Commit 74d65636 authored by Jukka Rissanen's avatar Jukka Rissanen
Browse files

net: apps: Add TCP support to echo-server application 



If TCP is activated, then the echo-server application starts
to listen TCP socket and reply data to caller. For TCP
connections, the application data is not reversed because
it would make it difficult to verify the data in the sender
side. The TCP packets might be split into smaller chunks by
echo-client so client would have some unnecessary difficult
to verify the data it sent if we reverse it.

Change-Id: Ic146dfc75825bc8057968c581bfe9d3285e0d7b4
Signed-off-by: default avatarJukka Rissanen <jukka.rissanen@linux.intel.com>
parent 67463d92

samples/net/echo_server/src/echo-server.c

+74 −29
Original line number Diff line number Diff line
@@ -26,6 +26,7 @@ 


#include <zephyr.h>

#include <sections.h>

#include <errno.h>



#include <net/ip_buf.h>

#include <net/net_core.h>
@@ -91,17 +92,20 @@ static inline void reverse(unsigned char *buf, int len) 


static inline struct net_buf *prepare_reply(const char *name,

					    const char *type,

					    struct net_buf *buf)

					    struct net_buf *buf,

					    int proto)

{

	PRINT("%s: %sreceived %d bytes\n", name, type,

	PRINT("%s: %sreceived %d bytes data\n", name, type,

	      ip_buf_appdatalen(buf));



	if (proto != IPPROTO_TCP) {

		/* In this test we reverse the received bytes.

		 * We could just pass the data back as is but

		 * this way it is possible to see how the app

		 * can manipulate the received data.

		 */

		reverse(ip_buf_appdata(buf), ip_buf_appdatalen(buf));

	}



#if defined(CONFIG_NET_TESTING)

	net_testing_set_reply_address(buf);
@@ -111,23 +115,26 @@ static inline struct net_buf *prepare_reply(const char *name, 
}



/* How many tics to wait for a network packet */

#if 0

#define WAIT_TIME 1

#define WAIT_TICKS (WAIT_TIME * sys_clock_ticks_per_sec)

#if 1

#define WAIT_TIME 0

#define WAIT_TICKS (WAIT_TIME * sys_clock_ticks_per_sec / 10)

#else

#define WAIT_TICKS TICKS_UNLIMITED

#endif



static inline void receive_and_reply(const char *name, struct net_context *recv,

struct nano_fifo *net_context_get_queue(struct net_context *context);

static inline void receive_and_reply(const char *name,

				     struct net_context *udp_recv,

				     struct net_context *tcp_recv,

				     struct net_context *mcast_recv)

{

	struct net_buf *buf;



	buf = net_receive(recv, WAIT_TICKS);

	buf = net_receive(udp_recv, WAIT_TICKS);

	if (buf) {

		prepare_reply(name, "unicast ", buf);

		prepare_reply(name, "unicast ", buf, IPPROTO_UDP);



		if (net_reply(recv, buf)) {

		if (net_reply(udp_recv, buf)) {

			ip_buf_unref(buf);

		}

		return;
@@ -135,22 +142,52 @@ static inline void receive_and_reply(const char *name, struct net_context *recv, 


	buf = net_receive(mcast_recv, WAIT_TICKS);

	if (buf) {

		prepare_reply(name, "multicast ", buf);

		prepare_reply(name, "multicast ", buf, IPPROTO_UDP);



		if (net_reply(mcast_recv, buf)) {

			ip_buf_unref(buf);

		}

		return;

	}



#if defined(CONFIG_NETWORKING_WITH_TCP)

	if (tcp_recv) {

		static struct net_buf *tcpbuf;



		if (tcpbuf) {

			int ret;

		reply:

			ret = net_reply(tcp_recv, tcpbuf);

			if (ret && ret != -EAGAIN) {

				ip_buf_unref(tcpbuf);

				tcpbuf = NULL;

			} else if (!ret) {

				tcpbuf = NULL;

			} else {

				PRINT("Retrying to send packet %p\n", tcpbuf);

			}

		} else {

			tcpbuf = net_receive(tcp_recv, WAIT_TICKS);

			if (tcpbuf) {

				PRINT("Received packet %p len %d\n",

				      tcpbuf, ip_buf_appdatalen(tcpbuf));

				prepare_reply(name, "tcp ", tcpbuf, IPPROTO_TCP);

				goto reply;

			}

		}

	}

#endif



	fiber_sleep(50);

}



static inline bool get_context(struct net_context **recv,

static inline bool get_context(struct net_context **udp_recv,

			       struct net_context **tcp_recv,

			       struct net_context **mcast_recv)

{

	static struct net_addr mcast_addr;

	static struct net_addr any_addr;

	static struct net_addr my_addr;

	int proto = IPPROTO_UDP;



#if defined(CONFIG_NETWORKING_WITH_IPV6)

	static const struct in6_addr in6addr_any = IN6ADDR_ANY_INIT;
@@ -179,17 +216,25 @@ static inline bool get_context(struct net_context **recv, 
	my_addr.family = AF_INET;

#endif



#ifdef CONFIG_NETWORKING_WITH_TCP

	proto = IPPROTO_TCP;

#endif /* CONFIG_NETWORKING_WITH_TCP */

	*udp_recv = net_context_get(IPPROTO_UDP,

				    &any_addr, 0,

				    &my_addr, MY_PORT);

	if (!*udp_recv) {

		PRINT("%s: Cannot get network context\n", __func__);

		return NULL;

	}



	*recv = net_context_get(proto,

#if defined(CONFIG_NETWORKING_WITH_TCP)

	if (tcp_recv) {

		*tcp_recv = net_context_get(IPPROTO_TCP,

					    &any_addr, 0,

					    &my_addr, MY_PORT);

	if (!*recv) {

		if (!*tcp_recv) {

			PRINT("%s: Cannot get network context\n", __func__);

			return NULL;

		}

	}

#endif



	*mcast_recv = net_context_get(IPPROTO_UDP,

				      &any_addr, 0,
@@ -210,16 +255,16 @@ char __noinit __stack fiberStack[STACKSIZE]; 


void receive(void)

{

	static struct net_context *recv;

	static struct net_context *udp_recv, *tcp_recv;

	static struct net_context *mcast_recv;



	if (!get_context(&recv, &mcast_recv)) {

	if (!get_context(&udp_recv, &tcp_recv, &mcast_recv)) {

		PRINT("%s: Cannot get network contexts\n", __func__);

		return;

	}



	while (1) {

		receive_and_reply(__func__, recv, mcast_recv);

		receive_and_reply(__func__, udp_recv, tcp_recv, mcast_recv);

	}

}

CRA Git | Maintained and supported by SUSTech CRA and CCSE