More trapframe panics

[Stacey Son]

On Jan 29, 2014, at 4:29 PM, Adrian Chadd <adrian at freebsd.org> wrote:

> isn't there a way to probe the valid pagemask values?
> 
> I thought I read some way of writing in pagemask values into a fixed
> TLB entry and then reading them back to see which bit(s) are set and
> cleared.
> 
> That way we could maybe do the above at boot-time with minimal evilness.

Yes, see page 107 of http://www.t-es-t.hu/download/mips/md00090c.pdf

"Software may determine which page sizes are supported by writing all ones to the PageMask register, then reading the value back. If a pair of bits reads back as ones, the processor implements that page size. The operation of the pro- cessor is UNDEFINED if software loads the Mask field with a value other than one of those listed in Table 9.16, even if the hardware returns a different value on read. Hardware may depend on this requirement in implementing hard- ware structures"

-stacey.