Leo Weppelman, the atari port maintainer, has put together a todo list for the NetBSD/atari port. If you're interested in helping on any of these please contact Leo and/or the port-atari mailing list.
The new driver code does not work at all and the old driver code only works partially. The problem seems to be somewhere in the DMA handling. Matching the cards cache-line size and buffer alignment to the DMA burst requirements of the Hades does not work however. Since there is absolutely no cooperation from adaptec in supplying info, the solution probably depends on black-arts. I am playing with it every once in a while...
Mostly ready for the Hades VGA part. The keyboard part is nearly finished, thanks to Thomas Gerner. Nothing has been done yet on the the framebuffer (TT/Falcon) side of things.
These pose a problem since I somehow these cards need to be tweaked
into standard VGA mode to make them work on NetBSD.
I have been working on this. Currently this is stalled.
I am currently working a bit on this. I used an x86 emulator
(ftp://ftp.scitechsoft.com/devel/x86emu/)
in conjunction with a
simplistic /dev/pci (to make pci spaces accessible in userland).
I have the emulator working up to a point that is correctly runs
the bios of my et6100 and S3Trio.
It turns out that the Milan does not need this, the Milan-BIOS
initializes the card to a pretty standard mode. On the hades with
a Mach64, it is unusable. They removed the BIOS rom from the card.
I punted on this one. The old driver had so much interaction with main memory (both DMA and CPU), that it was nearly impossible to tell when to byte-swap and when not. I haven't looked closely at Manuel's latest driver, but this situation might have improved. However, I traded my 810 card for an et6100. Hopefully someone else can pick this one up.
For systems with more than 4MB memory, I'd like to move to sysinst and
a 1.44MB floppy image. Systems with less then 4MB must stay on a 1MB
floppy image (memory constraint). I think they will have to do with
the current minirootfs-install with networking ripped out. The latter
is caused by a tendency for all tools to grow fatter over the years.
I cannot make the program set of the 1.3 miniroot fit in 1.5 :-(
This item involves the implementation of sysinst for the atari, create
atari specific docu and test it.
Update 11Mar. 2002: For small systems (works for bigger too), there
will come a bootstrap using the swap partition. I already created a
file2swap.ttp tool that transfers an image to a designated swap
partitions. This lifts some size contraints for the miniroot. Setting
up sysinst remains of course.
Although the Milan kernels are part of the 1.6 release, they are still experimental. Not all standard hardware is properly supported (like nvram).
Currently not supported. Porting efforts should be done by someone who owns one. I will help out in ideas and other secondary support, but I won't do the bulk of the porting effort.
It would be nice to have snapshots out on a somewhat regular basis.
This would be nice to have.
Implement the generic softint mechanism.
There are a few items here. Most of them should be repeatedly run over.
|
|