From mboxrd@z Thu Jan 1 00:00:00 1970 From: Christoph Hellwig Subject: odd sg_request munging code in i2o post-2.6.12 Date: Mon, 27 Jun 2005 22:55:20 +0200 Message-ID: <20050627205520.GA20272@lst.de> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Received: from verein.lst.de ([213.95.11.210]:34950 "EHLO mail.lst.de") by vger.kernel.org with ESMTP id S261783AbVF0Uzc (ORCPT ); Mon, 27 Jun 2005 16:55:32 -0400 Content-Disposition: inline Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: markus.lidel@shadowconnect.com Cc: linux-scsi@vger.kernel.org Markus, any chance you could try to send in changes that deal with scsi internals through James SCSI tree? You've sent some changes in that are rather odd: - created a include/scsi/sg_request.h header that duplicates struct sg_request, an internal data structure of the sg driver. This approach gets easily out of sync. - and poke into this data structure of an upper layer driver from a LLDD. That's a big "DON'T" While of the current drivers only SG uses that field it's open to any ULD, and you're also not supporting SG_IO on non-sg devices. Besides that we are actively killing scsi_requests before 2.6.13, so you code will have to be remove by then.