>> The package didn't seem to apply specifically for pmax (mipsel was
>> mentioned, and the pmax is a mipsel, but pmax wasn't). So I went with the
>> 2nd hint and softlinked ld.elf_so to ld.so in /usr/libexec.
>
> The secret with the package is it uses ${MACHINE_ARCH} and not
> ${MACHINE}.  On a pmax, ${MACHINE_ARCH} is "mipsel" and ${MACHINE}
> is "pmax".

Just out of a curiocity.  Both NetBSD/pmax and NetBSD/newsmips have

	machine xxx mips

designation in their std.xxx file in kernel conf/ directory.  Is it
meaningful for kernel build to have endianess distinguished, or ok to
leave them as they are?  (There is a technical possiblity to run
'opposite endian' user processes with MIPS processors)

Tohru Nishimura
Nara Institute of Science and Technology