From owner-freebsd-stable Mon May 6 10:30:24 1996 Return-Path: owner-stable Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id KAA26681 for stable-outgoing; Mon, 6 May 1996 10:30:24 -0700 (PDT) Received: from rocky.sri.MT.net (rocky.sri.MT.net [204.182.243.10]) by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id KAA26675 for ; Mon, 6 May 1996 10:30:21 -0700 (PDT) Received: (from nate@localhost) by rocky.sri.MT.net (8.6.12/8.6.12) id LAA18033; Mon, 6 May 1996 11:30:06 -0600 Date: Mon, 6 May 1996 11:30:06 -0600 From: Nate Williams Message-Id: <199605061730.LAA18033@rocky.sri.MT.net> To: John Capo Cc: freebsd-stable@freebsd.org Subject: Re: FS corruption during rm -fr In-Reply-To: <199605042232.SAA09640@irbs.irbs.com> References: <199605042232.SAA09640@irbs.irbs.com> Sender: owner-stable@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > play/stable: > total 2 > drwxr-xr-x 3 root wheel 512 May 4 15:47 . > drwxr-xr-x 3 root wheel 512 May 4 15:48 .. > irbs 2# rm -fr play > rm: play/stable: Directory not empty > rm: play: Directory not empty > > > The same type of corruption has happened 5 times in the last 2 > days. Always during a rm -fr on a directory tree checked out of > cvs. Several passes are needed to get the FS clean again. I see this *all* the time on many systems (IDE and SCSI). I'm not sure why, but it appears to be related to CVS, since it's only the CVS directories that appear to get messed up. An easy way to recreate this bug is to check out the world, and then remove it all. It will *always* leave at least one or two directories that an un-removable. 3 passes of fsck in single-user mode are necessary to clean this up, at which point I can # cd /usr/lost+found;rm -rf * To clean it up, and then reboot and all will be well until I no longer unmount the FS cleanly (laptop patches) when the directories will show up again. > The system is a 486/66 with an Adaptec 1542B that I have been using > since 0.1 days and it has always been rock solid. The kernel is > pretty much stock -stable. The only difference in this kernel and > the ones I have run for the last 4 months is DDB and KTRACE. This > is the first -stable kernel I have run with those options. I've seen this with/without DDB and KTRACE. Nate