[cc'd to port-i386 since the discussion started there]

> I'm trying to get an SCSI card a friend gave me to work, and someone
> mentioned the PR you sent in.
> 
> I'm having some problems with it. Most of them I describe in the note I
> am enclosing below (I thought Rene had written the code - I didn't look
> too close).
> 
> The main problem I have now is that patch doesn't seem to like the diff as
> it got to me. It seems to loose sight of the fact that there are multiple
> patches to one file, and keeps asking for the file to patch. :-( Can you
> mail me the patch directly?

No problem... patch coming by seperate post...

> > > A friend gave me an ISA SCSI interface, and I'm trying to get it to work.
> > > It's based on a DTCT 436P chip, and shows up as:
> > > 
> > > isapnp0: <DTC436E, DTC436E, , > port 0x240/32 irq 5 not configured
> > > 
> > > Does anyone know if that matches a "standard" scsi driver?
> > 
> >   The DTC436E is a NEC 53C80 clone that can be found with SCSI cards
> > that come with some scanners like the UMAX Astra 1200S.  There is
> > currently no native support for this device in NetBSD 1.3.X.  However,
> > in PR 6274 you should find a device driver patch for NetBSD 1.3.2 that
> > gives you an "nca" device for 53C80 compatible devices (see
> > http://www.NetBSD.org/cgi-bin/query-pr-single.pl?number=6274).  After
> > applying the patch you have to add
> 
> I'm a bit confused. The driver seems to want 16 i/o ports, but the isapnp
> mapping is wanting 32 ports. What's up with the other 16?

To tell tell the truth, I'm not sure... A real 53C80 has 8 ports while a
53C400 has 16. 

>From what I can tell, there are a number of chips which are based around
a 53C80 core, but seem to do things differently. For instance, NatSemi
had an "enhanced" chip which purported to be faster. OTOH, Trantor had
at least one custom varient (the one used in the T-160) which only
appears to be compatible at the lowest levels (i.e., it work fine
in polled mode, but seems to have its own ideas about DMA-driven modes).
(This is why T-160s don't seem to run with most "generic" 53C80/53C400
drivers.
 
> Why won't the M.I. testing code work? What would need changing? One reason
> I'm asking is that I'm interested in an isapnp driver rather than an isa
> one. If the M.I. chip test can be made to work, then dev/isa/nca.c becomes
> nca_isa.c and only needs isa attach code, and dev/isapnp/nca_pnp.c only
> needs pnp attach code. 

Well, to be honest, I didn't take enough of a look at the MI driver to even
know that there was a chip test. (Horrible confession)

Second, PNP NCR 53C80 is almost an oxymoron. Most of these cards are pretty
simple affairs with little more than some glue logic and a lot of jumpers.
 
> I'm confused about the testing state. The pr says it works, but your EMail
> says it doesn't. I gather that it works on everything other than the 436?
> :-(

I gather that the 436 needs some more special handling. Sigh...
 
<<<john>>>