Re: Blitz, Atomwide and STD

In article <[email protected]> you write:
>In message <[email protected]>
>          Ben Harris <[email protected]> wrote:
>> I've got the documentation for the Blitz, so if there's demand I'll have a
>> bash at adding support for it when I'm next poking around in icside.c. 
>> 32-bit mode is likely to be a bit of a pain, but 16-bit mode (both PIO and
>> DMA) is pretty similar to ARCIN v6.
>
>That would be good!  Consider this to be part of the demand!

OK.  I'll see what I can do.

>What makes 32-bit more difficult, just that it differs from the
>existing 16-bit mode, or is there some inherent problem?

The problem is that the Blitz doesn't behave quite like our standard IDE
driver expects 32-bit interfaces (e.g. on PCI) to behave.  I think the right
solution is for the generic IDE driver to provide a hook for bulk data
transfer to/from the drive so that cards with odd requirements (like the
Blitz and most minipodule IDE interfacces) can provide their own functions
for it.  I'll want to check this with other people first, though.

>Do any of
>the SCSI card drivers use 32-bit mode?

I don't think so, no.  I don't think we've currently got any drivers using
EASI, come to that.

>> The major problem is likely to be that the podulebus code doesn't
>> currently support podules whose ROM is in EASI space, which Blitz's is.
>> This means that a current kernel simply won't see the Blitz at all.
>
>Yep, I can confirm this - the Blitz is identified as an ICS IDE, so
>the manufacturer and podule ids must be the same as the ARCIN V6.

Ooh.  I hadn't expected it to even manage that.

>However, the identification string doesn't get displayed - presumably
>because it is in EASI space.

Possibly.  This isn't really a problem in any case.

>RO still works with the ARCIN, but BSD now panics, even when there is
>no drive attached (Yes, I have removed the Blitz!). Grrrrrr.
>I've tried a delete-power on, checked *Status, *Unplug's etc, but I
>can't see any obvious reason for the ARCIN to have stopped working.

How odd.  I'll try to make the icside driver generally more resilient to
finding the hardware in odd states and see if that helps.

-- 
Ben Harris                                                   <[email protected]>
Portmaster, NetBSD/acorn26           <URL:http://www.netbsd.org/Ports/acorn26/>