Address Spaces:

------------------------------------------

Recall:

Uniprogramming:
	run only one program at a time --
	we need to load/store all of a program's memory at once

multi-programming:
	many programs use the adress space at once.
	any system that isn't embedded uses multi-programming.
	MUST use address-space translation because programs can't know the
		physical address they will use at compile time.
	Can be resolved statically, by a linker/loader.
		Can't use register offset addressing because the running
			program can over-write another program's memory.
	Can be resolved dynamically, by a pager.
		Can do a dynamic check on the validity of the address
			Used in managed runtimes.
			In general, too slow, so must use hardware.
		Base-and-bounds was a method for this.
	

--------------------------------------------

Pro/Con's of Base and Bound

Pro:
Adding memory is easy -- increase the bound

Con:
Can't move the base down when another process finishes
	Must copy the entire physical memory elsewhere, then
	adjust the base, then move everything down.

Don't actually get virtual memory -- can't exceed physical memory with bound
Processes can't share memory usefully.
Over time, wastes lots of memory due to external fragmentation.

Fragmentation example:

1000	-----------------
	|		|
	|		|
	|		|
	|		|
	|		|
	|		|
	|		|
	|		|
0	|		|
	-------------------