X-Git-Url: http://git.onelab.eu/?a=blobdiff_plain;f=lib%2Fpackets.c;h=16f4fe63afe254f93e2678c4771013c9b7ac3566;hb=58cf43df9aa551b58970153d020728c1825f91c8;hp=03f92e365acd8840054dd1d4c5a2935d3eb11790;hpb=75a4ead16d571a4e30cb4b54c5456e30a20c3d78;p=sliver-openvswitch.git diff --git a/lib/packets.c b/lib/packets.c index 03f92e365..16f4fe63a 100644 --- a/lib/packets.c +++ b/lib/packets.c @@ -1,5 +1,5 @@ /* - * Copyright (c) 2009, 2010, 2011 Nicira Networks. + * Copyright (c) 2009, 2010, 2011, 2012 Nicira, Inc. * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. @@ -22,6 +22,9 @@ #include #include #include "byte-order.h" +#include "csum.h" +#include "flow.h" +#include "hmap.h" #include "dynamic-string.h" #include "ofpbuf.h" @@ -41,6 +44,83 @@ dpid_from_string(const char *s, uint64_t *dpidp) return *dpidp != 0; } +/* Returns true if 'ea' is a reserved address, that a bridge must never + * forward, false otherwise. + * + * If you change this function's behavior, please update corresponding + * documentation in vswitch.xml at the same time. */ +bool +eth_addr_is_reserved(const uint8_t ea[ETH_ADDR_LEN]) +{ + struct eth_addr_node { + struct hmap_node hmap_node; + uint64_t ea64; + }; + + static struct eth_addr_node nodes[] = { + /* STP, IEEE pause frames, and other reserved protocols. */ + { HMAP_NODE_NULL_INITIALIZER, 0x0108c2000000ULL }, + { HMAP_NODE_NULL_INITIALIZER, 0x0108c2000001ULL }, + { HMAP_NODE_NULL_INITIALIZER, 0x0108c2000002ULL }, + { HMAP_NODE_NULL_INITIALIZER, 0x0108c2000003ULL }, + { HMAP_NODE_NULL_INITIALIZER, 0x0108c2000004ULL }, + { HMAP_NODE_NULL_INITIALIZER, 0x0108c2000005ULL }, + { HMAP_NODE_NULL_INITIALIZER, 0x0108c2000006ULL }, + { HMAP_NODE_NULL_INITIALIZER, 0x0108c2000007ULL }, + { HMAP_NODE_NULL_INITIALIZER, 0x0108c2000008ULL }, + { HMAP_NODE_NULL_INITIALIZER, 0x0108c2000009ULL }, + { HMAP_NODE_NULL_INITIALIZER, 0x0108c200000aULL }, + { HMAP_NODE_NULL_INITIALIZER, 0x0108c200000bULL }, + { HMAP_NODE_NULL_INITIALIZER, 0x0108c200000cULL }, + { HMAP_NODE_NULL_INITIALIZER, 0x0108c200000dULL }, + { HMAP_NODE_NULL_INITIALIZER, 0x0108c200000eULL }, + { HMAP_NODE_NULL_INITIALIZER, 0x0108c200000fULL }, + + /* Extreme protocols. */ + { HMAP_NODE_NULL_INITIALIZER, 0x00e02b000000ULL }, /* EDP. */ + { HMAP_NODE_NULL_INITIALIZER, 0x00e02b000004ULL }, /* EAPS. */ + { HMAP_NODE_NULL_INITIALIZER, 0x00e02b000006ULL }, /* EAPS. */ + + /* Cisco protocols. */ + { HMAP_NODE_NULL_INITIALIZER, 0x01000c000000ULL }, /* ISL. */ + { HMAP_NODE_NULL_INITIALIZER, 0x01000cccccccULL }, /* PAgP, UDLD, CDP, + * DTP, VTP. */ + { HMAP_NODE_NULL_INITIALIZER, 0x01000ccccccdULL }, /* PVST+. */ + { HMAP_NODE_NULL_INITIALIZER, 0x01000ccdcdcdULL }, /* STP Uplink Fast, + * FlexLink. */ + + /* Cisco CFM. */ + { HMAP_NODE_NULL_INITIALIZER, 0x01000cccccc0ULL }, + { HMAP_NODE_NULL_INITIALIZER, 0x01000cccccc1ULL }, + { HMAP_NODE_NULL_INITIALIZER, 0x01000cccccc2ULL }, + { HMAP_NODE_NULL_INITIALIZER, 0x01000cccccc3ULL }, + { HMAP_NODE_NULL_INITIALIZER, 0x01000cccccc4ULL }, + { HMAP_NODE_NULL_INITIALIZER, 0x01000cccccc5ULL }, + { HMAP_NODE_NULL_INITIALIZER, 0x01000cccccc6ULL }, + { HMAP_NODE_NULL_INITIALIZER, 0x01000cccccc7ULL }, + }; + + static struct hmap addrs = HMAP_INITIALIZER(&addrs); + struct eth_addr_node *node; + uint64_t ea64; + + if (hmap_is_empty(&addrs)) { + for (node = nodes; node < &nodes[ARRAY_SIZE(nodes)]; node++) { + hmap_insert(&addrs, &node->hmap_node, + hash_2words(node->ea64, node->ea64 >> 32)); + } + } + + ea64 = eth_addr_to_uint64(ea); + HMAP_FOR_EACH_IN_BUCKET (node, hmap_node, hash_2words(ea64, ea64 >> 32), + &addrs) { + if (node->ea64 == ea64) { + return true; + } + } + return false; +} + bool eth_addr_from_string(const char *s, uint8_t ea[ETH_ADDR_LEN]) { @@ -53,31 +133,41 @@ eth_addr_from_string(const char *s, uint8_t ea[ETH_ADDR_LEN]) } } -/* Fills 'b' with an 802.2 SNAP packet with Ethernet source address 'eth_src', - * the Nicira OUI as SNAP organization and 'snap_type' as SNAP type. The text - * string in 'tag' is enclosed as the packet payload. - * +/* Fills 'b' with a Reverse ARP packet with Ethernet source address 'eth_src'. * This function is used by Open vSwitch to compose packets in cases where - * context is important but content doesn't (or shouldn't) matter. For this - * purpose, 'snap_type' should be a random number and 'tag' should be an - * English phrase that explains the purpose of the packet. (The English phrase - * gives hapless admins running Wireshark the opportunity to figure out what's - * going on.) */ + * context is important but content doesn't (or shouldn't) matter. + * + * The returned packet has enough headroom to insert an 802.1Q VLAN header if + * desired. */ void -compose_benign_packet(struct ofpbuf *b, const char *tag, uint16_t snap_type, - const uint8_t eth_src[ETH_ADDR_LEN]) +compose_rarp(struct ofpbuf *b, const uint8_t eth_src[ETH_ADDR_LEN]) { - size_t tag_size = strlen(tag) + 1; - char *payload; + struct eth_header *eth; + struct arp_eth_header *arp; - payload = snap_compose(b, eth_addr_broadcast, eth_src, 0x002320, snap_type, - tag_size + ETH_ADDR_LEN); - memcpy(payload, tag, tag_size); - memcpy(payload + tag_size, eth_src, ETH_ADDR_LEN); + ofpbuf_clear(b); + ofpbuf_prealloc_tailroom(b, ETH_HEADER_LEN + VLAN_HEADER_LEN + + ARP_ETH_HEADER_LEN); + ofpbuf_reserve(b, VLAN_HEADER_LEN); + eth = ofpbuf_put_uninit(b, sizeof *eth); + memcpy(eth->eth_dst, eth_addr_broadcast, ETH_ADDR_LEN); + memcpy(eth->eth_src, eth_src, ETH_ADDR_LEN); + eth->eth_type = htons(ETH_TYPE_RARP); + + arp = ofpbuf_put_uninit(b, sizeof *arp); + arp->ar_hrd = htons(ARP_HRD_ETHERNET); + arp->ar_pro = htons(ARP_PRO_IP); + arp->ar_hln = sizeof arp->ar_sha; + arp->ar_pln = sizeof arp->ar_spa; + arp->ar_op = htons(ARP_OP_RARP); + memcpy(arp->ar_sha, eth_src, ETH_ADDR_LEN); + arp->ar_spa = htonl(0); + memcpy(arp->ar_tha, eth_src, ETH_ADDR_LEN); + arp->ar_tpa = htonl(0); } /* Insert VLAN header according to given TCI. Packet passed must be Ethernet - * packet. + * packet. Ignores the CFI bit of 'tci' using 0 instead. * * Also sets 'packet->l2' to point to the new Ethernet header. */ void @@ -91,7 +181,7 @@ eth_push_vlan(struct ofpbuf *packet, ovs_be16 tci) memcpy(tmp.veth_dst, eh->eth_dst, ETH_ADDR_LEN); memcpy(tmp.veth_src, eh->eth_src, ETH_ADDR_LEN); tmp.veth_type = htons(ETH_TYPE_VLAN); - tmp.veth_tci = tci; + tmp.veth_tci = tci & htons(~VLAN_CFI); tmp.veth_next_type = eh->eth_type; veh = ofpbuf_push_uninit(packet, VLAN_HEADER_LEN); @@ -100,13 +190,82 @@ eth_push_vlan(struct ofpbuf *packet, ovs_be16 tci) packet->l2 = packet->data; } +/* Removes outermost VLAN header (if any is present) from 'packet'. + * + * 'packet->l2' must initially point to 'packet''s Ethernet header. */ +void +eth_pop_vlan(struct ofpbuf *packet) +{ + struct vlan_eth_header *veh = packet->l2; + if (packet->size >= sizeof *veh + && veh->veth_type == htons(ETH_TYPE_VLAN)) { + struct eth_header tmp; + + memcpy(tmp.eth_dst, veh->veth_dst, ETH_ADDR_LEN); + memcpy(tmp.eth_src, veh->veth_src, ETH_ADDR_LEN); + tmp.eth_type = veh->veth_next_type; + + ofpbuf_pull(packet, VLAN_HEADER_LEN); + packet->l2 = (char*)packet->l2 + VLAN_HEADER_LEN; + memcpy(packet->data, &tmp, sizeof tmp); + } +} + +/* Converts hex digits in 'hex' to an Ethernet packet in '*packetp'. The + * caller must free '*packetp'. On success, returns NULL. On failure, returns + * an error message and stores NULL in '*packetp'. */ +const char * +eth_from_hex(const char *hex, struct ofpbuf **packetp) +{ + struct ofpbuf *packet; + + packet = *packetp = ofpbuf_new(strlen(hex) / 2); + + if (ofpbuf_put_hex(packet, hex, NULL)[0] != '\0') { + ofpbuf_delete(packet); + *packetp = NULL; + return "Trailing garbage in packet data"; + } + + if (packet->size < ETH_HEADER_LEN) { + ofpbuf_delete(packet); + *packetp = NULL; + return "Packet data too short for Ethernet"; + } + + return NULL; +} + +void +eth_format_masked(const uint8_t eth[ETH_ADDR_LEN], + const uint8_t mask[ETH_ADDR_LEN], struct ds *s) +{ + ds_put_format(s, ETH_ADDR_FMT, ETH_ADDR_ARGS(eth)); + if (mask && !eth_mask_is_exact(mask)) { + ds_put_format(s, "/"ETH_ADDR_FMT, ETH_ADDR_ARGS(mask)); + } +} + +void +eth_addr_bitand(const uint8_t src[ETH_ADDR_LEN], + const uint8_t mask[ETH_ADDR_LEN], + uint8_t dst[ETH_ADDR_LEN]) +{ + int i; + + for (i = 0; i < ETH_ADDR_LEN; i++) { + dst[i] = src[i] & mask[i]; + } +} + /* Given the IP netmask 'netmask', returns the number of bits of the IP address - * that it specifies, that is, the number of 1-bits in 'netmask'. 'netmask' - * must be a CIDR netmask (see ip_is_cidr()). */ + * that it specifies, that is, the number of 1-bits in 'netmask'. + * + * If 'netmask' is not a CIDR netmask (see ip_is_cidr()), the return value will + * still be in the valid range but isn't otherwise meaningful. */ int ip_count_cidr_bits(ovs_be32 netmask) { - assert(ip_is_cidr(netmask)); return 32 - ctz(ntohl(netmask)); } @@ -204,7 +363,10 @@ ipv6_create_mask(int mask) /* Given the IPv6 netmask 'netmask', returns the number of bits of the IPv6 * address that it specifies, that is, the number of 1-bits in 'netmask'. - * 'netmask' must be a CIDR netmask (see ipv6_is_cidr()). */ + * 'netmask' must be a CIDR netmask (see ipv6_is_cidr()). + * + * If 'netmask' is not a CIDR netmask (see ipv6_is_cidr()), the return value + * will still be in the valid range but isn't otherwise meaningful. */ int ipv6_count_cidr_bits(const struct in6_addr *netmask) { @@ -212,8 +374,6 @@ ipv6_count_cidr_bits(const struct in6_addr *netmask) int count = 0; const uint8_t *netmaskp = &netmask->s6_addr[0]; - assert(ipv6_is_cidr(netmask)); - for (i=0; i<16; i++) { if (netmaskp[i] == 0xff) { count += 8; @@ -289,45 +449,156 @@ eth_compose(struct ofpbuf *b, const uint8_t eth_dst[ETH_ADDR_LEN], return data; } -/* Populates 'b' with an Ethernet LLC+SNAP packet headed with the given - * 'eth_dst', 'eth_src', 'snap_org', and 'snap_type'. A payload of 'size' - * bytes is allocated in 'b' and returned. This payload may be populated with - * appropriate information by the caller. +static void +packet_set_ipv4_addr(struct ofpbuf *packet, ovs_be32 *addr, ovs_be32 new_addr) +{ + struct ip_header *nh = packet->l3; + + if (nh->ip_proto == IPPROTO_TCP && packet->l7) { + struct tcp_header *th = packet->l4; + + th->tcp_csum = recalc_csum32(th->tcp_csum, *addr, new_addr); + } else if (nh->ip_proto == IPPROTO_UDP && packet->l7) { + struct udp_header *uh = packet->l4; + + if (uh->udp_csum) { + uh->udp_csum = recalc_csum32(uh->udp_csum, *addr, new_addr); + if (!uh->udp_csum) { + uh->udp_csum = htons(0xffff); + } + } + } + nh->ip_csum = recalc_csum32(nh->ip_csum, *addr, new_addr); + *addr = new_addr; +} + +/* Modifies the IPv4 header fields of 'packet' to be consistent with 'src', + * 'dst', 'tos', and 'ttl'. Updates 'packet''s L4 checksums as appropriate. + * 'packet' must contain a valid IPv4 packet with correctly populated l[347] + * markers. */ +void +packet_set_ipv4(struct ofpbuf *packet, ovs_be32 src, ovs_be32 dst, + uint8_t tos, uint8_t ttl) +{ + struct ip_header *nh = packet->l3; + + if (nh->ip_src != src) { + packet_set_ipv4_addr(packet, &nh->ip_src, src); + } + + if (nh->ip_dst != dst) { + packet_set_ipv4_addr(packet, &nh->ip_dst, dst); + } + + if (nh->ip_tos != tos) { + uint8_t *field = &nh->ip_tos; + + nh->ip_csum = recalc_csum16(nh->ip_csum, htons((uint16_t) *field), + htons((uint16_t) tos)); + *field = tos; + } + + if (nh->ip_ttl != ttl) { + uint8_t *field = &nh->ip_ttl; + + nh->ip_csum = recalc_csum16(nh->ip_csum, htons(*field << 8), + htons(ttl << 8)); + *field = ttl; + } +} + +static void +packet_set_port(ovs_be16 *port, ovs_be16 new_port, ovs_be16 *csum) +{ + if (*port != new_port) { + *csum = recalc_csum16(*csum, *port, new_port); + *port = new_port; + } +} + +/* Sets the TCP source and destination port ('src' and 'dst' respectively) of + * the TCP header contained in 'packet'. 'packet' must be a valid TCP packet + * with its l4 marker properly populated. */ +void +packet_set_tcp_port(struct ofpbuf *packet, ovs_be16 src, ovs_be16 dst) +{ + struct tcp_header *th = packet->l4; + + packet_set_port(&th->tcp_src, src, &th->tcp_csum); + packet_set_port(&th->tcp_dst, dst, &th->tcp_csum); +} + +/* Sets the UDP source and destination port ('src' and 'dst' respectively) of + * the UDP header contained in 'packet'. 'packet' must be a valid UDP packet + * with its l4 marker properly populated. */ +void +packet_set_udp_port(struct ofpbuf *packet, ovs_be16 src, ovs_be16 dst) +{ + struct udp_header *uh = packet->l4; + + if (uh->udp_csum) { + packet_set_port(&uh->udp_src, src, &uh->udp_csum); + packet_set_port(&uh->udp_dst, dst, &uh->udp_csum); + + if (!uh->udp_csum) { + uh->udp_csum = htons(0xffff); + } + } else { + uh->udp_src = src; + uh->udp_dst = dst; + } +} + +/* If 'packet' is a TCP packet, returns the TCP flags. Otherwise, returns 0. * - * The returned packet has enough headroom to insert an 802.1Q VLAN header if - * desired. */ -void * -snap_compose(struct ofpbuf *b, const uint8_t eth_dst[ETH_ADDR_LEN], - const uint8_t eth_src[ETH_ADDR_LEN], - unsigned int oui, uint16_t snap_type, size_t size) + * 'flow' must be the flow corresponding to 'packet' and 'packet''s header + * pointers must be properly initialized (e.g. with flow_extract()). */ +uint8_t +packet_get_tcp_flags(const struct ofpbuf *packet, const struct flow *flow) { - struct eth_header *eth; - struct llc_snap_header *llc_snap; - void *payload; + if ((flow->dl_type == htons(ETH_TYPE_IP) || + flow->dl_type == htons(ETH_TYPE_IPV6)) && + flow->nw_proto == IPPROTO_TCP && packet->l7) { + const struct tcp_header *tcp = packet->l4; + return TCP_FLAGS(tcp->tcp_ctl); + } else { + return 0; + } +} - /* Compose basic packet structure. (We need the payload size to stick into - * the 802.2 header.) */ - ofpbuf_clear(b); - ofpbuf_prealloc_tailroom(b, ETH_HEADER_LEN + VLAN_HEADER_LEN - + LLC_SNAP_HEADER_LEN + size); - ofpbuf_reserve(b, VLAN_HEADER_LEN); - eth = ofpbuf_put_zeros(b, ETH_HEADER_LEN); - llc_snap = ofpbuf_put_zeros(b, LLC_SNAP_HEADER_LEN); - payload = ofpbuf_put_uninit(b, size); +/* Appends a string representation of the TCP flags value 'tcp_flags' + * (e.g. obtained via packet_get_tcp_flags() or TCP_FLAGS) to 's', in the + * format used by tcpdump. */ +void +packet_format_tcp_flags(struct ds *s, uint8_t tcp_flags) +{ + if (!tcp_flags) { + ds_put_cstr(s, "none"); + return; + } - /* Compose 802.2 header. */ - memcpy(eth->eth_dst, eth_dst, ETH_ADDR_LEN); - memcpy(eth->eth_src, eth_src, ETH_ADDR_LEN); - eth->eth_type = htons(b->size - ETH_HEADER_LEN); - - /* Compose LLC, SNAP headers. */ - llc_snap->llc.llc_dsap = LLC_DSAP_SNAP; - llc_snap->llc.llc_ssap = LLC_SSAP_SNAP; - llc_snap->llc.llc_cntl = LLC_CNTL_SNAP; - llc_snap->snap.snap_org[0] = oui >> 16; - llc_snap->snap.snap_org[1] = oui >> 8; - llc_snap->snap.snap_org[2] = oui; - llc_snap->snap.snap_type = htons(snap_type); - - return payload; + if (tcp_flags & TCP_SYN) { + ds_put_char(s, 'S'); + } + if (tcp_flags & TCP_FIN) { + ds_put_char(s, 'F'); + } + if (tcp_flags & TCP_PSH) { + ds_put_char(s, 'P'); + } + if (tcp_flags & TCP_RST) { + ds_put_char(s, 'R'); + } + if (tcp_flags & TCP_URG) { + ds_put_char(s, 'U'); + } + if (tcp_flags & TCP_ACK) { + ds_put_char(s, '.'); + } + if (tcp_flags & 0x40) { + ds_put_cstr(s, "[40]"); + } + if (tcp_flags & 0x80) { + ds_put_cstr(s, "[80]"); + } }