Add new platform(s): Foenix F256 Jr & K

[bcampbell]

https://c256foenix.com/f256k and https://c256foenix.com/f256-jr
I think they're similar enough to just be a single target.

I've got a very rough first pass going at: https://github.com/bcampbell/llvm-mos-sdk/tree/f256k

It's pretty specific to my own current project, but it'd be good to get some critique and feedback and start it heading down the path toward upstream inclusion...

[sduensin]

I built this and made a simple binary to test with FoenixIDE. Fantastic!

[mysterymath]

The changes you have look more or less like I'd expect; I think we can hammer out the details in a PR if you're up for it.

[bcampbell]

I'm going to be traveling overseas for the next month, so I'm probably not going to have a whole lot of time to put into this. Happy for someone else to pick it up, or I'll get stuck into it properly when I have more time for it.

[bcampbell]

One thing that was a little odd - I had trouble with some for loops which just filled a block of memory with a value. They just weren't working. Turns out they were being optimised to use memset instead. For whatever reason, that seemed to not work right. Using -fno-builtin-memset got the loops behaving as they should.

So I suspect I've screwed up something with the runtime...
TBH I'm pretty shaky on all the merged utility routines to initialise bss, copy data into zp, etc etc... I just added the ones that sounded right and fiddled a bit. So I don't have a lot of confidence I got it 100% right :-)

[sduensin]

I had trouble with some for loops which just filled a block of memory with a value. They just weren't working. Turns out they were being optimised to use memset instead. For whatever reason, that seemed to not work right.

Just ran into this myself. -fno-builtin-memset also helped in my case.