 07:02:07]

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

[commit-bot: I haz the power, 2016-10-18 07:02:16]

Dry run: 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, 2016-10-18 08:14:53]

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

[commit-bot: I haz the power, 2016-10-18 08:14:53]

Dry run: This issue passed the CQ dry run.

[Sam McNally, 2016-10-19 05:50:45]

Patchset #1 (id:1) has been deleted

[Sam McNally, 2016-10-19 05:50:52]

Patchset #1 (id:20001) has been deleted

[Sam McNally, 2016-10-19 05:51:46]

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

[commit-bot: I haz the power, 2016-10-19 05:53:54]

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

[Sam McNally, 2016-10-19 06:05:54]

Description was changed from

==========
Use ChannelMojo between the remoting daemon and network processes.

BUG=604282
==========

to

==========
Use ChannelMojo between the remoting daemon and network processes.

This refactors LaunchProcessWithToken to take an explicit list of
handles for the child to inherit (copied from the equivalent process
launch code in //base).

This also changes WorkerProcessLauncherTest to use ChannelMojo.

BUG=604282
==========

[Sam McNally, 2016-10-19 06:29:11]

Patchset #1 (id:40001) has been deleted

[Sam McNally, 2016-10-19 06:37:26]

Patchset #1 (id:60001) has been deleted

[Sam McNally, 2016-10-19 07:47:38]

sammc@chromium.org changed reviewers:
+ rockot@chromium.org, sergeyu@chromium.org

[Sam McNally, 2016-10-19 07:47:39]

+sergeyu for //remoting

+rockot for DEPS changes and use of mojo

[Sergey Ulanov, 2016-10-20 17:21:29]

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

[Sergey Ulanov, 2016-10-20 17:21:30]

+joedow@ for second pair of eyes

looks good, but I have poor understanding of how the new Mojo-based channel
works and if there are any security implications. Particularly I'm not sure
about the role of the pipe token. Is there a design doc to read about it?

https://codereview.chromium.org/2424353002/diff/80001/remoting/host/switches.cc
File remoting/host/switches.cc (right):

https://codereview.chromium.org/2424353002/diff/80001/remoting/host/switches....
remoting/host/switches.cc:27: const char kMojoPipeToken[] = "mojo-pipe-token";
Is this token used to limit access to the pipe? If so is it a problem that other
processes will be able to see the command line?

https://codereview.chromium.org/2424353002/diff/80001/remoting/host/win/unpri...
File remoting/host/win/unprivileged_process_delegate.cc (right):

https://codereview.chromium.org/2424353002/diff/80001/remoting/host/win/unpri...
remoting/host/win/unprivileged_process_delegate.cc:279: base::CommandLine
command_line(target_command_->argv());
nit: Move this down, closer to where it's used.

[joedow, 2016-10-20 17:40:04]

Adding on to Sergey's comments.

Thanks for getting the IPC -> Mojo conversion process started!

Beyond the comments below, have you built a Windows MSI and tried this yet
(tested beyond running the unit tests).  If not I can help get one set up for
testing.

https://codereview.chromium.org/2424353002/diff/80001/remoting/base/auto_thre...
File remoting/base/auto_thread_task_runner.h (right):

https://codereview.chromium.org/2424353002/diff/80001/remoting/base/auto_thre...
remoting/base/auto_thread_task_runner.h:37: const
scoped_refptr<base::SingleThreadTaskRunner>& underlying_task_runner() {
What is the problem when passing an instance of this class directly to Mojo
(since it overrides SingleThreadTaskRunner)?  

I'd probably just call this member 'task_runner()'.  It feels odd to indicate in
the property name that encapsulation is being broken by stating 'underlying'.

https://codereview.chromium.org/2424353002/diff/80001/remoting/host/daemon_pr...
File remoting/host/daemon_process_win.cc (right):

https://codereview.chromium.org/2424353002/diff/80001/remoting/host/daemon_pr...
remoting/host/daemon_process_win.cc:128: // to shut down.
I'd move this comment to the declaration of the member var as it seems out of
place in the initializer list.  Can you make the comment a bit more clear as to
which 'it' should be ignored.

https://codereview.chromium.org/2424353002/diff/80001/remoting/host/remoting_...
File remoting/host/remoting_me2me_host.cc (right):

https://codereview.chromium.org/2424353002/diff/80001/remoting/host/remoting_...
remoting/host/remoting_me2me_host.cc:463: // shut down.
Copy/paste comment from the other class, is there any direct action/non-action
that falls out from this behavior?  As someone first reading through this code,
I'm not sure why this is important to note since nothing seems to have been
added or removed to adjust the service shutdown behavior.

https://codereview.chromium.org/2424353002/diff/80001/remoting/host/win/launc...
File remoting/host/win/launch_process_with_token.cc (right):

https://codereview.chromium.org/2424353002/diff/80001/remoting/host/win/launc...
remoting/host/win/launch_process_with_token.cc:154: DPLOG(ERROR);
PLOG(ERROR) here instead.  We often troubleshoot machines in the field using ETW
logs and skipping  logs for Windows functions that might fail makes debugging
more difficult.

https://codereview.chromium.org/2424353002/diff/80001/remoting/host/win/launc...
remoting/host/win/launch_process_with_token.cc:162: DPLOG(ERROR);
PLOG(ERROR)

https://codereview.chromium.org/2424353002/diff/80001/remoting/host/win/worke...
File remoting/host/win/worker_process_launcher_unittest.cc (left):

https://codereview.chromium.org/2424353002/diff/80001/remoting/host/win/worke...
remoting/host/win/worker_process_launcher_unittest.cc:47: const char
kIpcSecurityDescriptor[] = "D:(A;;GA;;;AU)";
How is the Mojo pipe secured?  Is there a default DACL applied?

https://codereview.chromium.org/2424353002/diff/80001/remoting/host/win/wts_s...
File remoting/host/win/wts_session_process_delegate.cc (right):

https://codereview.chromium.org/2424353002/diff/80001/remoting/host/win/wts_s...
remoting/host/win/wts_session_process_delegate.cc:419: session_token_.Get(),
security_attributes.get(), nullptr, {},
add some comments for the hard-coded params like /*inherited_handles=*/={}

[Sam McNally, 2016-10-20 22:40:23]

> On 2016/10/20 17:40:04, joedow wrote:
> Adding on to Sergey's comments.
> 
> Thanks for getting the IPC -> Mojo conversion process started!
> 
> Beyond the comments below, have you built a Windows MSI and tried this yet
> (tested beyond running the unit tests).  If not I can help get one set up for
> testing.
> 
Yes.

https://codereview.chromium.org/2424353002/diff/80001/remoting/base/auto_thre...
File remoting/base/auto_thread_task_runner.h (right):

https://codereview.chromium.org/2424353002/diff/80001/remoting/base/auto_thre...
remoting/base/auto_thread_task_runner.h:37: const
scoped_refptr<base::SingleThreadTaskRunner>& underlying_task_runner() {
On 2016/10/20 17:40:03, joedow wrote:
> What is the problem when passing an instance of this class directly to Mojo
> (since it overrides SingleThreadTaskRunner)?  
> 
> I'd probably just call this member 'task_runner()'.  It feels odd to indicate
in
> the property name that encapsulation is being broken by stating 'underlying'.

The mojo system holds onto its IO task runner forever so the IO thread never
shuts down and because the IO thread shutdown is set to run on the main thread,
it never stops and so the remoting host never shuts down.

https://codereview.chromium.org/2424353002/diff/80001/remoting/host/daemon_pr...
File remoting/host/daemon_process_win.cc (right):

https://codereview.chromium.org/2424353002/diff/80001/remoting/host/daemon_pr...
remoting/host/daemon_process_win.cc:128: // to shut down.
On 2016/10/20 17:40:03, joedow wrote:
> I'd move this comment to the declaration of the member var as it seems out of
> place in the initializer list.  Can you make the comment a bit more clear as
to
> which 'it' should be ignored.

Done.

https://codereview.chromium.org/2424353002/diff/80001/remoting/host/remoting_...
File remoting/host/remoting_me2me_host.cc (right):

https://codereview.chromium.org/2424353002/diff/80001/remoting/host/remoting_...
remoting/host/remoting_me2me_host.cc:463: // shut down.
On 2016/10/20 17:40:04, joedow wrote:
> Copy/paste comment from the other class, is there any direct action/non-action
> that falls out from this behavior?  As someone first reading through this
code,
> I'm not sure why this is important to note since nothing seems to have been
> added or removed to adjust the service shutdown behavior.

Done.

https://codereview.chromium.org/2424353002/diff/80001/remoting/host/switches.cc
File remoting/host/switches.cc (right):

https://codereview.chromium.org/2424353002/diff/80001/remoting/host/switches....
remoting/host/switches.cc:27: const char kMojoPipeToken[] = "mojo-pipe-token";
On 2016/10/20 17:21:30, Sergey Ulanov wrote:
> Is this token used to limit access to the pipe? If so is it a problem that
other
> processes will be able to see the command line?

Establishing a usable mojo connection between two processes has two layers. The
first establishes the physical connection over a pipe/socket; the second
(glossing over some details) has each process create one end of a message pipe
associated with a shared string, which the child sends to the parent to glue the
ends together. This token is that string.

The token could be used by another process, but only if it already had a
connection to the parent process.

The best reference I know if for this is the comments in
https://cs.chromium.org/chromium/src/mojo/edk/embedder/embedder.h?l=174.

https://codereview.chromium.org/2424353002/diff/80001/remoting/host/win/launc...
File remoting/host/win/launch_process_with_token.cc (right):

https://codereview.chromium.org/2424353002/diff/80001/remoting/host/win/launc...
remoting/host/win/launch_process_with_token.cc:154: DPLOG(ERROR);
On 2016/10/20 17:40:04, joedow wrote:
> PLOG(ERROR) here instead.  We often troubleshoot machines in the field using
ETW
> logs and skipping  logs for Windows functions that might fail makes debugging
> more difficult.

Done.

https://codereview.chromium.org/2424353002/diff/80001/remoting/host/win/launc...
remoting/host/win/launch_process_with_token.cc:162: DPLOG(ERROR);
On 2016/10/20 17:40:04, joedow wrote:
> PLOG(ERROR)

Done.

https://codereview.chromium.org/2424353002/diff/80001/remoting/host/win/unpri...
File remoting/host/win/unprivileged_process_delegate.cc (right):

https://codereview.chromium.org/2424353002/diff/80001/remoting/host/win/unpri...
remoting/host/win/unprivileged_process_delegate.cc:279: base::CommandLine
command_line(target_command_->argv());
On 2016/10/20 17:21:30, Sergey Ulanov wrote:
> nit: Move this down, closer to where it's used.

Done.

https://codereview.chromium.org/2424353002/diff/80001/remoting/host/win/worke...
File remoting/host/win/worker_process_launcher_unittest.cc (left):

https://codereview.chromium.org/2424353002/diff/80001/remoting/host/win/worke...
remoting/host/win/worker_process_launcher_unittest.cc:47: const char
kIpcSecurityDescriptor[] = "D:(A;;GA;;;AU)";
On 2016/10/20 17:40:04, joedow wrote:
> How is the Mojo pipe secured?  Is there a default DACL applied?

For pipes to be passed through handle inheritance the default security
descriptor is used:
https://cs.chromium.org/chromium/src/mojo/edk/embedder/platform_channel_pair_....
For pipes to be passed by name there is a DACL applied:
https://cs.chromium.org/chromium/src/mojo/edk/embedder/named_platform_handle_....

https://codereview.chromium.org/2424353002/diff/80001/remoting/host/win/wts_s...
File remoting/host/win/wts_session_process_delegate.cc (right):

https://codereview.chromium.org/2424353002/diff/80001/remoting/host/win/wts_s...
remoting/host/win/wts_session_process_delegate.cc:419: session_token_.Get(),
security_attributes.get(), nullptr, {},
On 2016/10/20 17:40:04, joedow wrote:
> add some comments for the hard-coded params like /*inherited_handles=*/={}

Done.

[Sergey Ulanov, 2016-10-21 18:49:03]

I don't any further comments. Will let Joe finish this review

[Ken Rockot(use gerrit already), 2016-10-21 22:38:20]

lgtm fwiw

[joedow, 2016-10-21 23:16:30]

FYI, This is my first exposure to Mojo so I am working through the review more. 
Overall I think the changes look good, but I want to understand the security
aspect a bit more before signing off on this first conversion CL.

Just to confirm Sam, you are building the MSI and connecting to the host
(preferably with curtain mode enabled) for your testing right?  Just
double-checking :)

[Sam McNally, 2016-10-24 06:33:41]

On 2016/10/21 23:16:30, joedow wrote:
> FYI, This is my first exposure to Mojo so I am working through the review
more. 
> Overall I think the changes look good, but I want to understand the security
> aspect a bit more before signing off on this first conversion CL.
> 
> Just to confirm Sam, you are building the MSI and connecting to the host
> (preferably with curtain mode enabled) for your testing right?  Just
> double-checking :)

Yes.

[joedow, 2016-10-24 17:42:51]

lgtm once my comments are addressed.

https://codereview.chromium.org/2424353002/diff/100001/remoting/host/host_mai...
File remoting/host/host_main.cc (right):

https://codereview.chromium.org/2424353002/diff/100001/remoting/host/host_mai...
remoting/host/host_main.cc:215: mojo::edk::Init();
I'm not sure how expensive this call is for the other platforms, but IPC/Mojo is
only used when we are in multi-process mode so we might want to scope it:
#if defined(REMOTING_MULTI_PROCESS)
    mojo::edk::Init();
#endif

https://codereview.chromium.org/2424353002/diff/100001/remoting/host/win/laun...
File remoting/host/win/launch_process_with_token.cc (right):

https://codereview.chromium.org/2424353002/diff/100001/remoting/host/win/laun...
remoting/host/win/launch_process_with_token.cc:153: if
(!startup_info_wrapper.InitializeProcThreadAttributeList(1)) {
nit: /*attribute_count=*/1  for the parameter

https://codereview.chromium.org/2424353002/diff/100001/remoting/host/win/laun...
remoting/host/win/launch_process_with_token.cc:154: PLOG(ERROR);
For the PLOGs, I'd prefer to have a simple message as well:
PLOG(ERROR) << "InitializeProcThreadAttributeList()";

It simplifies the debugging and troubleshooting process for us.

https://codereview.chromium.org/2424353002/diff/100001/remoting/host/win/laun...
remoting/host/win/launch_process_with_token.cc:162: PLOG(ERROR);
Here too: PLOG(ERROR) << "UpdateProcThreadAttribute()";

[Sergey Ulanov, 2016-10-24 17:53:15]

https://codereview.chromium.org/2424353002/diff/100001/remoting/host/host_mai...
File remoting/host/host_main.cc (right):

https://codereview.chromium.org/2424353002/diff/100001/remoting/host/host_mai...
remoting/host/host_main.cc:215: mojo::edk::Init();
On 2016/10/24 17:42:51, joedow wrote:
> I'm not sure how expensive this call is for the other platforms, but IPC/Mojo
is
> only used when we are in multi-process mode so we might want to scope it:
> #if defined(REMOTING_MULTI_PROCESS)
>     mojo::edk::Init();
> #endif

+1
It's also likely to affect binary size, as we don't currently use mojo on other
platforms

[Sam McNally, 2016-10-24 22:42:24]

Patchset #4 (id:140001) has been deleted

[Sam McNally, 2016-10-24 22:42:33]

Patchset #4 (id:160001) has been deleted

[Sam McNally, 2016-10-24 22:43:26]

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

[commit-bot: I haz the power, 2016-10-24 22:44:27]

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

[Sam McNally, 2016-10-24 22:54:04]

https://codereview.chromium.org/2424353002/diff/100001/remoting/host/host_mai...
File remoting/host/host_main.cc (right):

https://codereview.chromium.org/2424353002/diff/100001/remoting/host/host_mai...
remoting/host/host_main.cc:215: mojo::edk::Init();
On 2016/10/24 17:42:51, joedow wrote:
> I'm not sure how expensive this call is for the other platforms, but IPC/Mojo
is
> only used when we are in multi-process mode so we might want to scope it:
> #if defined(REMOTING_MULTI_PROCESS)
>     mojo::edk::Init();
> #endif

Done.

https://codereview.chromium.org/2424353002/diff/100001/remoting/host/win/laun...
File remoting/host/win/launch_process_with_token.cc (right):

https://codereview.chromium.org/2424353002/diff/100001/remoting/host/win/laun...
remoting/host/win/launch_process_with_token.cc:153: if
(!startup_info_wrapper.InitializeProcThreadAttributeList(1)) {
On 2016/10/24 17:42:51, joedow wrote:
> nit: /*attribute_count=*/1  for the parameter

Done.

https://codereview.chromium.org/2424353002/diff/100001/remoting/host/win/laun...
remoting/host/win/launch_process_with_token.cc:154: PLOG(ERROR);
On 2016/10/24 17:42:51, joedow wrote:
> For the PLOGs, I'd prefer to have a simple message as well:
> PLOG(ERROR) << "InitializeProcThreadAttributeList()";
> 
> It simplifies the debugging and troubleshooting process for us.

Done.

https://codereview.chromium.org/2424353002/diff/100001/remoting/host/win/laun...
remoting/host/win/launch_process_with_token.cc:162: PLOG(ERROR);
On 2016/10/24 17:42:51, joedow wrote:
> Here too: PLOG(ERROR) << "UpdateProcThreadAttribute()";

Done.

[commit-bot: I haz the power, 2016-10-24 22:58:31]

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

[commit-bot: I haz the power, 2016-10-24 22:58:32]

Dry run: Try jobs failed on following builders:
  chromeos_daisy_chromium_compile_only_ng on master.tryserver.chromium.linux
(JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/chromeos_daisy_...)

[Sam McNally, 2016-10-24 23:56:11]

The CQ bit was checked by sammc@chromium.org

[Sam McNally, 2016-10-24 23:56:11]

The patchset sent to the CQ was uploaded after l-g-t-m from rockot@chromium.org,
joedow@chromium.org
Link to the patchset: https://codereview.chromium.org/2424353002/#ps180001
(title: " ")

[commit-bot: I haz the power, 2016-10-24 23:56:58]

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, 2016-10-25 02:04:20]

Committed patchset #4 (id:180001)

[commit-bot: I haz the power, 2016-10-25 02:06:20]

Description was changed from

==========
Use ChannelMojo between the remoting daemon and network processes.

This refactors LaunchProcessWithToken to take an explicit list of
handles for the child to inherit (copied from the equivalent process
launch code in //base).

This also changes WorkerProcessLauncherTest to use ChannelMojo.

BUG=604282
==========

to

==========
Use ChannelMojo between the remoting daemon and network processes.

This refactors LaunchProcessWithToken to take an explicit list of
handles for the child to inherit (copied from the equivalent process
launch code in //base).

This also changes WorkerProcessLauncherTest to use ChannelMojo.

BUG=604282

Committed: https://crrev.com/41c99b466b9bbba9dcb002e8cb00f1d7da22dff4
Cr-Commit-Position: refs/heads/master@{#427238}
==========

[commit-bot: I haz the power, 2016-10-25 02:06:22]

Patchset 4 (id:??) landed as
https://crrev.com/41c99b466b9bbba9dcb002e8cb00f1d7da22dff4
Cr-Commit-Position: refs/heads/master@{#427238}

[dmazzoni, 2016-10-26 15:25:35]

A revert of this CL (patchset #4 id:180001) has been created in
https://codereview.chromium.org/2451203002/ by dmazzoni@chromium.org.

The reason for reverting is: http://crbug.com/659219
Causes WorkerProcessLauncherTest.Crash to fail
on multiple Win bots

.

[Sam McNally, 2016-10-26 20:59:54]

Description was changed from

==========
Use ChannelMojo between the remoting daemon and network processes.

This refactors LaunchProcessWithToken to take an explicit list of
handles for the child to inherit (copied from the equivalent process
launch code in //base).

This also changes WorkerProcessLauncherTest to use ChannelMojo.

BUG=604282

Committed: https://crrev.com/41c99b466b9bbba9dcb002e8cb00f1d7da22dff4
Cr-Commit-Position: refs/heads/master@{#427238}
==========

to

==========
Use ChannelMojo between the remoting daemon and network processes.

This refactors LaunchProcessWithToken to take an explicit list of
handles for the child to inherit (copied from the equivalent process
launch code in //base).

This also changes WorkerProcessLauncherTest to use ChannelMojo.

BUG=604282

Committed: https://crrev.com/41c99b466b9bbba9dcb002e8cb00f1d7da22dff4
Cr-Commit-Position: refs/heads/master@{#427238}
==========

[Sam McNally, 2016-10-26 21:08:16]

Description was changed from

==========
Use ChannelMojo between the remoting daemon and network processes.

This refactors LaunchProcessWithToken to take an explicit list of
handles for the child to inherit (copied from the equivalent process
launch code in //base).

This also changes WorkerProcessLauncherTest to use ChannelMojo.

BUG=604282

Committed: https://crrev.com/41c99b466b9bbba9dcb002e8cb00f1d7da22dff4
Cr-Commit-Position: refs/heads/master@{#427238}
==========

to

==========
Use ChannelMojo between the remoting daemon and network processes.

This refactors LaunchProcessWithToken to take an explicit list of
handles for the child to inherit (copied from the equivalent process
launch code in //base).

BUG=604282

Committed: https://crrev.com/41c99b466b9bbba9dcb002e8cb00f1d7da22dff4
Cr-Commit-Position: refs/heads/master@{#427238}
==========

[Sam McNally, 2016-10-26 21:08:56]

The CQ bit was checked by sammc@chromium.org

[Sam McNally, 2016-10-26 21:08:57]

The patchset sent to the CQ was uploaded after l-g-t-m from rockot@chromium.org,
joedow@chromium.org
Link to the patchset: https://codereview.chromium.org/2424353002/#ps200001
(title: "sans test changes")

[commit-bot: I haz the power, 2016-10-26 21:10:11]

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, 2016-10-26 22:10:09]

Description was changed from

==========
Use ChannelMojo between the remoting daemon and network processes.

This refactors LaunchProcessWithToken to take an explicit list of
handles for the child to inherit (copied from the equivalent process
launch code in //base).

BUG=604282

Committed: https://crrev.com/41c99b466b9bbba9dcb002e8cb00f1d7da22dff4
Cr-Commit-Position: refs/heads/master@{#427238}
==========

to

==========
Use ChannelMojo between the remoting daemon and network processes.

This refactors LaunchProcessWithToken to take an explicit list of
handles for the child to inherit (copied from the equivalent process
launch code in //base).

BUG=604282

Committed: https://crrev.com/41c99b466b9bbba9dcb002e8cb00f1d7da22dff4
Cr-Commit-Position: refs/heads/master@{#427238}
==========

[commit-bot: I haz the power, 2016-10-26 22:10:11]

Committed patchset #5 (id:200001)

[commit-bot: I haz the power, 2016-10-26 22:13:49]

Description was changed from

==========
Use ChannelMojo between the remoting daemon and network processes.

This refactors LaunchProcessWithToken to take an explicit list of
handles for the child to inherit (copied from the equivalent process
launch code in //base).

BUG=604282

Committed: https://crrev.com/41c99b466b9bbba9dcb002e8cb00f1d7da22dff4
Cr-Commit-Position: refs/heads/master@{#427238}
==========

to

==========
Use ChannelMojo between the remoting daemon and network processes.

This refactors LaunchProcessWithToken to take an explicit list of
handles for the child to inherit (copied from the equivalent process
launch code in //base).

BUG=604282

Committed: https://crrev.com/41c99b466b9bbba9dcb002e8cb00f1d7da22dff4
Committed: https://crrev.com/e5d78a2c2a7e226ebfd531f0c42ba39759f00043
Cr-Original-Commit-Position: refs/heads/master@{#427238}
Cr-Commit-Position: refs/heads/master@{#427843}
==========

[commit-bot: I haz the power, 2016-10-26 22:13:50]

Patchset 5 (id:??) landed as
https://crrev.com/e5d78a2c2a7e226ebfd531f0c42ba39759f00043
Cr-Commit-Position: refs/heads/master@{#427843}