From owner-freebsd-current Sat Mar 2 13:31:47 1996 Return-Path: owner-current Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id NAA18796 for current-outgoing; Sat, 2 Mar 1996 13:31:47 -0800 (PST) Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.19]) by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id NAA18789 for ; Sat, 2 Mar 1996 13:31:41 -0800 (PST) Received: (from bde@localhost) by godzilla.zeta.org.au (8.6.12/8.6.9) id HAA21506; Sun, 3 Mar 1996 07:26:55 +1000 Date: Sun, 3 Mar 1996 07:26:55 +1000 From: Bruce Evans Message-Id: <199603022126.HAA21506@godzilla.zeta.org.au> To: jhay@mikom.csir.co.za, terry@lambert.org Subject: Re: rename panics kernel Cc: freebsd-current@FreeBSD.ORG Sender: owner-current@FreeBSD.ORG Precedence: bulk Yet more on this... >> > Resently I got a "panic : vrele : negative reference count". >> > The vrele() was called from rename(). >> > >> > I tried a simple script to exercise rename (attached below) and a >> > current system seems to panic (trapped in ufs_rename). There's a race >> > condition lurking, it seems. I havent tried other than the sticky /tmp >> > directory as the source and target files parent directory. Also the >> > test was run under root's account, if it matters. >> > >> > Rename exercise: >> > ------ >> > #!/bin/sh >> > a=/tmp/foo.now >> > b=/tmp/foo.prev >> > while true >> > do >> > for n in 1 2 3 4 5 6 7 8 9 0 >> > do >> > (mv $a $b ; touch $a) & >> > done >> > wait >> > done >> > ------ This is a very versatile test :-). I still haven't got it to panic for rename with my own version of -current (my previous report about this was misleading), but it caused the following problems: On system b which has 16MB memory and 64MB swap, the script was killed after about 5 hours with a "swap_pager: out of swap space" error. The system was otherwise inactive apart from running standard daemons. On system c which has 8MB memory and 32MB swap, the script always causes a "vm_fork: u_map allocation failed" after about 1 minute. Bruce