Received: (from majordomo@localhost) by oss.sgi.com (8.11.2/8.11.3) id f7TAVpi20004 for linux-xfs-outgoing; Wed, 29 Aug 2001 03:31:51 -0700 Received: from atrey.karlin.mff.cuni.cz (root@atrey.karlin.mff.cuni.cz [195.113.31.123]) by oss.sgi.com (8.11.2/8.11.3) with SMTP id f7TAVld19985 for ; Wed, 29 Aug 2001 03:31:47 -0700 Received: (from jack@localhost) by atrey.karlin.mff.cuni.cz (8.9.3/8.9.3/Debian 8.9.3-21) id MAA15299; Wed, 29 Aug 2001 12:30:42 +0200 Date: Wed, 29 Aug 2001 12:30:42 +0200 From: Jan Kara To: Nathan Scott Cc: machack , linux-xfs@oss.sgi.com Subject: Re: Kernel Dump on quotaon Message-ID: <20010829123042.H8775@atrey.karlin.mff.cuni.cz> References: <20010828224031.JZQZ25401.femail45.sdc1.sfba.home.com@there> <20010829103155.G323889@wobbly.melbourne.sgi.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.3.15i In-Reply-To: <20010829103155.G323889@wobbly.melbourne.sgi.com>; from nathans@sgi.com on Wed, Aug 29, 2001 at 10:31:55AM +1100 Sender: owner-linux-xfs@oss.sgi.com Precedence: bulk Hello, > On Tue, Aug 28, 2001 at 06:21:59PM -0400, machack wrote: > > rc.sysinit cuases the kernel to dump when it hits quotaon. this problem only > > happens on kernels newer then 2.4.6 on these later kernels I have no > > problems when I pass devfs=nomount on boot. Am I the only one to have this > > problem or is it documented. my root partion is xfs. > > > > I'd be interested in seeing a kdb or ksymoops backtrace if you > can provide one. quotaon at startup is always ignored by XFS - > enabling quota is a mount time operation for XFS filesystems, so > this is likely to be a problem in the interaction between the > quotactl system call (takes a device special file as its second > argument) with Q_QUOTAON and devfs; and unrelated to XFS. > > If you only have XFS filesystems using quota (and no ext2/other > filesystems using quota), you can safely remove the "quotaon -a" > from your startup scripts (thats really not a very nice solution > though...) > > Jan, any chance you've come across this one before? I don't > remember there being anything new/changed in 2.4.6 in the quota > subsystem, so seems odd that this would suddenly start to fail. Hmm.. There were two minor changes in quota code in 2.4.6. The first one was just that remove_dquot_refs() gets super_block and not device. The second one was that we allow initialization of quotas on other inodes, than just symlinks, regular files and directories. The second one probably causes the problem but I can't see how... I'll look into it but the trace would be very helpful. Honza -- Jan Kara SuSE Labs