Re: [PATCH 02/11] t/lib-git-daemon: make sure to kill the 'git-daemon' process

From: Junio C Hamano <gitster@pobox.com>
Date: Mar 14, 2019, 02:35 UTC
Message-ID: <xmqqpnqub2tx.fsf@gitster-ct.c.googlers.com>
In-Reply-To: <20190313122419.2210-3-szeder.dev@gmail.com>

SZEDER Gábor <szeder.dev@gmail.com> writes:

Show 7 quoted lines

> After 'start_git_daemon' starts 'git daemon' (note the space in the
> middle) in the background, it saves the background process' PID, so
> the daemon can be stopped at the end of the test script.  However,
> 'git-daemon' is not a builtin but a dashed external command, which
> means that the dashless 'git daemon' executes the dashed 'git-daemon'
> command, and, consequently, the PID recorded is not the PID of the
> "real" daemon process, but that of the main 'git' wrapper.

Hmph. execv_dashed_external() does not quite execv() but uses run_command() and waits, which is the source of this issue. It feels almost possible to work it around by actually exec'ing, though.

	/*
	 * If we fail because the command is not found, it is
	 * OK to return. Otherwise, we just pass along the status code,
	 * or our usual generic code if we were not even able to exec
	 * the program.
	 */
	status = run_command(&cmd);
	if (status >= 0)
		exit(status);
	else if (errno != ENOENT)
		exit(128);

would become

	... prepare args from cmd ...
	execv(...);
	if (errno != ENOENT)
		exit(128);

and if exec succeeds, the whole process will exit with the status of dashed command, failing to exec for reasons other than ENOENT will exit with 128 and ENOENT will continue returning.

Show 10 quoted lines

> Now, if a
> test script involving 'git daemon' is interrupted by ctrl-C, then only
> the main 'git' process is stopped, but the real daemon process tends
> to survive somehow, and keeps on running in the background
> indefinitely, keeping the daemon's port to itself, and thus preventing
> subsequent runs of the same test script.
>
> Work this around by running 'git daemon' with the '--pidfile=...'
> option to save the PID of the real daemon process, and kill that
> process in 'stop_git_daemon' as well.

OK, so for the purpose of waiting and monitoring, we still use the pid of the git potty that is running and waiting for the daemon, but we'll send another signal to kill the daemon off. Makes sense.