/*
route.c -- routing
Copyright (C) 2000-2005 Ivo Timmermans,
- 2000-2009 Guus Sliepen <guus@tinc-vpn.org>
+ 2000-2010 Guus Sliepen <guus@tinc-vpn.org>
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
#include "system.h"
-#include "avl_tree.h"
+#include "splay_tree.h"
#include "connection.h"
+#include "control_common.h"
#include "ethernet.h"
#include "ipv4.h"
#include "ipv6.h"
#include "logger.h"
+#include "meta.h"
#include "net.h"
#include "protocol.h"
#include "route.h"
#include "utils.h"
rmode_t routing_mode = RMODE_ROUTER;
+fmode_t forwarding_mode = FMODE_INTERNAL;
+bool decrement_ttl = true;
+bool directonly = false;
bool priorityinheritance = false;
int macexpire = 600;
bool overwrite_mac = false;
+bool broadcast = true;
mac_t mymac = {{0xFE, 0xFD, 0, 0, 0, 0}};
+bool pcap = false;
/* Sizes of various headers */
static const size_t icmp6_size = sizeof(struct icmp6_hdr);
static const size_t ns_size = sizeof(struct nd_neighbor_solicit);
static const size_t opt_size = sizeof(struct nd_opt_hdr);
-#define max(a, b) ((a) > (b) ? (a) : (b))
+
+#ifndef MAX
+#define MAX(a, b) ((a) > (b) ? (a) : (b))
+#endif
+
+static struct event age_subnets_event;
/* RFC 1071 */
static bool ratelimit(int frequency) {
static time_t lasttime = 0;
static int count = 0;
+ time_t now = time(NULL);
if(lasttime == now) {
if(++count > frequency)
return true;
}
+static void clamp_mss(const node_t *source, const node_t *via, vpn_packet_t *packet) {
+ if(!source || !via || !(via->options & OPTION_CLAMP_MSS))
+ return;
+
+ uint16_t mtu = source->mtu;
+ if(via != myself && via->mtu < mtu)
+ mtu = via->mtu;
+
+ /* Find TCP header */
+ int start = 0;
+ uint16_t type = packet->data[12] << 8 | packet->data[13];
+
+ if(type == ETH_P_IP && packet->data[23] == 6)
+ start = 14 + (packet->data[14] & 0xf) * 4;
+ else if(type == ETH_P_IPV6 && packet->data[20] == 6)
+ start = 14 + 40;
+
+ if(!start || packet->len <= start + 20)
+ return;
+
+ /* Use data offset field to calculate length of options field */
+ int len = ((packet->data[start + 12] >> 4) - 5) * 4;
+
+ if(packet->len < start + 20 + len)
+ return;
+
+ /* Search for MSS option header */
+ for(int i = 0; i < len;) {
+ if(packet->data[start + 20 + i] == 0)
+ break;
+
+ if(packet->data[start + 20 + i] == 1) {
+ i++;
+ continue;
+ }
+
+ if(i > len - 2 || i > len - packet->data[start + 21 + i])
+ break;
+
+ if(packet->data[start + 20 + i] != 2) {
+ if(packet->data[start + 21 + i] < 2)
+ break;
+ i += packet->data[start + 21 + i];
+ continue;
+ }
+
+ if(packet->data[start + 21] != 4)
+ break;
+
+ /* Found it */
+ uint16_t oldmss = packet->data[start + 22 + i] << 8 | packet->data[start + 23 + i];
+ uint16_t newmss = mtu - start - 20;
+ uint16_t csum = packet->data[start + 16] << 8 | packet->data[start + 17];
+
+ if(oldmss <= newmss)
+ break;
+
+ ifdebug(TRAFFIC) logger(LOG_INFO, "Clamping MSS of packet from %s to %s to %d", source->name, via->name, newmss);
+
+ /* Update the MSS value and the checksum */
+ packet->data[start + 22 + i] = newmss >> 8;
+ packet->data[start + 23 + i] = newmss & 0xff;
+ csum ^= 0xffff;
+ csum -= oldmss;
+ csum += newmss;
+ csum ^= 0xffff;
+ packet->data[start + 16] = csum >> 8;
+ packet->data[start + 17] = csum & 0xff;
+ break;
+ }
+}
+
static void swap_mac_addresses(vpn_packet_t *packet) {
mac_t tmp;
memcpy(&tmp, &packet->data[0], sizeof tmp);
memcpy(&packet->data[6], &tmp, sizeof tmp);
}
+static void age_subnets(int fd, short events, void *data) {
+ subnet_t *s;
+ connection_t *c;
+ splay_node_t *node, *next, *node2;
+ bool left = false;
+ time_t now = time(NULL);
+
+ for(node = myself->subnet_tree->head; node; node = next) {
+ next = node->next;
+ s = node->data;
+ if(s->expires && s->expires < now) {
+ ifdebug(TRAFFIC) {
+ char netstr[MAXNETSTR];
+ if(net2str(netstr, sizeof netstr, s))
+ logger(LOG_INFO, "Subnet %s expired", netstr);
+ }
+
+ for(node2 = connection_tree->head; node2; node2 = node2->next) {
+ c = node2->data;
+ if(c->status.active)
+ send_del_subnet(c, s);
+ }
+
+ subnet_del(myself, s);
+ } else {
+ if(s->expires)
+ left = true;
+ }
+ }
+
+ if(left)
+ event_add(&age_subnets_event, &(struct timeval){10, 0});
+}
+
static void learn_mac(mac_t *address) {
subnet_t *subnet;
- avl_node_t *node;
+ splay_node_t *node;
connection_t *c;
subnet = lookup_subnet_mac(myself, address);
subnet = new_subnet();
subnet->type = SUBNET_MAC;
- subnet->expires = now + macexpire;
+ subnet->expires = time(NULL) + macexpire;
subnet->net.mac.address = *address;
subnet->weight = 10;
subnet_add(myself, subnet);
+ subnet_update(myself, subnet, true);
/* And tell all other tinc daemons it's our MAC */
if(c->status.active)
send_add_subnet(c, subnet);
}
- }
-
- if(subnet->expires)
- subnet->expires = now + macexpire;
-}
-
-void age_subnets(void) {
- subnet_t *s;
- connection_t *c;
- avl_node_t *node, *next, *node2;
- for(node = myself->subnet_tree->head; node; node = next) {
- next = node->next;
- s = node->data;
- if(s->expires && s->expires < now) {
- ifdebug(TRAFFIC) {
- char netstr[MAXNETSTR];
- if(net2str(netstr, sizeof netstr, s))
- logger(LOG_INFO, "Subnet %s expired", netstr);
- }
-
- for(node2 = connection_tree->head; node2; node2 = node2->next) {
- c = node2->data;
- if(c->status.active)
- send_del_subnet(c, s);
- }
-
- subnet_del(myself, s);
- }
+ if(!timeout_initialized(&age_subnets_event))
+ timeout_set(&age_subnets_event, age_subnets, NULL);
+ event_add(&age_subnets_event, &(struct timeval){10, 0});
+ } else {
+ if(subnet->expires)
+ subnet->expires = time(NULL) + macexpire;
}
}
todo = ntohs(ip.ip_len) - ip_size;
if(ether_size + ip_size + todo != packet->len) {
- ifdebug(TRAFFIC) logger(LOG_WARNING, "Length of packet (%d) doesn't match length in IPv4 header (%zd)", packet->len, ether_size + ip_size + todo);
+ ifdebug(TRAFFIC) logger(LOG_WARNING, "Length of packet (%d) doesn't match length in IPv4 header (%d)", packet->len, (int)(ether_size + ip_size + todo));
return;
}
}
if(!subnet->owner->status.reachable)
- route_ipv4_unreachable(source, packet, ICMP_DEST_UNREACH, ICMP_NET_UNREACH);
+ return route_ipv4_unreachable(source, packet, ICMP_DEST_UNREACH, ICMP_NET_UNREACH);
+
+ if(forwarding_mode == FMODE_OFF && source != myself && subnet->owner != myself)
+ return route_ipv4_unreachable(source, packet, ICMP_DEST_UNREACH, ICMP_NET_ANO);
if(priorityinheritance)
packet->priority = packet->data[15];
via = (subnet->owner->via == myself) ? subnet->owner->nexthop : subnet->owner->via;
- if(via && packet->len > max(via->mtu, 590) && via != myself) {
+ if(directonly && subnet->owner != via)
+ return route_ipv4_unreachable(source, packet, ICMP_DEST_UNREACH, ICMP_NET_ANO);
+
+ if(via && packet->len > MAX(via->mtu, 590) && via != myself) {
ifdebug(TRAFFIC) logger(LOG_INFO, "Packet for %s (%s) length %d larger than MTU %d", subnet->owner->name, subnet->owner->hostname, packet->len, via->mtu);
if(packet->data[20] & 0x40) {
- packet->len = max(via->mtu, 590);
+ packet->len = MAX(via->mtu, 590);
route_ipv4_unreachable(source, packet, ICMP_DEST_UNREACH, ICMP_FRAG_NEEDED);
} else {
fragment_ipv4_packet(via, packet);
return;
}
+ clamp_mss(source, via, packet);
+
send_packet(subnet->owner, packet);
}
if(!checklength(source, packet, ether_size + ip_size))
return;
- if(((packet->data[30] & 0xf0) == 0xe0) || (
+ if(broadcast && (((packet->data[30] & 0xf0) == 0xe0) || (
packet->data[30] == 255 &&
packet->data[31] == 255 &&
packet->data[32] == 255 &&
- packet->data[33] == 255))
+ packet->data[33] == 255)))
broadcast_packet(source, packet);
else
route_ipv4_unicast(source, packet);
/* Generate checksum */
- checksum = inet_checksum(&pseudo, sizeof(pseudo), ~0);
+ checksum = inet_checksum(&pseudo, sizeof pseudo, ~0);
checksum = inet_checksum(&icmp6, icmp6_size, checksum);
checksum = inet_checksum(packet->data + ether_size + ip6_size + icmp6_size, ntohl(pseudo.length) - icmp6_size, checksum);
}
if(!subnet->owner->status.reachable)
- route_ipv6_unreachable(source, packet, ICMP6_DST_UNREACH, ICMP6_DST_UNREACH_NOROUTE);
+ return route_ipv6_unreachable(source, packet, ICMP6_DST_UNREACH, ICMP6_DST_UNREACH_NOROUTE);
+
+ if(forwarding_mode == FMODE_OFF && source != myself && subnet->owner != myself)
+ return route_ipv6_unreachable(source, packet, ICMP6_DST_UNREACH, ICMP6_DST_UNREACH_ADMIN);
via = (subnet->owner->via == myself) ? subnet->owner->nexthop : subnet->owner->via;
- if(via && packet->len > max(via->mtu, 1294) && via != myself) {
+ if(directonly && subnet->owner != via)
+ return route_ipv6_unreachable(source, packet, ICMP6_DST_UNREACH, ICMP6_DST_UNREACH_ADMIN);
+
+ if(via && packet->len > MAX(via->mtu, 1294) && via != myself) {
ifdebug(TRAFFIC) logger(LOG_INFO, "Packet for %s (%s) length %d larger than MTU %d", subnet->owner->name, subnet->owner->hostname, packet->len, via->mtu);
- packet->len = max(via->mtu, 1294);
+ packet->len = MAX(via->mtu, 1294);
route_ipv6_unreachable(source, packet, ICMP6_PACKET_TOO_BIG, 0);
return;
}
+ clamp_mss(source, via, packet);
+
send_packet(subnet->owner, packet);
}
/* Generate checksum */
- checksum = inet_checksum(&pseudo, sizeof(pseudo), ~0);
+ checksum = inet_checksum(&pseudo, sizeof pseudo, ~0);
checksum = inet_checksum(&ns, ns_size, checksum);
if(has_opt) {
checksum = inet_checksum(&opt, opt_size, checksum);
/* Generate checksum */
- checksum = inet_checksum(&pseudo, sizeof(pseudo), ~0);
+ checksum = inet_checksum(&pseudo, sizeof pseudo, ~0);
checksum = inet_checksum(&ns, ns_size, checksum);
if(has_opt) {
checksum = inet_checksum(&opt, opt_size, checksum);
return;
}
- if(packet->data[38] == 255)
+ if(broadcast && packet->data[38] == 255)
broadcast_packet(source, packet);
else
route_ipv6_unicast(source, packet);
/* Check if this is a valid ARP request */
if(ntohs(arp.arp_hrd) != ARPHRD_ETHER || ntohs(arp.arp_pro) != ETH_P_IP ||
- arp.arp_hln != ETH_ALEN || arp.arp_pln != sizeof(addr) || ntohs(arp.arp_op) != ARPOP_REQUEST) {
+ arp.arp_hln != ETH_ALEN || arp.arp_pln != sizeof addr || ntohs(arp.arp_op) != ARPOP_REQUEST) {
ifdebug(TRAFFIC) logger(LOG_WARNING, "Cannot route packet: received unknown type ARP request");
return;
}
memcpy(packet->data, packet->data + ETH_ALEN, ETH_ALEN); /* copy destination address */
packet->data[ETH_ALEN * 2 - 1] ^= 0xFF; /* mangle source address so it looks like it's not from us */
- memcpy(&addr, arp.arp_tpa, sizeof(addr)); /* save protocol addr */
- memcpy(arp.arp_tpa, arp.arp_spa, sizeof(addr)); /* swap destination and source protocol address */
- memcpy(arp.arp_spa, &addr, sizeof(addr)); /* ... */
+ memcpy(&addr, arp.arp_tpa, sizeof addr); /* save protocol addr */
+ memcpy(arp.arp_tpa, arp.arp_spa, sizeof addr); /* swap destination and source protocol address */
+ memcpy(arp.arp_spa, &addr, sizeof addr); /* ... */
memcpy(arp.arp_tha, arp.arp_sha, ETH_ALEN); /* set target hard/proto addr */
memcpy(arp.arp_sha, packet->data + ETH_ALEN, ETH_ALEN); /* add fake source hard addr */
subnet = lookup_subnet_mac(NULL, &dest);
if(!subnet) {
- broadcast_packet(source, packet);
+ if(broadcast)
+ broadcast_packet(source, packet);
return;
}
return;
}
+ if(forwarding_mode == FMODE_OFF && source != myself && subnet->owner != myself)
+ return;
+
// Handle packets larger than PMTU
node_t *via = (subnet->owner->via == myself) ? subnet->owner->nexthop : subnet->owner->via;
+
+ if(directonly && subnet->owner != via)
+ return;
if(via && packet->len > via->mtu && via != myself) {
ifdebug(TRAFFIC) logger(LOG_INFO, "Packet for %s (%s) length %d larger than MTU %d", subnet->owner->name, subnet->owner->hostname, packet->len, via->mtu);
}
}
+ clamp_mss(source, via, packet);
+
send_packet(subnet->owner, packet);
}
+static void send_pcap(vpn_packet_t *packet) {
+ pcap = false;
+ for(splay_node_t *node = connection_tree->head; node; node = node->next) {
+ connection_t *c = node->data;
+ if(!c->status.pcap)
+ continue;
+ else
+ pcap = true;
+ if(send_request(c, "%d %d %d", CONTROL, REQ_PCAP, packet->len))
+ send_meta(c, (char *)packet->data, packet->len);
+ }
+}
+
+static bool do_decrement_ttl(node_t *source, vpn_packet_t *packet) {
+ uint16_t type = packet->data[12] << 8 | packet->data[13];
+
+ switch (type) {
+ case ETH_P_IP:
+ if(!checklength(source, packet, 14 + 32))
+ return false;
+
+ if(packet->data[22] < 1) {
+ route_ipv4_unreachable(source, packet, ICMP_TIME_EXCEEDED, ICMP_EXC_TTL);
+ return false;
+ }
+
+ uint16_t old = packet->data[22] << 8 | packet->data[23];
+ packet->data[22]--;
+ uint16_t new = packet->data[22] << 8 | packet->data[23];
+
+ uint32_t checksum = packet->data[24] << 8 | packet->data[25];
+ checksum += old + (~new & 0xFFFF);
+ while(checksum >> 16)
+ checksum = (checksum & 0xFFFF) + (checksum >> 16);
+ packet->data[24] = checksum >> 8;
+ packet->data[25] = checksum & 0xff;
+
+ return true;
+
+ case ETH_P_IPV6:
+ if(!checklength(source, packet, 14 + 40))
+ return false;
+
+ if(packet->data[21] < 1) {
+ route_ipv6_unreachable(source, packet, ICMP6_TIME_EXCEEDED, ICMP6_TIME_EXCEED_TRANSIT);
+ return false;
+ }
+
+ packet->data[21]--;
+
+ return true;
+
+ default:
+ return true;
+ }
+}
+
void route(node_t *source, vpn_packet_t *packet) {
+ if(pcap)
+ send_pcap(packet);
+
+ if(forwarding_mode == FMODE_KERNEL && source != myself) {
+ send_packet(myself, packet);
+ return;
+ }
+
if(!checklength(source, packet, ether_size))
return;
+ if(decrement_ttl && source != myself)
+ if(!do_decrement_ttl(source, packet))
+ return;
+
switch (routing_mode) {
case RMODE_ROUTER:
{
projects / meshlink / blobdiff