*BSD News Article 16085


Return to BSD News archive

Newsgroups: comp.os.386bsd.bugs
Path: sserve!newshost.anu.edu.au!munnari.oz.au!news.Hawaii.Edu!ames!think.com!cass.ma02.bull.com!melb.bull.oz.au!sjg
From: sjg@melb.bull.oz.au (Simon J Gerraty)
Subject: New SCSI does not read last blocks on tape: aha1542b, Archive Viper 150
Message-ID: <1993May15.074104.12425@melb.bull.oz.au>
Organization: Bull HN Information Systems Australia
Date: Sat, 15 May 1993 07:41:04 GMT
Lines: 62

I posted some time ago about problems with "soft errors" with an
aha1542b driving my old 60Mb Archive tape drive using Julian's
SCSI drivers.

I received a patch from one source which basically replicated my
own efforts, which experimentation had shown did not solve the
problem.

The basic problem is (I'd guess) that the drive is reporting a
partial write, but either it doesn't say or the driver does not
take note of how much data was written.

The only other suggestion I received was to get a newer tape
drive.  Well, I've been testing it today with an Archive 150Mb
"Viper" drive:

May 15 10:34:51 zen /386bsd:   dev5,lu0: type \
	0:1(sequential),removable 'ARCHIVE VIPER 150  21822-001' scsi1
May 15 10:34:51 zen /386bsd:    st1: scsi tape drive, 0 blocks \
	of 512 bytes

Using this drive and a stock scsi driver setup (as per patchkit
0.2.3) I get none of the "soft error" messages, my applications
read/write happily - BUT I still lose data.

dd bs=8k < usr.src.cpio > /dev/rst1
dd bs=8k < /dev/rst1 > /tmp/tape.data

-rw-rw-r--    1 root     wheel    42415616 Apr  6 20:23 \
	/l0/tmp/usr.src.cpio
-rw-rw-r--    1 root     wheel    42409984 May 15 16:28 \
	/tmp/tape.data

That's 5632 bytes (11 * 512 tape blocks) short.

tail /tmp/tape.data | cat -vt 

shows that we are definitely short of useful data (about half
way through sys.386bsd/sys/socketvar.h last file in the archive
apart from a symlink)

This is 300% better than the result I get with the older 60Mb
drive which typically hits a write error at about 11Mb.  The
60Mb drive was able to read all my old dump and cpio archives ok
except that it always failed to return the last few blocks of
data - as is the case above with the 150Mb drive.

Has anybody else seen a similar problem? 
Anyone tried restoring from their backups?

Has anyone patched st.c to cater for QIC24 drives? 
I would like to still be able to write 60Mb tapes, even if I buy
a 150/250 Mb drive. 

BTW, thanks to all who responded to my post about keyboard
hangs, the patch (from chmr@edvz.tu-graz.ac.at) worked nicely.
Unfortunately I was chasing hardware faults due to some info
supplied to a friend running the same hardware...
-- 
Simon J. Gerraty        <sjg@zen.void.oz.au>            (home)

#include <disclaimer>   /* imagine something *very* witty here */