MMUCAT - PowerCL_EXT NONEXISTENT XROM call

[Monte Dalrymple]

(02-15-2017 12:21 AM)mfleming Wrote:  I had the same problem with MMUCAT. Plugging in OSX3 solved the problem, but I did not have the extended catalog functionality of OSX3. The arrangement was YFNX in page C and OSX3 in page D.

When I swapped the order of the modules so that OSX3 appeared in the lower numbered page before YFNX, the extended catalog function worked again (really missed that!) Is there a reason that the order of these two modules is significant?

There are a handful of flag bits in the Nut CPU that retain their state during deep sleep. But in the NEWT CPU, since the FPGA is powered down during deep sleep, the state of these bits is lost. To combat this, YFNX saves the state of these bits in RAM before the machine turns off, using a polling point. But OSX3 uses some of these bits to hold status, and this status is updated by the OSX3 polling point. For this reason the OSX3 polling point code needs to run before the YFNX polling point code runs, and the way to guarantee this is to have OSX3 plugged into a lower page number. I didn't discover this until well after the 41CL had been released, which was actually a good thing, because if I knew about it before the release I might have abandoned the project, thinking that there was no way to preserve the state of those status bits. (Necessity is the mother of invention.)