> is anyone actively working on 64 bit mips3 support?

Hey, could someone out there just run scratch-and-built of pmap.c? 

Current approach of TLB/ASID is designed just wrong.  It costs severe
performance loss.  When PTE protection attribute is about to be
changed for a range of user address space, bumping ASID for the pmap
is the most effective solution with no cost.  Trash away TLBUpdate()!
It's the cause of why fork/exec runs miserably.

Tohru Nishimura