Add SystemSnapshotMac and its test

Add SystemSnapshotMac and its test.

TEST=snapshot_test SystemSnapshotMac.*
R=rsesek@chromium.org

Committed: https://chromium.googlesource.com/crashpad/crashpad/+/3f81599848a53bd46f82a58f2699e44c33df56d0

[Mark Mentovai, 2014-10-03 17:05:50]

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

[Mark Mentovai, 2014-10-03 17:05:51]

Test coverage isn’t total in here, because I think we’re going to reach a point
of diminishing marginal returns. Most of *SnapshotMac is just going to wrap the
existing well-tested stuff in util, and I’m not excited about duplicating all of
those tests just to exercise things at this layer.

[Robert Sesek, 2014-10-03 18:05:45]

https://codereview.chromium.org/626013002/diff/1/snapshot/system_snapshot_mac.cc
File snapshot/system_snapshot_mac.cc (right):

https://codereview.chromium.org/626013002/diff/1/snapshot/system_snapshot_mac...
snapshot/system_snapshot_mac.cc:150: #endif
#else NOTREACHED() or are you relying on the compiler to alert you that there's
no return?

https://codereview.chromium.org/626013002/diff/1/snapshot/system_snapshot_mac...
snapshot/system_snapshot_mac.cc:165: #endif
#else NOTREACHED() ?

https://codereview.chromium.org/626013002/diff/1/snapshot/system_snapshot_mac...
snapshot/system_snapshot_mac.cc:267: struct __attribute__((aligned(16))) {
Can this not be a CPUContextX86::Fxsave ?

https://codereview.chromium.org/626013002/diff/1/snapshot/system_snapshot_mac...
snapshot/system_snapshot_mac.cc:361: // This assumes that the offset between
standard and daylight saving time is
Technically this is not a safe assumption. http://www.timeanddate.com/time/dst/

https://codereview.chromium.org/626013002/diff/1/snapshot/system_snapshot_mac...
File snapshot/system_snapshot_mac_test.cc (right):

https://codereview.chromium.org/626013002/diff/1/snapshot/system_snapshot_mac...
snapshot/system_snapshot_mac_test.cc:37: class TestSystemSnapshotMac {
Why not do this in a testing::TestCase::SetUp() ? Then the assertions would
merely stop the test.

[Mark Mentovai, 2014-10-03 18:34:57]

Good suggestions, this removes a bunch of duplication.

https://codereview.chromium.org/626013002/diff/1/snapshot/system_snapshot_mac.cc
File snapshot/system_snapshot_mac.cc (right):

https://codereview.chromium.org/626013002/diff/1/snapshot/system_snapshot_mac...
snapshot/system_snapshot_mac.cc:150: #endif
Robert Sesek wrote:
> #else NOTREACHED() or are you relying on the compiler to alert you that
there's
> no return?

Exactly, this would need to be ported.

I made this more explicit than just relying on the compiler to notice the
absence of “return”.

https://codereview.chromium.org/626013002/diff/1/snapshot/system_snapshot_mac...
snapshot/system_snapshot_mac.cc:194: NOTREACHED();
I did #else NOTREACHED() on the x86-family ones because they shouldn’t be called
on non-x86-family, but the interface says that they need to be defined.

https://codereview.chromium.org/626013002/diff/1/snapshot/system_snapshot_mac...
snapshot/system_snapshot_mac.cc:361: // This assumes that the offset between
standard and daylight saving time is
Robert Sesek wrote:
> Technically this is not a safe assumption.
http://www.timeanddate.com/time/dst/

If you think it matters, I guess I could probe a few times during the year to
work out the actual DST and non-DST offsets.

[Robert Sesek, 2014-10-03 18:44:26]

LGTM

https://codereview.chromium.org/626013002/diff/1/snapshot/system_snapshot_mac.cc
File snapshot/system_snapshot_mac.cc (right):

https://codereview.chromium.org/626013002/diff/1/snapshot/system_snapshot_mac...
snapshot/system_snapshot_mac.cc:361: // This assumes that the offset between
standard and daylight saving time is
On 2014/10/03 18:34:57, Mark Mentovai wrote:
> Robert Sesek wrote:
> > Technically this is not a safe assumption.
> http://www.timeanddate.com/time/dst/
> 
> If you think it matters, I guess I could probe a few times during the year to
> work out the actual DST and non-DST offsets.

I think it's probably fine.

[Mark Mentovai, 2014-10-03 18:55:58]

Committed patchset #2 (id:20001) manually as
3f81599848a53bd46f82a58f2699e44c33df56d0 (presubmit successful).