Simple WolfSSL Server Side Example

The three steps to using WolfSSL in a client side FreeRTOS application
The three steps to using WolfSSL in a server side application

Introduction

This page describes how, in just a few easy steps, the WolfSSL library can be used to to ensure the security and integrity of a server side networking application.

The information on this page is nearly identical to that found on the Simple WolfSSL Client Side Example page. It is repeated here for consistency – with the differences highlighted where they exist.

Header files

wolfssl/ssl.h contains the WolfSSL structures, data definitions, and function prototypes. It must be included in all the source files that use the WolfSSL library. 



#include “wolfssl/ssl.h”




The header file that must be included in all source files that use the WolfSSL library


Initialising the library and creating a WolfSSL context

wolfSSL_Init() prepares the WolfSSL library for use, and must be called before any other WolfSSL API functions.

Next, a variable of type WOLFSSL_CTX is required to store context information, and can be created using wolfSSL_CTX_new(). The SSL or TLS protocol to use is specified as the context is created using the function’s parameter. Options include SSLv3, TLSv1, TLSv1.1, TLSv1.2, or DTLS. This example demonstrates the TLSv1 server protocol being selected (the client side example selects TLSv1 client). The values to use to select the other protocol options are listed in the user manual.

The client side example demonstrated a Certificate Authority (CA) file being loaded into the WolfSSL context. In addition to the CA, the server context must also be loaded with the server certificate, and a key file. This allows the server to send its certificate to the client for verification. wolfSSL_CTX_load_verify_locations() is called to load the CA, wolfSSL_CTX_use_certificate_file() to load the certificate, and wolfSSL_CTX_use_PrivateKey_file to load the private key file. 



/* Define a structure to hold the WolfSSL context. */

WOLFSSL_CTX* xWolfSSL_Context;



    /* Initialise WolfSSL.  This must be done before any other WolfSSL functions

    are called. */

    wolfSSL_Init();



    /* Attempt to create a context that uses the TLS V1 server protocol. */

    xWolfSSL_Context = wolfSSL_CTX_new( CyaTLSv1_server_method() );



    if( xWolfSSL_Context != NULL )

    {

        /* Load the CA certificate.  Real applications should ensure that

        wolfSSL_CTX_load_verify_locations() returns SSL_SUCCESS before proceeding. */

        wolfSSL_CTX_load_verify_locations( xWolfSSL_Context, “ca-cert.pem”, 0 );



        /* Again, checking of the return values is omitted from this example,

        just for clarity.  Real applications must ensure the following two

        functions return SSL_SUCCESS. */

        wolfSSL_CTX_use_certificate_file( xWolfSSL_Context, “server-cert.pem”, SSL_FILETYPE_PEM );

        wolfSSL_CTX_use_PrivateKey_file( xWolfSSL_Context, “server-key.pem”, SSL_FILETYPE_PEM );

    }




Library initialisation, protocol selection, and loading the CA certificate,




server certificate and private key into the WolfSSL context.


Associating a WolfSSL object with a connected socket

Each accepted connection must be associated with a WolfSSL object. WolfSSL objects are created using wolfSSL_new(), and associated with a TCP socket using wolfSSL_set_fd(). 



WOLFSSL* xWolfSSL_Object;



    /* A connection has been accepted by the server.  Create a WolfSSL

    object for use with the newly connected socket. */

    xWolfSSL_Object = wolfSSL_new( xWolfSSL_Context );



    if( xWolfSSL_Object != NULL )

    {

        /* Associate the created WolfSSL object with the connected socket

        (sockfd). */

        wolfSSL_set_fd( xWolfSSL_Object, sockfd );

    }




Creating a WolfSSL object, and associating it with an accepted connection


Using the socket

Secure communication can now be made through the socket by using wolfSSL_write() in place of the standard sockets write() or send() functions, and wolfSSL_read() in place of the standard sockets read() or recv().

Note that the first parameter to both wolfSSL_write() and wolfSSL_read() is not the socket descriptor, but the WolfSSL object that is associated with the socket descriptor. 



char ucTxBuf[ MAXLINE ], ucRxBuf[ MAXLINE ];



    if( wolfSSL_write( xWolfSSL_Object, ucTxBuf, strlen( ucTxBuf ) ) != strlen( ucTxBuf ) )

    {

        /* Send failed. */

    }



    if( wolfSSL_read( xWolfSSL_Object, ucRxBuf, MAXLINE ) <= 0 )

    {

        /* Read failed. */

    }




Writing to and reading from a socket using the WolfSSL API


Deleting allocated resources

WolfSSL API functions that result in dynamic resource allocation have a counterpart function that should be called to free the resource when it is no longer required. The code snippet below shows how the objects created in this small example should be freed. 





    /* WolfSSL objects should be deleted when they are no longer required. */

    wolfSSL_free( xWolfSSL_Object );



    /* The WolfSSL context should be deleted if it is no longer required.  However,

    because most deeply embedded applications will keep the context for the lifetime

    of the application, and only ever be restarted when the system is rebooted, it

    might be that the context is never explicitly freed. */

    wolfSSL_CTX_free( xWolfSSL_Context );



    /* The library itself should be shut down cleanly if it too is no longer

    required.  Again, because most deeply embedded applications will require the

    library for the lifetime of the application, and only ever be restarted when

    the system is rebooted, it might be that the library is never explicitly closed. */

    wolfSSL_Cleanup();






Deleting objects that were dynamically allocated in this example