Add Multiprocess::SetExpectedChildTermination()

Add Multiprocess::SetExpectedChildTermination().

An expectation of the child’s termination reason can now be specified.
It is now possible to test with children that do not exit cleanly, if
they are not expected to exit cleanly.

TEST=util_test Multiprocess.*
R=rsesek@chromium.org

Committed: https://chromium.googlesource.com/crashpad/crashpad/+/7fdc7f277edc61d56c97677466d14bd2e669ddac

[Mark Mentovai, 2014-09-09 20:26:19]

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

[Mark Mentovai, 2014-09-09 20:26:19]

This will be used by the exc_variants test for a “live” test that the handler
can deal with a crashing child. Since the child is crashing, Multiprocess needs
to be able to deal with a child that doesn’t exit with status 0 without causing
the test to fail.

[Robert Sesek, 2014-09-09 20:42:44]

LGTM

https://codereview.chromium.org/553153002/diff/1/util/test/multiprocess.h
File util/test/multiprocess.h (right):

https://codereview.chromium.org/553153002/diff/1/util/test/multiprocess.h#new...
util/test/multiprocess.h:163: int code_;
Flip the order of these two members to match the parameter order?

[Mark Mentovai, 2014-09-09 20:49:34]

https://codereview.chromium.org/553153002/diff/1/util/test/multiprocess.h
File util/test/multiprocess.h (right):

https://codereview.chromium.org/553153002/diff/1/util/test/multiprocess.h#new...
util/test/multiprocess.h:163: int code_;
rsesek wrote:
> Flip the order of these two members to match the parameter order?

That’s worse for structure packing, do you insist?

[Robert Sesek, 2014-09-09 20:50:24]

https://codereview.chromium.org/553153002/diff/1/util/test/multiprocess.h
File util/test/multiprocess.h (right):

https://codereview.chromium.org/553153002/diff/1/util/test/multiprocess.h#new...
util/test/multiprocess.h:163: int code_;
On 2014/09/09 20:49:33, Mark Mentovai wrote:
> rsesek wrote:
> > Flip the order of these two members to match the parameter order?
> 
> That’s worse for structure packing, do you insist?

No, but that's a pretty irrelevant argument for a test-only class IMO.

[Mark Mentovai, 2014-09-09 20:51:51]

I’ve been doing it on all classes.

I’d really like there to be an I_DONT_CARE_ABOUT_MY_LAYOUT attribute.

[Mark Mentovai, 2014-09-09 21:04:51]

Committed patchset #1 (id:1) manually as 7fdc7f2 (presubmit successful).