Remove named SharedMemory usage on Mac.

Remove named SharedMemory usage on Mac.

- Only users are unit tests.
- Ifdef out tests that are no longer relevant for Mac.

BUG=345734
Committed: https://crrev.com/d0deb21e9c8442fdab00f8a709817e6cff88709f
Cr-Commit-Position: refs/heads/master@{#348835}

[Lei Zhang, 2015-09-15 01:17:42]

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

[erikchen, 2015-09-15 01:44:31]

Your CL description says "Ifdef out tests that are no longer relevant for Mac."
Why are they no longer relevant? I understand that they test a creation path
which is not representative of the real behavior of base::SharedMemory, but the
test SharedMemoryAcrossProcesses is testing memory.Map(), which would otherwise
not be tested.

[Lei Zhang, 2015-09-15 02:08:51]

On 2015/09/15 01:44:31, erikchen wrote:
> Your CL description says "Ifdef out tests that are no longer relevant for
Mac."
> Why are they no longer relevant? I understand that they test a creation path
> which is not representative of the real behavior of base::SharedMemory, but
the
> test SharedMemoryAcrossProcesses is testing memory.Map(), which would
otherwise
> not be tested.

We are still testing Map() in the following tests, unless I'm reading the
#ifdefs wrong: ShareReadOnly, ShareToSelf, MapAt, MapTwice, and
AnonymousExecutable.

The only thing we are skipping are multi-process/thread access to shared memory.
Maybe the multi-thread test can be re-written to not use
CreateNamedDeprecated()?

[erikchen, 2015-09-15 02:13:36]

On 2015/09/15 02:08:51, Lei Zhang wrote:
> On 2015/09/15 01:44:31, erikchen wrote:
> > Your CL description says "Ifdef out tests that are no longer relevant for
> Mac."
> > Why are they no longer relevant? I understand that they test a creation path
> > which is not representative of the real behavior of base::SharedMemory, but
> the
> > test SharedMemoryAcrossProcesses is testing memory.Map(), which would
> otherwise
> > not be tested.
> 
> We are still testing Map() in the following tests, unless I'm reading the
> #ifdefs wrong: ShareReadOnly, ShareToSelf, MapAt, MapTwice, and
> AnonymousExecutable.
> 
> The only thing we are skipping are multi-process/thread access to shared
memory.
> Maybe the multi-thread test can be re-written to not use
> CreateNamedDeprecated()?

Testing functionality of a shared memory object within the context of a single
process is a lot less interesting than across processes. That being said, we're
not losing that much test coverage, and I plan on revamping this test suite in
the not-so-distant future, so lgtm.

[Lei Zhang, 2015-09-15 02:23:41]

On 2015/09/15 02:13:36, erikchen wrote:
> Testing functionality of a shared memory object within the context of a single
> process is a lot less interesting than across processes. That being said,
we're
> not losing that much test coverage, and I plan on revamping this test suite in
> the not-so-distant future, so lgtm.

Sounds good. I see you own several SharedMemory bugs already, so you've probably
thought about this more than I have.

[Lei Zhang, 2015-09-15 02:23:47]

The CQ bit was checked by thestig@chromium.org

[commit-bot: I haz the power, 2015-09-15 02:24:04]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1340223002/40001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1340223002/40001

[commit-bot: I haz the power, 2015-09-15 03:42:19]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2015-09-15 03:42:51]

Patchset 3 (id:??) landed as
https://crrev.com/d0deb21e9c8442fdab00f8a709817e6cff88709f
Cr-Commit-Position: refs/heads/master@{#348835}

[commit-bot: I haz the power, 2015-09-23 12:42:30]

Patchset 3 (id:??) landed as
https://crrev.com/d0deb21e9c8442fdab00f8a709817e6cff88709f
Cr-Commit-Position: refs/heads/master@{#348835}