Tim Hutt

Oh yeah I like that.

I don't think the old condition was wrong was it? These are just two different options for what to do when there's a misaligned access that overlaps the edge of...

Ah yes that looks correct.

I think normally you would have one PMA for your entire MMIO region that wouldn't support misaligned accesses at all.

I added #1389 to track making this all make sense, but this fix is at least correct - are you happy to merge it and defer the rest of the...

I assume this was intended just to make the log more informational? But yeah I agree with @jrtc27 - we should probably stick to assembly that actually works.

Yeah I think the proper way to solve this is with the planned callback API or machine readable log format. You can extract the instructions opcode and then use a...

I don't think we're going to do this so I'll close it. Hope that's ok!

That was changed to this in recent specs (including the latest ratified one): > Instructions that access a non-existent CSR are reserved. So I think we do need this flag.

(or at least, it isn't wrong, and probably implementations will behave like this)