From owner-aic7xxx Fri Nov 24 11: 4:24 2000 Delivered-To: aic7xxx@freebsd.org Received: from doral2.para-protect.com (unknown [141.156.68.130]) by hub.freebsd.org (Postfix) with ESMTP id D3BD537B4C5 for ; Fri, 24 Nov 2000 11:04:20 -0800 (PST) Received: by DORAL2 with Internet Mail Service (5.5.2650.21) id ; Fri, 24 Nov 2000 13:56:58 -0500 Message-ID: <59D960EEFB11D41188AF00B0D021E02AF972FA@DORAL2> From: "van Wyk, Ken" To: 'Mike Sharp' , AIC7xxx@FreeBSD.ORG Subject: RE: Solved? (was Re: Any known bugs in aic7xxx) Date: Fri, 24 Nov 2000 13:56:56 -0500 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2650.21) Content-Type: text/plain; charset="iso-8859-1" Sender: owner-aic7xxx@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org Hello, Thanks for all of the debugging effort, Mike! I can toss in just a little additional (hopefully useful) data. On my system, as I tried to describe in my previous email, I only have the one device on the SCSI/LVD bus. The one device is the Seagate SCSI 160 drive that I mentioned. So, it's not entirely an issue of mixing slow/fast hardware on the same bus. I'll plug in my SCSI UMAX scanner, which is a slower device, and see if that makes any difference. I'll send any useful follow-ups here. Cheers, Ken van Wyk -----Original Message----- From: Mike Sharp [mailto:mike_sharp@pacbell.net] Sent: Friday, November 24, 2000 11:34 AM To: AIC7xxx@FreeBSD.ORG Subject: Solved? (was Re: Any known bugs in aic7xxx) Hi, As some of you know, I've been trying to get a 29160 running on my system. Unfortunately I've had nothing but problems with SCSI bus resets... After reconfiguring my system 2^N different ways yesterday I had a surface mounted filter capacitor pop off the 29160. (sigh) As bad as this may seem, it actually helped me figure out what the potential problem is! After a few moments of dispair I decided a slow RAID was better than none so I got out my trusty old 2940 board. Strangely enough I started experiencing similar bus resets with this, but not as often. More than a large number of hours passed and I figured out that the bus resets are being caused by mixing the following devices on one 2940 adapter: two Seagate 318404LW (Cheetah 18 XL, 18g, ultra160 disk) two Seagate 318404LC (Cheetah 18XL, 18g, ultra 160 disk) Panasonic (Matshita) CD-7503 CD-R drive Seagate Peregrine 4586 DDS-2 DAT tape with autoloader (or alternately you can use a Quantum Fireball in place of the CD-R and DAT. this configuration doesn't work either) I was lucky enough to have *two* 2940 adapters laying around unused and I discovered that if I put the u160 disks on one and the CD-R and DAT on the other everything is stable and happy!! As proof of this, I'm currently writing this from the system in question after having run "bonnie -S 2047" and loading about 8g of data onto it's RAID without so much as a single SCSI problem. Currently I'm using the 2.4.0-test10 release straight off kernel.org. I am *NOT* currently using the alpha version of the 29160 driver. I also have had RH7.0 working out of the box, but I moved to 2.4.0 because I wanted to use my RAID as '/' and RH7.0 doesn't support this out of the box. SO: There seems to be a bug related to mixing widely different speed SCSI devices on one controller. In my case this was mixing 5 and 10mb devices with ones that really would prefer to run at 160mb. I find this a bit surprising, but I'm convinced of it now. I'd like to suggest to the "powers that be" (i.e. someone who knows how to debug the AIC7xxx driver further) that they configure a system with a 2940, a u160 disk and an old SCSI-1 CDROM and then pound on the system with "bonnie -s 2047". Before I separated my devices onto different controllers this would repeatably cause the bus resets to occur. Even though I/O wasn't targeting the CDROM it would occasionally freak out (why?) and cause the reset to occur. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe aic7xxx" in the body of the message