Winsock tutorial – Socket programming in C on windows
Socket programming with winsock
This is a quick guide/tutorial to learning socket programming in C language on Windows. «Windows» because the code snippets shown over here will work only on Windows. The windows api to socket programming is called winsock.
Sockets are the fundamental «things» behind any kind of network communications done by your computer. For example when you type www.google.com in your web browser, it opens a socket and connects to google.com to fetch the page and show it to you.
Same with any chat client like gtalk or skype. Any network communication goes through a socket.
Before you begin
This tutorial assumes that you have basic knowledge of C and pointers. Also download Visual C++ 2010 Express Edition.
Winsock first needs to be initialiased like this :
winsock2.h is the header file to be included for winsock functions. ws2_32.lib is the library file to be linked with the program to be able to use winsock functions.
The WSAStartup function is used to start or initialise winsock library. It takes 2 parameters ; the first one is the version we want to load and second one is a WSADATA structure which will hold additional information after winsock has been loaded.
If any error occurs then the WSAStartup function would return a non zero value and WSAGetLastError can be used to get more information about what error happened.
OK , so next step is to create a socket.
Creating a socket
The socket() function is used to create a socket.
Here is a code sample :
Function socket() creates a socket and returns a socket descriptor which can be used in other network commands. The above code will create a socket of :
Address Family : AF_INET (this is IP version 4)
Type : SOCK_STREAM (this means connection oriented TCP protocol)
Protocol : 0 [ or IPPROTO_TCP , IPPROTO_UDP ]
It would be a good idea to read some documentation here
Ok , so you have created a socket successfully. But what next ? Next we shall try to connect to some server using this socket. We can connect to www.google.com
Apart from SOCK_STREAM type of sockets there is another type called SOCK_DGRAM which indicates the UDP protocol. This type of socket is non-connection socket. In this tutorial we shall stick to SOCK_STREAM or TCP sockets.
Connect to a Server
We connect to a remote server on a certain port number. So we need 2 things , IP address and port number to connect to.
To connect to a remote server we need to do a couple of things. First is create a sockaddr_in structure with proper values filled in. Lets create one for ourselves :
Have a look at the structures
The sockaddr_in has a member called sin_addr of type in_addr which has a s_addr which is nothing but a long. It contains the IP address in long format.
Function inet_addr is a very handy function to convert an IP address to a long format. This is how you do it :
So you need to know the IP address of the remote server you are connecting to. Here we used the ip address of google.com as a sample. A little later on we shall see how to find out the ip address of a given domain name.
The last thing needed is the connect function. It needs a socket and a sockaddr structure to connect to. Here is a code sample.
It cannot be any simpler. It creates a socket and then connects. If you run the program it should show Connected.
Try connecting to a port different from port 80 and you should not be able to connect which indicates that the port is not open for connection.
OK , so we are now connected. Lets do the next thing , sending some data to the remote server.
The concept of «connections» apply to SOCK_STREAM/TCP type of sockets. Connection means a reliable «stream» of data such that there can be multiple such streams each having communication of its own. Think of this as a pipe which is not interfered by other data.
Other sockets like UDP , ICMP , ARP dont have a concept of «connection». These are non-connection based communication. Which means you keep sending or receiving packets from anybody and everybody.
Function send will simply send data. It needs the socket descriptor , the data to send and its size.
Here is a very simple example of sending some data to google.com ip :
In the above example , we first connect to an ip address and then send the string message «GET / HTTP/1.1\r\n\r\n» to it.
The message is actually a http command to fetch the mainpage of a website.
Now that we have send some data , its time to receive a reply from the server. So lets do it.
Function recv is used to receive data on a socket. In the following example we shall send the same message as the last example and receive a reply from the server.
Here is the output of the above code :
We can see what reply was send by the server. It looks something like Html, well IT IS html. Google.com replied with the content of the page we requested. Quite simple!
Now that we have received our reply, its time to close the socket.
Function closesocket is used to close the socket. Also WSACleanup must be called to unload the winsock library (ws2_32.dll).
So in the above example we learned how to :
1. Create a socket
2. Connect to remote server
3. Send some data
4. Receive a reply
Its useful to know that your web browser also does the same thing when you open www.google.com
This kind of socket activity represents a CLIENT . A client is a system that connects to a remote system to fetch or retrieve data.
The other kind of socket activity is called a SERVER. A server is a system that uses sockets to receive incoming connections and provide them with data. It is just the opposite of Client. So www.google.com is a server and your web browser is a client. Or more technically www.google.com is a HTTP Server and your web browser is an HTTP client.
Now its time to do some server tasks using sockets. But before we move ahead there are a few side topics that should be covered just incase you need them.
Get IP address of a hostname/domain
When connecting to a remote host , it is necessary to have its IP address. Function gethostbyname is used for this purpose. It takes the domain name as the parameter and returns a structure of type hostent. This structure has the ip information. It is present in netdb.h . Lets have a look at this structure
The h_addr_list has the IP addresses. So now lets have some code to use them.
Output of the code would look like :
So the above code can be used to find the ip address of any domain name. Then the ip address can be used to make a connection using a socket.
Function inet_ntoa will convert an IP address in long format to dotted format. This is just the opposite of inet_addr .
So far we have see some important structures that are used. Lets revise them :
1. sockaddr_in — Connection information. Used by connect , send , recv etc.
2. in_addr — Ip address in long format
4. hostent — The ip addresses of a hostname. Used by gethostbyname
OK now onto server things. Servers basically do the following :
1. Open a socket
2. Bind to a address(and port).
3. Listen for incoming connections.
4. Accept connections
We have already learnt how to open a socket. So the next thing would be to bind it.
Bind a socket
Function bind can be used to bind a socket to a particular address and port. It needs a sockaddr_in structure similar to connect function.
Lets see a code example :
Now that bind is done, its time to make the socket listen to connections. We bind a socket to a particular IP address and a certain port number. By doing this we ensure that all incoming data which is directed towards this port number is received by this application.
This makes it obvious that you cannot have 2 sockets bound to the same port.
Listen for connections
After binding a socket to a port the next thing we need to do is listen for connections. For this we need to put the socket in listening mode. Function listen is used to put the socket in listening mode. Just add the following line after bind.
Thats all. Now comes the main part of accepting new connections.
Function accept is used for this. Here is the code
Run the program. It should show
So now this program is waiting for incoming connections on port 8888. Dont close this program , keep it running.
Now a client can connect to it on this port. We shall use the telnet client for testing this. Open a terminal and type
And the server output will show
So we can see that the client connected to the server. Try the above process till you get it perfect.
You can get the ip address of client and the port of connection by using the sockaddr_in structure passed to accept function. It is very simple :
We accepted an incoming connection but closed it immediately. This was not very productive. There are lots of things that can be done after an incoming connection is established. Afterall the connection was established for the purpose of communication. So lets reply to the client.
Here is an example :
Run the above code in 1 terminal. And connect to this server using telnet from another terminal and you should see this :
So the client(telnet) received a reply from server. We had to use a getchar because otherwise the output would scroll out of the client terminal without waiting
We can see that the connection is closed immediately after that simply because the server program ends after accepting and sending reply. A server like www.google.com is always up to accept incoming connections.
It means that a server is supposed to be running all the time. Afterall its a server meant to serve. So we need to keep our server RUNNING non-stop. The simplest way to do this is to put the accept in a loop so that it can receive incoming connections all the time.
So a live server will be alive for all time. Lets code this up :
We havent done a lot there. Just the accept was put in a loop.
Now run the program in 1 terminal , and open 3 other terminals. From each of the 3 terminal do a telnet to the server port.
Run telnet like this. It will launch the interactive prompt.
At the telnet shell, run the command «open localhost 8888». This command will try to connect to localhost on port number 8888.
Next you should see the following message at the telnet prompt. This message is received from the socket server running on port 8888.
On the other hand, the server terminal would show the following messages, indicating that a client connected to it.
So now the server is running nonstop and the telnet terminals are also connected nonstop. Now close the server program.
All telnet terminals would show «Connection to host lost.»
Good so far. But still there is not effective communication between the server and the client.
The server program accepts connections in a loop and just send them a reply, after that it does nothing with them. Also it is not able to handle more than 1 connection at a time. So now its time to handle the connections , and handle multiple connections together.
To handle every connection we need a separate handling code to run along with the main server accepting connections.
One way to achieve this is using threads. The main server program accepts a connection and creates a new thread to handle communication for the connection, and then the server goes back to accept more connections.
We shall now use threads to create handlers for each connection the server accepts. Lets do it pal.
Run the above server and open 3 terminals like before. Now the server will create a thread for each client connecting to it.
The telnet terminals would show :
This one looks good , but the communication handler is also quite dumb. After the greeting it terminates. It should stay alive and keep communicating with the client.
One way to do this is by making the connection handler wait for some message from a client as long as the client is connected. If the client disconnects , the connection handler ends.
So the connection handler can be rewritten like this :
The above connection handler takes some input from the client and replies back with the same. Simple! Here is how the telnet output might look
So now we have a server thats communicative. Thats useful now.
The winsock api is quite similar to Linux sockets in terms of function name and structures. Few differences exist like :
1. Winsock needs to be initialised with the WSAStartup function. No such thing in linux.
2. Header file names are different. Winsock needs winsock2.h , whereas Linux needs socket.h , apra/inet.h , unistd.h and many others.
3. Winsock function to close a socket is closesocket , whereas on Linux it is close .
On Winsock WSACleanup must also be called to unload the winsock dll.
4. On winsock the error number is fetched by the function WSAGetLastError() . On Linux the errno variable from errno.h file is filled with the error number.
And there are many more differences as we go deep.
By now you must have learned the basics of socket programming in C. You can try out some experiments like writing a chat client or something similar.
If you think that the tutorial needs some addons or improvements or any of the code snippets above dont work then feel free to make a comment below so that it gets fixed.
71 thoughts on “ Winsock tutorial – Socket programming in C on windows ”
window alert % test >
Hey Silver Moon,
Thank you very much for sharing your codes and the introduction – keep going :-)!
I´m an absolute beginner in “socket” programming and I have to say that your explanation is easy to understand (for me).
Good and newbie-friendly tutorial.
in the chapter “Handling Connections” no sample source code and no output in the boxes is shown.
Might we trouble you for the last two pieces of multithreading code?
Why don’t you update all you typos? The HTML versions of (“,) Take about 30 minutes to edit. It’s a pain to have to edit this code. If not just take this site down, it’s not worth trying to figure this all out.
Then you use ‘s’ as a variable for ‘socket’, why not use something more descriptive such as sock_p , for primary socket. Your examples are hard to follow.
For crying out loud all you are doing is creating a socket, connecting to server and sending info. If your variables were more discriptive it wouldn’t took take two days to debug your code to getting running for the absolute beginer.
Also, the client application has an outdated inet address for google. It doesn’t connect and it took me two days to figure that out, since I have zero previous knowledge of Socket programing.
I had to ping Google.com to get a current Inet Address.
Why don’t you update all you typos? The HTML versions of (“,) Take about 30 minutes to edit. It’s a pain to have to edit this code. If not just take this site down, it’s not worth trying to figure this all out. (Without generating income from this site, the user is under no obligation to update the content presented. Go elsewhere if it is not satisfactory.)
Then you use ‘s’ as a variable for ‘socket’, why not use something more descriptive such as sock_p , for primary socket. Your examples are hard to follow. (That falls down to coder preference. I used “sock” some prefer “s”. Their is no particularly wrong answer.)
For crying out loud all you are doing is creating a socket, connecting to server and sending info. If your variables were more discriptive it wouldn’t took take two days to debug your code to getting running for the absolute beginer. (In C code, that is quite the accomplishment. Especially on a windows machine. As for the debugging, that is a natural part of coding and most of your time will be spent doing it.)
Also, the client application has an outdated inet address for google. It doesn’t connect and it took me two days to figure that out, since I have zero previous knowledge of Socket programing. (Great! You just learned the first rule of using other people’s code: trust nothing.)
I had to ping Google.com to get a current Inet Address. (And you found the solution yourself. I’d bet fifty bucks that you won’t be making this mistake again.)