From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1423215AbWF1IPY (ORCPT ); Wed, 28 Jun 2006 04:15:24 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1423217AbWF1IPY (ORCPT ); Wed, 28 Jun 2006 04:15:24 -0400 Received: from aa002msr.fastwebnet.it ([85.18.95.65]:8873 "EHLO aa002msr.fastwebnet.it") by vger.kernel.org with ESMTP id S1423215AbWF1IPW (ORCPT ); Wed, 28 Jun 2006 04:15:22 -0400 Date: Wed, 28 Jun 2006 10:14:59 +0200 From: Paolo Ornati To: Jens Axboe Cc: Linux Kernel Mailing List Subject: scary fcache trouble - silent data corruption(?) Message-ID: <20060628101459.1375a27a@localhost> X-Mailer: Sylpheed-Claws 2.3.1 (GTK+ 2.8.17; x86_64-pc-linux-gnu) Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Hi, It's always me :) This time I had a problem that scared me a bit... Yesterday I've tested fcache for a while. I have primed system boot + preferred applications startup, and then I've used the system for a while. Today I launch my mail-client (Sylpheed-Claws) and the Setup Wizard dialog appears :-O! (usually it appears when it isn't configured) After a bit of reserch I've found that an XML config file (.sylpheed-claws/folderlist.xml) was corrupted. But it wasn't corrupted randomly, it was "replaced" with a peace of the firefox cookies file :-X! Now, Firefox and Sylpheed-Claws are 2 of the few application I've primed... so my suspect is that there's some subtle BUG in fcache, or maybe it's elsewhere since it is an experimental kernel after all... Luckily Sylpheed-Claws makes backup copy of every config file :) The size of the corrupted file is almost identical to the backup one (so it is probably equal to the size of the original one). Even if there is really a BUG in fcache so that it can supply wrong data I don't know how it got written to the real filesystem (since the corruption persisted booting without fcache). Maybe Sylpheed-Claws have read the file and then rewritten to it... I don't know. Every time something like this happens I ask myself why I test experimental things on my main PC... I think this time I was lucky :) -- Paolo Ornati Linux 2.6.17-ga39727f2-dirty on x86_64