Date: Mon, 07 Jun 1999 08:42:16 -0400 From: "Robert W. Rowe" <rrowe@winstar.com> To: dledford@redhat.com Cc: Jonathan Stimmel <jstimmel@taos.com>, "J. Alan Eldridge" <alane@wozzle.geeksrus.net>, aic7xxx@FreeBSD.ORG Subject: Re: aha2930u2 Message-ID: <3.0.5.32.19990607084216.009ca780@mail.winstar.com> In-Reply-To: <375AE8BD.D3ADC165@redhat.com> References: <Pine.A41.3.96.990606095531.48952A-100000@vcmr-19.rcs.rpi.edu>
next in thread | previous in thread | raw e-mail | index | archive | help
Thank you very much for a comprehensive reply. My problem is an on-the-motherboard aic7895 controller. I have two SCSI3 disks on the wide channel A and two narrow devices, a CD-RW and a Zip drive, on the 50-pin side. As many people who have aic789x cards have complained, I have timing problems. The SCSI bus continually resets, etc. I was unable to install RH 6.0 without disconnecting the narrow devices. With the narrow devices attached, the installation sequence went off into limbo during the disk-formatting phase. The SCSI card (on the motherboard) boots up in this order, with everything connected: Wide SCSI Disk 0 CD-RW Disk 2 Zip Disk 5 Wide SCSI Disk 6 which Linux recognizes as /dev/sda, /dev/scd0, /dev/sdb and /dev/sdc. However, when I removed the narrow devices so I could load RH 6.0, the second wide SCSI, still ID'd as Disk 6, became /dev/sdb in Linux. When everything was loaded and I wanted to see if the timing problems still were a factor, I plugged the narrow devices back in and turned on the machine. The Zip drive came up as /dev/sdb and the second wide SCSI as /dev/sdc. Naturally, Linux couldn't boot. I have three options: 1. Figure out how to get the zip drive to come up third in the SCSI disk sequence and be /dev/sdc. However, I will most likely still have timing problems. 2. I have an Adaptec ISA 1520 card. If it will work, if I can figure out how to get Linux to recognize it, put that in the machine and attach the narrow devices to it. 3. Turn off the Adaptec 7895 and put in a Buslogic (or whatever they are called now) card. The 7895 has been a problem from the time I got the computer. I have two EIDE drives, one now used for Win95 and the other used for daily backup. I also have an ATAPI CD reader because I couldn't get the SCSI CD-RW to work with the 7895. When I had RH 5.2, I ran Linux off the smaller EIDE drive and set up raid0 on two narrow SCSI fireballs. Now I have RH 6.0 set up on two IBM wide SCSI's, no raid. I want to get narrow SCSI working because I want to be able to burn a CD now and then, and because I want to put a SCSI scanner on the system. Thanks. At 05:31 PM 6/6/99 -0400, dledford@redhat.com wrote: > >The ordering is determined in two ways. Amongst different drivers (eg, >aic7xxx driver and ncr53c8xx driver) the order will be based upon the driver >startup sequence that you can find in /usr/src/linux/drivers/scsi/hosts.h. >Controlling this order is usually as simple as compiling one driver into the >kernel and the rest as modules so that you will automatically get the compiled >in driver first and then all the modular drivers in the order you insert >them. Or, do all the drivers as modules and simply control the insertion >order. The second form of ordering is the multiple instances of cards >controlled by a single driver (eg, two 2940U2W cards). In that case the >ordering is up to the driver. In the case of the aic7xxx driver we order >things as VLB/EISA cards with BIOS enabled first (sorted by BIOS address from >lowest to highest). PCI cards with their BIOS enabled second (sorted by PCI >slot #, except dual channel controllers that are always A then B or B then A >depending on the BIOS setting, and sorted from lowest to highest PCI slot #). >VLB/EISA cards with BIOS disabled are third and are sorted the same as those >with BIOS enabled. PCI cards with BIOS disabled are fourth and are sorted the >same as those with BIOS enabled. > >Since we know that in an EISA/PCI machine that the BIOS will scan and activate >all EISA BIOSes before PCI BIOSes, we know that we can safely always put EISA >cards with BIOS enabled in front of PCI cards with BIOS enabled. So, the >order of those four groups above never changes. However, in the event that >your motherboard scans PCI cards from highest PCI slot # to lowest PCI slot #, >there is the option aic7xxx=reverse_scan which will cause us to reverse our >sort ordering for PCI cards that we find. We will still preserve things like >channel A to B relationships on dual channel controllers, but we may move a >2940U2W in front of a 7880 controller for instance. The one condition that >this doesn't catch (and for which there is no clean way to catch these cases) >are BIOSes on motherboards that have built in aic7xxx controllers and also >have some aic7xxx cards plugged into slots. I've seen BIOSes that give the >options of 1) Built in controllers first, then scan from highest to lowest for >others or 2) Built in controllers first, then scan from lowest to highest for >others or 3) scan PCI slots first, then do built in controllers. Obviously, >there is no clean way to detect this and there is no discernable ordering >present in all cases. > >-- > Doug Ledford <dledford@redhat.com> > Opinions expressed are my own, but > they should be everybody's. > > To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe aic7xxx" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3.0.5.32.19990607084216.009ca780>