Return to BSD News archive
Newsgroups: comp.os.386bsd.questions Path: sserve!newshost.anu.edu.au!munnari.oz.au!news.Hawaii.Edu!ames!elroy.jpl.nasa.gov!usc!cs.utexas.edu!cactus.org!wixer!dpm From: dpm@wixer.bga.com (David Maynard) Subject: IDE vs. bad144 Message-ID: <1993Sep27.033241.15874@wixer.bga.com> Organization: Real/Time Communications Date: Mon, 27 Sep 1993 03:32:41 GMT Lines: 57 I am currently running NetBSD-0.9 on my 486/33 with IDE & SCSI disks. With the proper magic, my IDE drive (Conner CP30174E) does seem to automatically repair grown defects. Unfortunately, the proper magic does not seem to include normal accesses by *BSD. Twice in the past few weeks, my drive as developed defects that register as hard read errors under both FreeBSD and NetBSD. (I also tried putting a MSDOS filesystem on the partition and using Norton disk doctor, but that doesn't repair the defects either.) The hard errors persist until I use the BIOS disk utilities to run a surface analysis. The surface analysis is apparently enough to make the drive repair itself since it completes without finding bad blocks and *BSD no longer finds an error after the scan. Unfortunately, the surface analysis wipes the disk clean and I have to re-install enough of BSD to restore from tape. Since it seems to be becoming a habit, I thought I would try using bad144 to map out grown defects instead of doing the wipe/reinstall process ad nauseum. Since the FreeBSD install disks can supposedly be used to install NetBSD (with a little tweaking?), I decided to try their simplified bad144 support. I used the FreeBSD install disks and told them that my disk was an ST506 drive that did NOT automatically repair defects. The bad144 scan appeared to run successfully as did the copy operation for the files on the install filesystem floppy. However, when I halted the system and rebooted to copy the kernel to the hard drive, the boot sequence reported: wd0: cannot find label (bad sector table corrupted) I get essentially the same error from either the NetBSD or the FreeBSD kernel copy floppy. The only workaround I've found is to redo the *BSD install specifying the IDE drive type without bad144 mapping. I have tried the equivalent sequence using the NetBSD install floppies with identical results. Can anyone shed some light on: 1) why my IDE drive is so reluctant to repair errors and how I might coax it to fix them without destroying my BSD partitions, and 2) why I can't get bad144 to work on my drive? NetBSD runs well otherwise. Thanks, David P.S. I really like the new FreeBSD install floppies! The package management stuff is also a big win (and works fine under NetBSD 0.9). -- David P. Maynard, Carnegie Mellon University & Dependable Solutions USMail: 14312 Richard Walker Blvd, Austin, TX 78728-6862 EMail: dpm@depend.com, Tel: +1 512 251 8122, Fax: +1 512 251 8308