khval
commented
3 years ago But that can have some bad side effects, we might need to provide EUAE cpu core, maybe it can be in its own library as well? The UAE CPU core has the advantage it allows virtualized memory, and you have different callback for different memory pages. Dis advantage its slow. But can it be used for some routines in some conditions, can 68K programs be broken up into parts, where some runs native on AmigaOS4, and some run on Petunia, and other routines runs on UAE cpu core, more dynamically?
Technically we can have chipset.library as back end, that can give the mode the ability to use fake virtualized chipsets, that again can give it cooper, and bitter, provided the 680x0 program knows that chipset is not located at $DFF000 but at some other address. (Petunia / OS4 interpreter does not do stuff on write or read, and MMU interrupts is slow.), Macros works as proven before, but is lot work to rewrite old code, and code can break because routines need more space.