Here is a code snippet taken from stunnel (from client.c in the local_bind function if you want to look at all the code).
#ifdef IP_TRANSPARENT
int on=1;
if(c->opt->option.transparent) {
if(setsockopt(c->fd, SOL_IP, IP_TRANSPARENT, &on, sizeof on))
sockerror("setsockopt IP_TRANSPARENT");
/* ignore the error to retain Linux 2.2 compatibility */
/* the error will be handled by bind(), anyway */
}
#endif /* IP_TRANSPARENT */
memcpy(&addr, &c->bind_addr.addr[0], sizeof addr);
if(ntohs(addr.in.sin_port)>=1024) { /* security check */
if(!bind(c->fd, &addr.sa, addr_len(addr))) {
s_log(LOG_INFO, "local_bind succeeded on the original port");
return; /* success */
}
if(get_last_socket_error()!=EADDRINUSE
#ifndef USE_WIN32
|| !c->opt->option.transparent
#endif /* USE_WIN32 */
) {
sockerror("local_bind (original port)");
longjmp(c->err, 1);
}
}
Earlier, c->bind_addr was set to the address of the connecting peer with this code:
else if(c->opt->option.transparent)
memcpy(&c->bind_addr, &c->peer_addr, sizeof(SOCKADDR_LIST));
The stunnel documentation contains this advice for recent Linux kernels:
Remote mode (either 2.2.x and >=2.6.28) requires stunnel to be executed as root. setuid option will also break this functionality.
Linux >=2.6.28 requires the following setup for iptables and routing (possibly in /etc/rc.local or equivalent file):
iptables -t mangle -N DIVERT
iptables -t mangle -A PREROUTING -p tcp -m socket -j DIVERT
iptables -t mangle -A DIVERT -j MARK --set-mark 1
iptables -t mangle -A DIVERT -j ACCEPT
ip rule add fwmark 1 lookup 100
ip route add local 0.0.0.0/0 dev lo table 100