From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S267423AbUBROZz (ORCPT ); Wed, 18 Feb 2004 09:25:55 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S267425AbUBROZz (ORCPT ); Wed, 18 Feb 2004 09:25:55 -0500 Received: from host-64-65-253-246.alb.choiceone.net ([64.65.253.246]:18390 "EHLO gaimboi.tmr.com") by vger.kernel.org with ESMTP id S267423AbUBROZv (ORCPT ); Wed, 18 Feb 2004 09:25:51 -0500 Message-ID: <40337199.2060609@tmr.com> Date: Wed, 18 Feb 2004 09:07:21 -0500 From: Bill Davidsen User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.6b) Gecko/20031208 X-Accept-Language: en-us, en MIME-Version: 1.0 Newsgroups: mail.linux-kernel To: James Morris CC: Jari Ruusu , Jan Rychter , linux-kernel@vger.kernel.org, Andrew Morton Subject: Re: Oopsing cryptoapi (or loop device?) on 2.6.* References: <402F877C.C9B693C1@users.sourceforge.net> In-Reply-To: Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org James Morris wrote: > On Sun, 15 Feb 2004, Jari Ruusu wrote: > > >>Jan Rychter wrote: >> >>>FWIW, I've just tried loop-AES with 2.4.24, after using cryptoapi for a >>>number of years. My machine froze dead in the midst of copying 2.8GB of >>>data onto my file-backed reiserfs encrypted loopback mount. >>> >>>Since the system didn't ever freeze on me before and since I've had zero >>>problems with cryptoapi, I attribute the freeze to loop-AES. >>> >>>Yes, I know this isn't a good bugreport... >> >>Is there any particular reason why you insist on using file backed loops? >> >>File backed loops have hard to fix re-entry problem: GFP_NOFS memory >>allocations that cause dirty pages to written out to file backed loop, will >>have to re-enter the file system anyway to complete the write. This causes >>deadlocks. Same deadlocks are there in mainline loop+cryptoloop combo. > > > Given the security issues, and the above problems, we should probably just > remove cryptoloop from the kernel and wait for something with a better > design. I hope you're kidding... one of the reasons for going to 2.6 is that you no longer have to patch your kernel to get cryptoloop. That is a real issue in some organizations, which only allow vendor or kernel.org kernels. If you start dropping features which work for most people but aren't perfect, you will wind up with a microkernel indeed. -- bill davidsen CTO TMR Associates, Inc Doing interesting things with small computers since 1979