*BSD News Article 5656


Return to BSD News archive

Xref: sserve comp.unix.bsd:5704 comp.os.linux:11028
Path: sserve!manuel!munnari.oz.au!spool.mu.edu!uunet!mcsun!uknet!doc.ic.ac.uk!ajt
From: ajt@doc.ic.ac.uk (Tony Travis)
Newsgroups: comp.unix.bsd,comp.os.linux
Subject: Problems reading partition table in 386bsd
Message-ID: <1a2o7mINNah3@frigate.doc.ic.ac.uk>
Date: 26 Sep 92 22:29:10 GMT
Reply-To: ajt@rri.sari.ac.uk
Organization: Dept. of Computing, Imperial College, University of London, UK.
Lines: 30
NNTP-Posting-Host: swan.doc.ic.ac.uk
X-Newsreader: TIN [version 1.1 PL6]

I've just re-installed 386bsd onto a 100Mb Quantum IDE drive on my
Elonex PC-320X (20MHz 386SX with 4Mb RAM) from the distribution
floppies and I noticed that, occasionally, 386bsd fails to read the
partition table correctly and reports that the 'b' partition overlaps
the 'a' partition so refuses to use it as a swap area.

I've checked using "disklabel wd0" and the 'b' partition appears to
start at cylinder 0 in the kernel's copy, when the error is reported.
But re-reading the label with '-r' shows that the label is intact on
the disk.

I suspect that this may have been the cause of the poor response I
observed from shell commands in the foreground whilst compiling the
GENERICISA kernel in the background.  With only 4Mb of memory and no
swap space it is likely that my foreground processes blocked.

As far as I am aware, this problem only occurs at boot-time and
repeatedly re-reading the label with "disklabel -r wd0" worked
correctly every time.

The remarks I made about the relative performance of Linux and 386bsd
on comp.os.linux were honest, but inaccurate.  Bill Jolitz was right
not to believe me and I apologise for misleading everyone.

	Tony.
--
Dr. A.J.Travis,                       |  Tony Travis
Rowett Research Institute,            |  JANET: <ajt@uk.ac.sari.rri>
Greenburn Road, Bucksburn,            |  other: <ajt@rri.sari.ac.uk>
Aberdeen, AB2 9SB. UK.                |  phone: 0224-712751