Re: [PATCH v10 2/2] selftests: add tests for clone3() with *set_tid

[Adrian Reber <areber@redhat.com>]

On Thu, Nov 14, 2019 at 10:34:21AM -0800, Andrei Vagin wrote:
> > +		/*
> > +		 * This should work and from the parent we should see
> > +		 * something like 'NSpid:	pid	42	1'.
> > +		 */
> > +		test_clone3_set_tid(set_tid, 3, CLONE_NEWPID, 0, 42, true);
> > +
> > +		_exit(ksft_cnt.ksft_pass);
> > +	}
> > +
> > +	close(pipe_1[1]);
> > +	close(pipe_2[0]);
> > +	while (read(pipe_1[0], &buf, 1) > 0) {
> 
> If a child process will crash, this will be a busyloop.

No. To be honest the whole loop is unnecessary. If the loop is entered
there is immediately a break.

> > +		ksft_print_msg("[%d] Child is ready and waiting\n", getpid());
> > +		break;
> > +	}

If the loop is not entered it does not loop. So the whole thing is
useless. I think I was expecting it to block, but it doesn't work that
way.

> > +	snprintf(proc_path, sizeof(proc_path), "/proc/%d/status", pid);
> > +	f = fopen(proc_path, "r");
> > +	if (f == NULL)
> > +		ksft_exit_fail_msg(
> > +			"%s - Could not open %s\n",
> > +			strerror(errno), proc_path);

If the child does not exist anymore, the test will fail here and exit.

Besides this while() I tried to address all your comments in v11. Any
further comments on the test?

		Adrian