From owner-freebsd-hackers Tue Jul 29 23:32:09 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id XAA12759 for hackers-outgoing; Tue, 29 Jul 1997 23:32:09 -0700 (PDT) Received: from bunyip.cc.uq.edu.au (daemon@bunyip.cc.uq.edu.au [130.102.2.1]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id XAA12700 for ; Tue, 29 Jul 1997 23:31:59 -0700 (PDT) Received: (from daemon@localhost) by bunyip.cc.uq.edu.au (8.8.5/8.8.6) id QAA03133; Wed, 30 Jul 1997 16:31:03 +1000 Received: from localhost.dtir.qld.gov.au by ogre.dtir.qld.gov.au (8.7.5/DEVETIR-E0.3a) with SMTP id QAA05299; Wed, 30 Jul 1997 16:21:56 +1000 (EST) Message-Id: <199707300621.QAA05299@ogre.dtir.qld.gov.au> To: Nelson Murilo cc: freebsd-hackers@freebsd.org, syssgm@dtir.qld.gov.au Subject: Re: SCSI 1542 error References: In-Reply-To: from Nelson Murilo at "Tue, 29 Jul 1997 15:11:28 +0000" Date: Wed, 30 Jul 1997 16:21:56 +1000 From: Stephen McKay Sender: owner-freebsd-hackers@freebsd.org X-Loop: FreeBSD.org Precedence: bulk On Tuesday, 29th July 1997, Nelson Murilo wrote: >One hour after install FreeBSD 2.2.2 on 486, with 32Mb and SCSI 1542C, >my console say: > >/kernel: sd0 (aha0:0:0): timeout >/kernel: adapter not talking commands... frozen?! >/kernel: aha0: MBO 02 and not 00 (free) >[...] I used to have this problem all the time running 2.0.5 (I think) using this driver with a BT-545S (which can run in 1542 compatible mode). My problem was that the 486DX4/100 I was using wasn't completely compatible with my old Dell motherboard. When I went back to a 33 (and then to a DX2/66) everything was fine. While I never proved my theory conclusively, I think it was a cache coherency problem with the DX4/100. Now I run 2.2.1 using the bt driver with a DX2/66 cpu. There are no problems. >Next step as change speed for 5.0 (default), disable all shadow and >disable check parity, but the problem not stop. My problem was solved by changing the CPU. You could try disabling L1 and L2 caches to see if your problem could be the same as my old problem. If this has no effect, try using only 16Mb of memory. If this helps, there must be bit-rot in the bounce buffer code. Of course, I'd be pissed if it turns out that it was just a code bug all along, and triggered by the faster CPU. I basically gave it away! Stephen.