Re: [hatari-devel] Adding cache support for the MegaSTE -> DONE

[Nicolas Pomarède <npomarede@xxxxxxxxxxxx>]

Le 04/09/2024 à 21:30, Eero Tamminen a écrit :


> PS. Wouldn't most correct behavior for SafeCopy/SafeClear would actually 
> be to have last pieces of data written to memory, going also to cache, 
> instead of cache having been flushed?

That would be rather complicated depending on the cache type. For 
MegaSTE's cache the structure is rather simple and it has the advantage 
of being unified between code and data, so you don't really have to 
worry about updating one data cache or one instr cache and pushing into 
cache every read/write made by stMemory functions could be doable.

for cpu >= 68020 it would be more complex due to having 2 caches.

also many programs or even TOS expect the cache to be wrong after doing 
some IO access and will explicitely flush the cache. So trying to 
maintain the cache's consistancy anyway would be a lost effort.

I'm not sure it would be worth the effort of adding code for that, or 
even speding too much time thinking about how to do it :)

Nicolas