1 /* Copyright (c) 2016 Facebook 2 * 3 * This program is free software; you can redistribute it and/or 4 * modify it under the terms of version 2 of the GNU General Public 5 * License as published by the Free Software Foundation. 6 */ 7 #include <linux/bpf.h> 8 #include <linux/if_link.h> 9 #include <assert.h> 10 #include <errno.h> 11 #include <signal.h> 12 #include <stdio.h> 13 #include <stdlib.h> 14 #include <string.h> 15 #include <sys/resource.h> 16 #include <arpa/inet.h> 17 #include <netinet/ether.h> 18 #include <unistd.h> 19 #include <time.h> 20 #include "bpf_load.h" 21 #include <bpf/bpf.h> 22 #include "bpf_util.h" 23 #include "xdp_tx_iptunnel_common.h" 24 25 #define STATS_INTERVAL_S 2U 26 27 static int ifindex = -1; 28 static __u32 xdp_flags = 0; 29 30 static void int_exit(int sig) 31 { 32 if (ifindex > -1) 33 bpf_set_link_xdp_fd(ifindex, -1, xdp_flags); 34 exit(0); 35 } 36 37 /* simple per-protocol drop counter 38 */ 39 static void poll_stats(unsigned int kill_after_s) 40 { 41 const unsigned int nr_protos = 256; 42 unsigned int nr_cpus = bpf_num_possible_cpus(); 43 time_t started_at = time(NULL); 44 __u64 values[nr_cpus], prev[nr_protos][nr_cpus]; 45 __u32 proto; 46 int i; 47 48 memset(prev, 0, sizeof(prev)); 49 50 while (!kill_after_s || time(NULL) - started_at <= kill_after_s) { 51 sleep(STATS_INTERVAL_S); 52 53 for (proto = 0; proto < nr_protos; proto++) { 54 __u64 sum = 0; 55 56 assert(bpf_map_lookup_elem(map_fd[0], &proto, values) == 0); 57 for (i = 0; i < nr_cpus; i++) 58 sum += (values[i] - prev[proto][i]); 59 60 if (sum) 61 printf("proto %u: sum:%10llu pkts, rate:%10llu pkts/s\n", 62 proto, sum, sum / STATS_INTERVAL_S); 63 memcpy(prev[proto], values, sizeof(values)); 64 } 65 } 66 } 67 68 static void usage(const char *cmd) 69 { 70 printf("Start a XDP prog which encapsulates incoming packets\n" 71 "in an IPv4/v6 header and XDP_TX it out. The dst <VIP:PORT>\n" 72 "is used to select packets to encapsulate\n\n"); 73 printf("Usage: %s [...]\n", cmd); 74 printf(" -i <ifindex> Interface Index\n"); 75 printf(" -a <vip-service-address> IPv4 or IPv6\n"); 76 printf(" -p <vip-service-port> A port range (e.g. 433-444) is also allowed\n"); 77 printf(" -s <source-ip> Used in the IPTunnel header\n"); 78 printf(" -d <dest-ip> Used in the IPTunnel header\n"); 79 printf(" -m <dest-MAC> Used in sending the IP Tunneled pkt\n"); 80 printf(" -T <stop-after-X-seconds> Default: 0 (forever)\n"); 81 printf(" -P <IP-Protocol> Default is TCP\n"); 82 printf(" -S use skb-mode\n"); 83 printf(" -N enforce native mode\n"); 84 printf(" -h Display this help\n"); 85 } 86 87 static int parse_ipstr(const char *ipstr, unsigned int *addr) 88 { 89 if (inet_pton(AF_INET6, ipstr, addr) == 1) { 90 return AF_INET6; 91 } else if (inet_pton(AF_INET, ipstr, addr) == 1) { 92 addr[1] = addr[2] = addr[3] = 0; 93 return AF_INET; 94 } 95 96 fprintf(stderr, "%s is an invalid IP\n", ipstr); 97 return AF_UNSPEC; 98 } 99 100 static int parse_ports(const char *port_str, int *min_port, int *max_port) 101 { 102 char *end; 103 long tmp_min_port; 104 long tmp_max_port; 105 106 tmp_min_port = strtol(optarg, &end, 10); 107 if (tmp_min_port < 1 || tmp_min_port > 65535) { 108 fprintf(stderr, "Invalid port(s):%s\n", optarg); 109 return 1; 110 } 111 112 if (*end == '-') { 113 end++; 114 tmp_max_port = strtol(end, NULL, 10); 115 if (tmp_max_port < 1 || tmp_max_port > 65535) { 116 fprintf(stderr, "Invalid port(s):%s\n", optarg); 117 return 1; 118 } 119 } else { 120 tmp_max_port = tmp_min_port; 121 } 122 123 if (tmp_min_port > tmp_max_port) { 124 fprintf(stderr, "Invalid port(s):%s\n", optarg); 125 return 1; 126 } 127 128 if (tmp_max_port - tmp_min_port + 1 > MAX_IPTNL_ENTRIES) { 129 fprintf(stderr, "Port range (%s) is larger than %u\n", 130 port_str, MAX_IPTNL_ENTRIES); 131 return 1; 132 } 133 *min_port = tmp_min_port; 134 *max_port = tmp_max_port; 135 136 return 0; 137 } 138 139 int main(int argc, char **argv) 140 { 141 unsigned char opt_flags[256] = {}; 142 unsigned int kill_after_s = 0; 143 const char *optstr = "i:a:p:s:d:m:T:P:SNh"; 144 int min_port = 0, max_port = 0; 145 struct iptnl_info tnl = {}; 146 struct rlimit r = {RLIM_INFINITY, RLIM_INFINITY}; 147 struct vip vip = {}; 148 char filename[256]; 149 int opt; 150 int i; 151 152 tnl.family = AF_UNSPEC; 153 vip.protocol = IPPROTO_TCP; 154 155 for (i = 0; i < strlen(optstr); i++) 156 if (optstr[i] != 'h' && 'a' <= optstr[i] && optstr[i] <= 'z') 157 opt_flags[(unsigned char)optstr[i]] = 1; 158 159 while ((opt = getopt(argc, argv, optstr)) != -1) { 160 unsigned short family; 161 unsigned int *v6; 162 163 switch (opt) { 164 case 'i': 165 ifindex = atoi(optarg); 166 break; 167 case 'a': 168 vip.family = parse_ipstr(optarg, vip.daddr.v6); 169 if (vip.family == AF_UNSPEC) 170 return 1; 171 break; 172 case 'p': 173 if (parse_ports(optarg, &min_port, &max_port)) 174 return 1; 175 break; 176 case 'P': 177 vip.protocol = atoi(optarg); 178 break; 179 case 's': 180 case 'd': 181 if (opt == 's') 182 v6 = tnl.saddr.v6; 183 else 184 v6 = tnl.daddr.v6; 185 186 family = parse_ipstr(optarg, v6); 187 if (family == AF_UNSPEC) 188 return 1; 189 if (tnl.family == AF_UNSPEC) { 190 tnl.family = family; 191 } else if (tnl.family != family) { 192 fprintf(stderr, 193 "The IP version of the src and dst addresses used in the IP encapsulation does not match\n"); 194 return 1; 195 } 196 break; 197 case 'm': 198 if (!ether_aton_r(optarg, 199 (struct ether_addr *)tnl.dmac)) { 200 fprintf(stderr, "Invalid mac address:%s\n", 201 optarg); 202 return 1; 203 } 204 break; 205 case 'T': 206 kill_after_s = atoi(optarg); 207 break; 208 case 'S': 209 xdp_flags |= XDP_FLAGS_SKB_MODE; 210 break; 211 case 'N': 212 xdp_flags |= XDP_FLAGS_DRV_MODE; 213 break; 214 default: 215 usage(argv[0]); 216 return 1; 217 } 218 opt_flags[opt] = 0; 219 } 220 221 for (i = 0; i < strlen(optstr); i++) { 222 if (opt_flags[(unsigned int)optstr[i]]) { 223 fprintf(stderr, "Missing argument -%c\n", optstr[i]); 224 usage(argv[0]); 225 return 1; 226 } 227 } 228 229 if (setrlimit(RLIMIT_MEMLOCK, &r)) { 230 perror("setrlimit(RLIMIT_MEMLOCK, RLIM_INFINITY)"); 231 return 1; 232 } 233 234 snprintf(filename, sizeof(filename), "%s_kern.o", argv[0]); 235 236 if (load_bpf_file(filename)) { 237 printf("%s", bpf_log_buf); 238 return 1; 239 } 240 241 if (!prog_fd[0]) { 242 printf("load_bpf_file: %s\n", strerror(errno)); 243 return 1; 244 } 245 246 signal(SIGINT, int_exit); 247 signal(SIGTERM, int_exit); 248 249 while (min_port <= max_port) { 250 vip.dport = htons(min_port++); 251 if (bpf_map_update_elem(map_fd[1], &vip, &tnl, BPF_NOEXIST)) { 252 perror("bpf_map_update_elem(&vip2tnl)"); 253 return 1; 254 } 255 } 256 257 if (bpf_set_link_xdp_fd(ifindex, prog_fd[0], xdp_flags) < 0) { 258 printf("link set xdp fd failed\n"); 259 return 1; 260 } 261 262 poll_stats(kill_after_s); 263 264 bpf_set_link_xdp_fd(ifindex, -1, xdp_flags); 265 266 return 0; 267 } 268