Hi,
we have just replaced an ADSP-2185M on a particular board with a 5V
ADSP-2185 (don't ask....it'd take too long), changed the psu etc. We
expected it to boot through the IDMA port just the same as the 2185M,
but found that it always crashes after the first couple of data
writes. After a lot of debug, we sussed that it seems to be treating
the second address write (the boot file expects to do an address
write followed by an overlay page write before starting on the data
writes), which happened to be '0x8000', as a write to PM address 0,
and therefore the CPU tried to start running as soon as the following
data write completed. - Still following? Good - not sure I am :-)
We 'cured' it by not doing the second write to the address space, so
just doing a single address write, followed by the data writes to
that patch of memory. Seems to work, but cannot find it documented
ANYWHERE. There are still some strange effects in the code....doesn't
look like some of the maths is working quite the same...so I am not
confident that we are quite there yet.
Does anyone have similar experience, or good advice on
other 'surprises' we may have in store?
Thanks,
regards
Jim
