From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from flinx.ccr.net (ccr@alogconduit1ae.ccr.net [208.130.159.5]) by kvack.org (8.8.7/8.8.7) with ESMTP id GAA22546 for ; Wed, 24 Mar 1999 06:43:01 -0500 Subject: Re: [Fwd: LINUX-MM] References: <36F8974A.D4EB4BAD@imsid.uni-jena.de> From: ebiederm+eric@ccr.net (Eric W. Biederman) Date: 24 Mar 1999 05:49:08 -0600 In-Reply-To: Matthias Arnold's message of "Wed, 24 Mar 1999 08:42:02 +0100" Message-ID: Sender: owner-linux-mm@kvack.org To: Matthias Arnold Cc: linux-mm@kvack.org List-ID: >>>>> "MA" == Matthias Arnold writes: >> >> On Tue, 23 Mar 1999 10:49:11 EST, Kev said: >> >> >> IIRC there's a slight bug in some of the newer kernels >> >> where the swap cache isn't being freed when you exit >> >> your program, but only later on when the system tries >> >> to reclaim memory... >> >> > I believe the problem lies in the fact that there is not enough >> > SysV shared memory available. >> >> It's nothing to do with SysV shared memory. >> >> The behaviour is there, but the only impact on the normal user will be >> that "free" lies a little. No big deal: it just shows up as cache. The >> effect is only a matter of when we recover the memory, not whether we >> recover it. MA> Thanks to the community for dealing with my problems.Before we go on to MA> discuss this as bug or feature I emphasize MA> that this memory effect is really disastrous for me. MA> I hope that changing the kernel (as Rik told me) helps. MA> Any suggestion is very welcome. I missed the start of this, (I just monitor linux-mm), what is the problem you have? Eric -- To unsubscribe, send a message with 'unsubscribe linux-mm my@address' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://humbolt.geo.uu.nl/Linux-MM/