Dear Bertram,
   
   You wrote:
> >    I have always been hearing that both MFM and SCSI controllers on
> > KA410 can DMA only to a fixed buffer and not to arbitrary memory
> > addresses, and that this buffer is shared between the two.
>
> The 2000's manual clearly states these facts. RTFM.
   
   "I have always been hearing that ..." above includes this manual. I
haven't got it yet myself, but I have heard other people (including you)
say that it does say this.
   
> Reading sources is "reverse-engineering" ????
   
   Engineering is going from ideas to a usable product. Reverse-engineering
is taking a product and discerning what ideas were in the mind of the guy
who has engineered it. It doesn't matter whether the product is in the
source or the binary form, it's still reverse-engineering. The difference
is in the difficulty of the process.
   
> Ey, you've seen nothing else but what I've written into these sources.
> Away from some mis-understandings this doesn't prove anything.
   
   and
   
> You didn't show anything and you didn't prove anything.
> Initializing these registers with zeroes means nothing else but setting
> a base-address or base-offset (which happens to be zero) to the HDC9224.
> Writing zeroes into these registers is something I decided to do, there's
> no need to do it that way. Thus this can't prove anything.
   
   Nowhere does the code load the address of the buffer in the main memory
into any registers, and this does prove that the controller can't DMA
directly to main memory, only to a special buffer. NetBSD's hdc driver
moves the data between the main memory and the special buffer with bcopy(),
i.e., with CPU instructions.
   
> >    Bertram was talking about SCSI there, not about MFM. (He has chosen
> > not to support MFM on KA42. I promise not to repeat this grave mistake
> > in my work.)
>
> You don't understand. Period.
   
   What do you think I don't understand?
   
   Sincerely,
   Michael Sokolov
   Phone: 440-449-0299
   ARPA Internet SMTP mail: sokolov@alpha.ces.cwru.edu