question about Fastlane Z3 problems

[Stefan Hensen <hensen%wpos4.physik.uni-wuppertal.de@localhost>]

Hi,

I have a question concerning the problems of the Fastlane Z3 with multiple
SCSI devices on the bus, which are mentioned in the installation
instructions. (Well, it's probably more about the general Emulex FAS
problems.) There is something that surprised me about this:

Around April 1995, after the first experimental (pre-1.1) kernels with
Fastlane Z3 support came out, I had my first look at NetBSD. At that time,
I could do the installation on a SCSI harddisk (Quantum Empire 1080S) in my
A4000 from a SCSI DAT tape drive (HP35480A), which were both connected to
the Fastlane, without any problems. I also could do tape backups and
restores without problems.
(After a while, I removed NetBSD from the system, because the system
recources were a bit too limited to do much useful with NetBSD and I had no
real need for it then ... )

Meanwhile, I have upgraded my system (more RAM, second SCSI harddisk,
CyberVision 64, ...), and recently I started again with NetBSD 1.2, because
I needed (and still need) to use certain unix programs. But this time, it
was not possible to do a tape installation as before, because the SCSI bus
hangs after a few 10kB of data transfer. The same happens when trying to
install from an AmigaOS partition on the other harddisk. Just the problems
mentioned in the installation instructions :-(
Finally, I have managed to install NetBSD on my second harddisk (Quantum
Fireball 540S) with help of the BFFS filesystem for AmigaOS. (That was a
bit tricky, but it worked :-)
All drives may be turned on and are recognized by NetBSD and the system
runs stable as long as only the NetBSD drive is accessed.

The big question is: What has happened to the Fastlane driver, that things
that used to work with an older version do not work any more?
(Or could it occasionally be that the Quantum Fireball causes problems that
the Empire does not?)

I have been thinking of problems with syncronous data transfer, but even
using the option '-I ff' for loadbsd does not help. (Under AmigaOS, all
devices seem to work in syncronous mode.)

The other thing I thought of was the Zorro III DMA, but even with a Buster
chip revision 11 and after letting rework the Fastlane card for the latest
board revision, the problems still exist. But I find this not so surprising
with the problems seeming to depend on the number of active SCSI devices ...

Unfortunately, I have no experiences in kernel programming and no technical
documentation (and my time is very limited at the moment), so I can't do
much myself about the problem. But maybe the answer to the above question
could help somehow to find out the reasons. I still have the old sources
somewhere around, in case somebody should need them.

Best regards,

Stefan Hensen