Subject: Re: NetBSD/arm32 on CATS
To: Jason Thorpe <thorpej@nas.nasa.gov>
From: Mark Brinicombe <mark@causality.com>
List: port-arm32
Date: 08/18/1998 21:28:23
On Tue, 18 Aug 1998, Jason Thorpe wrote:

> On Tue, 18 Aug 1998 16:03:49 +0100 (BST) 
>  Mark Brinicombe <mark@causality.com> wrote:
> 
>  > A number of MI PCI drivers can be used. The proviso is that they MUST be
>  > bus space and bus dma based drivers. Drivers that support bus mastering
>  > but not bus dma will not work with the ARM.
> 
> FWIW, I think this is all of them except:
> 
> 	SCSI: NCR driver.  "Good luck."  This driver is so badly structured,
> 	I'm not sure it's worth trying unless you just rewrite it.

Well in that case I'll put off looking at adding bus dma support. If I can
get my current to-do list down to a manageable size I may take a look at
this. (For backup purposed etc NCR SCSI cards offer a very cheap interface
~20 UKP) Adaptec or qlogic cards tend to be in the 80+ range.

> 	Ethernet: ThunderLAN driver.  Should be ~trivial to convert.
> 
> Two PCI network drivers currently slightly misuse bus_dma (and I feel
> somewhat guilty about the latter, because I wrote the bloody thing :-):
> 
> 	fxp (Intel EtherExpress PRO 10+/100B/100+), epic (SMC EtherPower-II)
>
> ...they don't do partial map syncs on the descriptor memory, because they
> were bus_dma'd or written before the partial sync support existed, and I
> have been too lazy to fix that :-)  Should be ~trivial to fix.  Anyhow,
> because of the missing sync support, you'll get coherency problems in the
> descriptor memory, possible, although I believe these drivers map it w/
> BUS_DMA_COHERENT, which should cause them to be mapped non-cached on the
> arm32, so they might work OK.

So far the main cards I have been using are all tulip based cards i.e.
dlink, netgear, lantronix etc.

Need to find a good source of PCI cards I can borrow here in the UK for
testing.

I really need to take a trolley round frys and get one of every PCI card
to try with the CATS board ;-))))

Cheers,
				Mark