From: Tom Lane <tgl@sss.pgh.pa.us>
To: Thomas Munro <thomas.munro@gmail.com>
cc: Robert Haas <robertmhaas@gmail.com>, Andrew Dunstan <andrew@dunslane.net>,
        Andres Freund <andres@anarazel.de>,
        pgsql-hackers <pgsql-hackers@postgresql.org>
Subject: Re: Cleaning up historical portability baggage
In-reply-to: 
 <CA+hUKGKdmC4A7cJNNSmY6KTfUsnX0b2AnjJu52Wd3_uHO1bBpg@mail.gmail.com>
References: 
 <CA+hUKGJ3LHeP9w5Fgzdr4G8AnEtJ=z=p6hGDEm4qYGEUX5B6fQ@mail.gmail.com>
 <CA+Tgmoazusui5EoU_7Z=AgwC5pTAqDndTa7d-owjvQwG3zoRhg@mail.gmail.com>
 <CA+hUKGLLJUMDgCzOK5K4NE6PpU87Ka7t7RQsZoZAZ_EKfL5L+A@mail.gmail.com>
 <CA+Tgmoaw6VBC9Xzr4vOaC0kc-94jb9nzH8MY38Gqn7=wy=i_3w@mail.gmail.com>
 <20220712173352.trdeq66asbv6vf2q@awork3.anarazel.de>
 <1662559.1657648462@sss.pgh.pa.us>
 <CA+hUKG+MOzmDkr27ceG8+uwgFxbJUUv699V3VP3DCioQigrf7A@mail.gmail.com>
 <1249825.1658622186@sss.pgh.pa.us>
 <CA+Tgmob_5AUNCzyFGJX6quYSnQnKCHW6DGGJa1noofJqSu+weg@mail.gmail.com>
 <3920417.1659710931@sss.pgh.pa.us>
 <CA+TgmoZriceWqD4WD2vzEgPPcbAoY6OPR0N9fQJ2ZPVdH_AzYQ@mail.gmail.com>
 <CA+hUKGKdmC4A7cJNNSmY6KTfUsnX0b2AnjJu52Wd3_uHO1bBpg@mail.gmail.com>
Comments: In-reply-to Thomas Munro <thomas.munro@gmail.com>
	message dated "Sun, 07 Aug 2022 10:23:17 +1200"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <146575.1659825723.1@sss.pgh.pa.us>
Date: Sat, 06 Aug 2022 18:42:03 -0400
Message-ID: <146576.1659825723@sss.pgh.pa.us>
Archived-At: 
 <https://www.postgresql.org/message-id/146576.1659825723%40sss.pgh.pa.us>
Precedence: bulk

Thomas Munro <thomas.munro@gmail.com> writes:
> Did I understand correctly that the places that do kill(-pid) followed
> by kill(pid) really only need the kill(-pid)?

Uh ... did you read the comment right above signal_child?

 * There is a race condition for recently-forked children: they might not
 * have executed setsid() yet.  So we signal the child directly as well as
 * the group.  We assume such a child will handle the signal before trying
 * to spawn any grandchild processes.  We also assume that signaling the
 * child twice will not cause any problems.

It might be that this is wrong and signaling -pid will work even if
the child hasn't yet done setsid(), but I doubt it: the kill(2) man
page is pretty clear that it'll fail if "the process group doesn't
exist".

Perhaps we could finesse that by signaling -pid first, and then
signaling pid if that fails, but offhand it seems like that has
the described race condition w.r.t. grandchild processes.

			regards, tom lane