From owner-freebsd-fs Fri Mar 1 17:50:40 2002 Delivered-To: freebsd-fs@freebsd.org Received: from apollo.backplane.com (apollo.backplane.com [216.240.41.2]) by hub.freebsd.org (Postfix) with ESMTP id 28DD337B400 for ; Fri, 1 Mar 2002 17:50:36 -0800 (PST) Received: (from dillon@localhost) by apollo.backplane.com (8.11.6/8.9.1) id g221oWS48392; Fri, 1 Mar 2002 17:50:32 -0800 (PST) (envelope-from dillon) Date: Fri, 1 Mar 2002 17:50:32 -0800 (PST) From: Matthew Dillon Message-Id: <200203020150.g221oWS48392@apollo.backplane.com> To: Kris Kennaway , Ian Dowse , Kirk McKusick , Finch , fs@FreeBSD.ORG, fanf@chiark.greenend.org.uk Subject: Re: UFS panic on -stable References: <200202260141.g1Q1f8i28365@beastie.mckusick.com> <200202260237.aa51774@salmon.maths.tcd.ie> <20020226184340.A78562@xor.obsecurity.org> <20020228135233.B5586@xor.obsecurity.org> <20020301173740.A21179@xor.obsecurity.org> Sender: owner-freebsd-fs@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.org :One of the other machines panicked today; it wasn't even under load at :the time (last reported load average was 0.00). A different panic :this time (I've seen it before, but I couldn't get a crashdump that :time). : :A common thread in all of these panics is that they all seem to :involve operations on MFS filesystems. Even though the two UFS disk :filesystems on the clients have the most read/write load, I don't :think I've seen a panic involving them. : :/var: bad dir ino 24 at offset 0: mangled entry :panic: ufs_dirbad: bad dir :Debugger("panic") :Stopped at Debugger+0x35: movb $0,in_Debugger.426 :db> : :Here's the backtrace. I wonder if this has anything to do with MFS's VM pages getting paged in and out. Could you try hacking mount_mfs to mlock() the MFS memory? As shown below. I'm not entirely sure this will actually do anything, but it's worth a shot (assuming you have enough memory to cover all your MFS mounted filesystems). If this does the trick I'll take a look at the way MFS accesses the process's VM space. If it doesn't at least we will know it isn't related to MFS's paging. -Matt Index: mkfs.c =================================================================== RCS file: /home/ncvs/src/sbin/newfs/mkfs.c,v retrieving revision 1.29.2.6 diff -u -r1.29.2.6 mkfs.c --- mkfs.c 21 Sep 2001 19:15:21 -0000 1.29.2.6 +++ mkfs.c 2 Mar 2002 01:49:33 -0000 @@ -236,6 +236,8 @@ 0); if(membase == MAP_FAILED) err(12, "mmap"); + mlock(membase, fssize * sectorsize); + bzero(membase, fssize * sectorsize); close(fd); } else { #ifndef STANDALONE To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-fs" in the body of the message