register translation about 6502-cpp HOT 6 CLOSED

In his talk I believe @lefticus mentions that the leftover registers are assigned some fixed addresses in the zero page which the 6502 will access faster than other RAM.

from 6502-cpp.

Oh, right. But how can we be sure that the program doesn't use it?

from 6502-cpp.

That is correct. I use zero page mapping of x86 registers. The mapping is here: https://github.com/lefticus/x86-to-6502/blob/master/src/main.cpp#L75-L92

I'm relying on two things specifically

1. I'm targeting the C64, so I know that the ROM/hardware isn't using those locations: http://sta.c64.org/cbm64mem.html
2. I'm assuming that the entire program executing is the one that was translated from x86

from 6502-cpp.

Ok, thank.

from 6502-cpp.

How best to implement alternative zero-page mappings? Implement get_register as a function pointer and write multiple implementations? Make new C64 and A2 classes that inherit from a the base 6502 class and implement their own get_register methods?

I'm specifically thinking of targeting Apple II.

http://apple2.org.za/gswv/a2zine/faqs/csa2pfaq.html#017

from 6502-cpp.

Since this is just a huge experiment anyhow and I have no idea what would be the best answer in the long run - I say go with the function pointer (or std::function, or template Callable thing?) and we'll see how it ends up working out.

from 6502-cpp.