From mboxrd@z Thu Jan 1 00:00:00 1970 From: James Bottomley Subject: Re: [PATCH] Move SG_GET_SCSI_ID from sg to scsi Date: Mon, 27 Mar 2006 09:40:01 -0600 Message-ID: <1143474001.3334.11.camel@mulgrave.il.steeleye.com> References: <20060326200522.GA3486@parisc-linux.org> <20060327152026.GG3486@parisc-linux.org> Mime-Version: 1.0 Content-Type: text/plain Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <20060327152026.GG3486@parisc-linux.org> Sender: linux-kernel-owner@vger.kernel.org To: Matthew Wilcox Cc: Bodo Eggert <7eggert@gmx.de>, linux-scsi@vger.kernel.org, linux-kernel@vger.kernel.org, torvalds@osdl.org List-Id: linux-scsi@vger.kernel.org On Mon, 2006-03-27 at 08:20 -0700, Matthew Wilcox wrote: > So we don't currently save the scsi_lun anywhere. We, er, translate it > into a 32-bit int, and carry that around. Obviously this is going to > have to change for things like iSCSI, but this is totally offtopic for > your patch. So I agree with you, whoever adds the 64-bit LUN support > gets to patch the ioctl struct too. Actually, this is precisely the argument for why we don't want such an ioctl at all. This area is one of the ones there should be significant changes in for the forseeable future. 32->64 bit LUNs are probably only the start; then there'll be elimination of the target ID (or replacing it with a string). If we have this ioctl, every time this area changes, so now does the user ABI. Finally, I really don't think we want to encourage anybody in userspace to rely on a string of internally generated numbers ... that's what sysfs paths are for, but at least they're simply strings opaque, so we can change them. James