> > > From what I read the 3000 is a slightly earlier generation, not sure how close > it will be. Sounds like a good idea to get it going anyway. The thing I am > struggling with at the moment is getting sys$crmpsc_pfn_64 to work so I can > use C to test the NVRAM. I seem to have a problem with the service writing > back the address and length, but I don't yet know why. >
My program to test the NVRAM is not going well. When I try to write to what I think is the flashbus index register, I kill the machine stone dead. I'd just like to validate my approach, see if it makes sense: The docs says the index register is at 0x1.0000.0000 (physical I assume). On the assumption that the Alpha has an 8K page size that means the Page Frame Number for the physical page will be at 0x8.0000. So I am using sys$crmpsc_pfn_64 to start at that PFN. Correct? When I get the address back from the system, that is actually the address of the index register, because it is right at the start of page, so that is the address I access from my code. I was also unsure which region to use and I have tried P0 and P2, P2 seemed to not crash the machine but would give me ACCVIO instead, except that in my latest test P2 also crashed the machine. I am also specifying USER mode. I can post the code if that helps. Any ideas where I am going wrong here? Thanks Rob