paper evalutions

Kiran Tati (ktati@csag.ucsd.edu)
Tue, 25 Apr 2000 01:22:18 -0700

On the Duality of Operating System Structures:

The main point of this paper is that two models of operating systems are

dual to each other. This means for every program implement on one model
there exists an equivalent program on another model. This paper defines
a precise
characterization of these models and then presents how one can transform

a program implemented on one model to another while maintaining the same

functionality as well as performace for a given machine architecture.
This
means the designer of operating system can ignore about how the
operating
functionality is offered to the applications rather he should
concentrate
on what the architecture provide for the operating system
implementation.
This duality might be true for the two ends of the spectrum but the
operating
systems in between these two spectrums may be much better for the
applications.

The Structure of Systems Using Upcalls:
The goal of this paper is to describe and discuss about the upcall
mechanism
that can be used in implementation of various modules of operating
system.
The approach is to use the simple function call mechanism for the inter
layer
communication and use of some inter process communication for intra
layer
communication. The advantages with this are improved performance,
ease of programming (by taking advice) and it provide more opportunities

for the piggybacking. The performance improvement is debatable because
the
intra layer communication costs may out weigh the inter layer
communication
costs, if one try to reduce the intra layer communication it may leads
too
much code replication. In monolithic kernel this is not a problem at all

because all the kernel activities are running in a single thread of
control.
Another disadvantage is that it violate the "depends on" relation so
designing
and debugging the is very hard, as expected. The paper presents various
adhoc
solutions for the above problem but bug implementation of such design is
very
hard because it is very hard to reason about the correctness. This paper
also
presents about how this mechanism can be used in task scheduling and
address
space management.

--
Kiran Tati (Graduate Student)            Kiran Tati
UCSD/CSE AP&M 4402                       9450 Gilman Drive
9500, Gilman Drive, Dept. 0114           UCSD#922563
La Jolla, CA, 92093-0114                 La Jolla, CA 92092-2563
Phone:  858 5345486 (Office)             858 552 9291 (Home)
E-Mail: ktati@csag.ucsd.edu

Office Address: AP & M 6426