Don't attempt to register render processes with the shell

Don't attempt to register render processes with the shell

This CL simply deletes the code which initiates the
child-to-shell registration process, as it serves no practial
purpose in M50. The process has already been made much less
terrible since M50, so it is neither possible nor necessary
to land this change further upstream.

BUG=597287
R=jam@chromium.org

Committed: https://chromium.googlesource.com/chromium/src/+/2b58a7118b948076fb9c9e2f11c1dbd123808160

[Ken Rockot(use gerrit already), 2016-03-23 15:47:53]

PTAL - Fixes an FD leak in M50.

We've since stopped allocating new FDs at all for this behavior, but I don't
think it would be wise to try merging all the relevant changes.

[Ken Rockot(use gerrit already), 2016-03-23 15:51:05]

Description was changed from

==========
Don't attempt to register render processes with the shell

Registering a child with the shell allocates a new socket pair
and stashes one end (unscoped!) in the process host user data.
In a normal configuration (i.e. not Mash) we never pass it onto
the child to complete initialization. Thus the child end of the
pipe leaks and the parent end waits indefinitely for the child
to do something. The result is that every render process launch
leaks two FDs.

This CL simply deletes the code which initiates the registration
process, as it serves no practial purpose in M50. The process
has already been made much less terrible since M50, so neither
possible nor necessary to land this change further upstream.

BUG=582143
R=jam@chromium.org
==========

to

==========
Don't attempt to register render processes with the shell

Registering a child with the shell allocates a new socket pair
and stashes one end (unscoped!) in the process host user data.
In a normal configuration (i.e. not Mash) we never pass it onto
the child to complete initialization. Thus the child end of the
pipe leaks and the parent end waits indefinitely for the child
to do something. The result is that every render process launch
leaks two FDs.

This CL simply deletes the code which initiates the registration
process, as it serves no practial purpose in M50. The process
has already been made much less terrible since M50, so it is
neither possible nor necessary to land this change further
upstream.

BUG=582143
R=jam@chromium.org
==========

[jam, 2016-03-23 15:52:04]

lgtm

[Ken Rockot(use gerrit already), 2016-03-23 16:19:11]

Description was changed from

==========
Don't attempt to register render processes with the shell

Registering a child with the shell allocates a new socket pair
and stashes one end (unscoped!) in the process host user data.
In a normal configuration (i.e. not Mash) we never pass it onto
the child to complete initialization. Thus the child end of the
pipe leaks and the parent end waits indefinitely for the child
to do something. The result is that every render process launch
leaks two FDs.

This CL simply deletes the code which initiates the registration
process, as it serves no practial purpose in M50. The process
has already been made much less terrible since M50, so it is
neither possible nor necessary to land this change further
upstream.

BUG=582143
R=jam@chromium.org
==========

to

==========
Don't attempt to register render processes with the shell

This CL simply deletes the code which initiates the
child-to-shell registration process, as it serves no practial
purpose in M50. The process has already been made much less
terrible since M50, so it is neither possible nor necessary
to land this change further upstream.

BUG=582143
R=jam@chromium.org
==========

[Ken Rockot(use gerrit already), 2016-03-23 16:19:46]

Description was changed from

==========
Don't attempt to register render processes with the shell

This CL simply deletes the code which initiates the
child-to-shell registration process, as it serves no practial
purpose in M50. The process has already been made much less
terrible since M50, so it is neither possible nor necessary
to land this change further upstream.

BUG=582143
R=jam@chromium.org
==========

to

==========
Don't attempt to register render processes with the shell

This CL simply deletes the code which initiates the
child-to-shell registration process, as it serves no practial
purpose in M50. The process has already been made much less
terrible since M50, so it is neither possible nor necessary
to land this change further upstream.

BUG=597287
R=jam@chromium.org
==========

[Ken Rockot(use gerrit already), 2016-03-23 17:38:35]

Description was changed from

==========
Don't attempt to register render processes with the shell

This CL simply deletes the code which initiates the
child-to-shell registration process, as it serves no practial
purpose in M50. The process has already been made much less
terrible since M50, so it is neither possible nor necessary
to land this change further upstream.

BUG=597287
R=jam@chromium.org
==========

to

==========
Don't attempt to register render processes with the shell

This CL simply deletes the code which initiates the
child-to-shell registration process, as it serves no practial
purpose in M50. The process has already been made much less
terrible since M50, so it is neither possible nor necessary
to land this change further upstream.

BUG=597287
R=jam@chromium.org

Committed:
https://chromium.googlesource.com/chromium/src/+/2b58a7118b948076fb9c9e2f11c1...
==========

[Ken Rockot(use gerrit already), 2016-03-23 17:38:36]

Committed patchset #1 (id:1) manually as
2b58a7118b948076fb9c9e2f11c1dbd123808160 (presubmit successful).