From mboxrd@z Thu Jan 1 00:00:00 1970 From: James Bottomley Subject: Re: Forcing use of sd driver on Suse 10.3 (2.6.22) Date: Thu, 17 Jan 2008 09:40:21 -0600 Message-ID: <1200584422.3255.7.camel@localhost.localdomain> References: <1200308514.4729.25.camel@desk.ty-rhos-barn> <478BAA8E.7020703@s5r6.in-berlin.de> <1200567191.5778.6.camel@desk.ty-rhos-barn> Mime-Version: 1.0 Content-Type: text/plain Content-Transfer-Encoding: 7bit Return-path: Received: from accolon.hansenpartnership.com ([76.243.235.52]:37590 "EHLO accolon.hansenpartnership.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752114AbYAQPkf (ORCPT ); Thu, 17 Jan 2008 10:40:35 -0500 In-Reply-To: <1200567191.5778.6.camel@desk.ty-rhos-barn> Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: Bill Adair Cc: linux-scsi@vger.kernel.org On Thu, 2008-01-17 at 10:53 +0000, Bill Adair wrote: > On Mon, 2008-01-14 at 19:31 +0100, Stefan Richter wrote: > > Bill Adair wrote: > > > Is there any way under Linux of forcing use of the sd driver for a > > > device on the bus instead of sg? > > > > The INQUIRY data which the SCSI core gets from the device have to > > indicate that the device implements SBC or RBC (is of peripheral device > > type 00h or 0Eh). > I have altered the sd.c file by patching sd_probe to check for the > particular unit I have and accepting pdt 03h (Processor) as a valid > return and the unit works fine. That's a bit of a nasty. We can't blindly attach sd to a processor otherwise it will pick up all sorts of stuff its not supposed to bind to (like scanners, enclosure devices and other initiators). I suppose a firmware upgrade for the device to make it actually comply vaguely with the spec is out of the question? If not, there's a trick we do in scsi_scan.c for Toshiba CDROMS that fail to set their type correctly (it's the code under BLIST_ISROM), we could do the same for this device. What are its inquiry strings? And does anyone have any clue whether it should be RBC or DISK type? > I suspect the sd file is considered very important though (and I don't > know how many of these card readers exist) so I assume a patch is not > required ;-). Would there be another approach to creating a "driver" for > my Spyrus RD300 in it's guise as a card reader? James