From owner-freebsd-hackers Thu Jul 13 22:29:15 1995 Return-Path: hackers-owner Received: (from majordom@localhost) by freefall.cdrom.com (8.6.10/8.6.6) id WAA16736 for hackers-outgoing; Thu, 13 Jul 1995 22:29:15 -0700 Received: from wiley.csusb.edu (wiley.csusb.edu [139.182.2.2]) by freefall.cdrom.com (8.6.10/8.6.6) with SMTP id WAA16730 for ; Thu, 13 Jul 1995 22:29:03 -0700 Received: by wiley.csusb.edu (5.67a/1.34) id AA12615; Thu, 13 Jul 1995 22:31:58 -0700 From: rmallory@wiley.csusb.edu (Rob Mallory) Message-Id: <199507140531.AA12615@wiley.csusb.edu> Subject: Re: SCSI disk wedge To: msmith@atrad.adelaide.edu.au (Michael Smith) Date: Thu, 13 Jul 1995 22:31:58 -0700 (PDT) Cc: karl@mcs.net, gary@palmer.demon.co.uk, tom@misery.sdf.com, rgrimes@gndrsh.aac.dev.com, freebsd-hackers@freebsd.org In-Reply-To: <199507132336.JAA15005@genesis.atrad.adelaide.edu.au> from "Michael Smith" at Jul 14, 95 09:06:39 am X-Mailer: ELM [version 2.4 PL22] Content-Type: text Content-Length: 3419 Sender: hackers-owner@freebsd.org Precedence: bulk > Still haven't seen a list of disk model numbers, firmware revisions & > the like. The Identify strings at startup would be a start. This may get a bit long. Its just my views, I'm to busy now to hack on it at work, ...but I noticed some strange-ness with 2.0.5R with a 2740 at work, with the aha1 driver found by GENERIC, it would give instantly recoverable timeouts,, but worked long enough to compile an AIC kernel with tagged queueing,, which worked flawlessly. 3 days later, i reinstalled, after getting disgusted with Slowaris-2.5B (CDE is cool though),, I tried again; this time, with a new drive in the chain. Now, it could be out-o-spec cable length, I have a mix of centronics, micro-50, and big-50 cables,(4 total, all 18 inches each). Here's the setup: 1 st15150N barracuda. (freebsd) scsi-2 cmd-queueable 1 sun/archive viper qic150 scsi1 1 sun/sony external cdrom scsi1 1 old segate 670 scsi1 1 semi-new conner 540 not-scsi-2 when I added the conner (the spec sheet only mentions scsi) freebsd gives it 5MB/s on probe. (I'll send boot msgs later if requested) The kernel panic's (didnt write down the error) due to a scsi problem. This happens at the begenning of extract media when it begins ftp session. before the conner, like i said before,, it worked lkong enough to compile an AIC kernel. which ran for 3 days before playing with suns "fine" product. Slowaris runs fine (6 days so far) (and slow as a dog on a dx50 with 16MB) I will try to find time tommorow to play. eg: what will happen if I turn the conner off, and try to install,, if freebsd works as before, then possibly a conner which wont disconnect nicely? ooooH duh.. i just remembered I turned *off* the disonnect switch in the eisa cfg, as per recomendation in solaris manual for dealing with the old sun cdroms... eeek. <8-) anyways... I'll get back to you guys if I find anything. Im also willing to test any quickie-patches at work next week.. Rob Mallory [mallorrp@sce.com][rmallory@csusb.edu] ps: with reguards to firmware problems, the baracudas (my 15150n is at level 19, latest) are some of the touchiest drives I have ever seen. we have 4, 2 each on sbus FSBE cards, with 6 feet total cable on each bus, striped. 2 drives per stripe, seperate busses. 4.1.3U1b,O:DS They are *garunteed* to hang if dumping both filesystems at same time. they hang every other night now that i dump them sequentialy. comp.scsi shows one or three others with 15150n's like this...:( Don't it just suck theese days, as drive densities are getting tighter and tighter, techs dont get to spend enough time writing firmware which works the first time!!! > > > Do you really mean to try to tell me that two vendors have *identical* > > firmware problems? > > Firmware problems with similar symptoms, sure. > > > I rate the probability of that somewhere close to a comet hitting the earth > > today, especially when 2.0BSDI (which uses large transfer sizes and > > contiguous operations) has no such problem. > > If you weren't in such a responsible position, (and I covered in shame > after my recent outburst) I'd fall off my seat reading this. > > > Karl Denninger (karl@MCS.Net)| MCSNet - The Finest Internet Connectivity > > You're still not helping; even after Rod has enumerated _painfully_ > the list of things that are _needed_ to be known before anything > can usefully be done about your problem. >