[mojo-edk] Invalidate message handle when rewrite fails

[mojo-edk] Invalidate message handle when rewrite fails

On Windows we rewrite (i.e. duplicate-and-close) outgoing
handles before sending to another process. This operation
can fail if the target process is closing.

This CL invalidates a stored handle if DuplicateHandle fails,
preventing PlatformHandle from later trying to close a
non-existent handle during Message destruction.

BUG=608767
R=jam@chromium.org

Committed: https://crrev.com/a4bd42ba95dfa2be77c4a9db6388e97270046dea
Cr-Commit-Position: refs/heads/master@{#391287}

[Ken Rockot(use gerrit already), 2016-05-03 16:17:34]

Description was changed from

==========
[mojo-edk] Invalidate message handle when rewrite fails

We rewrite (i.e. duplicate-and-close) outgoing HANDLEs before
sending to another process. This operation can fail if the
target process is closing.

If the operation fails the local HANDLE is closed anyway.

Currently we're leaving the (now invalid) HANDLE in the
Message, and PlatformHandle will then DCHECK when trying
to close it.

This CL fixes that by invalidating the stored HANDLE on
DuplicateHandle failure.

BUG=608767
R=jam@chromium.org
==========

to

==========
[mojo-edk] Invalidate message handle when rewrite fails

On Windows we rewrite (i.e. duplicate-and-close) outgoing
handles before sending to another process. This operation
can fail if the target process is closing.

This CL invalidates a stored handle if DuplicateHandle fails,
preventing PlatformHandle from later trying to close a
non-existent handle later on Message destruction.

BUG=608767
R=jam@chromium.org
==========

[Ken Rockot(use gerrit already), 2016-05-03 16:38:07]

per offline discussion, added checks to avoid duplicating the current process
handle to a remote process

[Ken Rockot(use gerrit already), 2016-05-03 16:38:56]

On 2016/05/03 at 16:38:07, Ken Rockot wrote:
> per offline discussion, added checks to avoid duplicating the current process
handle to a remote process

also note - AFAICT the call in platform_handle.cc
(https://code.google.com/p/chromium/codesearch#chromium/src/mojo/edk/embedder/...)
doesn't need to be checked

[Ken Rockot(use gerrit already), 2016-05-03 16:42:13]

Description was changed from

==========
[mojo-edk] Invalidate message handle when rewrite fails

On Windows we rewrite (i.e. duplicate-and-close) outgoing
handles before sending to another process. This operation
can fail if the target process is closing.

This CL invalidates a stored handle if DuplicateHandle fails,
preventing PlatformHandle from later trying to close a
non-existent handle later on Message destruction.

BUG=608767
R=jam@chromium.org
==========

to

==========
[mojo-edk] Invalidate message handle when rewrite fails

On Windows we rewrite (i.e. duplicate-and-close) outgoing
handles before sending to another process. This operation
can fail if the target process is closing.

This CL invalidates a stored handle if DuplicateHandle fails,
preventing PlatformHandle from later trying to close a
non-existent handle during Message destruction.

BUG=608767
R=jam@chromium.org
==========

[Ken Rockot(use gerrit already), 2016-05-03 16:46:49]

https://codereview.chromium.org/1940353002/diff/20001/mojo/edk/system/channel.cc
File mojo/edk/system/channel.cc (right):

https://codereview.chromium.org/1940353002/diff/20001/mojo/edk/system/channel...
mojo/edk/system/channel.cc:334: CHECK_NE(handles[i].handle,
INVALID_HANDLE_VALUE);
Just noticing that this can't be reached given the test above. It should be OK
for an outgoing message to contain invalid handles as long as we don't try to
dupe them.

[jam, 2016-05-03 17:09:04]

lgtm

[Ken Rockot(use gerrit already), 2016-05-03 17:12:03]

The CQ bit was checked by rockot@chromium.org

[commit-bot: I haz the power, 2016-05-03 17:12:28]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1940353002/40001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1940353002/40001

[commit-bot: I haz the power, 2016-05-03 17:45:51]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2016-05-03 17:46:50]

Description was changed from

==========
[mojo-edk] Invalidate message handle when rewrite fails

On Windows we rewrite (i.e. duplicate-and-close) outgoing
handles before sending to another process. This operation
can fail if the target process is closing.

This CL invalidates a stored handle if DuplicateHandle fails,
preventing PlatformHandle from later trying to close a
non-existent handle during Message destruction.

BUG=608767
R=jam@chromium.org
==========

to

==========
[mojo-edk] Invalidate message handle when rewrite fails

On Windows we rewrite (i.e. duplicate-and-close) outgoing
handles before sending to another process. This operation
can fail if the target process is closing.

This CL invalidates a stored handle if DuplicateHandle fails,
preventing PlatformHandle from later trying to close a
non-existent handle during Message destruction.

BUG=608767
R=jam@chromium.org

Committed: https://crrev.com/a4bd42ba95dfa2be77c4a9db6388e97270046dea
Cr-Commit-Position: refs/heads/master@{#391287}
==========

[commit-bot: I haz the power, 2016-05-03 17:46:50]

Patchset 3 (id:??) landed as
https://crrev.com/a4bd42ba95dfa2be77c4a9db6388e97270046dea
Cr-Commit-Position: refs/heads/master@{#391287}