As a demonstration that you have successfully completed the primary goal (implementing locks and condition variables) and in order to get more practice in the use of monitors and condition variables, you will complete and run a demonstration program which makes use of these facilities; I will provide most of this program, but have left some of the operations in one class for you to write.
The class I provide only parts of implements a bounded-buffer communication channel abstraction, into which a producer can put integers (waiting if the buffer is full), and from which a consumer can then get the integers (waiting if there are none). One additional feature is that the producer can explicitly "shutdown" the communications channel, indicating that no more integers will ever be put into the channel. The consumer's get operation indicates when all the values have been gotten and the channel has been shutdown; if the channel isn't shutdown, the consumer's get operation waits for the producer to either put another value in or shutdown the channel. We'll already have looked at the partial implementation in class, and discussed in general terms the operations that remain for you to implement.
The bounded buffers in turn will be used (by code I provide all of) to support a pipelined style of communication for a large ensemble of threads that are ``sieving'' for prime numbers. (This is purely to demonstrate threads and synchronization, not because this is a very efficient way to find primes.)
~max/www-docs/courses/F98/MC78/labs/lab2/code. The simplest thing for you
to do is to make a copy of the whole directory by in a shell window doing
the following command:
cp -pr ~/www-docs/courses/F98/MC78/labs/lab2/code .This will give you your own
code directory as a
subdirectory of whatever directory you were in when you did the
cp command. The relevant files are all in the threads
subdirectory. The most interesting are the following:
synch.h and synch.cc - These are the two that you will need to change
to implement locks and condition variables. (Printouts are
attached.)
boundedbuf.h and
boundedbuf.cc - You will need to change the latter of
these to finish the bounded buffer implementation. (Printouts are
attached.)
sieve.cc - This is the main program, which makes use of the BBuf class
from boundedbuf.{h,cc}. You won't need to modify this, but might find
it interesting. (Printout attached.)
thread.h and thread.cc, scheduler.h and scheduler.cc, switch.h and
switch.s - These are the core (together with the synch module) of the
threads system, and you should understand how they work. (Printouts are
attached.)
cd to code/threads and then do the
command make. This will recompile/assemble/link
everything, resulting in a program called nachos, unless,
of course, the compiler gives error messages on your code. Once you
have fixed the compiler's complaints and are getting an executable
nachos program, you can try running it from the shell.
It should print out a list of primes, ending with 997 as the
168th prime, and the an explicit message ``end of primes.'' If you
get all 168 primes, but no ``end of primes'' message, something is
wrong with the way you shutdown the bounded buffers.
ThreadTest
procedure in sieve.cc:
DEBUG('t', "Entering SieveTest");
Normally these lines produce no output, but if you run
nachos with the command line arguments -d t,
then they will be printed out as the program runs. That is, to run
the program with debugging output you do:
nachos -d t
Instructor: Max Hailperin