mac: Add auto-close and share-read-only functionality to SharedMemory.

mac: Add auto-close and share-read-only functionality to Mach based SharedMemory.

base::FileDescriptor has a property |auto_close| that is used to indicate that
when the object is passed to an IPC message, the message takes ownership of the
underlying OS handle. Since SharedMemoryHandle needs to be interchangeable with
base::FileDescriptor, I added a property with similar functionality named
|ownership_passes_to_ipc_|.

The method ShareToProcess() is used to lower the current and maximum protection
of the underlying OS handle before it is transferred to a different process. I
implemented this functionality for Mach memory objects.

BUG=535711
Committed: https://crrev.com/033bbbcb63cab781552dfb435c035131c423de30
Cr-Commit-Position: refs/heads/master@{#355880}

[erikchen, 2015-10-23 00:28:53]

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

[commit-bot: I haz the power, 2015-10-23 00:29:02]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1418113003/60001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1418113003/60001

[erikchen, 2015-10-23 00:34:28]

erikchen@chromium.org changed reviewers:
+ mark@chromium.org

[erikchen, 2015-10-23 00:34:29]

mark: Please review.

[erikchen, 2015-10-23 00:34:54]

Description was changed from

==========
mac: Add auto-close and share-read-only functionality to SharedMemory.

base::FileDescriptor has a property |auto_close| that is used to indicate that
when the object is passed to an IPC message, the message takes ownership of the
underlying OS handle. Since SharedMemoryHandle needs to be interchangeable with
base::FileDescriptor, I added a property with similar functionality named
|ownership_passes_to_ipc_|.

The method ShareToProcess() is used to lower the current and maximum protection
of the underlying OS handle before it is transferred to a different process. I
implemented this functionality for Mach memory objects.

BUG=535711
==========

to

==========
mac: Add auto-close and share-read-only functionality to SharedMemory.

base::FileDescriptor has a property |auto_close| that is used to indicate that
when the object is passed to an IPC message, the message takes ownership of the
underlying OS handle. Since SharedMemoryHandle needs to be interchangeable with
base::FileDescriptor, I added a property with similar functionality named
|ownership_passes_to_ipc_|.

The method ShareToProcess() is used to lower the current and maximum protection
of the underlying OS handle before it is transferred to a different process. I
implemented this functionality for Mach memory objects.

BUG=535711
==========

[commit-bot: I haz the power, 2015-10-23 01:47:21]

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

[commit-bot: I haz the power, 2015-10-23 01:47:21]

Dry run: This issue passed the CQ dry run.

[Mark Mentovai, 2015-10-23 14:24:03]

LGTM

https://codereview.chromium.org/1418113003/diff/60001/base/memory/shared_memo...
File base/memory/shared_memory_handle_mac.cc (right):

https://codereview.chromium.org/1418113003/diff/60001/base/memory/shared_memo...
base/memory/shared_memory_handle_mac.cc:190: VM_PROT_WRITE | VM_PROT_READ |
VM_PROT_IS_MASK,  // Maximum protection
This bit wasn’t defined in 10.6. Does its kernel mind being passed it?

Does it make sense to use the IS_MASK bit for current protection too?

https://codereview.chromium.org/1418113003/diff/60001/base/memory/shared_memo...
File base/memory/shared_memory_mac.cc (right):

https://codereview.chromium.org/1418113003/diff/60001/base/memory/shared_memo...
base/memory/shared_memory_mac.cc:410: // Unmap() and Close().
So do it?

[erikchen, 2015-10-23 17:13:31]

https://codereview.chromium.org/1418113003/diff/60001/base/memory/shared_memo...
File base/memory/shared_memory_handle_mac.cc (right):

https://codereview.chromium.org/1418113003/diff/60001/base/memory/shared_memo...
base/memory/shared_memory_handle_mac.cc:190: VM_PROT_WRITE | VM_PROT_READ |
VM_PROT_IS_MASK,  // Maximum protection
On 2015/10/23 14:24:03, Mark Mentovai wrote:
> This bit wasn’t defined in 10.6. Does its kernel mind being passed it?
> 
> Does it make sense to use the IS_MASK bit for current protection too?

mark: 10.6.8 doesn't have VM_PROT_IS_MASK and will have the wrong behaior. 

Options: 
(1) don't support Mach-based shared memory on 10.6.8
(2) find a way to determine the max prot of a memory object
(3) Add a max_prot member to SharedMemoryHandle, and plumb it through ipc/.

I couldn't find a way to do (2). (3) is not difficult, but honestly I'm leaning
towards (1). Thoughts?

[Mark Mentovai, 2015-10-23 17:16:31]

(1) is mostly only sucky because 10.6, which nobody tests, will behave
differently than other OSes.

But whatever.

[erikchen, 2015-10-23 17:25:56]

mark: PTAL

https://codereview.chromium.org/1418113003/diff/60001/base/memory/shared_memo...
File base/memory/shared_memory_mac.cc (right):

https://codereview.chromium.org/1418113003/diff/60001/base/memory/shared_memo...
base/memory/shared_memory_mac.cc:410: // Unmap() and Close().
On 2015/10/23 14:24:03, Mark Mentovai wrote:
> So do it?

Will do so in a separate CL. removing the comment.

[Mark Mentovai, 2015-10-23 17:36:19]

LGTM. OK, no 10.6 support it is.

[erikchen, 2015-10-23 17:38:17]

erikchen@chromium.org changed reviewers:
+ tsepez@chromium.org

[erikchen, 2015-10-23 17:38:18]

tsepez: Please review.

[erikchen, 2015-10-23 17:39:05]

On 2015/10/23 17:38:18, erikchen wrote:
> tsepez: Please review.

*Please review ipc/

[Tom Sepez, 2015-10-23 18:28:29]

lgtm

[erikchen, 2015-10-23 18:34:26]

The CQ bit was checked by erikchen@chromium.org

[commit-bot: I haz the power, 2015-10-23 18:34:57]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1418113003/80001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1418113003/80001

[commit-bot: I haz the power, 2015-10-23 21:05:32]

Committed patchset #5 (id:80001)

[commit-bot: I haz the power, 2015-10-23 21:07:18]

Patchset 5 (id:??) landed as
https://crrev.com/033bbbcb63cab781552dfb435c035131c423de30
Cr-Commit-Position: refs/heads/master@{#355880}

[erikchen, 2015-10-23 22:29:53]

A revert of this CL (patchset #5 id:80001) has been created in
https://codereview.chromium.org/1421933002/ by erikchen@chromium.org.

The reason for reverting is: New unit-tests failing on 10.8

http://build.chromium.org/p/chromium.mac/builders/Mac10.8%20Tests/builds/1013....