[PATCH] [sanitizer] Fix log_path behavior with StopTheWorld.

[Alexey Samsonov]

  I don't fully understand the problem you're solving. Is it the case that parent process and tracer task simultaneously call RawWrite()? Do they have the shared memory (and thus shared report_fd/report_fd_pid, but different internal_getpid())?

http://llvm-reviews.chandlerc.com/D2306