> With the pipe, there is still a potential race after the parent writes to the pipe and before sigwaitinfo() is invoked, versus the child sleep() call.

Yeah, I know, that's why I kept the ugly 100 ms sleep.

> What do you think of my suggestion to block the signal? Then (in theory) it should be robust, rather than relying on timing.
> https://bugs.python.org/issue25277#msg256278

I read your comment, but with this change, I'm not sure anymore that we test that sigwaitinfo() retries when it's interrupted by a signal (EINTR).