From owner-freebsd-hackers Wed Jan 4 22:49:00 1995 Return-Path: hackers-owner Received: (from root@localhost) by freefall.cdrom.com (8.6.9/8.6.6) id WAA26653 for hackers-outgoing; Wed, 4 Jan 1995 22:49:00 -0800 Received: from perth.DIALix.oz.au (uucp@perth.DIALix.oz.au [192.203.228.2]) by freefall.cdrom.com (8.6.9/8.6.6) with ESMTP id WAA26646 for ; Wed, 4 Jan 1995 22:48:52 -0800 Received: from jules.UUCP (uucp@localhost) by perth.DIALix.oz.au (8.6.9/8.6.9/DIALix) with UUCP id OAA24045; Thu, 5 Jan 1995 14:25:30 +0800 Received: by jules.DIALix.oz.au (8.6.9/25-eef) id OAA00515; Thu, 5 Jan 1995 14:21:46 GMT From: Julian Elischer Message-Id: <199501051421.OAA00515@jules.DIALix.oz.au> Subject: Re: Bad sectors on SCSI drive! To: dufault@hda.com (Peter Dufault) Date: Thu, 5 Jan 1995 14:21:46 +0000 () Cc: jkh@time.cdrom.com, mtaylor@gateway.cybernet.com, hackers@FreeBSD.org In-Reply-To: <199501041055.FAA01628@hda.com> from "Peter Dufault" at Jan 4, 95 05:55:38 am X-Mailer: ELM [version 2.4 PL24] Content-Type: text Content-Length: 1468 Sender: hackers-owner@FreeBSD.org Precedence: bulk > > Jordan K. Hubbard writes: > > > > > Allright, so what can be done? How can I map out bad sectors on a scsi > > > drive? Do I have to newfs the drive? > > > > Return it to the manufacturer. A good SCSI drive shouldn't develop faults > > like this, and when it does, it's time to shoot it. > > If you can't return the drive: > > I don't think Jordan's statement is universally true. Mode page > 1 for a direct access device has a "AWRE" (automatic write > reallocation) and "ARRE" (automatic read reallocation) selections > that can be turned on and off. > > I'm not sure why these would be turned off. > > Also the spec reads that the automatic reallocation will be performed > only if the target has the valid data. Maybe on some drives there > are circumstances where that can happen. when the read error is sobad it can't be reconstructed by the drive...... > > If you're running -current you can apply the patches in > freefall:pub/incoming/sutar7.gz , rebuild the kernel and run the > "ssec" command to slip the sectors by block number and the "mode_sense" > command to see what the mode page settings are. this command is in the driver... just needs an ioctl to call it...... of course then you have a bad block in a file.. use tar cf /dev/null / to find the bad file BEFORE you fix the error the bad file will be mentionned on stderr.. otherwise you'll never know which file is corruptted with a block of nulls in it's middle.. > julian