From owner-freebsd-hackers Thu Sep 28 12:30:56 1995 Return-Path: owner-hackers Received: (from root@localhost) by freefall.freebsd.org (8.6.12/8.6.6) id MAA26343 for hackers-outgoing; Thu, 28 Sep 1995 12:30:56 -0700 Received: from grunt.grondar.za (grunt.grondar.za [196.7.18.129]) by freefall.freebsd.org (8.6.12/8.6.6) with ESMTP id MAA26336 for ; Thu, 28 Sep 1995 12:30:42 -0700 Received: from grumble.grondar.za (grumble.grondar.za [196.7.18.130]) by grunt.grondar.za (8.6.12/8.6.9) with ESMTP id VAA13909; Thu, 28 Sep 1995 21:30:26 +0200 Received: from localhost (localhost [127.0.0.1]) by grumble.grondar.za (8.6.12/8.6.9) with SMTP id VAA18103; Thu, 28 Sep 1995 21:30:24 +0200 Message-Id: <199509281930.VAA18103@grumble.grondar.za> X-Authentication-Warning: grumble.grondar.za: Host localhost didn't use HELO protocol To: "Justin T. Gibbs" cc: Mark Murray , Andreas Klemm , hackers@freebsd.org Subject: Re: make world on FreeBSD-stable impossible. cc1: ... signal 11 Date: Thu, 28 Sep 1995 21:30:24 +0200 From: Mark Murray Sender: owner-hackers@freebsd.org Precedence: bulk > >You are not alone. I have a 486DX4/100/PCI and an adaptec 2940, and I get > >these the whole time, as well as the signal 11's. I just tried a reboot, > >ant the files that were corrupted before are now OK???!! They are corrupted > >in exactly that same sort of way you are reporting. > > > >My kernel is stable, the rest is a mix (a lot hand-installed). > > > >M > > The only known problem with the aic7xxx driver has to do with losing bytes > (the transfer leaves a residual of 1-13 bytes). This only happens when > the transfer is going faster then 10MB/s (like a wide cappella or atlas), > so a narrow device shouldn't show this problem. Single bit errors are > almost always ram or cache problems. The driver supports full parity > checking on its data up to the point that it is transfered to host memory, > so I don't think this is the driver's fault. I don't think you are right. I have slowed my Adaptec down to 5MB/s and I get (got) 2 errors: 1) Single bit errors - fixed by banging the box a bit an reseating the chips (cache etc) 2) Missing chunks of code - 1-13 looks about right. This is with an 2940 and an HP scsi2 (narrow) disk. Replacing the 2940 with a 1542 fixes the problem. The problem only occurs after a great deal of activity (like a big system build), and is repeatable on the broken file until a reboot. After the reboot the file is fine. If a sig11 occurs, it is also repeatable until a reboot. M -- Mark Murray 46 Harvey Rd, Claremont, Cape Town 7700, South Africa +27 21 61-3768 GMT+0200 Finger mark@grumble.grondar.za for PGP key