On Fri, 22 Nov 2002, Matthew Jacob wrote:

> [ I'm the isp author ]
> 
> I wouldn't necessarily try and reflash this card- it's a special for
> MACs I believe.

	I don't see how, it is my understanding that the same card should 
work in every platfrom, x86, Sun OpenFirmware and Macs, all you had to do 
was flash it for OpenFirmware or x86 if you wanted to boot from it on a 
PC. I'm probably wrong though.

> Since you have some cards, can you loan me one to put into one of my G4s
> so I can try and put this to bead once and for all?

	Sure! Just hit me up off list with preferred shipping method and 
where to send it.

> A 'RAM Checksum failure' has to do with downloading f/w to the card and
> then issuing a mailbox command to ensure that the download worked. The
> fact that it didn't means that SRAM has been corrupted.

	Well it did download something to the card that worked. When I 
first installed the card, the open firmware device tree reported it as 
/bandit/1077,1020 (I think). After I tried to boot NetBSD on it, halted 
back to openfirmware, it is now listed in the device tree as 
FWB,JackHammer and has several more entries in .properties. No words still 
though, blah. There's gotta be space in there somewhere for OpenFirmware 
boot code :) Scrap the Apple ROM driver!

> There are two possible areas where we might need to adjust:
> 
> a) For 'jackhammer', do *not* download newer f/w. Live with whatever OBP
> or other PROMS have set up.
> 
> b) For 'jackhammer', the memory timing register (MTR) on the ISP chip
> should *not* be reset to that which is normal for any other QLogic- it
> has to be left alone (because somebody did a completely different memory
> I/F for the SRAM that the chip is talking to). This is an area of h/w
> support that is not something that is easy to infer from any parameters
> or first principles.
> 
> But to check all of this out, I need a bloody card. Can you loan me one?
> 
> -matt