Warner Losh wrote:
> In message <email@example.com>
> "Bradley D. LaRonde" writes:
> : Clio's processor is Vr4111, and I've been treating it like a R3000.
> You should be treating it as if it is a R4000. The TLB entries are
> exactly like a R4000... The locking functions may be different, but
> if you treat the MMU as a R3000, bad things will likely happen.
I'm starting to agree with Warner about it being closer to R4000 than R3000
as far as the kernel is concerned, but the fact is that we cannot simply
treat it as either an R4000 or R3000. It's not exactly like either one.
What we need to do is define the CPU as a VR41xx and then look at every
piece of the kernel code that is CPU-type specific, adding the appropriate
For the MMU code, it'll be R4000-like. For the exception-handling code,
it'll be mostly R4000-like, but parts will probably be neither exactly
R4000-like nor exactly R3000-like. Anywhere there's an LL/SC instruction
pair, it should be R3000-like.