From owner-freebsd-hackers Tue Jul 13 12:10: 5 1999 Delivered-To: freebsd-hackers@freebsd.org Received: from nightfly.apk.net (nightfly.apk.net [207.54.149.151]) by hub.freebsd.org (Postfix) with SMTP id AC94B15233 for ; Tue, 13 Jul 1999 12:10:00 -0700 (PDT) (envelope-from rme@nightfly.apk.net) Received: (qmail 16973 invoked by uid 1000); 13 Jul 1999 19:11:26 -0000 To: freebsd-hackers@freebsd.org Subject: Re: Replacement for grep(1) (part 2) References: X-Attribution: rme From: rme@nightfly.apk.net (R. Matthew Emerson) Date: 13 Jul 1999 15:11:26 -0400 In-Reply-To: "Brian F. Feldman"'s message of "Tue, 13 Jul 1999 14:47:20 -0400 (EDT)" Message-ID: <87pv1wjrfl.fsf@nightfly.apk.net> Lines: 18 X-Mailer: Gnus v5.5/Emacs 20.3 Sender: owner-freebsd-hackers@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG "Brian F. Feldman" writes: > But I have a valid point: can we do something better than posting a SIGKILL > to the largest process? If I remember correctly, AIX sends a signal to all processes asking them to free up memory. (Processes ignore this signal by default.) If nobody responds, then it selects a victim and blows it away. I think it is nice to processes that catch the signal and try to free up memory in that it won't kill them first. (What would you call such a signal, SIGDIET?) I guess memory overcommit is desirable in a "worse is better" kind of way, because it is clearly not "the right thing." :-( (see section 2.1 of http://naggum.no/worse-is-better.html) -matt [cc:'s trimmed] To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message