Amiga GigaPD 3/Amiga_GigaPD_v3_3of3.iso/netbsd/docs/mailinglist-archive/1993-12/text0113.txt

Text File | 1993-06-25 | 1.7 KB | 35 lines

On Dec  7, 12:33am, Ty Sarna wrote:
> Not quite true...  Series II '030 Accelerator's controller can.  It'd be
> nice not to have to go through bounce buffers when it isn't neccesary.
> How to tell when it is and when it isn't? Well...  I would suggest
> looking at the autoconfig list.  If there is manufacturer 2017 memory
> starting at >0x00FFFFFF, it's safe to do 32bit DMA.  If there isn't 2017
> ram in the >24 bit space, it's either not safe or it doesn't matter.
> (alternately, I suppose you could look at the Mask values in the RDB, or
> make it binpatchable (blech).).

  I think for now it will probably have to be binpatchable :-(.  The RDB
values wouldn't necessarily indicate anything.  The HC8+ card driver will
accept any address and handle it correctly.

> Hmm, come to think of it, that method isn't 100% correct either.  If
> someone has a SII '030 and a plain SII, (do multiple SCSI controllers
> work anyway?) one needs bounce buffers and the other doesn't...  Hmm,
> maybe we can just get GVP to tell us how THEY tell the difference :-)

  I think multiple controllers will work, but I think you would have to
configure the kernel to only have support for the GVP controller and
specify an entry for each.  I'm not real clear on how everything fits
together, but it looked like defining one of each for the A3000, A2091,
and GVP11 controllers could cause problems if there are two or more
of the same controller type.

  I suspect that GVP may not have a problem telling the controllers apart.
When the SII '030 configures, it probably uses the driver from the '030
board.  When the Zorro II board configures, it would probably use the
driver contained on that board.  Not having a GVP '030 board, I can't
tell if the driver is exactly the same or not.

Michael