[ ... Disregard the previous one mistakenly splushed ... ]

>> R3000A processor seems to do runtime endian-switching as well.  Some of
>> Japaneses once toyed an idea of 'endian reversing emul layer', but it
>> didn't be realized.
>
> R3000A can do it on an "entire processor" scale, right?  R4000 can do it
> as "user in one mode, supervisor in another".  At least, that's how I
> remember it.

I checked out one of IDT pdf manuals.  At least their implementation of
R3000A derivative CPUs can run 'endian-revesing' per process basis.
Yes, I remember I was surprised at it when I first read it.

  RE	reverse endianness.   [ ... ] If this bit is cleared, the
	endianness defined at reset is used for the current user task.
	If this bit is set, then the user task will operate with the
	opposite byte ordering convention from that determined by reset.
	This bit has no effect on kernel mode operation.

Tohru Nishimura