On Tue, Aug 12, 2003 at 09:11:39AM -0700, Matthew Jacob wrote:
> Yes, this is likely a flakey scsi bus- a data overrun for a write
> generally means that the bus stayed in DATA OUT phase after all data
> scheduled to go had been sent.
> 
> The 'adapter resource shortage' has already been explained.
> 
> There's nothing wrong with isp cards. The 1020 is old and has signalling
> issues. The 1040s on forward are pretty damned solid.

Aren't "there's nothing wrong" and "is old and has signalling issues"
mutually exclusive..?

Could the issue be that the 1020 is using the 1040 'risc code' (whatever
that is, I assume its microcode of some sort) and thus isn't running the
correct driver, even though it does work? After all, all the pci isp driver
does is prints out that it is an isp 1020 Fast Wide (just recently fixed so
it no longer incorrectly says Ultra Wide) and proceeds to load the 1040 code
for the chip. Ultra-Wide and Fast-Wide are different enough that I think
that could at least *possibly* be an issue.

Well, I'll be looking into getting an LSI LSIU160 53c1010-based card this
weekend, since SRM's support for different scsi chipsets is lacking, and I've
been told that the 53c1010 works in an AXPpci33, so hopefully it should in an
AS600 with 1999-dated firmware.

-Sean

-- 
/~\ The ASCII
\ / Ribbon Campaign                   Sean Davis
 X  Against HTML                       aka dive
/ \ Email!