CSCI 4210 — Operating Systems
Homework 4
Network Programming and Multi-Threaded Programming using C
Overview
This homework is due by 11:59:59 PM on Wednesday, 29, 2020.
If you have late days available, you can use them.
This homework is to be completed individually. Do not share your code with anyone else.
You must use C for this homework assignment, and your code must successfully compile
via gcc with absolutely no warning messages when the -Wall (i.e., warn all) compiler option
is used. We will also use -Werror, which will treat all warnings as critical errors.
Your code must successfully compile and run on Submitty, which uses Ubuntu v18.04.1 LTS.
Note that the gcc compiler is version 7.4.0 (Ubuntu 7.4.0-1ubuntu1~18.04.1).
Further, you must use the Pthread library. Remember, to compile your code, use -pthread
to link the Pthread library.
Homework Specifications
In this fourth and final homework assignment, you will use C to write server code to implement
a multi-threaded chat server using sockets. Clients will be able to send and receive short text
messages from one another. Note that all communication between clients must be handled and
sent via your server.
Clients communicate with your server via TCP or UDP via the port number specified as the first
command-line argument. For TCP, clients connect to this port number (i.e., the TCP listener
port). For UDP, clients send datagram(s) to this port number.
To support both TCP and UDP at the same time, you must use the select() system call in the
main thread to poll for incoming TCP connection requests and UDP datagrams.
Your server must not be a single-threaded iterative server. Instead, your server must allocate a
child thread for each TCP connection. Further, your server must be parallelized to the extent
possible. As such, be sure you handle all potential synchronization issues.
Note that your server must support clients implemented in any language (e.g., Java, C, Python,
MIPS, etc.); therefore, only handle streams of bytes as opposed to any language-specific structures.
And though you will only submit your server code for this assignment, plan to create one or more
test clients. Test clients will not be provided, but feel free to share test clients with others via
Submitty. Do not share server code. Also note that you can use netcat to test your server; but
do not use telnet. Also remember the -u flag for UDP in netcat.
Simultaneously supporting TCP and UDP
To provide flexibility to clients, clients can either establish a connection via TCP or simply
send/receive datagrams via UDP. As noted above, to support both TCP and UDP at the same time,
you must use the select() system call in the main thread to poll for incoming TCP connection
requests and UDP datagrams.
Your server must support at least 32 concurrently connected clients (i.e., TCP connections, with
each connection corresponding to a child thread).
For TCP, use a dedicated child thread to handle each TCP connection. In other words, after the
accept() call, immediately create a child thread to handle that connection, thereby enabling the
main thread to loop back around and call select() again. For TCP, you are only guaranteed to
receive data up to and including the first newline character in the first packet received; therefore,
expect to implement multiple read() calls.
Since UDP is connectionless, for UDP, use an iterative approach (i.e., handle incoming UDP data-
grams in the main thread, then immediately loop back to the select() system call).
Application-layer protocol
The application-layer protocol between client and server is a line-based protocol. Streams of bytes
(i.e., characters) are transmitted between clients and your server. Note that all commands are
specified in upper-case and end with a newline ('\n') character, as shown below.
In general, when the server receives a request, it responds with either a four-byte “OK!\n” response
or an error. When an error occurs, the server must respond with:
ERROR \n
For any error message not specified below, use a short human-readable description matching the
simple one-line format shown above. Expect clients to display these error messages to users.
The following subsections define the application-layer protocol commands.
LOGIN
If a client uses TCP, the user must first log in (though authentication is not required). A user
identifies itself as follows:
LOGIN \n
Note that a valid is a string of alphanumeric characters with a length in the range [4,16].
Upon receiving a LOGIN request, if successful, the server sends the four-byte “OK!\n” string.
If the given is already connected via TCP, the server responds with an
of “Already connected”; if is invalid, the server responds with an of
“Invalid userid” (and in both cases keeps the TCP connection open).
2
WHO
A user may send a WHO request to obtain a list of all users currently active within the chat server.
When your server receives this request, in addition to sending “OK!\n” to the client, the response
should consist of an ASCII-based sorted list of all users, with users delimited by newline ('\n')
characters.
As an example, the server may respond with the following:
OK!\nMorty\nRick\nShirley\nSummer\nmeme\nqwerty\n
Note that the WHO command must be supported for both TCP and UDP.
LOGOUT
A user connected via TCP may send a LOGOUT request to ensure the server marks the user as being
completely logged out and inactive. Note that this is recommended but not required for TCP, since
the client can simply close its connection to indicate it is logging out.
When a LOGOUT command is sent, the server is required to send an “OK!\n” response. And the
connection should stay open (until the remote side closes the connection).
SEND
A user connected via TCP may attempt to send a private message to another user via the SEND
command. The required format of the SEND command is as follows:
SEND \n
To be a valid SEND request, the must be a currently active user and the
(i.e., length of the portion) must be an integer in the range [1,990].
Note that the can contain any bytes whatsoever. You may assume that the number of
bytes will always match the given value.
If the request is valid, the server responds by sending an “OK!\n” response. Further, the server
attempts to send the message to the by sending a TCP packet using the
following format:
FROM \n
If the request is invalid, send the appropriate error message from among the following:
“Unknown userid”
“Invalid msglen”
“Invalid SEND format”
“SEND not supported over UDP”
3
BROADCAST
If a user wishes to send a message to all active users, the BROADCAST command can be used. This
command can also be used via UDP, i.e., without the need to first log in. The format of this
command is as follows:
BROADCAST \n
The and parameters match that of the SEND command above. For UDP, use
a hard-coded “UDP-client” string for the that all UDP clients use.
Handling system call errors
In general, if a system call fails, use perror() to display the appropriate error message on stderr,
then exit the program and return EXIT_FAILURE. If a system or library call does not set the
global errno, use fprintf() instead of perror() to write an error message to stderr.
Error messages must be one line only and use one of the appropriate formats shown below:
MAIN: ERROR
Or:
CHILD : ERROR
4
Required Output
Your server is required to output one or more lines describing each request that it receives. Required
output is illustrated in the example below. As per usual, output lines may be interleaved as multiple
clients interact with the server simultaneously.
bash$ ./a.out 9876
MAIN: Started server
MAIN: Listening for TCP connections on port: 9876
MAIN: Listening for UDP datagrams on port: 9876
...
MAIN: Rcvd incoming TCP connection from:
CHILD : Rcvd LOGIN request for userid Rick
...
MAIN: Rcvd incoming TCP connection from:
CHILD : Rcvd LOGIN request for userid Morty
CHILD : Rcvd WHO request
CHILD : Rcvd SEND request to userid Rick
CHILD : Rcvd SEND request to userid Summer
CHILD : Sent ERROR (Unknown userid)
CHILD : Rcvd WHO request
MAIN: Rcvd incoming UDP datagram from:
MAIN: Rcvd BROADCAST request
CHILD : Client disconnected
CHILD : Rcvd LOGOUT request
CHILD : Client disconnected
...
MAIN: Rcvd incoming TCP connection from:
CHILD : Rcvd LOGIN request for userid Rick
CHILD : Rcvd WHO request
CHILD : Rcvd SEND request to userid Rick
CHILD : Rcvd SEND request to userid Rick
CHILD : Rcvd SEND request to userid Morty
MAIN: Rcvd incoming UDP datagram from:
MAIN: Rcvd WHO request
CHILD : Rcvd SEND request to userid Morty
CHILD : Rcvd BROADCAST request
CHILD : Client disconnected
...
Note that the required output above is certainly different than the specific data sent and received
via the application-layer protocol. On Submitty, test clients will connect to your server and test
whether you have correctly implemented all aspects of the application-layer protocol.
5
Submission instructions
To submit your assignment (and also perform final testing of your code), please use Submitty, the
homework submission server.
Note that this assignment will be available on Submitty a minimum of three days before the due
date. Please do not ask when Submitty will be available, as you should perform adequate testing
on your own Ubuntu platform.
That said, to make sure that your program does execute properly everywhere, including Submitty,
use the techniques below.
First, as discussed in class, use the DEBUG_MODE technique to make sure you do not submit any
debugging code. Here is an example:
#ifdef DEBUG_MODE
printf( "the value of x is %d\n", x );
printf( "the value of q is %d\n", q );
printf( "why is my program crashing here?!" );
fflush( stdout );
#endif
And to compile this code in “debug” mode, use the -D flag as follows:
bash$ gcc -Wall -Werror -D DEBUG_MODE hw4.c -pthread
Second, as also discussed in class, output to standard output (stdout) is buffered. To disable
buffered output for grading on Submitty, use setvbuf() as follows:
setvbuf( stdout, NULL, _IONBF, 0 );
You would not generally do this in practice, as this can substantially slow down your program, but
to ensure correctness on Submitty, this is a good technique to use.