From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S272924AbTHKULS (ORCPT ); Mon, 11 Aug 2003 16:11:18 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S272948AbTHKULS (ORCPT ); Mon, 11 Aug 2003 16:11:18 -0400 Received: from e2.ny.us.ibm.com ([32.97.182.102]:40861 "EHLO e2.ny.us.ibm.com") by vger.kernel.org with ESMTP id S272924AbTHKULR (ORCPT ); Mon, 11 Aug 2003 16:11:17 -0400 Date: Mon, 11 Aug 2003 13:14:58 -0700 From: "Martin J. Bligh" To: linux-kernel cc: ak@suse.de Subject: [Bug 1083] New: JFS corrupts file systems on 64bit architectures Message-ID: <872950000.1060632898@flay> X-Mailer: Mulberry/2.1.2 (Linux/x86) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Content-Disposition: inline Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org http://bugme.osdl.org/show_bug.cgi?id=1083 Summary: JFS corrupts file systems on 64bit architectures Kernel Version: 2.6.0test2 Status: NEW Severity: high Owner: shaggy@austin.ibm.com Submitter: ak@suse.de On AMD64 JFS seems to corrupt file systems quickly. I created a JFS file system, rsync'ed an BKCVS kernel checkout from another box then did two cvs updates. Result was that several directory inodes in the kernel tree were unreadable on the next access (always returned EPERM even for a stat). I did a fsck then, which found some bad inodes and resulted in a endless lost+found (ls ran out of memory trying to list it) 2.4 JFS still worked. I also got feedback from one user that they are seeing similar problems on sparc64.