*BSD News Article 18625


Return to BSD News archive

Path: sserve!newshost.anu.edu.au!munnari.oz.au!news.Hawaii.Edu!ames!agate!howland.reston.ans.net!ux1.cso.uiuc.edu!roundup.crhc.uiuc.edu!uivlsisd.csl.uiuc.edu!terry
From: terry@uivlsisd.csl.uiuc.edu (Terry Lee)
Newsgroups: comp.os.386bsd.bugs
Subject: Re: pk0.2.4: error with DKFL_QUIET (wd.c), w/ patch
Date: 19 Jul 93 22:28:51 GMT
Organization: Center for Reliable and High-Performance Computing, University of Illinois at Urbana-Champaign
Lines: 61
Message-ID: <terry.743120931@uivlsisd.csl.uiuc.edu>
References: <1993Jul16.125843.6987@hq.demos.su> <GENE.93Jul17073841@stark.uucp> <22br6bINNi58@kralizec.zeta.org.au>
NNTP-Posting-Host: uivlsisd.csl.uiuc.edu

bde@kralizec.zeta.org.au (Bruce Evans) writes:

>In <GENE.93Jul17073841@stark.uucp> gene@cs.sunysb.edu!stark (Gene Stark) writes:

>>In article <1993Jul16.125843.6987@hq.demos.su> apg@demos.su (Paul Antonov) writes:

>>   With newest patchkit 0.2.4 it's virtually impossible to install
>>   386bsd to empty disk drive because of small bug in wd.c: DKFL_QUIET
>>   must be set if proper disklabel doesn't exist.

>>   [patch omitted]

>>Actually, wouldn't it be better to revert to the way the old driver was:
>>to set DKFL_QUIET *only* if one is attempting to open the raw drive
>>(d partition)?

>No.

>All these versions are buggy:

>The pre-patchkit-0.2.4 version only cleared the DKFL_QUIET flag for the
>'d' partition, and then only if the 'd' partition happens to be opened
>first.  (The comment in the source still mentions the 'c' partition, but
>is out of date.)  The 'a' partition is usually opened first so the error
>messages are usually turned off.

>The patchkit-0.2.4 version never turns on the DKFL_QUIET flag, so the
>error messages are always turned on for all partitions.

>The patch for the patchkit-0.2.4 version always turns on the DKFL_QUIET
>flag, so the error messages are always turned off.

>I prefer getting an error message for all errors.  The 0.2.4 version
>can be salvaged by not setting errno after reading the disk label fails
>iff the partition being opened is the raw partition.

>Ignoring the error like the other versions do is completely wrong.  At
>best the unititialized copy of the disklabel contains harmless data.
>-- 
>Bruce Evans  bde@kralizec.zeta.org.au

	First, an apology from the one who submitted the original patch to the
patch kit:  I didn't think of testing the patch on a drive that didn't already
have a disk label, and I'm sorry about the grief it caused for people who tried
to install 386BSD using the patchkit-0.2.4 floppies.  Now I finally know what
the DKFL_QUIET flag is for.  :-)  I have made a hopefully correct patch to
solve both the disk label problem and the bad block mapping problem, but I
haven't had a chance to test it out (hopefully tonight).  As for a temporary
work around for people who are trying to install 386BSD, use the original
dist-0.1 floppy (should still be on agate) to create the disk label ONLY.
After that, it should be OK to use the patchkit-0.2.4 floppies to finish the
installation.

----------
Terry Lee
terry@uivlsi.csl.uiuc.edu
-- 
----------
Terry Lee
Coordinated Science Lab
University of Illinois