Subject: tape drive woes (cf. PR #2727)
To: None <port-sun3@NetBSD.ORG>
From: Jari Kokko <jkokko@tnso13.ntc.nokia.com>
List: port-sun3
Date: 01/19/1997 10:22:35
In article <5bkovc$vo7$1@bilbo.ntc.nokia.com>,
Jari Kokko  <jkokko@cpcemail.ntc.nokia.com> wrote:
> Setting options to 5 removed the bug on my system completely. 

But it made the tape drive problem I've had with NetBSD since 1.1
worse:

options = 5 (PILVI #8)
panic: ncr5380_scsi_cmd: polled request, abort failed
panic: ncr5380_machine: no current cmd

options = 1 (GENERIC #89 (1.2 release))
si: DMA timeout (while polling)
si0: aborting but phase=DATA_OUT (reset)
si0: reset SCSI bus for TID=4 LUN=0
panic: ncr5380_scsi_cmd: polled request, abort failed

options = 7 (PILVI #7)
st0(si0:4:0): soft error (corrected), data = 00 00 00
st0(si0:4:0): soft error (corrected), data = 00 00 00

This last kernel has been patched with stuff (st.c) from Volker
Seebode, to prevent it from panicking with: 'strategy broken done < 0'

Also, options = 5 doesn't seem to enable disconnect/reselect. When
using the tape drive, the whole system freezes for instance when the
tape is rewinding.

On an otherwise great OS, this tape problem is painful, prevents
backups.

Jari Kokko

PS. PR #2727 deals with this.