[net.arch] 386 paging, et al.

kds@intelca.UUCP (Ken Shoemaker) (12/14/85)

actually, since the 386 provides both segmentation and paging, you could
use a global page table (i.e., not change page tables on a context
switch), and use segmentation to relocate things around in the linear
space before you do the page translation.  If you ran out of the 4Gbyte
"linear" space (the space between the segmentation unit and the paging
unit), then you could swap entire segments, although if you didn't want
to do this, you would limit the size of the total number of running
processes on the system to something like 4Gbyte.  Pardon my ignorance,
but if I remember correctly, this sounds a little like what our favorite
VAX os does...page until it gets out of hand, then start swapping
whole processes.  I'm sure someone out there can point out the problems
with scheme, and I'd like to know if there have been any systems
constructed this way.  Surely, someone has built hardware that would
support a similar mechanism in the past...
-- 
yes, some uncomplicated peoples still believe this myth...

Ken Shoemaker, Santa Clara, Ca.
{pur-ee,hplabs,amd,scgvaxd,dual,qantel}!intelca!kds
	
---the above views are personal.