Add CrashpadClient

Add CrashpadClient.

CrashpadClient is the primary interface for an application to have
Crashpad monitor it for crashes. It contains StartHandler(), which
starts a Crashpad handler process, and UseHandler(), which configures
the process to direct its crashes to a handler process.

R=rsesek@chromium.org

Committed: https://chromium.googlesource.com/crashpad/crashpad/+/2602e9d5d0501923999c8462033024309dce0e27

[Mark Mentovai, 2014-12-19 22:48:41]

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

[Mark Mentovai, 2014-12-19 22:48:41]

A generic crashpad_handler executable is proposed for review separately at
https://codereview.chromium.org/789693005. There is no strict dependency
relationship between these two changes, although neither is fully usable without
the other.

[Robert Sesek, 2014-12-29 16:34:04]

LGTM

Did you consider using posix_spawn for this instead?

https://codereview.chromium.org/785233011/diff/20001/client/crashpad_client_m...
File client/crashpad_client_mac.cc (right):

https://codereview.chromium.org/785233011/diff/20001/client/crashpad_client_m...
client/crashpad_client_mac.cc:91: // It’s not desirable for the handler to have
a controlling terminale. The
sp: terminale

[Mark Mentovai, 2014-12-29 17:20:37]

Since I want to double-fork(), I’ve got to either fork() at least once, or
double-posix_spawn(). Double-posix_spawn() means that I’d need to pick up the
first posix_spawn() in main() somewhere and do it again, and that seems like
it’s a lot of trouble for nothing. I thought about doing one fork() and one
posix_spawn(), but we’d need a way to close unwanted file descriptors, and
that’s https://crbug.com/179923#c31 all over again (this problem would also
exist for double-posix_spawn()). Presumably the race you mentioned on that bug
wouldn’t be a problem between fork() and posix_spawn() because things are
single-threaded at that point, so it could work, maybe.

I’ll investigate that as a possible enhancement (with the ability to use
POSIX_SPAWN_CLOEXEC_DEFAULT on 10.8 where it exists and doesn’t cause the kernel
to panic). Do you mind if I land this first so that there’s a checked-in basis
for further improvements?

It would be nice to use posix_spawn() for something finally…

[Robert Sesek, 2014-12-30 15:46:01]

Right, I thought there may be a posix_spawn() flag that would daemonize the
process for you. The closest thing is POSIX_SPAWN_SETPGROUP, but that won't get
you all the way there.

[Mark Mentovai, 2014-12-30 19:24:56]

Committed patchset #3 (id:40001) manually as
2602e9d5d0501923999c8462033024309dce0e27 (presubmit successful).