Subject: Conference results
To: None <amiga@NetBSD.ORG, amiga-dev@NetBSD.ORG>
From: Markus Illenseer <markus@TechFak.Uni-Bielefeld.DE>
Date: 01/16/1995 18:49:14
This weekend has quite been eventfull for me. We had sort of Amiga
developer conference, among them a dozen of NetBSD-freaks. Markus Wild,
Hubert Feyrer, Klaus Burkert, Carsten Hammer, Ignatios Souvatzis and
even Michael v. Elst to name some.
Besides some hacking round the world (we had Internet access) we even did
something purposefull and tried to hack together something usefull.
The ethernet-driver for the Ariadne card has been tested throughout the
meeting and has been proved to be somehow stable. It does as yet not work
properly on Bernd Ernesti's cyberstorm, but I strongly believe this is due
to cyberstorm, beause he has quite a few problems with this setup. The driver
was able to blast at a speed of upto 360K/s (peak, tested with NFS and
ftp transferring a *large* file between two Amiga 3000). This rates the
Ariadne faster than the A2065.
ISA adaption from Niklas Hallquist (who has been on IRC for the
first time to 'join' us virtually :-) proved to be very functional though
not yet stable. Carsten Hammer adapted his ISA-stuff to the Crosslink
card and ran a kernel with support for ISA-Bus ethernet card. Last minor
changes must be implemented though. Transfer of large files (many packets)
fails yet. Same to the serial driver which was not able to drive SLIP yet,
though this might be a configuration problem.
For graphic boards users we have two news.
The white-screen problem of the Cirrus-graphic board driver has been
pointed out and most probably been fixed even. Klaus told me that Dagge and
me forgot that we initialize (we have to) the card twice but only once
did it correctly.
Bad News for A2000/040 with Picasso (maybe even Piccolo and Spectrum)
users. If you encountered the so called black-screen problem, you are out
of luck for now. We need a major rewrite of the grf_cl driver to make
it possible to even initialize the card (black = not enabled or initialized).
The problem is quite serious: Both Fusion40 and GVP040 have a special way
to implement traffic from and to the Z-Bus and therefor memory. Due to
this addressing scheme the board is not addressed at the real place in
the memory, we need to implement an offset for this purpose. The problem
now is to first get the offset (not simple as far as i understood) and
then use it in all the access-routines (simple by using macros). This
process may take quite a while, as for I have lot of other tings with
higher priority in my ToDo-queue.
Good news for Arcnet-fans among the readership. Ignatios has hacked a
first version of if_sa.c, a driver for the Arcnet 2060 card. It will
support almost all features of several RFCs for Arcnet, but not yet the
latest one with the proxy features. As for yet the driver can send packets
but not yet correctly receive them, we still seek out for lost packets
from the meeting, if you find them, please call Ignatios .-)
Last but not least i must again send big thanks to the core team
of the release distribution - the default distribtion again turned out
to be very usefull. ie. I had not even to bother with sendmail.cf -
it worked out of the box.