ffi-tox 0.0.1

data/ProjectTox-Core/core/DHT.c

@@ -0,0 +1,1104 @@
+/* DHT.c
+ *
+ * An implementation of the DHT as seen in docs/DHT.txt
+ *
+ *  Copyright (C) 2013 Tox project All Rights Reserved.
+ *
+ *  This file is part of Tox.
+ *
+ *  Tox is free software: you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License as published by
+ *  the Free Software Foundation, either version 3 of the License, or
+ *  (at your option) any later version.
+ *
+ *  Tox is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with Tox.  If not, see <http://www.gnu.org/licenses/>.
+ *
+ */
+
+#include "DHT.h"
+
+typedef struct {
+    uint8_t client_id[CLIENT_ID_SIZE];
+    IP_Port ip_port;
+    uint32_t timestamp;
+    uint32_t last_pinged;
+    IP_Port ret_ip_port;/* The ip_port returned by this node for the friend
+                           (for nodes in friends_list) or us (for nodes in close_clientlist) */
+    uint32_t ret_timestamp;
+} Client_data;
+
+/* maximum number of clients stored per friend. */
+#define MAX_FRIEND_CLIENTS 8
+
+typedef struct {
+    uint8_t client_id[CLIENT_ID_SIZE];
+    Client_data client_list[MAX_FRIEND_CLIENTS];
+    uint32_t lastgetnode; /* time at which the last get_nodes request was sent. */
+
+    /*Symetric NAT hole punching stuff*/
+    uint8_t hole_punching; /*0 if not hole punching, 1 if currently hole punching */
+    uint32_t punching_index;
+    uint32_t punching_timestamp;
+    uint32_t recvNATping_timestamp;
+    uint64_t NATping_id;
+    uint32_t NATping_timestamp;
+} Friend;
+
+typedef struct {
+    uint8_t client_id[CLIENT_ID_SIZE];
+    IP_Port ip_port;
+} Node_format;
+
+typedef struct {
+    IP_Port ip_port;
+    uint64_t ping_id;
+    uint32_t timestamp;
+
+} Pinged;
+
+/* Our client id/public key */
+uint8_t self_public_key[CLIENT_ID_SIZE];
+uint8_t self_secret_key[crypto_box_SECRETKEYBYTES];
+
+/* TODO: Move these out of here and put them into the .c file.
+   A list of the clients mathematically closest to ours. */
+#define LCLIENT_LIST 32
+static Client_data close_clientlist[LCLIENT_LIST];
+
+static Friend * friends_list;
+static uint16_t num_friends;
+
+/* The list of ip ports along with the ping_id of what we sent them and a timestamp */
+#define LPING_ARRAY 256
+
+static Pinged pings[LPING_ARRAY];
+
+#define LSEND_NODES_ARRAY LPING_ARRAY/2
+
+static Pinged send_nodes[LSEND_NODES_ARRAY];
+
+/* Compares client_id1 and client_id2 with client_id
+   return 0 if both are same distance
+   return 1 if client_id1 is closer
+   return 2 if client_id2 is closer */
+int id_closest(uint8_t * client_id, uint8_t * client_id1, uint8_t * client_id2) /* tested */
+{
+    uint32_t i;
+    for(i = 0; i < CLIENT_ID_SIZE; ++i) {
+        if(abs(client_id[i] ^ client_id1[i]) < abs(client_id[i] ^ client_id2[i]))
+            return 1;
+        else if(abs(client_id[i] ^ client_id1[i]) > abs(client_id[i] ^ client_id2[i]))
+            return 2;
+    }
+    return 0;
+}
+
+/* check if client with client_id is already in list of length length.
+   if it is set it's corresponding timestamp to current time.
+   if the id is already in the list with a different ip_port, update it.
+   return True(1) or False(0)
+   TODO: maybe optimize this. */
+int client_in_list(Client_data * list, uint32_t length, uint8_t * client_id, IP_Port ip_port)
+{
+    uint32_t i;
+    uint32_t temp_time = unix_time();
+
+    for(i = 0; i < length; ++i) {
+        /*If ip_port is assigned to a different client_id replace it*/
+        if(list[i].ip_port.ip.i == ip_port.ip.i &&
+           list[i].ip_port.port == ip_port.port) {
+            memcpy(list[i].client_id, client_id, CLIENT_ID_SIZE);
+        }
+
+        if(memcmp(list[i].client_id, client_id, CLIENT_ID_SIZE) == 0) {
+            /* Refresh the client timestamp. */
+            list[i].timestamp = temp_time;
+            list[i].ip_port.ip.i = ip_port.ip.i;
+            list[i].ip_port.port = ip_port.port;
+            return 1;
+        }
+    }
+    return 0;
+
+}
+
+/* check if client with client_id is already in node format list of length length.
+   return True(1) or False(0) */
+int client_in_nodelist(Node_format * list, uint32_t length, uint8_t * client_id)
+{
+    uint32_t i;
+    for(i = 0; i < length; ++i)
+        if(memcmp(list[i].client_id, client_id, CLIENT_ID_SIZE) == 0)
+            return 1;
+    return 0;
+}
+
+/*Return the friend number from the client_id
+ Return -1 if failure, number of friend if success*/
+static int friend_number(uint8_t * client_id)
+{
+    uint32_t i;
+    for(i = 0; i < num_friends; ++i)
+        if(memcmp(friends_list[i].client_id, client_id, CLIENT_ID_SIZE) == 0) /* Equal */
+            return i;
+    return -1;
+}
+
+/* the number of seconds for a non responsive node to become bad. */
+#define BAD_NODE_TIMEOUT 70
+/* the max number of nodes to send with send nodes. */
+#define MAX_SENT_NODES 8
+
+/* Find MAX_SENT_NODES nodes closest to the client_id for the send nodes request:
+   put them in the nodes_list and return how many were found.
+   TODO: Make this function much more efficient. */
+int get_close_nodes(uint8_t * client_id, Node_format * nodes_list)
+{
+    uint32_t i, j, k;
+    int num_nodes=0;
+    uint32_t temp_time = unix_time();
+    for(i = 0; i < LCLIENT_LIST; ++i)
+        if(close_clientlist[i].timestamp + BAD_NODE_TIMEOUT > temp_time &&
+           !client_in_nodelist(nodes_list, MAX_SENT_NODES,close_clientlist[i].client_id)) {
+            /* if node is good and not already in list. */
+            if(num_nodes < MAX_SENT_NODES) {
+                memcpy(nodes_list[num_nodes].client_id, close_clientlist[i].client_id, CLIENT_ID_SIZE);
+                nodes_list[num_nodes].ip_port = close_clientlist[i].ip_port;
+                num_nodes++;
+            } else for(j = 0; j < MAX_SENT_NODES; ++j)
+                    if(id_closest(client_id, nodes_list[j].client_id, close_clientlist[i].client_id) == 2) {
+                        memcpy(nodes_list[j].client_id, close_clientlist[i].client_id, CLIENT_ID_SIZE);
+                        nodes_list[j].ip_port = close_clientlist[i].ip_port;
+                        break;
+                    }
+        }
+
+    for(i = 0; i < num_friends; ++i)
+        for(j = 0; j < MAX_FRIEND_CLIENTS; ++j)
+            if(friends_list[i].client_list[j].timestamp + BAD_NODE_TIMEOUT > temp_time &&
+               !client_in_nodelist(nodes_list, MAX_SENT_NODES,friends_list[i].client_list[j].client_id)) {
+                /* if node is good and not already in list. */
+                if(num_nodes < MAX_SENT_NODES) {
+                    memcpy(nodes_list[num_nodes].client_id, friends_list[i].client_list[j].client_id, CLIENT_ID_SIZE);
+                    nodes_list[num_nodes].ip_port = friends_list[i].client_list[j].ip_port;
+                    num_nodes++;
+                } else for(k = 0; k < MAX_SENT_NODES; ++k)
+                        if(id_closest(client_id, nodes_list[k].client_id, friends_list[i].client_list[j].client_id) == 2) {
+                            memcpy(nodes_list[k].client_id, friends_list[i].client_list[j].client_id, CLIENT_ID_SIZE);
+                            nodes_list[k].ip_port = friends_list[i].client_list[j].ip_port;
+                            break;
+                        }
+            }
+    return num_nodes;
+}
+
+/* replace first bad (or empty) node with this one
+   return 0 if successful
+   return 1 if not (list contains no bad nodes) */
+int replace_bad(Client_data * list, uint32_t length, uint8_t * client_id, IP_Port ip_port) /* tested */
+{
+    uint32_t i;
+    uint32_t temp_time = unix_time();
+    for(i = 0; i < length; ++i)
+        if(list[i].timestamp + BAD_NODE_TIMEOUT < temp_time) { /* if node is bad. */
+            memcpy(list[i].client_id, client_id, CLIENT_ID_SIZE);
+            list[i].ip_port = ip_port;
+            list[i].timestamp = temp_time;
+            list[i].ret_ip_port.ip.i = 0;
+            list[i].ret_ip_port.port = 0;
+            list[i].ret_timestamp = 0;
+            return 0;
+        }
+
+    return 1;
+}
+
+/* replace the first good node that is further to the comp_client_id than that of the client_id in the list */
+int replace_good(Client_data * list, uint32_t length, uint8_t * client_id, IP_Port ip_port, uint8_t * comp_client_id)
+{
+    uint32_t i;
+    uint32_t temp_time = unix_time();
+
+    for(i = 0; i < length; ++i)
+        if(id_closest(comp_client_id, list[i].client_id, client_id) == 2) {
+            memcpy(list[i].client_id, client_id, CLIENT_ID_SIZE);
+            list[i].ip_port = ip_port;
+            list[i].timestamp = temp_time;
+            list[i].ret_ip_port.ip.i = 0;
+            list[i].ret_ip_port.port = 0;
+            list[i].ret_timestamp = 0;
+            return 0;
+        }
+
+    return 1;
+}
+
+/* Attempt to add client with ip_port and client_id to the friends client list and close_clientlist */
+void addto_lists(IP_Port ip_port, uint8_t * client_id)
+{
+    uint32_t i;
+
+    /* NOTE: current behavior if there are two clients with the same id is to replace the first ip by the second. */
+    if(!client_in_list(close_clientlist, LCLIENT_LIST, client_id, ip_port))
+        if(replace_bad(close_clientlist, LCLIENT_LIST, client_id, ip_port))
+            /* if we can't replace bad nodes we try replacing good ones */
+            replace_good(close_clientlist, LCLIENT_LIST, client_id, ip_port, self_public_key);
+
+    for(i = 0; i < num_friends; ++i)
+        if(!client_in_list(friends_list[i].client_list, MAX_FRIEND_CLIENTS, client_id, ip_port))
+            if(replace_bad(friends_list[i].client_list, MAX_FRIEND_CLIENTS, client_id, ip_port))
+                /* if we can't replace bad nodes we try replacing good ones. */
+                replace_good(friends_list[i].client_list, MAX_FRIEND_CLIENTS, client_id, ip_port, friends_list[i].client_id);
+}
+
+/* If client_id is a friend or us, update ret_ip_port
+   nodeclient_id is the id of the node that sent us this info */
+void returnedip_ports(IP_Port ip_port, uint8_t * client_id, uint8_t * nodeclient_id)
+{
+    uint32_t i, j;
+    uint32_t temp_time = unix_time();
+    if(memcmp(client_id, self_public_key, CLIENT_ID_SIZE) == 0) {
+        for(i = 0; i < LCLIENT_LIST; ++i)
+            if(memcmp(nodeclient_id, close_clientlist[i].client_id, CLIENT_ID_SIZE) == 0) {
+                close_clientlist[i].ret_ip_port = ip_port;
+                close_clientlist[i].ret_timestamp = temp_time;
+                return;
+            }
+    } else
+        for(i = 0; i < num_friends; ++i)
+            if(memcmp(client_id, friends_list[i].client_id, CLIENT_ID_SIZE) == 0)
+                for(j = 0; j < MAX_FRIEND_CLIENTS; ++j)
+                    if(memcmp(nodeclient_id, friends_list[i].client_list[j].client_id, CLIENT_ID_SIZE) == 0) {
+                        friends_list[i].client_list[j].ret_ip_port = ip_port;
+                        friends_list[i].client_list[j].ret_timestamp = temp_time;
+                        return;
+                    }
+}
+
+/* ping timeout in seconds */
+#define PING_TIMEOUT 5
+
+/* check if we are currently pinging an ip_port and/or a ping_id
+   variables with values of zero will not be checked.
+   if we are already, return 1
+   else return 0
+TODO: optimize this */
+int is_pinging(IP_Port ip_port, uint64_t ping_id)
+{
+    uint32_t i;
+    uint8_t pinging;
+    uint32_t temp_time = unix_time();
+
+    for(i = 0; i < LPING_ARRAY; ++i )
+        if((pings[i].timestamp + PING_TIMEOUT) > temp_time) {
+            pinging = 0;
+            if(ip_port.ip.i != 0)
+                if(pings[i].ip_port.ip.i == ip_port.ip.i &&
+                   pings[i].ip_port.port == ip_port.port)
+                    ++pinging;
+            if(ping_id != 0)
+                if(pings[i].ping_id == ping_id)
+                    ++pinging;
+            if(pinging == (ping_id != 0) + (ip_port.ip.i != 0))
+                return 1;
+        }
+
+    return 0;
+}
+
+/* Same as last function but for get_node requests. */
+int is_gettingnodes(IP_Port ip_port, uint64_t ping_id)
+{
+    uint32_t i;
+    uint8_t pinging;
+    uint32_t temp_time = unix_time();
+
+    for(i = 0; i < LSEND_NODES_ARRAY; ++i )
+        if((send_nodes[i].timestamp + PING_TIMEOUT) > temp_time) {
+            pinging = 0;
+            if(ip_port.ip.i != 0)
+                if(send_nodes[i].ip_port.ip.i == ip_port.ip.i &&
+                   send_nodes[i].ip_port.port == ip_port.port)
+                    ++pinging;
+            if(ping_id != 0)
+                if(send_nodes[i].ping_id == ping_id)
+                    ++pinging;
+            if(pinging == (ping_id != 0) + (ip_port.ip.i != 0))
+                return 1;
+
+        }
+
+    return 0;
+}
+
+/* Add a new ping request to the list of ping requests
+   returns the ping_id to put in the ping request
+   returns 0 if problem.
+   TODO: optimize this */
+uint64_t add_pinging(IP_Port ip_port)
+{
+    uint32_t i, j;
+    uint64_t ping_id = ((uint64_t)random_int() << 32) + random_int();
+    uint32_t temp_time = unix_time();
+
+    for(i = 0; i < PING_TIMEOUT; ++i )
+        for(j = 0; j < LPING_ARRAY; ++j )
+            if((pings[j].timestamp + PING_TIMEOUT - i) < temp_time) {
+                pings[j].timestamp = temp_time;
+                pings[j].ip_port = ip_port;
+                pings[j].ping_id = ping_id;
+                return ping_id;
+            }
+
+    return 0;
+}
+
+/* Same but for get node requests */
+uint64_t add_gettingnodes(IP_Port ip_port)
+{
+    uint32_t i, j;
+    uint64_t ping_id = ((uint64_t)random_int() << 32) + random_int();
+    uint32_t temp_time = unix_time();
+
+    for(i = 0; i < PING_TIMEOUT; ++i )
+        for(j = 0; j < LSEND_NODES_ARRAY; ++j )
+            if((send_nodes[j].timestamp + PING_TIMEOUT - i) < temp_time) {
+                send_nodes[j].timestamp = temp_time;
+                send_nodes[j].ip_port = ip_port;
+                send_nodes[j].ping_id = ping_id;
+                return ping_id;
+            }
+
+    return 0;
+}
+
+/* send a ping request
+   Ping request only works if none has been sent to that ip/port in the last 5 seconds. */
+static int pingreq(IP_Port ip_port, uint8_t * public_key)
+{
+    if(memcmp(public_key, self_public_key, CLIENT_ID_SIZE) == 0) /* check if packet is gonna be sent to ourself */
+        return 1;
+
+    if(is_pinging(ip_port, 0))
+        return 1;
+
+    uint64_t ping_id = add_pinging(ip_port);
+    if(ping_id == 0)
+        return 1;
+
+    uint8_t data[1 + CLIENT_ID_SIZE + crypto_box_NONCEBYTES + sizeof(ping_id) + ENCRYPTION_PADDING];
+    uint8_t encrypt[sizeof(ping_id) + ENCRYPTION_PADDING];
+    uint8_t nonce[crypto_box_NONCEBYTES];
+    random_nonce(nonce);
+
+    int len = encrypt_data(public_key, self_secret_key, nonce, (uint8_t *)&ping_id, sizeof(ping_id), encrypt);
+    if(len != sizeof(ping_id) + ENCRYPTION_PADDING)
+        return -1;
+    data[0] = 0;
+    memcpy(data + 1, self_public_key, CLIENT_ID_SIZE);
+    memcpy(data + 1 + CLIENT_ID_SIZE, nonce, crypto_box_NONCEBYTES);
+    memcpy(data + 1 + CLIENT_ID_SIZE + crypto_box_NONCEBYTES, encrypt, len);
+
+    return sendpacket(ip_port, data, sizeof(data));
+}
+
+/* send a ping response */
+static int pingres(IP_Port ip_port, uint8_t * public_key, uint64_t ping_id)
+{
+    /* check if packet is gonna be sent to ourself */
+    if(memcmp(public_key, self_public_key, CLIENT_ID_SIZE) == 0)
+        return 1;
+
+    uint8_t data[1 + CLIENT_ID_SIZE + crypto_box_NONCEBYTES + sizeof(ping_id) + ENCRYPTION_PADDING];
+    uint8_t encrypt[sizeof(ping_id) + ENCRYPTION_PADDING];
+    uint8_t nonce[crypto_box_NONCEBYTES];
+    random_nonce(nonce);
+
+    int len = encrypt_data(public_key, self_secret_key, nonce, (uint8_t *)&ping_id, sizeof(ping_id), encrypt);
+    if(len != sizeof(ping_id) + ENCRYPTION_PADDING)
+        return -1;
+    data[0] = 1;
+    memcpy(data + 1, self_public_key, CLIENT_ID_SIZE);
+    memcpy(data + 1 + CLIENT_ID_SIZE, nonce, crypto_box_NONCEBYTES);
+    memcpy(data + 1 + CLIENT_ID_SIZE + crypto_box_NONCEBYTES, encrypt, len);
+
+    return sendpacket(ip_port, data, sizeof(data));
+}
+
+/* send a getnodes request */
+static int getnodes(IP_Port ip_port, uint8_t * public_key, uint8_t * client_id)
+{
+    /* check if packet is gonna be sent to ourself */
+    if(memcmp(public_key, self_public_key, CLIENT_ID_SIZE) == 0)
+        return 1;
+
+    if(is_gettingnodes(ip_port, 0))
+        return 1;
+
+    uint64_t ping_id = add_gettingnodes(ip_port);
+
+    if(ping_id == 0)
+        return 1;
+
+    uint8_t data[1 + CLIENT_ID_SIZE + crypto_box_NONCEBYTES + sizeof(ping_id) + CLIENT_ID_SIZE + ENCRYPTION_PADDING];
+    uint8_t plain[sizeof(ping_id) + CLIENT_ID_SIZE];
+    uint8_t encrypt[sizeof(ping_id) + CLIENT_ID_SIZE + ENCRYPTION_PADDING];
+    uint8_t nonce[crypto_box_NONCEBYTES];
+    random_nonce(nonce);
+
+    memcpy(plain, &ping_id, sizeof(ping_id));
+    memcpy(plain + sizeof(ping_id), client_id, CLIENT_ID_SIZE);
+
+    int len = encrypt_data(public_key, self_secret_key, nonce, plain, sizeof(ping_id) + CLIENT_ID_SIZE, encrypt);
+
+    if(len != sizeof(ping_id) + CLIENT_ID_SIZE + ENCRYPTION_PADDING)
+        return -1;
+    data[0] = 2;
+    memcpy(data + 1, self_public_key, CLIENT_ID_SIZE);
+    memcpy(data + 1 + CLIENT_ID_SIZE, nonce, crypto_box_NONCEBYTES);
+    memcpy(data + 1 + CLIENT_ID_SIZE + crypto_box_NONCEBYTES, encrypt, len);
+    return sendpacket(ip_port, data, sizeof(data));
+}
+
+/* send a send nodes response */
+static int sendnodes(IP_Port ip_port, uint8_t * public_key, uint8_t * client_id, uint64_t ping_id)
+{
+    if(memcmp(public_key, self_public_key, CLIENT_ID_SIZE) == 0) /* check if packet is gonna be sent to ourself */
+        return 1;
+
+    uint8_t data[1 + CLIENT_ID_SIZE + crypto_box_NONCEBYTES + sizeof(ping_id)
+                 + sizeof(Node_format) * MAX_SENT_NODES + ENCRYPTION_PADDING];
+
+    Node_format nodes_list[MAX_SENT_NODES];
+    int num_nodes = get_close_nodes(client_id, nodes_list);
+
+    if(num_nodes == 0)
+        return 0;
+
+    uint8_t plain[sizeof(ping_id) + sizeof(Node_format) * MAX_SENT_NODES];
+    uint8_t encrypt[sizeof(ping_id) + sizeof(Node_format) * MAX_SENT_NODES + ENCRYPTION_PADDING];
+    uint8_t nonce[crypto_box_NONCEBYTES];
+    random_nonce(nonce);
+
+    memcpy(plain, &ping_id, sizeof(ping_id));
+    memcpy(plain + sizeof(ping_id), nodes_list, num_nodes * sizeof(Node_format));
+
+    int len = encrypt_data(public_key, self_secret_key, nonce, plain,
+                           sizeof(ping_id) + num_nodes * sizeof(Node_format), encrypt);
+
+    if(len != sizeof(ping_id) + num_nodes * sizeof(Node_format) + ENCRYPTION_PADDING)
+        return -1;
+
+    data[0] = 3;
+    memcpy(data + 1, self_public_key, CLIENT_ID_SIZE);
+    memcpy(data + 1 + CLIENT_ID_SIZE, nonce, crypto_box_NONCEBYTES);
+    memcpy(data + 1 + CLIENT_ID_SIZE + crypto_box_NONCEBYTES, encrypt, len);
+
+    return sendpacket(ip_port, data, 1 + CLIENT_ID_SIZE + crypto_box_NONCEBYTES + len);
+}
+
+/* Packet handling functions
+   One to handle each types of packets we receive
+   return 0 if handled correctly, 1 if packet is bad. */
+int handle_pingreq(uint8_t * packet, uint32_t length, IP_Port source)
+{
+    uint64_t ping_id;
+    if(length != 1 + CLIENT_ID_SIZE + crypto_box_NONCEBYTES + sizeof(ping_id) + ENCRYPTION_PADDING)
+        return 1;
+    /* check if packet is from ourself. */
+    if(memcmp(packet + 1, self_public_key, CLIENT_ID_SIZE) == 0)
+        return 1;
+
+    int len = decrypt_data(packet + 1, self_secret_key, packet + 1 + CLIENT_ID_SIZE,
+                           packet + 1 + CLIENT_ID_SIZE + crypto_box_NONCEBYTES,
+                           sizeof(ping_id) + ENCRYPTION_PADDING, (uint8_t *)&ping_id);
+    if(len != sizeof(ping_id))
+        return 1;
+
+    pingres(source, packet + 1, ping_id);
+
+    pingreq(source, packet + 1); /* TODO: make this smarter? */
+
+    return 0;
+}
+
+int handle_pingres(uint8_t * packet, uint32_t length, IP_Port source)
+{
+    uint64_t ping_id;
+    if(length != 1 + CLIENT_ID_SIZE + crypto_box_NONCEBYTES + sizeof(ping_id) + ENCRYPTION_PADDING)
+        return 1;
+    if(memcmp(packet + 1, self_public_key, CLIENT_ID_SIZE) == 0) /* check if packet is from ourself. */
+        return 1;
+
+    int len = decrypt_data(packet + 1, self_secret_key, packet + 1 + CLIENT_ID_SIZE,
+                           packet + 1 + CLIENT_ID_SIZE + crypto_box_NONCEBYTES,
+                           sizeof(ping_id) + ENCRYPTION_PADDING, (uint8_t *)&ping_id);
+    if(len != sizeof(ping_id))
+        return 1;
+
+    if(is_pinging(source, ping_id)) {
+        addto_lists(source, packet + 1);
+        return 0;
+    }
+    return 1;
+
+}
+
+int handle_getnodes(uint8_t * packet, uint32_t length, IP_Port source)
+{
+    uint64_t ping_id;
+    if(length != 1 + CLIENT_ID_SIZE + crypto_box_NONCEBYTES + sizeof(ping_id) + CLIENT_ID_SIZE + ENCRYPTION_PADDING)
+        return 1;
+    /* check if packet is from ourself. */
+    if(memcmp(packet + 1, self_public_key, CLIENT_ID_SIZE) == 0)
+        return 1;
+
+    uint8_t plain[sizeof(ping_id) + CLIENT_ID_SIZE];
+
+    int len = decrypt_data(packet + 1, self_secret_key, packet + 1 + CLIENT_ID_SIZE,
+                           packet + 1 + CLIENT_ID_SIZE + crypto_box_NONCEBYTES,
+                           sizeof(ping_id) + CLIENT_ID_SIZE + ENCRYPTION_PADDING, plain);
+
+    if(len != sizeof(ping_id) + CLIENT_ID_SIZE)
+        return 1;
+
+    memcpy(&ping_id, plain, sizeof(ping_id));
+    sendnodes(source, packet + 1, plain + sizeof(ping_id), ping_id);
+
+    pingreq(source, packet + 1); /* TODO: make this smarter? */
+
+    return 0;
+
+}
+
+int handle_sendnodes(uint8_t * packet, uint32_t length, IP_Port source)
+{
+    uint64_t ping_id;
+    /* TODO: make this more readable */
+    if(length > (1 + CLIENT_ID_SIZE + crypto_box_NONCEBYTES + sizeof(ping_id)
+                 + sizeof(Node_format) * MAX_SENT_NODES + ENCRYPTION_PADDING) ||
+       (length - (1 + CLIENT_ID_SIZE + crypto_box_NONCEBYTES + sizeof(ping_id)
+                  + ENCRYPTION_PADDING)) % (sizeof(Node_format)) != 0 ||
+       length < 1 + CLIENT_ID_SIZE + crypto_box_NONCEBYTES + sizeof(ping_id)
+       + sizeof(Node_format) + ENCRYPTION_PADDING) {
+        return 1;
+    }
+    uint32_t num_nodes = (length - (1 + CLIENT_ID_SIZE + crypto_box_NONCEBYTES
+                                    + sizeof(ping_id) + ENCRYPTION_PADDING)) / sizeof(Node_format);
+
+    uint8_t plain[sizeof(ping_id) + sizeof(Node_format) * MAX_SENT_NODES];
+
+    int len = decrypt_data(packet + 1, self_secret_key, packet + 1 + CLIENT_ID_SIZE,
+                           packet + 1 + CLIENT_ID_SIZE + crypto_box_NONCEBYTES,
+                           sizeof(ping_id) + num_nodes * sizeof(Node_format) + ENCRYPTION_PADDING, plain);
+
+    if(len != sizeof(ping_id) + num_nodes * sizeof(Node_format))
+        return 1;
+
+    memcpy(&ping_id, plain, sizeof(ping_id));
+    if(!is_gettingnodes(source, ping_id))
+        return 1;
+
+    Node_format nodes_list[MAX_SENT_NODES];
+    memcpy(nodes_list, plain + sizeof(ping_id), num_nodes * sizeof(Node_format));
+
+    addto_lists(source, packet + 1);
+
+    uint32_t i;
+    for(i = 0; i < num_nodes; ++i)  {
+        pingreq(nodes_list[i].ip_port, nodes_list[i].client_id);
+        returnedip_ports(nodes_list[i].ip_port, nodes_list[i].client_id, packet + 1);
+    }
+
+    return 0;
+}
+
+/* END of packet handling functions */
+
+int DHT_addfriend(uint8_t * client_id)
+{
+    Friend * temp;
+    temp = realloc(friends_list, sizeof(Friend) * (num_friends + 1));
+    if(temp == NULL)
+        return 1;
+
+    friends_list = temp;
+    memset(&friends_list[num_friends], 0, sizeof(Friend));
+    memcpy(friends_list[num_friends].client_id, client_id, CLIENT_ID_SIZE);
+    friends_list[num_friends].NATping_id = ((uint64_t)random_int() << 32) + random_int();
+    ++num_friends;
+    return 0;
+}
+
+int DHT_delfriend(uint8_t * client_id)
+{
+    uint32_t i;
+    Friend * temp;
+    for(i = 0; i < num_friends; ++i)
+        /* Equal */
+        if(memcmp(friends_list[i].client_id, client_id, CLIENT_ID_SIZE) == 0) {
+            --num_friends;
+            if(num_friends != i)
+                memcpy(friends_list[i].client_id, friends_list[num_friends].client_id, CLIENT_ID_SIZE);
+            temp = realloc(friends_list, sizeof(Friend) * (num_friends));
+            if(temp != NULL)
+                friends_list = temp;
+            return 0;
+        }
+
+    return 1;
+}
+
+/* TODO: Optimize this. */
+IP_Port DHT_getfriendip(uint8_t * client_id)
+{
+    uint32_t i, j;
+    IP_Port empty = {{{0}}, 0};
+    uint32_t temp_time = unix_time();
+    for(i = 0; i < num_friends; ++i)
+        /* Equal */
+        if(memcmp(friends_list[i].client_id, client_id, CLIENT_ID_SIZE) == 0) {
+            for(j = 0; j < MAX_FRIEND_CLIENTS; ++j)
+                if(memcmp(friends_list[i].client_list[j].client_id, client_id, CLIENT_ID_SIZE) == 0 &&
+                   friends_list[i].client_list[j].timestamp + BAD_NODE_TIMEOUT > temp_time)
+                    return friends_list[i].client_list[j].ip_port;
+
+            return empty;
+        }
+    empty.ip.i = 1;
+    return empty;
+
+}
+
+/* The timeout after which a node is discarded completely. */
+#define Kill_NODE_TIMEOUT 300
+
+/* ping interval in seconds for each node in our lists. */
+#define PING_INTERVAL 60
+
+/* ping interval in seconds for each random sending of a get nodes request. */
+#define GET_NODE_INTERVAL 10
+
+/* Ping each client in the "friends" list every 60 seconds.
+   Send a get nodes request every 20 seconds to a random good node for each "friend" in our "friends" list. */
+void doDHTFriends()
+{
+    uint32_t i, j;
+    uint32_t temp_time = unix_time();
+    uint32_t rand_node;
+    uint32_t index[MAX_FRIEND_CLIENTS];
+
+    for(i = 0; i < num_friends; ++i) {
+        uint32_t num_nodes = 0;
+        for(j = 0; j < MAX_FRIEND_CLIENTS; ++j)
+            if(friends_list[i].client_list[j].timestamp + Kill_NODE_TIMEOUT > temp_time) { /* if node is not dead. */
+                if((friends_list[i].client_list[j].last_pinged + PING_INTERVAL) <= temp_time) {
+                    pingreq(friends_list[i].client_list[j].ip_port, friends_list[i].client_list[j].client_id);
+                    friends_list[i].client_list[j].last_pinged = temp_time;
+                }
+                if(friends_list[i].client_list[j].timestamp + BAD_NODE_TIMEOUT > temp_time) { /* if node is good. */
+                    index[num_nodes] = j;
+                    ++num_nodes;
+                }
+            }
+        if(friends_list[i].lastgetnode + GET_NODE_INTERVAL <= temp_time && num_nodes != 0) {
+            rand_node = rand() % num_nodes;
+            getnodes(friends_list[i].client_list[index[rand_node]].ip_port,
+                     friends_list[i].client_list[index[rand_node]].client_id,
+                     friends_list[i].client_id);
+            friends_list[i].lastgetnode = temp_time;
+        }
+    }
+}
+
+static uint32_t close_lastgetnodes;
+
+/* Ping each client in the close nodes list every 60 seconds.
+   Send a get nodes request every 20 seconds to a random good node in the list. */
+void doClose() /* tested */
+{
+    uint32_t i;
+    uint32_t temp_time = unix_time();
+    uint32_t num_nodes = 0;
+    uint32_t rand_node;
+    uint32_t index[LCLIENT_LIST];
+
+    for(i = 0; i < LCLIENT_LIST; ++i)
+        /* if node is not dead. */
+        if(close_clientlist[i].timestamp + Kill_NODE_TIMEOUT > temp_time) {
+            if((close_clientlist[i].last_pinged + PING_INTERVAL) <= temp_time) {
+                pingreq(close_clientlist[i].ip_port, close_clientlist[i].client_id);
+                close_clientlist[i].last_pinged = temp_time;
+            }
+            /* if node is good. */
+            if(close_clientlist[i].timestamp + BAD_NODE_TIMEOUT > temp_time) {
+                index[num_nodes] = i;
+                ++num_nodes;
+            }
+        }
+
+    if(close_lastgetnodes + GET_NODE_INTERVAL <= temp_time && num_nodes != 0) {
+        rand_node = rand() % num_nodes;
+        getnodes(close_clientlist[index[rand_node]].ip_port,
+                 close_clientlist[index[rand_node]].client_id,
+                 self_public_key);
+        close_lastgetnodes = temp_time;
+    }
+}
+
+void DHT_bootstrap(IP_Port ip_port, uint8_t * public_key)
+{
+    getnodes(ip_port, public_key, self_public_key);
+}
+
+/* send the given packet to node with client_id
+   returns -1 if failure */
+int route_packet(uint8_t * client_id, uint8_t * packet, uint32_t length)
+{
+    uint32_t i;
+    for(i = 0; i < LCLIENT_LIST; ++i)
+        if(memcmp(client_id, close_clientlist[i].client_id, CLIENT_ID_SIZE) == 0)
+            return sendpacket(close_clientlist[i].ip_port, packet, length);
+    return -1;
+}
+
+/* Puts all the different ips returned by the nodes for a friend_num into array ip_portlist
+   ip_portlist must be at least MAX_FRIEND_CLIENTS big
+   returns the number of ips returned
+   return 0 if we are connected to friend or if no ips were found.
+   returns -1 if no such friend*/
+static int friend_iplist(IP_Port * ip_portlist, uint16_t friend_num)
+{
+    int num_ips = 0;
+    uint32_t i;
+    uint32_t temp_time = unix_time();
+    if(friend_num >= num_friends)
+        return -1;
+    for(i = 0; i < MAX_FRIEND_CLIENTS; ++i)
+        /*If ip is not zero and node is good */
+        if(friends_list[friend_num].client_list[i].ret_ip_port.ip.i != 0 &&
+           friends_list[friend_num].client_list[i].ret_timestamp + BAD_NODE_TIMEOUT > temp_time) {
+            if(memcmp(friends_list[friend_num].client_list[i].client_id, friends_list[friend_num].client_id, CLIENT_ID_SIZE) == 0 )
+                return 0;
+            ip_portlist[num_ips] = friends_list[friend_num].client_list[i].ret_ip_port;
+            ++num_ips;
+        }
+    return num_ips;
+}
+
+/* Send the following packet to everyone who tells us they are connected to friend_id
+   returns the number of nodes it sent the packet to */
+int route_tofriend(uint8_t * friend_id, uint8_t * packet, uint32_t length)
+{
+    uint32_t i, j;
+    uint32_t sent = 0;
+    uint32_t temp_time = unix_time();
+    for(i = 0; i < num_friends; ++i)
+        /* Equal */
+        if(memcmp(friends_list[i].client_id, friend_id, CLIENT_ID_SIZE) == 0)  {
+            for(j = 0; j < MAX_FRIEND_CLIENTS; ++j)
+                /*If ip is not zero and node is good */
+                if(friends_list[i].client_list[j].ret_ip_port.ip.i != 0 &&
+                   friends_list[i].client_list[j].ret_timestamp + BAD_NODE_TIMEOUT > temp_time)
+                    if(sendpacket(friends_list[i].client_list[j].ip_port, packet, length) == length)
+                        ++sent;
+            return sent;
+        }
+    return 0;
+}
+
+/* Send the following packet to one random person who tells us they are connected to friend_id
+   returns the number of nodes it sent the packet to */
+int routeone_tofriend(uint8_t * friend_id, uint8_t * packet, uint32_t length)
+{
+    int num = friend_number(friend_id);
+    if(num == -1)
+        return 0;
+
+    IP_Port ip_list[MAX_FRIEND_CLIENTS];
+    int n = 0;
+    uint32_t i;
+    uint32_t temp_time = unix_time();
+    for(i = 0; i < MAX_FRIEND_CLIENTS; ++i)
+        /*If ip is not zero and node is good */
+        if(friends_list[num].client_list[i].ret_ip_port.ip.i != 0 &&
+           friends_list[num].client_list[i].ret_timestamp + BAD_NODE_TIMEOUT > temp_time) {
+            ip_list[n] = friends_list[num].client_list[i].ip_port;
+            ++n;
+        }
+    if(n < 1)
+        return 0;
+    if(sendpacket(ip_list[rand() % n], packet, length) == length)
+        return 1;
+    return 0;
+}
+
+/* Puts all the different ips returned by the nodes for a friend_id into array ip_portlist
+   ip_portlist must be at least MAX_FRIEND_CLIENTS big
+   returns the number of ips returned
+   return 0 if we are connected to friend or if no ips were found.
+   returns -1 if no such friend*/
+int friend_ips(IP_Port * ip_portlist, uint8_t * friend_id)
+{
+
+    uint32_t i;
+    for(i = 0; i < num_friends; ++i)
+        /* Equal */
+        if(memcmp(friends_list[i].client_id, friend_id, CLIENT_ID_SIZE) == 0)
+            return friend_iplist(ip_portlist, i);
+    return -1;
+}
+
+/*BEGINNING OF NAT PUNCHING FUNCTIONS*/
+
+int send_NATping(uint8_t * public_key, uint64_t ping_id, uint8_t type)
+{
+    uint8_t data[sizeof(uint64_t) + 1];
+    data[0] = type;
+    memcpy(data + 1, &ping_id, sizeof(uint64_t));
+
+    uint8_t packet[MAX_DATA_SIZE];
+    int len = create_request(packet, public_key, data, sizeof(uint64_t) + 1, 254); /* 254 is NAT ping request packet id */
+    if(len == -1)
+        return -1;
+
+    int num = 0;
+
+    if(type == 0)
+        num = route_tofriend(public_key, packet, len);/*If packet is request use many people to route it*/
+    else if(type == 1)
+        num = routeone_tofriend(public_key, packet, len);/*If packet is response use only one person to route it*/
+    if(num == 0)
+        return -1;
+    return num;
+}
+
+/* Handle a recieved ping request for */
+int handle_NATping(uint8_t * packet, uint32_t length, IP_Port source)
+{
+    if(length <= crypto_box_PUBLICKEYBYTES * 2 + crypto_box_NONCEBYTES + 1 + ENCRYPTION_PADDING &&
+       length > MAX_DATA_SIZE + ENCRYPTION_PADDING)
+        return 1;
+    /* check if request is for us. */
+    if(memcmp(packet + 1, self_public_key, crypto_box_PUBLICKEYBYTES) == 0) {
+        uint8_t public_key[crypto_box_PUBLICKEYBYTES];
+        uint8_t data[MAX_DATA_SIZE];
+        int len = handle_request(public_key, data, packet, length);
+        if(len != sizeof(uint64_t) + 1)
+            return 1;
+        uint64_t ping_id;
+        memcpy(&ping_id, data + 1, sizeof(uint64_t));
+
+        int friendnumber = friend_number(public_key);
+        if(friendnumber == -1)
+            return 1;
+
+        if(data[0] == 0) {
+            send_NATping(public_key, ping_id, 1); /*1 is reply*/
+            friends_list[friendnumber].recvNATping_timestamp = unix_time();
+            return 0;
+        } else if (data[0] == 1)
+            if(friends_list[friendnumber].NATping_id == ping_id) {
+                friends_list[friendnumber].NATping_id = ((uint64_t)random_int() << 32) + random_int();
+                friends_list[friendnumber].hole_punching = 1;
+                return 0;
+            }
+        return 1;
+    }
+    /* if request is not for us, try routing it. */
+    else if(route_packet(packet + 1, packet, length) == length)
+        return 0;
+    return 0;
+}
+
+/*Get the most common ip in the ip_portlist
+  Only return ip if it appears in list min_num or more
+  len must not be bigger than MAX_FRIEND_CLIENTS
+  return ip of 0 if failure */
+static IP NAT_commonip(IP_Port * ip_portlist, uint16_t len, uint16_t min_num)
+{
+    IP zero = {{0}};
+    if(len > MAX_FRIEND_CLIENTS)
+        return zero;
+
+    uint32_t i, j;
+    uint16_t numbers[MAX_FRIEND_CLIENTS] = {0};
+    for(i = 0; i < len; ++i) {
+        for(j = 0; j < len; ++j)
+            if(ip_portlist[i].ip.i == ip_portlist[j].ip.i)
+                ++numbers[i];
+        if(numbers[i] >= min_num)
+            return ip_portlist[i].ip;
+    }
+    return zero;
+}
+
+/*Return all the ports for one ip in a list
+  portlist must be at least len long
+  where len is the length of ip_portlist
+  returns the number of ports and puts the list of ports in portlist*/
+static uint16_t NAT_getports(uint16_t * portlist, IP_Port * ip_portlist, uint16_t len, IP ip)
+{
+    uint32_t i;
+    uint16_t num = 0;
+    for(i = 0; i < len; ++i)
+        if(ip_portlist[i].ip.i == ip.i) {
+            portlist[num] = ntohs(ip_portlist[i].port);
+            ++num;
+        }
+    return num;
+}
+
+#define MAX_PUNCHING_PORTS 32
+
+static void punch_holes(IP ip, uint16_t * port_list, uint16_t numports, uint16_t friend_num)
+{
+    if(numports > MAX_FRIEND_CLIENTS || numports == 0)
+        return;
+    uint32_t i;
+    uint32_t top = friends_list[friend_num].punching_index + MAX_PUNCHING_PORTS;
+    for(i = friends_list[friend_num].punching_index; i != top; i++) {
+        /*TODO: improve port guessing algorithm*/
+        uint16_t port = port_list[(i/2) % numports] + (i/(2*numports))*((i % 2) ? -1 : 1);
+        IP_Port pinging = {ip, htons(port)};
+        pingreq(pinging, friends_list[friend_num].client_id);
+    }
+    friends_list[friend_num].punching_index = i;
+}
+
+/*Interval in seconds between punching attempts*/
+#define PUNCH_INTERVAL 10
+
+static void doNAT()
+{
+    uint32_t i;
+    uint32_t temp_time = unix_time();
+    for(i = 0; i < num_friends; ++i) {
+        IP_Port ip_list[MAX_FRIEND_CLIENTS];
+        int num = friend_iplist(ip_list, i);
+        /*If we are connected to friend or if friend is not online don't try to hole punch with him*/
+        if(num < MAX_FRIEND_CLIENTS/2)
+            continue;
+
+
+        if(friends_list[i].NATping_timestamp + PUNCH_INTERVAL < temp_time) {
+            send_NATping(friends_list[i].client_id, friends_list[i].NATping_id, 0); /*0 is request*/
+            friends_list[i].NATping_timestamp = temp_time;
+        }
+        if(friends_list[i].hole_punching == 1 &&
+            friends_list[i].punching_timestamp + PUNCH_INTERVAL < temp_time && 
+            friends_list[i].recvNATping_timestamp + PUNCH_INTERVAL*2 >= temp_time) {
+            IP ip = NAT_commonip(ip_list, num, MAX_FRIEND_CLIENTS/2);
+            if(ip.i == 0)
+                continue;
+
+            uint16_t port_list[MAX_FRIEND_CLIENTS];
+            uint16_t numports = NAT_getports(port_list, ip_list, num, ip);
+            punch_holes(ip, port_list, numports, i);
+
+            friends_list[i].punching_timestamp = temp_time;
+            friends_list[i].hole_punching = 0;
+        }
+    }
+}
+
+/*END OF NAT PUNCHING FUNCTIONS*/
+
+int DHT_handlepacket(uint8_t * packet, uint32_t length, IP_Port source)
+{
+    switch (packet[0]) {
+    case 0:
+        return handle_pingreq(packet, length, source);
+
+    case 1:
+        return handle_pingres(packet, length, source);
+
+    case 2:
+        return handle_getnodes(packet, length, source);
+
+    case 3:
+        return handle_sendnodes(packet, length, source);
+
+    case 254:
+        return handle_NATping(packet, length, source);
+
+    default:
+        return 1;
+
+    }
+
+    return 0;
+}
+
+void doDHT()
+{
+    doClose();
+    doDHTFriends();
+    doNAT();
+}
+
+/* get the size of the DHT (for saving) */
+uint32_t DHT_size()
+{
+    return sizeof(close_clientlist) + sizeof(Friend) * num_friends;
+}
+
+/* save the DHT in data where data is an array of size DHT_size() */
+void DHT_save(uint8_t * data)
+{
+    memcpy(data, close_clientlist, sizeof(close_clientlist));
+    memcpy(data + sizeof(close_clientlist), friends_list, sizeof(Friend) * num_friends);
+}
+
+/* load the DHT from data of size size;
+   return -1 if failure
+   return 0 if success */
+int DHT_load(uint8_t * data, uint32_t size)
+{
+    if(size < sizeof(close_clientlist))
+        return -1;
+    if((size - sizeof(close_clientlist)) % sizeof(Friend) != 0)
+        return -1;
+    uint32_t i, j;
+    /* uint32_t temp_time = unix_time(); */
+    uint16_t temp;
+
+    temp = (size - sizeof(close_clientlist))/sizeof(Friend);
+
+    if(temp != 0) {
+        Friend * tempfriends_list = (Friend *)(data + sizeof(close_clientlist));
+
+        for(i = 0; i < temp; ++i) {
+            DHT_addfriend(tempfriends_list[i].client_id);
+            for(j = 0; j < MAX_FRIEND_CLIENTS; ++j)
+                if(tempfriends_list[i].client_list[j].timestamp != 0) {
+                    getnodes(tempfriends_list[i].client_list[j].ip_port,
+                             tempfriends_list[i].client_list[j].client_id, tempfriends_list[i].client_id);
+                }
+        }
+    }
+    Client_data * tempclose_clientlist = (Client_data *)data;
+
+    for(i = 0; i < LCLIENT_LIST; ++i)
+        if(tempclose_clientlist[i].timestamp != 0)
+            DHT_bootstrap(tempclose_clientlist[i].ip_port, tempclose_clientlist[i].client_id);
+    return 0;
+}
+
+/* returns 0 if we are not connected to the DHT
+   returns 1 if we are */
+int DHT_isconnected()
+{
+    uint32_t i;
+    uint32_t temp_time = unix_time();
+    for(i = 0; i < LCLIENT_LIST; ++i)
+        if(close_clientlist[i].timestamp + BAD_NODE_TIMEOUT > temp_time)
+            return 1;
+    return 0;
+}