From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jens Axboe Subject: Re: writable mmc profiles actually are writable Date: Tue, 7 Oct 2003 09:04:04 +0200 Sender: linux-scsi-owner@vger.kernel.org Message-ID: <20031007070404.GU972@suse.de> References: <20031006182510.GG972@suse.de> <1065469831.5185.44.camel@patehci2> <20031006203816.GJ972@suse.de> <1065473899.6835.41.camel@patehci2> <1065478448.2361.1.camel@patehci2> <1065484607.2899.7.camel@patehci2> <20031007053858.GL972@suse.de> <20031006234543.A11937@one-eyed-alien.net> <20031007064829.GQ972@suse.de> <20031007000018.B11937@one-eyed-alien.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Received: from ns.virtualhost.dk ([195.184.98.160]:19936 "EHLO virtualhost.dk") by vger.kernel.org with ESMTP id S261802AbTJGHEJ (ORCPT ); Tue, 7 Oct 2003 03:04:09 -0400 Content-Disposition: inline In-Reply-To: <20031007000018.B11937@one-eyed-alien.net> List-Id: linux-scsi@vger.kernel.org To: Pat LaVarre , linux-scsi@vger.kernel.org On Tue, Oct 07 2003, Matthew Dharm wrote: > On Tue, Oct 07, 2003 at 08:48:29AM +0200, Jens Axboe wrote: > > On Mon, Oct 06 2003, Matthew Dharm wrote: > > > On Tue, Oct 07, 2003 at 07:38:58AM +0200, Jens Axboe wrote: > > > > On Mon, Oct 06 2003, Pat LaVarre wrote: > > > > > Yep, still happy, same in -test6 as in -test5 e.g. > > > > > > > > > > $ dmesg | grep ^sr > > > > > sr0: scsi3-mmc drive: 125x/125x caddy > > > > > sr0: scsi3-mmc maybe not writeable > > > > > sr0: scsi3-mmc writable profile: 0x0002 > > > > > $ > > > > > $ grep MMC /proc/sys/dev/cdrom/info > > > > > Can write other MMC-RW: 1 ... > > > > > > > > I don't quite agree with this patch. Please add the GET_CONFIGURATION > > > > stuff to cdrom.c instead so it works with ide-cd as well (ide-scsi is > > > > basically dead, and the number of scsi drives out there isn't exactly > > > > big and increasing). GPCMD_GET_CONFIGURATION already exists. Also, make > > > > DVD-RAM drives set CDC_MMC_RW as well and only check for that at open. > > > > > > While the number of SCSI CD-ROM drives may be shrinking, the ATAPI drives > > > behind a USB/ATAPI bridge is skyrocketing. > > > > Doesn't change the picture, if you put it in cdrom.c it will work for > > both. > > It will? I'll have to take your word for that, as I don't immediately see > how that works... then again, much of the higher layers of SCSI are a > mystery to me. > > I think I see how this work.... sr.c registers with cdrom.c, and cdrom.c > actually provides the interface to the /dev/scdN node, right? Precisely. > If that's the case, why is there so much code in sr.c anyway? The > get_capabilities() function really should be moved, as well as media-change > detection, much of the probe function (dealing with initialization of the > new structre), reading sector sizes, and sr_ioctl.c should probably be > renamed (as it doesn't really do IOCTL processing). > > Am I on the right track? That seems like an awfully long list of things > that suggest against my interpretation.... Yes you are essentially correct. Some things have been killed from sr a long time ago, but I generally resist adding new things to sr and ide-cd unless I really have to. And I've been doing so since 2.2, so I really don't want to reverse that right now. It's also confusing why then some things only work if you have ide-cd loaded, or if you use ide-scsi and sr. It's just a bad idea. -- Jens Axboe