Justin R. Smith wrote:
> 
> I installed the HWDIRECT.356 kernel. It boots exactly the same as all the
> previous kernels I tried: the boot appears to be normal up to the point
> where the screen goes white and the message
> 
> Preserving 310892 bytes of NetBSD symbol table
> 
> then the system hangs.
> 
> As I understand it, there were always two problems with the P475:
> 
> a. the 68LC040 problem (which causes bus errors during operation)
> 
> b. the ADB problem -- which causes the kind of boots I have experienced.
> 
> It appears that the ADB problem still remains.

As Justin noticed, and I forgot to mention to all of you before
I went on vacation last week, there is a new HWDIRECT kernel.
You can get details at http://www.macbsd.com/~jpw/adb.html

Highlights are:
1. Cuda support is complete. This means that ADB should work
   properly on all desktop machines now.
2. Logitech Mouseman mice that pretend to be Extended Mouse
   Protocol mice but really are NOT are now supported.
   (Specifically, this means Logitech Model Number M-AC13-4MD)
3. If booting with serial console, ADB device setup is still
   done so that IIsi-based and Cuda-based ADB models can still
   get the proper time and date. It doesn't matter if you don't 
   have any ADB devices attached, it case you are making a
   "headless" machine.

These changes have been submitted for inclusion into the -current
tree, for those that roll their own kernels.

To answer Justin's specific issue, the ADB _should_ be working
in the P475. The failure that you are seeing, which also happens
on several other machines (like the Q605) is WAY WAY before the
ADB setup, it is not caused by lack of or broken ADB support 
as you suggest. I think it's some sort of memory mapping issue 
or something else very early in the kernel boot sequence. Since
I don't know much about that code, I can't really tell you exactly
what the problem is, though.

The reason I say the P475 _should_ work (ADB wise) is that it
has the Cuda ADB chip in it, and I think I finally nailed down
the Cuda related ADB problems. Once the other early boot problem
is solved, we'll be able to confirm that the ADB on the P475 
works, and I'm fairly confident that it will without any further
changes to the ADB code.

Later,

	--John