*BSD News Article 2081


Return to BSD News archive

Path: sserve!manuel!munnari.oz.au!spool.mu.edu!yale.edu!ira.uka.de!smurf.sub.org!rnihd!rnivh!andrew
From: andrew@rnivh.rni.sub.org (Andrew Wheadon)
Newsgroups: comp.unix.bsd
Subject: MFM bug? and particial success
Keywords: bug,386bsd
Message-ID: <2304@rnivh.rni.sub.org>
Date: 19 Jul 92 22:20:30 GMT
Article-I.D.: rnivh.2304
Organization: Rhein-Neckar-Informationsdienst Viernheim
Lines: 39

After getting 386BSD on Friday I've spent my time during the
last few days trying to install it. After figuring out how
to partition before hand to force it to use the cylinders on
the disk up to 1019 instead of ca. 965. I ran into a major
difficulty.
  Whenever a file on a bad sector is read the kernel just 
hangs. The Problem was my first setup put compress on a 
bad sector causing me to think the install script was buggy.
  After moving the beginning of the partition past the two
bad sectors on the HD the installation seemed to work as far
as installing the Tiny-Disk was concerned yet as soon as I
tried to extract the bin01.?? files which had apparently been
flawlessly transferred to my Disk  extract would cause the
machine to hang. After a complete unsuccessful reinstallation
I discoverd that I could also reproduce the effect by cat'ing
the files to /dev/null Those files which caused the system
to hang were moved to a /.bad directory and extract got as far
as unpacking the files, but would complete due to a similar
problem. 
  Finally I have succeded in finding all the bad sectors at
the beginning and middle of the disk by writing files (which
all get written without an Errormessage) and catting them and
moving the bad ones. Of course for some reason something which
is read by fsck is also on a bad sector so I can't fsck at the
moment but otherwise everything except two header files in the
include/g++ directory seems to be fine.

I tried to tell 386bsd about the bad sectors with bad144 before
and after finding them with the above discribed procedure but to
no avail. 
Does anyone know a way of telling the system where the bad-sectors
are so that it doesn't try to write or read them ?
My Config: 386DX/387(IIT sp?) MFM Seagate st4096 80MB.
(I would have mailed a bug report to ljolitz but I can only
send/receive international mail from my own machine (andrew@acwbust.sub.org)
and my sendmail.cf isn't working yet under 386BSD 0.1. )
So please reply by news or to andrew@acwbust.sub.org not
rnivh.rni.sub.org. 
Thanks for your patience.Bye