🚀 go-pugleaf

In article <1992Jul27.134542.4845@ryn.mro4.dec.com>, bhandarkar@wrksys.enet.dec.com (Dileep Bhandarkar) writes:
> The P0 space begins at virtual address 0 and grows in the direction of larger
> addresses. P1 starts at virtual address 2**31-1 and grows in the direction of
> smaller addresses. Since the P1 region grows towards smaller addresses, the
> base and length registers describe the part of P1 space that is not accessible.

I think that the reason for this is so that you can put the user stack
way the heck up there in the address space without having to have a page
table large enough to map the entire virtual space for each process. If you
put the stack at the top of P0 space, for instance, the page table for the
process would have to have enough entries to map all of P0 space. Since most
of P0 space is unused, it doesn't make sense to require the page table to
include all of it.

Separating the stack into P1 space and making P1 space grow down lets the
machine keep the stack in a separate page table.

RISC processors sidestep the issue by pushing page table lookup into
softwareland.

Roger Ivie
ivie@cc.usu.edu