[TUHS] history of virtual address space

[Noel Chiappa via TUHS]

    > From: Ron Minnich via TUHS tuhs at tuhs.org

    > on the 11, trap vectors chewed up a chunk of low physical memory

I don't think so: I have a note that "Trap and interrupt vectors are fetched
from Kernel Data space, when memory management is enabled" (can't cite a
manual page, sorry, but I'm pretty sure that's right). So they could be
anywhere, physically.

Of course, in UNIX, Kernel Data space is at physical 0, so in _practice_, in
UNIX, they are in low physical memory.


Actually, there's:

  #define	B_RELOC	0200	/* no longer used */

in buf.h, and I have a vague memory of reading that at same point, buffers'
physical address was _not_ the same as their virtual addresses (which is what
all the hair with 'sysfix', etc, is needed to do), so who knows if that was
always true (if buffers were not always in low physical memory).


    > What other Unix systems used a layout where the kernel was in low
    > virtual?

In MINI-UNIX, user programs are loaded and run at 060000; the kernel was in
low memory. (Probably LSX too, but I don't know much about that.) Not sure if
that's what you meant! 

	Noel