Hi Stephen, I posted a similar report on linux-kernel a month ago, but didn't hear anything new from you. So I try it again in this list, this time with some additional info: Using a stock 2.2.17 kernel with NFS-patches and ext-0.0.3b I can reliably lockup the kjournald when extracting two or three tar.gz files (e.g. kernel tar-balls) onto a ext3 filesystem with activated quotas. This happens almost immediately when the fs is mounted with -osync. When mounted without sync it completes the run of three different 'tar -xzf' once, but the next time I tried it, one of the tar processes segfaulted and I got the attached Oops. The machine is running a plain RedHat 6.2 with updates and the e2fsprogs and quota-tools you provided with your ext-0.0.3b patch. Greetings, Juri -- juri.haberland@innominate.com system engineer innominate AG clustering & security the linux architects tel: +49-30-308806-45 fax: -77 http://www.innominate.com
Hi, On Tue, Nov 28, 2000 at 01:24:59PM +0100, Juri Haberland wrote:> > I posted a similar report on linux-kernel a month ago, but didn't hear > anything new from you. So I try it again in this list, this time with > some additional info: > > Using a stock 2.2.17 kernel with NFS-patches and ext-0.0.3b I can > reliably lockup the kjournald when extracting two or three tar.gz filesYes, I know, and it's on my list of things to look at. I've been dealing with major modifications to the core filesystem for ext3-0.0.5 with metadata-only journaling, and I've just about got the user level tools done for that. Once those are out, I'll start looking at the quota problems you have mentioned. Cheers, Stephen