From owner-freebsd-geom@FreeBSD.ORG Thu Jan 17 01:52:19 2008 Return-Path: Delivered-To: freebsd-geom@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id D80C316A41A; Thu, 17 Jan 2008 01:52:19 +0000 (UTC) (envelope-from afields@ncf.ca) Received: from saruman.ncf.ca (saruman.ncf.ca [134.117.136.37]) by mx1.freebsd.org (Postfix) with ESMTP id AFCFC13C45B; Thu, 17 Jan 2008 01:52:19 +0000 (UTC) (envelope-from afields@ncf.ca) Received: from [10.0.0.55] (CPE000d88cacd09-CM00159a09ff6e.cpe.net.cable.rogers.com [99.224.17.202]) by saruman.ncf.ca (iPlanet Messaging Server 5.2 Patch 2 (built Jul 14 2004)) with ESMTPSA id <0JUR00HJQE5RMH@saruman.ncf.ca>; Wed, 16 Jan 2008 17:51:29 -0500 (EST) Date: Wed, 16 Jan 2008 17:51:21 -0500 From: Allan Fields In-reply-to: To: Ivan Voras Message-id: <75FB90A1-5053-42C6-8466-1C4BF2208EF5@ncf.ca> MIME-version: 1.0 X-Mailer: Apple Mail (2.752.2) Content-type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Content-transfer-encoding: 7BIT References: <9e77bdb50801160832p39619f1fm85bf1454fead3357@mail.gmail.com> Cc: freebsd-geom@freebsd.org Subject: Re: Authentication with geom_eli X-BeenThere: freebsd-geom@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: GEOM-specific discussions and implementations List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 17 Jan 2008 01:52:20 -0000 On 16-Jan-08, at 2:31 PM, Ivan Voras wrote: > Cyrus Rahman wrote: > >> With this in mind, the addition of a less expensive authentication >> algorithm, say a CRC, which would provide still provide a check on >> the >> channel between geom_eli and the physical disk sectors without the >> overhead of md5 or sha256, would be extremely useful. > > I think this discussion was held in relation with ZFS (which by > default does strong hashing of ALL data ALWAYS) and that somebody > concluded from experiments that, given the difference in speed > between modern CPUs and modern drives, there wasn't much difference > between using CRC32 and using a strong hash. > > Of course, on slower / embedded devices the situation is much > different. Mind you perhaps this is best implemented as a separate GEOM class all-together. I have had difficulty getting the GELI SHA and MD5 hashing to perform as expected, though it initializes with-out error. Perhaps this works in a new release, I will verify, if not I'll file pr. Thanks, Allan Fields