1997 Problem Set
Problem 1 (24 Points)
A major focus of operating system research over the last ten
years has been on building extensible operating
systems.
(a) Provide the justification/motivation for building
extensible operating systems. (i.e., What are the problems being
solved?)
(b) Describe the criteria you would use to determine if
an operating system is indeed an extensible operating
system. (i.e., How do you tell an extensible operating system
from a non-extensible one?)
Problem 2 (24 Points)
Operating system algorithms are frequently influenced by
hardware technology. For example, since DRAM memory loses its
contents when electrical power is lost, operating systems
typically do not assume that memory contains valuable
information after a power failure. Consider what would happen if
some new non-volatile memory technology replaces DRAMs. In this
new technology, a memory location holds its current value even
when power is lost.
Since the contents of main memory represent a large and
costly part of the state of a modern computer system, the
non-volatile memory technology appears to hold promise for
building systems that can recover from a power failure crash
nearly instantly after the power is restored.
(a) Describe the challenges facing an operating system
that wants to use the non-volatile memory to speed recover from
power failure crashes. What are the difficult problems that need
to be solved?
(b) Describe how you might use this new memory to speed
recovery from power failure crashes.
Problem 3 (12 Points)
A technique used in most operating systems developed since
Multics is device independant input/output. Describe
the advantages and disadvantages of this technique.