Palacios Public Git Repository

To checkout Palacios execute

  git clone http://v3vee.org/palacios/palacios.web/palacios.git
This will give you the master branch. You probably want the devel branch or one of the release branches. To switch to the devel branch, simply execute
  cd palacios
  git checkout --track -b devel origin/devel
The other branches are similar.


Merge branch 'devel' of ssh://palacios@newskysaw.cs.northwestern.edu/home/palacios...
Lei Xia [Mon, 16 Nov 2009 23:21:28 +0000 (17:21 -0600)]
Kconfig
palacios/include/palacios/vmm_vnet.h [new file with mode: 0644]
palacios/src/palacios/Makefile
palacios/src/palacios/vmm_vnet.c [new file with mode: 0644]

diff --git a/Kconfig b/Kconfig
index 9597a1a..c2a6762 100644 (file)
--- a/Kconfig
+++ b/Kconfig
@@ -101,6 +101,13 @@ config EXPERIMENTAL
          The purpose of this option is to allow features under development to be committed to the mainline
          to more easily track changes and provide access to multiple developers
 
+config VNET
+       depends on EXPERIMENTAL
+       bool "Enable Vnet in Palacios"
+       default n
+       help
+         Enable the Vnet in Palacios
+
 config BUILT_IN_STDLIB
        bool "Enable Built in versions of stdlib functions"
        default n
diff --git a/palacios/include/palacios/vmm_vnet.h b/palacios/include/palacios/vmm_vnet.h
new file mode 100644 (file)
index 0000000..0c2ded9
--- /dev/null
@@ -0,0 +1,149 @@
+/* 
+ * This file is part of the Palacios Virtual Machine Monitor developed
+ * by the V3VEE Project with funding from the United States National 
+ * Science Foundation and the Department of Energy.  
+ *
+ * The V3VEE Project is a joint project between Northwestern University
+ * and the University of New Mexico.  You can find out more at 
+ * http://www.v3vee.org
+ *
+ * Copyright (c) 2009, Lei Xia <lxia@northwestern.edu> 
+ * Copyright (c) 2009, Yuan Tang <ytang@northwestern.edu> 
+ * Copyright (c) 2009, Jack Lange <jarusl@cs.northwestern.edu> 
+ * Copyright (c) 2009, Peter Dinda <pdinda@northwestern.edu
+ * Copyright (c) 2009, The V3VEE Project <http://www.v3vee.org> 
+ * All rights reserved.
+ *
+ * Author: Lei Xia <lxia@northwestern.edu>
+ *               Yuan Tang <ytang@northwestern.edu>
+ *               Jack Lange <jarusl@cs.northwestern.edu> 
+ *               Peter Dinda <pdinda@northwestern.edu
+ *
+ * This is free software.  You are permitted to use,
+ * redistribute, and modify it as specified in the file "V3VEE_LICENSE".
+ */
+
+#ifndef __VNET_H__
+#define __VNET_H__
+
+#include <palacios/vmm.h>
+#include <palacios/vmm_string.h>
+#include <palacios/vmm_types.h>
+#include <palacios/vmm_queue.h>
+#include <palacios/vmm_socket.h>
+#include <palacios/vmm_hashtable.h>
+#include <palacios/vmm_sprintf.h>
+
+
+#define ETHERNET_HEADER_LEN 14
+#define ETHERNET_DATA_MIN   46
+#define ETHERNET_DATA_MAX   1500
+#define ETHERNET_PACKET_LEN (ETHERNET_HEADER_LEN + ETHERNET_DATA_MAX)
+
+#define SOCK int
+
+//the routing entry
+struct routing{
+    char src_mac[6];
+    char dest_mac[6];
+
+    int src_mac_qual;
+    int dest_mac_qual;
+
+    int dest; //link[dest] is the link to be used to send pkt
+    int type; //EDGE|INTERFACE|ANY
+    int src;
+    int src_type; //EDGE|INTERFACE|ANY
+
+    int use;
+
+    int next;
+    int prev;
+  //  struct list_head entry_list;
+};
+
+//struct  gen_route {
+ //   uint_t num_entries;
+  //  struct list_head entries;
+//}
+
+ //This is the structure that stores the topology 
+struct topology {
+    SOCK link_sock;
+
+    unsigned long dest;
+
+    // Port for UDP
+    unsigned short remote_port;
+
+    int use;
+    int type; //TCP=0, UDP=1,VTP=2, can be extended so on
+
+    int next;
+    int prev;
+};
+
+struct sock_list {
+    SOCK sock;
+
+    int next;
+    int prev;
+};
+
+
+#define GENERAL_NIC 0
+
+struct ethAddr{
+  char addr[6];
+};
+
+struct vnet_if_device {
+    char name[50];
+    struct ethAddr device_addr;
+    
+    int (*input)(uchar_t * pkt, uint_t size);
+    
+    void * data;
+};
+
+
+struct device_list {
+    struct vnet_if_device *device;
+
+    int use;
+    int type;
+
+    int next;
+    int prev;
+};
+
+// 14 (ethernet frame) + 20 bytes
+struct HEADERS {
+    char ethernetdest[6];
+    char ethernetsrc[6];
+    unsigned char ethernettype[2]; // indicates layer 3 protocol type
+    char ip[20];
+};
+
+#define FOREACH(iter, list, start) for (iter = start; iter != -1; iter = list[iter].next)
+#define FOREACH_SOCK(iter, socks, start) FOREACH(iter, socks, start)
+#define FOREACH_LINK(iter, links, start) FOREACH(iter, links, start)
+#define FOREACH_ROUTE(iter, routes, start) FOREACH(iter, routes, start)
+#define FOREACH_DEVICE(iter, devices, start) FOREACH(iter, devices, start)
+
+
+int v3_Send_pkt(uchar_t *buf, int length);
+int v3_Register_pkt_event(int (*netif_input)(uchar_t * pkt, uint_t size));
+
+
+int vnet_send_pkt(char *buf, int length);
+int vnet_register_pkt_event(char *dev_name, int (*netif_input)(uchar_t * pkt, uint_t size), void *data);
+
+int v3_vnet_pkt_process();
+
+void v3_vnet_init();
+
+#endif
+
+
index f5621bd..7dacc0f 100644 (file)
@@ -56,6 +56,7 @@ obj-$(CONFIG_VMX) +=          vmx.o \
 obj-$(CONFIG_INSTRUMENT_VMM) += vmm_instrument.o
 obj-$(CONFIG_TELEMETRY) += vmm_telemetry.o 
 obj-$(CONFIG_SOCKET) +=  vmm_socket.o
+obj-$(CONFIG_VNET) += vmm_vnet.o
 
 obj-$(CONFIG_SYMBIOTIC) += vmm_sym_iface.o
 
diff --git a/palacios/src/palacios/vmm_vnet.c b/palacios/src/palacios/vmm_vnet.c
new file mode 100644 (file)
index 0000000..02f5182
--- /dev/null
@@ -0,0 +1,1239 @@
+/* 
+ * This file is part of the Palacios Virtual Machine Monitor developed
+ * by the V3VEE Project with funding from the United States National 
+ * Science Foundation and the Department of Energy.  
+ *
+ * The V3VEE Project is a joint project between Northwestern University
+ * and the University of New Mexico.  You can find out more at 
+ * http://www.v3vee.org
+ *
+ * Copyright (c) 2009, Lei Xia <lxia@northwestern.edu> 
+ * Copyright (c) 2009, Yuan Tang <ytang@northwestern.edu> 
+ * Copyright (c) 2009, Jack Lange <jarusl@cs.northwestern.edu> 
+ * Copyright (c) 2009, Peter Dinda <pdinda@northwestern.edu>
+ * Copyright (c) 2009, The V3VEE Project <http://www.v3vee.org> 
+ * All rights reserved.
+ *
+ * Author: Lei Xia <lxia@northwestern.edu>
+ *        Yuan Tang <ytang@northwestern.edu>
+ *        Jack Lange <jarusl@cs.northwestern.edu> 
+ *        Peter Dinda <pdinda@northwestern.edu
+ *
+ * This is free software.  You are permitted to use,
+ * redistribute, and modify it as specified in the file "V3VEE_LICENSE".
+ */
+#include <palacios/vmm_vnet.h>
+
+static const char any_type_str[] = "any";
+static const char not_type_str[] = "not";
+static const char none_type_str[] = "none";
+static const char empty_type_str[] = "empty";
+
+static const char link_interface_str[] = "INTERFACE";
+static const char link_edge_str[] = "EDGE";
+static const char link_any_str[] = "ANY";
+
+static const char link_prot_tcp_str[] = "tcp";
+static const char link_prot_udp_str[] = "udp";
+
+typedef enum {MAC_ANY, MAC_NOT, MAC_NONE, MAC_EMPTY} mac_type_t; //for 'src_mac_qual' and 'dst_mac_qual'
+typedef enum {LINK_INTERFACE, LINK_EDGE, LINK_ANY} link_type_t; //for 'type' and 'src_type' in struct routing
+typedef enum {TCP_TYPE, UDP_TYPE} link_prot_type_t;
+
+static const uint_t hash_key_size = 16;
+static SOCK g_udp_sockfd;
+static struct gen_queue * g_inpkt_q;
+static bool use_tcp = false;
+static uint_t vnet_udp_port = 1022;
+
+struct raw_ethernet_pkt {
+  int  size;
+  int type; // vm or link type:  INTERFACE|EDGE
+  char  data[ETHERNET_PACKET_LEN];
+};
+
+//static char *vnet_version = "0.9";
+//static int vnet_server = 0;
+
+
+#define MAX_LINKS 1
+#define MAX_ROUTES 1
+#define MAX_DEVICES 16
+
+static struct topology g_links[MAX_LINKS];
+static int g_num_links; //The current number of links
+static int g_first_link;
+static int g_last_link;
+
+static struct routing g_routes[MAX_ROUTES];
+static int g_num_routes; //The current number of routes
+static int g_first_route;
+static int g_last_route;
+
+static struct device_list g_devices[MAX_DEVICES];
+static int g_num_devices;
+static int g_first_device;
+static int g_last_device;
+
+
+static void print_packet(char *pkt, int size) {
+    PrintDebug("Vnet: print_data_packet: size: %d\n", size);
+    v3_hexdump(pkt, size, NULL, 0);
+}
+
+#if 0
+static void print_packet_addr(char *pkt) {
+    PrintDebug("Vnet: print_packet_destination_addr: ");
+    v3_hexdump(pkt + 8, 6, NULL, 0);
+    
+    PrintDebug("Vnet: print_packet_source_addr: ");
+    v3_hexdump(pkt + 14, 6, NULL, 0);
+}
+
+static void print_device_addr(char *ethaddr) {
+    PrintDebug("Vnet: print_device_addr: ");
+    v3_hexdump(ethaddr, 6, NULL, 0);
+} 
+#endif
+
+
+//network connection functions 
+static inline void raw_ethernet_packet_init(struct raw_ethernet_pkt * pt, const char * data, const size_t size) {
+    pt->size = size;
+    memcpy(pt->data, data, size);
+}
+
+
+
+/* Hash key format:
+ * 0-5:     src_eth_addr
+ * 6-11:    dest_eth_addr
+ * 12:      src type
+ * 13-16:   src index
+ */
+typedef char * route_hashkey_t;
+
+// This is the hash value, Format: 0: num_matched_routes, 1...n: matches[] -- TY
+struct route_cache_entry {
+    int num_matched_routes;
+    int * matches; 
+};
+
+#define HASH_KEY_LEN 16
+#define MIN_CACHE_SIZE 100
+
+
+//Header of the route cache
+static struct hashtable * g_route_cache; 
+
+static uint_t hash_from_key_fn(addr_t hashkey) {    
+    uint8_t * key = (uint8_t *)hashkey;
+    return v3_hash_buffer(key, HASH_KEY_LEN);
+}
+
+static int hash_key_equal(addr_t key1, addr_t key2) {
+    uint8_t * buf1 = (uint8_t *)key1;
+    uint8_t * buf2 = (uint8_t *)key2;
+    return (memcmp(buf1, buf2, HASH_KEY_LEN) == 0);
+}
+
+static int init_route_cache() {
+    g_route_cache = v3_create_htable(MIN_CACHE_SIZE, &hash_from_key_fn, &hash_key_equal);
+
+    if (g_route_cache == NULL){
+        PrintError("Vnet: Route Cache Initiate Failurely\n");
+        return -1;
+    }
+
+    return 0;
+}
+
+static void make_hash_key(route_hashkey_t hashkey, char src_addr[6], char dest_addr[6], char src_type, int src_index) {
+    int j;
+
+    for (j = 0; j < 6; j++) {
+       hashkey[j] = src_addr[j];
+       hashkey[j + 6] = dest_addr[j] + 1;
+    }
+
+    hashkey[12] = src_type;
+
+    *(int *)(hashkey + 12) = src_index;
+}
+static int add_route_to_cache(route_hashkey_t hashkey, int num_matched_r, int * matches) {
+    struct route_cache_entry * new_entry = NULL;
+    int i;
+    
+    new_entry = (struct route_cache_entry *)V3_Malloc(sizeof(struct route_cache_entry));
+    if (new_entry == NULL){
+       PrintError("Vnet: Malloc fails\n");
+       return -1;
+    }
+    
+    new_entry->num_matched_routes = num_matched_r;
+
+    new_entry->matches = (int *)V3_Malloc(num_matched_r * sizeof(int));
+    
+    if (new_entry->matches == NULL){
+       PrintError("Vnet: Malloc fails\n");
+       return -1;
+    }
+    
+    for (i = 0; i < num_matched_r; i++) {
+       new_entry->matches[i] = matches[i];
+    }
+    
+    //here, when v3_htable_insert return 0, it means insert fails
+    if (v3_htable_insert(g_route_cache, (addr_t)hashkey, (addr_t)new_entry) == 0){
+       PrintError("Vnet: Insert new route entry to cache failed\n");
+       V3_Free(new_entry->matches);
+       V3_Free(new_entry);
+    }
+    
+    return 0;
+}
+
+static int clear_hash_cache() {
+    v3_free_htable(g_route_cache, 1, 1);
+               
+    g_route_cache = v3_create_htable(MIN_CACHE_SIZE, hash_from_key_fn, hash_key_equal);
+    
+    if (g_route_cache == NULL){
+        PrintError("Vnet: Route Cache Create Failurely\n");
+        return -1;
+    }
+
+    return 0;
+}
+
+static int look_into_cache(route_hashkey_t hashkey, int * matches) {
+    int n_matches = -1;
+    int i = 0;
+    struct route_cache_entry * found = NULL;
+    
+    found = (struct route_cache_entry *)v3_htable_search(g_route_cache, (addr_t)hashkey);
+   
+    if (found != NULL) {
+        n_matches = found->num_matched_routes;
+
+        for (i = 0; i < n_matches; i++) {
+            matches[i] = found->matches[i];
+       }
+    }
+
+    return n_matches;
+}
+
+
+
+
+static inline uint8_t hex_nybble_to_nybble(const uint8_t hexnybble) {
+    uint8_t x = toupper(hexnybble);
+
+    if (isdigit(x)) {
+       return x - '0';
+    } else {
+       return 10 + (x - 'A');
+    }
+}
+
+static inline uint8_t hex_byte_to_byte(const uint8_t hexbyte[2]) {
+    return ((hex_nybble_to_nybble(hexbyte[0]) << 4) + 
+           (hex_nybble_to_nybble(hexbyte[1]) & 0xf));
+}
+
+
+static inline void string_to_mac(const char * str, uint8_t mac[6]) {
+    int k;
+
+    for (k = 0; k < 6; k++) {
+       mac[k] = hex_byte_to_byte(&(str[(2 * k) + k]));
+    }
+}
+
+static inline void mac_to_string(int mac[6], char * buf) {
+    snprintf(buf, 20, "%x:%x:%x:%x:%x:%x", 
+            mac[0], mac[1], mac[2],
+            mac[3], mac[4], mac[5]);
+}
+
+#if 0
+static void ip_to_string(uint32_t addr, char * buf) {
+    uint32_t addr_st;
+    char * tmp_str;
+    
+    addr_st = v3_htonl(addr);
+    tmp_str = v3_inet_ntoa(addr_st);
+    
+    memcpy(buf, tmp_str, strlen(tmp_str));
+}
+#endif
+
+int find_link_by_fd(SOCK sock) {
+    int i;
+
+    FOREACH_LINK(i, g_links, g_first_link) {
+       if (g_links[i].link_sock == sock) {
+           return i;
+       }
+    }
+    
+    return -1;
+}
+
+int vnet_add_link_entry(unsigned long dest, int type, int data_port,  SOCK fd) {
+    int i;
+    
+    for (i = 0; i < MAX_LINKS; i++) {
+       if (g_links[i].use == 0) {
+           g_links[i].dest = dest;
+           g_links[i].type = type;
+           g_links[i].link_sock = fd;
+           g_links[i].remote_port = data_port;
+           g_links[i].use = 1;
+
+           if (g_first_link == -1) {
+               g_first_link = i;
+           }
+
+           g_links[i].prev = g_last_link;
+           g_links[i].next = -1;
+           
+           if (g_last_link != -1) {
+               g_links[g_last_link].next = i;
+           }
+           
+           g_last_link = i;
+           
+           g_num_links++;
+           
+           return i;
+       }
+    }
+    
+    return -1;
+}
+
+
+int add_sock(struct sock_list *socks, int  len, int *first_sock, int *last_sock, SOCK fd) {
+    int i;
+
+    for (i = 0; i < len; i++) {
+       if (socks[i].sock == -1) {
+           socks[i].sock = fd;
+           
+           if (*first_sock == -1) {
+               *first_sock = i;
+           }
+           
+           socks[i].prev = *last_sock;
+           socks[i].next = -1;
+           
+           if (*last_sock != -1) {
+               socks[*last_sock].next = i;
+           }
+           
+           *last_sock = i;
+           
+           return i;
+       }
+    }
+    return -1;
+}
+
+
+int vnet_add_route_entry(char src_mac[6], char dest_mac[6], int src_mac_qual, int dest_mac_qual, int dest, int type, int src, int src_type) {
+    int i;
+    
+    for(i = 0; i < MAX_ROUTES; i++) {
+       if (g_routes[i].use == 0) {
+           
+           if ((src_mac_qual != MAC_ANY) && (src_mac_qual != MAC_NONE)) {
+               memcpy(g_routes[i].src_mac, src_mac, 6);
+           } else {
+               memset(g_routes[i].src_mac, 0, 6);
+           }
+           
+           if ((dest_mac_qual != MAC_ANY) && (dest_mac_qual != MAC_NONE)) {
+               memcpy(g_routes[i].dest_mac, dest_mac, 6);
+           } else {
+               memset(g_routes[i].dest_mac, 0, 6);
+           }
+           
+           g_routes[i].src_mac_qual = src_mac_qual;
+           g_routes[i].dest_mac_qual = dest_mac_qual;
+           g_routes[i].dest = dest;
+           g_routes[i].type = type;
+           g_routes[i].src = src;
+           g_routes[i].src_type = src_type;
+           g_routes[i].use = 1;
+           
+           if (g_first_route == -1) {
+               g_first_route = i;
+           }
+           
+           g_routes[i].prev = g_last_route;
+           g_routes[i].next = -1;
+           
+           if (g_last_route != -1) {
+               g_routes[g_last_route].next = i;
+           }
+           
+           g_last_route = i;
+           
+           g_num_routes++;
+           
+           return i;
+       }
+    }
+    
+    clear_hash_cache();
+    
+    return -1;
+}
+
+
+static int find_link_entry(unsigned long dest, int type) {
+    int i;
+
+    FOREACH_LINK(i, g_links, g_first_link) {
+       if ((g_links[i].dest == dest) && 
+           ((type == -1) || (g_links[i].type == type)) ) {
+           return i;
+       }
+    } 
+    
+    return -1;
+}
+
+static int delete_link_entry(int index) {
+    int next_i;
+    int prev_i;
+  
+    if (g_links[index].use == 0) {
+       return -1;
+    }
+
+    g_links[index].dest = 0;
+    g_links[index].type = 0;
+    g_links[index].link_sock = -1;
+    g_links[index].use = 0;
+
+    prev_i = g_links[index].prev;
+    next_i = g_links[index].next;
+
+    if (prev_i != -1) {
+       g_links[prev_i].next = g_links[index].next;
+    }    
+
+    if (next_i != -1) {
+       g_links[next_i].prev = g_links[index].prev;
+    }
+    
+    if (g_first_link == index) {
+       g_first_link = g_links[index].next;
+    }    
+
+    if (g_last_link == index) {
+       g_last_link = g_links[index].prev;
+    }    
+
+    g_links[index].next = -1;
+    g_links[index].prev = -1;
+    
+    g_num_links--;
+    
+    return 0;
+}
+
+int vnet_delete_link_entry_by_addr(unsigned long dest, int type) {
+    int index = find_link_entry(dest, type);
+  
+    if (index == -1) {
+       return -1;
+    }
+
+    return delete_link_entry(index);
+}
+
+
+static int find_route_entry(char src_mac[6], 
+                           char dest_mac[6], 
+                           int src_mac_qual, 
+                           int dest_mac_qual, 
+                           int dest, 
+                           int type, 
+                           int src, 
+                           int src_type) {
+    int i;
+    char temp_src_mac[6];
+    char temp_dest_mac[6];
+  
+    if ((src_mac_qual != MAC_ANY) && (src_mac_qual != MAC_NONE)) {
+       memcpy(temp_src_mac, src_mac, 6);
+    } else {
+       memset(temp_src_mac, 0, 6);
+    }
+    
+    if ((dest_mac_qual != MAC_ANY) && (dest_mac_qual != MAC_NONE)) {
+       memcpy(temp_dest_mac, dest_mac, 6);
+    } else {
+       memset(temp_dest_mac, 0, 6);
+    }
+    
+    FOREACH_LINK(i, g_routes, g_first_route) {
+       if ( (memcmp(temp_src_mac, g_routes[i].src_mac, 6) == 0) && 
+            (memcmp(temp_dest_mac, g_routes[i].dest_mac, 6) == 0) &&
+            (g_routes[i].src_mac_qual == src_mac_qual) &&
+            (g_routes[i].dest_mac_qual == dest_mac_qual)  &&
+            ( (type == -1) || 
+              ( (type == g_routes[i].type) && (g_routes[i].dest == dest)) ) &&
+            ( (src_type == -1) || 
+              ( (src_type == g_routes[i].src_type) && (g_routes[i].src == src)) ) ) {
+           return i;
+       }
+    } 
+    
+    return -1;
+}
+
+static int delete_route_entry(int index) {
+    int next_i;
+    int prev_i;
+
+    memset(g_routes[index].src_mac, 0, 6);
+    memset(g_routes[index].dest_mac, 0, 6);
+
+    g_routes[index].dest = 0;
+    g_routes[index].src = 0;
+    g_routes[index].src_mac_qual = 0;
+    g_routes[index].dest_mac_qual = 0;
+    g_routes[index].type = -1;
+    g_routes[index].src_type = -1;
+    g_routes[index].use = 0;
+
+    prev_i = g_routes[index].prev;
+    next_i = g_routes[index].next;
+  
+    if (prev_i != -1) {
+       g_routes[prev_i].next = g_routes[index].next;
+    }
+    
+    if (next_i != -1) {
+       g_routes[next_i].prev = g_routes[index].prev;
+    }
+    
+    if (g_first_route == index) {
+       g_first_route = g_routes[index].next;
+    }    
+
+    if (g_last_route == index) {
+       g_last_route = g_routes[index].prev;
+    }    
+
+    g_routes[index].next = -1;
+    g_routes[index].prev = -1;
+    
+    g_num_routes--;
+    
+    clear_hash_cache();
+    
+    return 0;
+}
+
+int vnet_delete_route_entry_by_addr(char src_mac[6], 
+                                   char dest_mac[6], 
+                                   int src_mac_qual, 
+                                   int dest_mac_qual, 
+                                   int dest, 
+                                   int type, 
+                                   int src, 
+                                   int src_type) {
+    int index = find_route_entry(src_mac, dest_mac, src_mac_qual, 
+                                dest_mac_qual, dest, type, src, src_type);
+    
+    if (index == -1) {
+       return -1;
+    }
+    
+    delete_route_entry(index);
+    
+    return 0;
+}
+
+int delete_sock(struct sock_list * socks, int * first_sock, int * last_sock, SOCK fd) {
+    int i;
+    int prev_i;
+    int next_i;
+    
+  
+    FOREACH_SOCK(i, socks, (*first_sock)) {
+       if (socks[i].sock == fd) {
+           V3_Close_Socket(socks[i].sock);
+           socks[i].sock = -1;
+           
+           prev_i = socks[i].prev;
+           next_i = socks[i].next;
+           
+           if (prev_i != -1) {
+               socks[prev_i].next = socks[i].next;
+           }
+           
+           if (next_i != -1) {
+               socks[next_i].prev = socks[i].prev;
+           }
+           
+           if (*first_sock == i) {
+               *first_sock = socks[i].next;
+           }
+           
+           if (*last_sock == i) {
+               *last_sock = socks[i].prev;
+           }
+           
+           socks[i].next = -1;
+           socks[i].prev = -1;
+           
+           return 0;
+       }
+    }
+    return -1;
+}
+
+//setup the topology of the testing network
+static void store_topologies(SOCK fd) {
+    int i;
+    int src_mac_qual = MAC_ANY;
+    int dest_mac_qual = MAC_ANY;
+    uint_t dest;
+#ifndef VNET_SERVER
+    dest = (0 | 172 << 24 | 23 << 16 | 1 );
+    PrintDebug("VNET: store_topologies. NOT VNET_SERVER, dest = %x\n", dest);
+#else
+    dest = (0 | 172 << 24 | 23 << 16 | 2 );
+    PrintDebug("VNET: store_topologies. VNET_SERVER, dest = %x\n", dest);
+#endif
+
+    int type = UDP_TYPE;
+    int src = 0;
+    int src_type= LINK_ANY; //ANY_SRC_TYPE
+    int data_port = 22;
+    
+    //store link table
+    for (i = 0; i < MAX_LINKS; i++) {
+       if (g_links[i].use == 0) {
+           g_links[i].dest = (int)dest;
+           g_links[i].type = type;
+           g_links[i].link_sock = fd;
+           g_links[i].remote_port = data_port;
+           g_links[i].use = 1;
+           
+           if (g_first_link == -1) {
+               g_first_link = i;
+           }       
+
+           g_links[i].prev = g_last_link;
+           g_links[i].next = -1;
+           
+           if (g_last_link != -1) {
+               g_links[g_last_link].next = i;
+           }
+           
+           g_last_link = i;
+           
+           g_num_links++;
+           PrintDebug("VNET: store_topologies. new link: socket: %d, remote %x:[%d]\n", g_links[i].link_sock, (uint_t)g_links[i].dest, g_links[i].remote_port);
+       }
+    }
+    
+    
+    //store route table
+    
+    type = LINK_EDGE;
+    dest = 0;
+    
+    for (i = 0; i < MAX_ROUTES; i++) {
+       if (g_routes[i].use == 0) {
+           if ((src_mac_qual != MAC_ANY) && (src_mac_qual != MAC_NONE)) {
+               //                  memcpy(g_routes[i].src_mac, src_mac, 6);
+           } else {
+               memset(g_routes[i].src_mac, 0, 6);
+           }
+           
+           if ((dest_mac_qual != MAC_ANY) && (dest_mac_qual != MAC_NONE)) {
+               //                  memcpy(g_routes[i].dest_mac, dest_mac, 6);
+           } else {
+               memset(g_routes[i].dest_mac, 0, 6);
+           }
+           
+           g_routes[i].src_mac_qual = src_mac_qual;
+           g_routes[i].dest_mac_qual = dest_mac_qual;
+           g_routes[i].dest = (int)dest;
+           g_routes[i].type = type;
+           g_routes[i].src = src;
+           g_routes[i].src_type = src_type;
+           
+           g_routes[i].use = 1;
+           
+           if (g_first_route == -1) {
+               g_first_route = i;
+           }
+           
+           g_routes[i].prev = g_last_route;
+           g_routes[i].next = -1;
+           
+           if (g_last_route != -1) {
+               g_routes[g_last_route].next = i;
+           }
+           
+           g_last_route = i;
+           
+           g_num_routes++;
+           
+           PrintDebug("VNET: store_topologies. new route: src_mac: %s, dest_mac: %s, dest: %d\n", g_routes[i].src_mac, g_routes[i].dest_mac, dest);
+           
+       }
+    }
+}
+
+static int match_route(uint8_t * src_mac, uint8_t * dst_mac, int src_type, int src_index, int * matches) { 
+    int values[MAX_ROUTES];
+    int matched_routes[MAX_ROUTES];
+    
+    int num_matches = 0;
+    int i;
+    int max = 0;
+    int no = 0;
+    int exact_match = 0;
+    
+    FOREACH_ROUTE(i, g_routes, g_first_route) {
+       if ((g_routes[i].src_type != LINK_ANY) &&
+           ((g_routes[i].src_type != src_type) ||
+            ((g_routes[i].src != src_index) &&
+             (g_routes[i].src != -1)))) {
+           PrintDebug("Vnet: MatchRoute: Source route is on and does not match\n");
+           continue;
+       }
+       
+       if ( (g_routes[i].dest_mac_qual == MAC_ANY) &&
+            (g_routes[i].src_mac_qual == MAC_ANY) ) {      
+           matched_routes[num_matches] = i;
+           values[num_matches] = 3;
+           num_matches++;
+       }
+       
+       if (memcmp((void *)&g_routes[i].src_mac, (void *)src_mac, 6) == 0) {
+           if (g_routes[i].src_mac_qual !=  MAC_NOT) {
+               if (g_routes[i].dest_mac_qual == MAC_ANY) {
+                   matched_routes[num_matches] = i;
+                   values[num_matches] = 6;
+                   
+                   num_matches++;
+               } else if (memcmp((void *)&g_routes[i].dest_mac, (void *)dst_mac, 6) == 0) {
+                   if (g_routes[i].dest_mac_qual != MAC_NOT) {   
+                       matched_routes[num_matches] = i;
+                       values[num_matches] = 8;    
+                       exact_match = 1;
+                       num_matches++;
+                   }
+               }
+           }
+       }
+       
+       if (memcmp((void *)&g_routes[i].dest_mac, (void *)dst_mac, 6) == 0) {
+           if (g_routes[i].dest_mac_qual != MAC_NOT) {
+               if (g_routes[i].src_mac_qual == MAC_ANY) {
+                   matched_routes[num_matches] = i;
+                   values[num_matches] = 6;
+                   
+                   num_matches++;
+               } else if (memcmp((void *)&g_routes[i].src_mac, (void *)src_mac, 6) == 0) {
+                   if (g_routes[i].src_mac_qual != MAC_NOT) {
+                       if (exact_match == 0) {
+                           matched_routes[num_matches] = i;
+                           values[num_matches] = 8;
+                           num_matches++;
+                       }
+                   }
+               }
+           }
+       }
+       
+       if ((g_routes[i].dest_mac_qual == MAC_NOT) &&
+           (memcmp((void *)&g_routes[i].dest_mac, (void *)dst_mac, 6) != 0)) {
+           if (g_routes[i].src_mac_qual == MAC_ANY) {
+               matched_routes[num_matches] = i;
+               values[num_matches] = 5;                    
+               num_matches++;    
+           } else if (memcmp((void *)&g_routes[i].src_mac, (void *)src_mac, 6) == 0) {
+               if (g_routes[i].src_mac_qual != MAC_NOT) {      
+                   matched_routes[num_matches] = i;
+                   values[num_matches] = 7;                  
+                   num_matches++;
+               }
+           }
+       }
+       
+       if ((g_routes[i].src_mac_qual == MAC_NOT) &&
+           (memcmp((void *)&g_routes[i].src_mac, (void *)src_mac, 6) != 0)) {
+           if (g_routes[i].dest_mac_qual == MAC_ANY) {
+               matched_routes[num_matches] = i;
+               values[num_matches] = 5;            
+               num_matches++;
+           } else if (memcmp((void *)&g_routes[i].dest_mac, (void *)dst_mac, 6) == 0) {
+               if (g_routes[i].dest_mac_qual != MAC_NOT) { 
+                   matched_routes[num_matches] = i;
+                   values[num_matches] = 7;
+                   num_matches++;
+               }
+           }
+       }
+    }
+    //end FOREACH_ROUTE
+    
+    FOREACH_ROUTE(i, g_routes, g_first_route) {
+       if ((memcmp((void *)&g_routes[i].src_mac, (void *)src_mac, 6) == 0) &&
+           (g_routes[i].dest_mac_qual == MAC_NONE) &&
+           ((g_routes[i].src_type == LINK_ANY) ||
+            ((g_routes[i].src_type == src_type) &&
+             ((g_routes[i].src == src_index) ||
+              (g_routes[i].src == -1))))) {
+           matched_routes[num_matches] = i;
+           values[num_matches] = 4;
+           PrintDebug("Vnet: MatchRoute: We matched a default route (%d)\n", i);
+           num_matches++;
+       }
+    }
+    
+    //If many rules have been matched, we choose one which has the highest value rating
+    if (num_matches == 0) {
+       return 0;
+    }
+    
+    for (i = 0; i < num_matches; i++) {
+       if (values[i] > max) {
+           no = 0;
+           max = values[i];
+           matches[no] = matched_routes[i];
+           no++;
+       } else if (values[i] == max) {
+           matches[no] = matched_routes[i];
+           no++;
+       }
+    }
+    
+    return no;
+}
+
+
+static int process_udpdata() {
+    struct raw_ethernet_pkt * pt;
+
+    uint32_t dest = 0;
+    uint16_t remote_port = 0;
+    SOCK link_sock = g_udp_sockfd;
+    int length = sizeof(struct raw_ethernet_pkt) - (2 * sizeof(int));   //minus the "size" and "type" 
+
+    //run in a loop to get packets from outside network, adding them to the incoming packet queue
+    while (1) {
+       pt = (struct raw_ethernet_pkt *)V3_Malloc(sizeof(struct raw_ethernet_pkt));
+
+       if (pt == NULL){
+           PrintError("Vnet: process_udp: Malloc fails\n");
+           continue;
+       }
+       
+       PrintDebug("Vnet: route_thread: socket: [%d]. ready to receive from ip [%x], port [%d] or from VMs\n", link_sock, (uint_t)dest, remote_port);
+       pt->size = V3_RecvFrom_IP( link_sock, dest, remote_port, pt->data, length);
+       PrintDebug("Vnet: route_thread: socket: [%d] receive from ip [%x], port [%d]\n", link_sock, (uint_t)dest, remote_port);
+       
+       if (pt->size <= 0) {
+           PrintDebug("Vnet: process_udp: receiving packet from UDP fails\n");
+           V3_Free(pt);
+           return -1;
+       }
+       
+       PrintDebug("Vnet: process_udp: get packet\n");
+       print_packet(pt->data, pt->size);
+
+       
+       //V3_Yield();
+    }
+}
+
+
+
+static int indata_handler( )
+{
+      if (!use_tcp)
+          process_udpdata();     
+
+      return 0;   
+}
+
+static int start_recv_data()
+{
+       if (use_tcp){
+               
+       } else {
+               SOCK udp_data_socket;
+  
+               if ((udp_data_socket = V3_Create_UDP_Socket()) < 0){
+                       PrintError("VNET: Can't setup udp socket\n");
+                       return -1; 
+               }
+               PrintDebug("Vnet: vnet_setup_udp: get socket: %d\n", udp_data_socket);
+               g_udp_sockfd = udp_data_socket;
+
+               store_topologies(udp_data_socket);
+
+               if (V3_Bind_Socket(udp_data_socket, vnet_udp_port) < 0){ 
+                       PrintError("VNET: Can't bind socket\n");
+                       return -1;
+               }
+               PrintDebug("VNET: vnet_setup_udp: bind socket successful\n");
+       }
+
+       V3_CREATE_THREAD(&indata_handler, NULL, "VNET_DATA_HANDLER");
+       return 0;
+}
+
+
+static inline int if_write_pkt(struct vnet_if_device *iface, struct raw_ethernet_pkt * pkt) {
+    return iface->input((uchar_t *)pkt->data, pkt->size);
+}
+
+static int handle_one_pkt(struct raw_ethernet_pkt * pkt) {
+    int src_link_index = 0;    //the value of src_link_index of udp always is 0
+    int i;
+    char src_mac[6];
+    char dst_mac[6];
+
+    int matches[g_num_routes];
+    int num_matched_routes = 0;
+
+    struct HEADERS headers;
+  
+    // get the ethernet and ip headers from the packet
+    memcpy((void *)&headers, (void *)pkt->data, sizeof(headers));
+
+    int j;
+    for (j = 0;j < 6; j++) {
+       src_mac[j] = headers.ethernetsrc[j];
+       dst_mac[j] = headers.ethernetdest[j];
+    }
+
+
+#ifdef DEBUG
+    char dest_str[18];
+    char src_str[18];
+    
+    mac_to_string(src_mac, src_str);  
+    mac_to_string(dst_mac, dest_str);
+    
+    PrintDebug("Vnet: HandleDataOverLink. SRC(%s), DEST(%s)\n", src_str, dest_str);
+#endif
+    
+    char hash_key[hash_key_size];
+    make_hash_key(hash_key, src_mac, dst_mac, LINK_EDGE, src_link_index);//link_edge -> pt->type???
+    
+    num_matched_routes = look_into_cache((route_hashkey_t)hash_key, matches);
+    
+    if (num_matched_routes == -1) {  //no match
+        num_matched_routes = match_route(src_mac, dst_mac, pkt->type, src_link_index, matches);
+       
+       if (num_matched_routes > 0) {
+           add_route_to_cache(hash_key, num_matched_routes,matches);      
+       }
+    }
+    
+    PrintDebug("Vnet: HandleDataOverLink: Matches=%d\n", num_matched_routes);
+    
+    for (i = 0; i < num_matched_routes; i++) {
+        int route_index = -1;
+        int link_index = -1;
+        int dev_index = -1;
+       
+        route_index = matches[i];
+       
+        PrintDebug("Vnet: HandleDataOverLink: Forward packet from link according to Route entry %d\n", route_index);
+       
+        if (g_routes[route_index].type == LINK_EDGE) {
+            link_index = g_routes[route_index].dest;
+           
+            if(g_links[link_index].type == UDP_TYPE) {
+                int size;
+
+                               
+                 PrintDebug("===Vnet: HandleDataOverLink: Serializing UDP Packet to link_sock [%d], dest [%x], remote_port [%d], size [%d]\n", g_links[link_index].link_sock, (uint_t)g_links[link_index].dest,  g_links[link_index].remote_port, (int)pkt->size);
+               
+                if ((size = V3_SendTo_IP(g_links[link_index].link_sock,  g_links[link_index].dest,  g_links[link_index].remote_port, pkt->data, pkt->size)) != pkt->size)  {
+                    PrintError("Vnet: sending by UDP Exception, %x\n", size);
+                    return -1;
+                }
+               
+                PrintDebug("Vnet: HandleDataOverLink: Serializing UDP Packet to link_sock [%d], dest [%x], remote_port [%d], size [%d]\n", g_links[link_index].link_sock, (uint_t)g_links[link_index].dest,  g_links[link_index].remote_port, (int)pkt->size);
+               
+            } else if (g_links[link_index].type == TCP_TYPE) {
+               
+            }
+        } else if (g_routes[route_index].type == LINK_INTERFACE) {
+            dev_index = g_routes[route_index].dest;
+      
+            PrintDebug("Writing Packet to device=%s\n", g_devices[dev_index].device->name);
+
+            if (if_write_pkt(g_devices[dev_index].device, pkt) == -1) {
+               PrintDebug("Can't write output packet to link\n");
+                return -1;
+            }
+        } else {
+            PrintDebug("Vnet: Wrong Edge type\n");
+        }
+    }
+
+     return 0;
+}
+
+static int send_ethernet_pkt(char * buf, int length) {
+       struct raw_ethernet_pkt * pt;
+
+       pt = (struct raw_ethernet_pkt *)V3_Malloc(sizeof(struct raw_ethernet_pkt));
+       raw_ethernet_packet_init(pt, buf, length);  //====here we copy sending data once 
+       
+       PrintDebug("VNET: vm_send_pkt: transmitting packet: (size:%d)\n", (int)pt->size);
+       print_packet((char *)buf, length);
+       
+       v3_enqueue(g_inpkt_q, (addr_t)pt);
+       return 0;
+       
+}
+
+int v3_Send_pkt(uchar_t *buf, int length) {
+    PrintDebug("VNET: In V3_Send_pkt: pkt length %d\n", length);
+    
+    return send_ethernet_pkt((char *)buf, length);
+}
+
+static int add_device_to_table(struct vnet_if_device*device, int type) {
+    int i;
+    
+    for (i = 0; i < MAX_DEVICES; i++) {
+       if (g_devices[i].use == 0) {
+           g_devices[i].type = type;
+           g_devices[i].use = 1;
+           
+           if (g_first_device == -1) {
+               g_first_device = i;
+           }       
+
+           g_devices[i].prev = g_last_device;
+           g_devices[i].next = -1;
+           
+           if (g_last_device != -1) {
+               g_devices[g_last_device].next = i;
+           }
+
+           g_last_device = i;
+           g_num_devices++;
+           
+           return i;
+       }
+    }
+    
+    return -1;
+}
+
+static int search_device(char *device_name) {
+    int i;
+
+    for (i = 0; i < MAX_DEVICES; i++) {
+        if (g_devices[i].use == 1) {
+           if (!strcmp(device_name, g_devices[i].device->name)) {
+               return i;
+           }
+        }
+    }
+    
+    return -1;
+}
+
+static struct vnet_if_device * delete_device_from_table(int index) {
+    int next_i;
+    int prev_i;
+    struct vnet_if_device * device = NULL;
+
+    if (g_devices[index].use == 0) {
+       return NULL;
+    }
+
+    g_devices[index].use = 0;
+
+    prev_i = g_devices[index].prev;
+    next_i = g_devices[index].next;
+
+    if (prev_i != -1) {
+        g_devices[prev_i].next = g_devices[index].next;
+    }
+
+    if (next_i != -1) {
+        g_devices[next_i].prev = g_devices[index].prev;
+    }
+
+    if (g_first_device == index) {
+        g_first_device = g_devices[index].next;
+    }
+
+    if (g_last_device == index) {
+        g_last_device = g_devices[index].prev;
+    }
+
+    g_devices[index].next = -1;
+    g_devices[index].prev = -1;
+
+    device = g_devices[index].device;
+    g_devices[index].device = NULL;
+
+    g_num_devices--;
+
+    return device;
+}
+
+
+int vnet_register_device(char * dev_name, int (*netif_input)(uchar_t * pkt, uint_t size), void * data) {
+    struct vnet_if_device * dev;
+    
+    dev = (struct vnet_if_device *)V3_Malloc(sizeof(struct vnet_if_device));
+    
+    if (dev == NULL){
+       PrintError("VNET: Malloc fails\n");
+       return -1;
+    }
+    
+    strncpy(dev->name, dev_name, 50);
+    dev->input = netif_input;
+    dev->data = data;
+    
+    if (add_device_to_table(dev, GENERAL_NIC) == -1) {
+       return -1;
+    }
+    
+    return 0;
+}
+
+int vnet_unregister_device(char * dev_name) {
+    int i;
+
+    i = search_device(dev_name);
+    
+    if (i == -1) {
+        return -1;
+    }
+
+    struct vnet_if_device * device = delete_device_from_table(i);
+
+    if (device == NULL) {
+       return -1;
+    }
+
+    V3_Free(device);
+
+    return 0;
+}
+
+int v3_Register_pkt_event(int (*netif_input)(uchar_t * pkt, uint_t size)) {
+    return vnet_register_device("NE2000", netif_input, NULL);
+}
+
+int v3_vnet_pkt_process() {
+    struct raw_ethernet_pkt * pt;
+
+    PrintDebug("VNET: In vnet_check\n");
+       
+    while ((pt = (struct raw_ethernet_pkt *)v3_dequeue(g_inpkt_q)) != NULL) {
+       PrintDebug("VNET: In vnet_check: pt length %d, pt type %d\n", (int)pt->size, (int)pt->type);
+       v3_hexdump(pt->data, pt->size, NULL, 0);
+       
+       if(handle_one_pkt(pt)) {
+           PrintDebug("VNET: vnet_check: handle one packet!\n");  
+       }
+       
+       V3_Free(pt); //be careful here
+    }
+    
+    return 0;
+}
+
+static void init_link_table() {
+    int i;
+
+    for (i = 0; i < MAX_LINKS; i++) {
+        g_links[i].use = 0;
+        g_links[i].next = -1;
+        g_links[i].prev = -1;
+    }
+       
+    g_first_link = -1;
+    g_last_link = -1;
+    g_num_links = 0;
+}
+
+static void init_device_table() {
+    int i;
+
+    for (i = 0; i < MAX_DEVICES; i++) {
+        g_devices[i].use = 0;
+        g_devices[i].next = -1;
+        g_devices[i].prev = -1;
+    }
+       
+    g_first_device = -1;
+    g_last_device = -1;
+    g_num_devices = 0;
+}
+
+static void init_route_table() {       
+    int i;
+
+    for (i = 0; i < MAX_ROUTES; i++) {
+        g_routes[i].use = 0;
+        g_routes[i].next = -1;
+        g_routes[i].prev = -1;
+    }
+       
+     g_first_route = -1;
+     g_last_route = -1;
+     g_num_routes = 0;
+}
+
+static void init_tables() {
+    init_link_table();
+    init_device_table();
+    init_route_table();
+    init_route_cache();
+}
+
+static void init_pkt_queue()
+{
+    PrintDebug("VNET Init package receiving queue\n");
+
+    g_inpkt_q = v3_create_queue();
+    v3_init_queue(g_inpkt_q);
+}
+
+
+
+void v3_vnet_init() {  
+
+       PrintDebug("VNET Init: Vnet input queue successful.\n");
+
+       init_tables();
+
+       init_pkt_queue();
+       
+       //store_topologies(udp_data_socket);
+
+       start_recv_data();
+}
+
+