Fix some potential Mach port leaks from mach_host_self using a new ScopedMachPort class.

Fix some potential Mach port leaks from mach_host_self using a new ScopedMachPort class.

BUG=119379
TEST=none


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=162225

[Mark Mentovai, 2012-10-16 16:17:51]

https://codereview.chromium.org/11188003/diff/1/base/mac/scoped_mach_port.h
File base/mac/scoped_mach_port.h (right):

https://codereview.chromium.org/11188003/diff/1/base/mac/scoped_mach_port.h#n...
base/mac/scoped_mach_port.h:18: // for specialized Mach port types.
But practically speaking, at least in userland, everything’s a mach_port_t, and
even routines like mach_host_self are declared as returning a mach_port_t and
not a host_name_port_t. Would we realize a code size savings if we did this as a
straight class with the implementation in a .cc file, rather than having to
instantiate template code all over the place?

https://codereview.chromium.org/11188003/diff/1/base/mac/scoped_mach_port.h#n...
base/mac/scoped_mach_port.h:49: class ScopedMachPortT : public
ScopedMachPort<mach_port_t> {
If you’re going to do the templated thing, why bother with this? It’s never
used. Every time you did use the template, you used it for host_name_port_t.

[Robert Sesek, 2012-10-16 16:30:59]

https://codereview.chromium.org/11188003/diff/1/base/mac/scoped_mach_port.h
File base/mac/scoped_mach_port.h (right):

https://codereview.chromium.org/11188003/diff/1/base/mac/scoped_mach_port.h#n...
base/mac/scoped_mach_port.h:18: // for specialized Mach port types.
On 2012/10/16 16:17:52, Mark Mentovai wrote:
> But practically speaking, at least in userland, everything’s a mach_port_t,
and
> even routines like mach_host_self are declared as returning a mach_port_t and
> not a host_name_port_t. Would we realize a code size savings if we did this as
a
> straight class with the implementation in a .cc file, rather than having to
> instantiate template code all over the place?

Done.

https://codereview.chromium.org/11188003/diff/1/base/mac/scoped_mach_port.h#n...
base/mac/scoped_mach_port.h:49: class ScopedMachPortT : public
ScopedMachPort<mach_port_t> {
On 2012/10/16 16:17:52, Mark Mentovai wrote:
> If you’re going to do the templated thing, why bother with this? It’s never
> used. Every time you did use the template, you used it for host_name_port_t.

Removed.

[Mark Mentovai, 2012-10-16 16:34:28]

LGTM

https://codereview.chromium.org/11188003/diff/5001/base/mac/scoped_mach_port.cc
File base/mac/scoped_mach_port.cc (right):

https://codereview.chromium.org/11188003/diff/5001/base/mac/scoped_mach_port....
base/mac/scoped_mach_port.cc:19: port_ = MACH_PORT_NULL;
Not strictly necessary for the destructor.

https://codereview.chromium.org/11188003/diff/5001/base/mac/scoped_mach_port.h
File base/mac/scoped_mach_port.h (right):

https://codereview.chromium.org/11188003/diff/5001/base/mac/scoped_mach_port....
base/mac/scoped_mach_port.h:20: // Creates a scoper with MACH_PORT_NULL.
This is unused and can be gotten rid of.

https://codereview.chromium.org/11188003/diff/5001/base/mac/scoped_mach_port....
base/mac/scoped_mach_port.h:28: operator mach_port_t() const;
These two getters should remain inlined in the header. Accessing the field is
cheaper than calling (execution-wise) and just as small (code size-wise).

[Robert Sesek, 2012-10-16 16:40:23]

https://codereview.chromium.org/11188003/diff/5001/base/mac/scoped_mach_port.cc
File base/mac/scoped_mach_port.cc (right):

https://codereview.chromium.org/11188003/diff/5001/base/mac/scoped_mach_port....
base/mac/scoped_mach_port.cc:19: port_ = MACH_PORT_NULL;
On 2012/10/16 16:34:28, Mark Mentovai wrote:
> Not strictly necessary for the destructor.

Done.

https://codereview.chromium.org/11188003/diff/5001/base/mac/scoped_mach_port.h
File base/mac/scoped_mach_port.h (right):

https://codereview.chromium.org/11188003/diff/5001/base/mac/scoped_mach_port....
base/mac/scoped_mach_port.h:20: // Creates a scoper with MACH_PORT_NULL.
On 2012/10/16 16:34:28, Mark Mentovai wrote:
> This is unused and can be gotten rid of.

Done.

https://codereview.chromium.org/11188003/diff/5001/base/mac/scoped_mach_port....
base/mac/scoped_mach_port.h:28: operator mach_port_t() const;
On 2012/10/16 16:34:28, Mark Mentovai wrote:
> These two getters should remain inlined in the header. Accessing the field is
> cheaper than calling (execution-wise) and just as small (code size-wise).

Done.

[Mark Mentovai, 2012-10-16 16:57:09]

LGTM

[commit-bot: I haz the power, 2012-10-16 17:06:50]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/rsesek@chromium.org/11188003/10001

[commit-bot: I haz the power, 2012-10-16 20:18:00]

Change committed as 162225