X-Git-Url: http://git.cascardo.eti.br/?a=blobdiff_plain;f=lib%2Fpackets.c;h=d82341d71829e6b7b5ae52bb6fa846e1367b026c;hb=8f79bb4d3999d993424e9578342b4130d10a556c;hp=d15c402fbc671d89f121093b2a9744a664a6588a;hpb=bb622f82d709e7468a5bc9eb4f038ce9343f74f6;p=cascardo%2Fovs.git diff --git a/lib/packets.c b/lib/packets.c index d15c402fb..d82341d71 100644 --- a/lib/packets.c +++ b/lib/packets.c @@ -1,5 +1,5 @@ /* - * Copyright (c) 2009, 2010, 2011, 2012, 2013 Nicira, Inc. + * Copyright (c) 2009, 2010, 2011, 2012, 2013, 2014, 2015 Nicira, Inc. * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. @@ -20,6 +20,7 @@ #include #include #include +#include #include #include "byte-order.h" #include "csum.h" @@ -27,11 +28,25 @@ #include "flow.h" #include "hmap.h" #include "dynamic-string.h" -#include "ofpbuf.h" #include "ovs-thread.h" +#include "odp-util.h" +#include "dp-packet.h" #include "unaligned.h" const struct in6_addr in6addr_exact = IN6ADDR_EXACT_INIT; +const struct in6_addr in6addr_all_hosts = IN6ADDR_ALL_HOSTS_INIT; + +struct in6_addr +flow_tnl_dst(const struct flow_tnl *tnl) +{ + return tnl->ip_dst ? in6_addr_mapped_ipv4(tnl->ip_dst) : tnl->ipv6_dst; +} + +struct in6_addr +flow_tnl_src(const struct flow_tnl *tnl) +{ + return tnl->ip_src ? in6_addr_mapped_ipv4(tnl->ip_src) : tnl->ipv6_src; +} /* Parses 's' as a 16-digit hexadecimal number representing a datapath ID. On * success stores the dpid into '*dpidp' and returns true, on failure stores 0 @@ -53,7 +68,7 @@ dpid_from_string(const char *s, uint64_t *dpidp) * 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]) +eth_addr_is_reserved(const struct eth_addr ea) { struct eth_addr_node { struct hmap_node hmap_node; @@ -111,15 +126,13 @@ eth_addr_is_reserved(const uint8_t ea[ETH_ADDR_LEN]) if (ovsthread_once_start(&once)) { hmap_init(&addrs); for (node = nodes; node < &nodes[ARRAY_SIZE(nodes)]; node++) { - hmap_insert(&addrs, &node->hmap_node, - hash_2words(node->ea64, node->ea64 >> 32)); + hmap_insert(&addrs, &node->hmap_node, hash_uint64(node->ea64)); } ovsthread_once_done(&once); } ea64 = eth_addr_to_uint64(ea); - HMAP_FOR_EACH_IN_BUCKET (node, hmap_node, hash_2words(ea64, ea64 >> 32), - &addrs) { + HMAP_FOR_EACH_IN_BUCKET (node, hmap_node, hash_uint64(ea64), &addrs) { if (node->ea64 == ea64) { return true; } @@ -128,13 +141,12 @@ eth_addr_is_reserved(const uint8_t ea[ETH_ADDR_LEN]) } bool -eth_addr_from_string(const char *s, uint8_t ea[ETH_ADDR_LEN]) +eth_addr_from_string(const char *s, struct eth_addr *ea) { - if (sscanf(s, ETH_ADDR_SCAN_FMT, ETH_ADDR_SCAN_ARGS(ea)) - == ETH_ADDR_SCAN_COUNT) { + if (ovs_scan(s, ETH_ADDR_SCAN_FMT, ETH_ADDR_SCAN_ARGS(*ea))) { return true; } else { - memset(ea, 0, ETH_ADDR_LEN); + *ea = eth_addr_zero; return false; } } @@ -146,123 +158,93 @@ eth_addr_from_string(const char *s, uint8_t ea[ETH_ADDR_LEN]) * The returned packet has enough headroom to insert an 802.1Q VLAN header if * desired. */ void -compose_rarp(struct ofpbuf *b, const uint8_t eth_src[ETH_ADDR_LEN]) +compose_rarp(struct dp_packet *b, const struct eth_addr eth_src) { struct eth_header *eth; struct arp_eth_header *arp; - ofpbuf_clear(b); - ofpbuf_prealloc_tailroom(b, 2 + ETH_HEADER_LEN + VLAN_HEADER_LEN + dp_packet_clear(b); + dp_packet_prealloc_tailroom(b, 2 + ETH_HEADER_LEN + VLAN_HEADER_LEN + ARP_ETH_HEADER_LEN); - ofpbuf_reserve(b, 2 + 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); + dp_packet_reserve(b, 2 + VLAN_HEADER_LEN); + eth = dp_packet_put_uninit(b, sizeof *eth); + eth->eth_dst = eth_addr_broadcast; + eth->eth_src = eth_src; eth->eth_type = htons(ETH_TYPE_RARP); - arp = ofpbuf_put_uninit(b, sizeof *arp); + arp = dp_packet_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_sha = eth_src; put_16aligned_be32(&arp->ar_spa, htonl(0)); - memcpy(arp->ar_tha, eth_src, ETH_ADDR_LEN); + arp->ar_tha = eth_src; put_16aligned_be32(&arp->ar_tpa, htonl(0)); + + dp_packet_reset_offsets(b); + dp_packet_set_l3(b, arp); } /* Insert VLAN header according to given TCI. Packet passed must be Ethernet * packet. Ignores the CFI bit of 'tci' using 0 instead. * - * Also sets 'packet->l2' to point to the new Ethernet header. */ + * Also adjusts the layer offsets accordingly. */ void -eth_push_vlan(struct ofpbuf *packet, ovs_be16 tci) +eth_push_vlan(struct dp_packet *packet, ovs_be16 tpid, ovs_be16 tci) { - struct eth_header *eh = packet->data; struct vlan_eth_header *veh; /* Insert new 802.1Q header. */ - struct vlan_eth_header tmp; - 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 & htons(~VLAN_CFI); - tmp.veth_next_type = eh->eth_type; - - veh = ofpbuf_push_uninit(packet, VLAN_HEADER_LEN); - memcpy(veh, &tmp, sizeof tmp); - - packet->l2 = packet->data; + veh = dp_packet_resize_l2(packet, VLAN_HEADER_LEN); + memmove(veh, (char *)veh + VLAN_HEADER_LEN, 2 * ETH_ADDR_LEN); + veh->veth_type = tpid; + veh->veth_tci = tci & htons(~VLAN_CFI); } /* Removes outermost VLAN header (if any is present) from 'packet'. * - * 'packet->l2_5' should initially point to 'packet''s outer-most MPLS header - * or may be NULL if there are no MPLS headers. */ + * 'packet->l2_5' should initially point to 'packet''s outer-most VLAN header + * or may be NULL if there are no VLAN headers. */ void -eth_pop_vlan(struct ofpbuf *packet) +eth_pop_vlan(struct dp_packet *packet) { - struct vlan_eth_header *veh = packet->l2; - if (packet->size >= sizeof *veh - && veh->veth_type == htons(ETH_TYPE_VLAN)) { - struct eth_header tmp; + struct vlan_eth_header *veh = dp_packet_l2(packet); - 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; + if (veh && dp_packet_size(packet) >= sizeof *veh + && eth_type_vlan(veh->veth_type)) { - ofpbuf_pull(packet, VLAN_HEADER_LEN); - packet->l2 = (char*)packet->l2 + VLAN_HEADER_LEN; - memcpy(packet->data, &tmp, sizeof tmp); + memmove((char *)veh + VLAN_HEADER_LEN, veh, 2 * ETH_ADDR_LEN); + dp_packet_resize_l2(packet, -VLAN_HEADER_LEN); } } -/* Return depth of mpls stack. - * - * 'packet->l2_5' should initially point to 'packet''s outer-most MPLS header - * or may be NULL if there are no MPLS headers. */ -uint16_t -eth_mpls_depth(const struct ofpbuf *packet) +/* Set ethertype of the packet. */ +static void +set_ethertype(struct dp_packet *packet, ovs_be16 eth_type) { - struct mpls_hdr *mh = packet->l2_5; - uint16_t depth; + struct eth_header *eh = dp_packet_l2(packet); - if (!mh) { - return 0; - } - - depth = 0; - while (packet->size >= ((char *)mh - (char *)packet->data) + sizeof *mh) { - depth++; - if (mh->mpls_lse & htonl(MPLS_BOS_MASK)) { - break; - } - mh++; + if (!eh) { + return; } - return depth; -} - -/* Set ethertype of the packet. */ -void -set_ethertype(struct ofpbuf *packet, ovs_be16 eth_type) -{ - struct eth_header *eh = packet->data; - - if (eh->eth_type == htons(ETH_TYPE_VLAN)) { + if (eth_type_vlan(eh->eth_type)) { ovs_be16 *p; + char *l2_5 = dp_packet_l2_5(packet); + p = ALIGNED_CAST(ovs_be16 *, - (char *)(packet->l2_5 ? packet->l2_5 : packet->l3) - 2); + (l2_5 ? l2_5 : (char *)dp_packet_l3(packet)) - 2); *p = eth_type; } else { eh->eth_type = eth_type; } } -static bool is_mpls(struct ofpbuf *packet) +static bool is_mpls(struct dp_packet *packet) { - return packet->l2_5 != NULL; + return packet->l2_5_ofs != UINT16_MAX; } /* Set time to live (TTL) of an MPLS label stack entry (LSE). */ @@ -311,60 +293,44 @@ set_mpls_lse_values(uint8_t ttl, uint8_t tc, uint8_t bos, ovs_be32 label) return lse; } -/* Push an new MPLS stack entry onto the MPLS stack and adjust 'packet->l2' and - * 'packet->l2_5' accordingly. The new entry will be the outermost entry on - * the stack. - * - * Previous to calling this function, 'packet->l2_5' must be set; if the MPLS - * label to be pushed will be the first label in 'packet', then it should be - * the same as 'packet->l3'. */ -static void -push_mpls_lse(struct ofpbuf *packet, struct mpls_hdr *mh) -{ - char * header; - size_t len; - header = ofpbuf_push_uninit(packet, MPLS_HLEN); - len = (char *)packet->l2_5 - (char *)packet->l2; - memmove(header, packet->l2, len); - memcpy(header + len, mh, sizeof *mh); - packet->l2 = (char*)packet->l2 - MPLS_HLEN; - packet->l2_5 = (char*)packet->l2_5 - MPLS_HLEN; -} - /* Set MPLS label stack entry to outermost MPLS header.*/ void -set_mpls_lse(struct ofpbuf *packet, ovs_be32 mpls_lse) +set_mpls_lse(struct dp_packet *packet, ovs_be32 mpls_lse) { - struct mpls_hdr *mh = packet->l2_5; - /* Packet type should be MPLS to set label stack entry. */ if (is_mpls(packet)) { + struct mpls_hdr *mh = dp_packet_l2_5(packet); + /* Update mpls label stack entry. */ - mh->mpls_lse = mpls_lse; + put_16aligned_be32(&mh->mpls_lse, mpls_lse); } } -/* Push MPLS label stack entry 'lse' onto 'packet' as the the outermost MPLS +/* Push MPLS label stack entry 'lse' onto 'packet' as the outermost MPLS * header. If 'packet' does not already have any MPLS labels, then its * Ethertype is changed to 'ethtype' (which must be an MPLS Ethertype). */ void -push_mpls(struct ofpbuf *packet, ovs_be16 ethtype, ovs_be32 lse) +push_mpls(struct dp_packet *packet, ovs_be16 ethtype, ovs_be32 lse) { - struct mpls_hdr mh; + char * header; + size_t len; if (!eth_type_mpls(ethtype)) { return; } if (!is_mpls(packet)) { - /* Set ethtype and MPLS label stack entry. */ - set_ethertype(packet, ethtype); - packet->l2_5 = packet->l3; + /* Set MPLS label stack offset. */ + packet->l2_5_ofs = packet->l3_ofs; } + set_ethertype(packet, ethtype); + /* Push new MPLS shim header onto packet. */ - mh.mpls_lse = lse; - push_mpls_lse(packet, &mh); + len = packet->l2_5_ofs; + header = dp_packet_resize_l2_5(packet, MPLS_HLEN); + memmove(header, header + MPLS_HLEN, len); + memcpy(header + len, &lse, sizeof lse); } /* If 'packet' is an MPLS packet, removes its outermost MPLS label stack entry. @@ -372,25 +338,19 @@ push_mpls(struct ofpbuf *packet, ovs_be16 ethtype, ovs_be32 lse) * Ethertype to 'ethtype' (which ordinarily should not be an MPLS * Ethertype). */ void -pop_mpls(struct ofpbuf *packet, ovs_be16 ethtype) +pop_mpls(struct dp_packet *packet, ovs_be16 ethtype) { - struct mpls_hdr *mh = NULL; - if (is_mpls(packet)) { - size_t len; - mh = packet->l2_5; - len = (char*)packet->l2_5 - (char*)packet->l2; + struct mpls_hdr *mh = dp_packet_l2_5(packet); + size_t len = packet->l2_5_ofs; + set_ethertype(packet, ethtype); - if (mh->mpls_lse & htonl(MPLS_BOS_MASK)) { - packet->l2_5 = NULL; - } else { - packet->l2_5 = (char*)packet->l2_5 + MPLS_HLEN; + if (get_16aligned_be32(&mh->mpls_lse) & htonl(MPLS_BOS_MASK)) { + dp_packet_set_l2_5(packet, NULL); } /* Shift the l2 header forward. */ - memmove((char*)packet->data + MPLS_HLEN, packet->data, len); - packet->size -= MPLS_HLEN; - packet->data = (char*)packet->data + MPLS_HLEN; - packet->l2 = (char*)packet->l2 + MPLS_HLEN; + memmove((char*)dp_packet_data(packet) + MPLS_HLEN, dp_packet_data(packet), len); + dp_packet_resize_l2_5(packet, -MPLS_HLEN); } } @@ -400,21 +360,21 @@ pop_mpls(struct ofpbuf *packet, ovs_be16 ethtype) * * Aligns the L3 header of '*packetp' on a 32-bit boundary. */ const char * -eth_from_hex(const char *hex, struct ofpbuf **packetp) +eth_from_hex(const char *hex, struct dp_packet **packetp) { - struct ofpbuf *packet; + struct dp_packet *packet; /* Use 2 bytes of headroom to 32-bit align the L3 header. */ - packet = *packetp = ofpbuf_new_with_headroom(strlen(hex) / 2, 2); + packet = *packetp = dp_packet_new_with_headroom(strlen(hex) / 2, 2); - if (ofpbuf_put_hex(packet, hex, NULL)[0] != '\0') { - ofpbuf_delete(packet); + if (dp_packet_put_hex(packet, hex, NULL)[0] != '\0') { + dp_packet_delete(packet); *packetp = NULL; return "Trailing garbage in packet data"; } - if (packet->size < ETH_HEADER_LEN) { - ofpbuf_delete(packet); + if (dp_packet_size(packet) < ETH_HEADER_LEN) { + dp_packet_delete(packet); *packetp = NULL; return "Packet data too short for Ethernet"; } @@ -423,24 +383,12 @@ eth_from_hex(const char *hex, struct ofpbuf **packetp) } void -eth_format_masked(const uint8_t eth[ETH_ADDR_LEN], - const uint8_t mask[ETH_ADDR_LEN], struct ds *s) +eth_format_masked(const struct eth_addr eth, + const struct eth_addr *mask, 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]; + if (mask && !eth_mask_is_exact(*mask)) { + ds_put_format(s, "/"ETH_ADDR_FMT, ETH_ADDR_ARGS(*mask)); } } @@ -452,14 +400,14 @@ eth_addr_bitand(const uint8_t src[ETH_ADDR_LEN], int ip_count_cidr_bits(ovs_be32 netmask) { - return 32 - ctz(ntohl(netmask)); + return 32 - ctz32(ntohl(netmask)); } void ip_format_masked(ovs_be32 ip, ovs_be32 mask, struct ds *s) { ds_put_format(s, IP_FMT, IP_ARGS(ip)); - if (mask != htonl(UINT32_MAX)) { + if (mask != OVS_BE32_MAX) { if (ip_is_cidr(mask)) { ds_put_format(s, "/%d", ip_count_cidr_bits(mask)); } else { @@ -468,44 +416,201 @@ ip_format_masked(ovs_be32 ip, ovs_be32 mask, struct ds *s) } } +/* Parses string 's', which must be an IP address. Stores the IP address into + * '*ip'. Returns true if successful, otherwise false. */ +bool +ip_parse(const char *s, ovs_be32 *ip) +{ + return inet_pton(AF_INET, s, ip) == 1; +} + +/* Parses string 's', which must be an IP address with an optional netmask or + * CIDR prefix length. Stores the IP address into '*ip' and the netmask into + * '*mask'. (If 's' does not contain a netmask, 255.255.255.255 is + * assumed.) + * + * Returns NULL if successful, otherwise an error message that the caller must + * free(). */ +char * OVS_WARN_UNUSED_RESULT +ip_parse_masked(const char *s, ovs_be32 *ip, ovs_be32 *mask) +{ + int prefix; + int n; + + if (ovs_scan(s, IP_SCAN_FMT"/"IP_SCAN_FMT"%n", + IP_SCAN_ARGS(ip), IP_SCAN_ARGS(mask), &n) && !s[n]) { + /* OK. */ + } else if (ovs_scan(s, IP_SCAN_FMT"/%d%n", IP_SCAN_ARGS(ip), &prefix, &n) + && !s[n]) { + if (prefix <= 0 || prefix > 32) { + return xasprintf("%s: network prefix bits not between 0 and " + "32", s); + } + *mask = be32_prefix_mask(prefix); + } else if (ip_parse(s, ip)) { + *mask = OVS_BE32_MAX; + } else { + return xasprintf("%s: invalid IP address", s); + } + return NULL; +} + +/* Similar to ip_parse_masked(), but the mask, if present, must be a CIDR mask + * and is returned as a prefix length in '*plen'. */ +char * OVS_WARN_UNUSED_RESULT +ip_parse_cidr(const char *s, ovs_be32 *ip, unsigned int *plen) +{ + ovs_be32 mask; + char *error; + + error = ip_parse_masked(s, ip, &mask); + if (error) { + return error; + } + + if (!ip_is_cidr(mask)) { + return xasprintf("%s: CIDR network required", s); + } + *plen = ip_count_cidr_bits(mask); + return NULL; +} + +/* Parses string 's', which must be an IPv6 address. Stores the IPv6 address + * into '*ip'. Returns true if successful, otherwise false. */ +bool +ipv6_parse(const char *s, struct in6_addr *ip) +{ + return inet_pton(AF_INET6, s, ip) == 1; +} + +/* Parses string 's', which must be an IPv6 address with an optional netmask or + * CIDR prefix length. Stores the IPv6 address into '*ip' and the netmask into + * '*mask'. (If 's' does not contain a netmask, all-one-bits is assumed.) + * + * Returns NULL if successful, otherwise an error message that the caller must + * free(). */ +char * OVS_WARN_UNUSED_RESULT +ipv6_parse_masked(const char *s, struct in6_addr *ip, struct in6_addr *mask) +{ + char ipv6_s[IPV6_SCAN_LEN + 1]; + int prefix; + int n; + + if (ovs_scan(s, IPV6_SCAN_FMT"%n", ipv6_s, &n) && ipv6_parse(ipv6_s, ip)) { + s += n; + if (!*s) { + *mask = in6addr_exact; + } else if (ovs_scan(s, "/%d%n", &prefix, &n) && !s[n]) { + if (prefix <= 0 || prefix > 128) { + return xasprintf("%s: IPv6 network prefix bits not between 0 " + "and 128", s); + } + *mask = ipv6_create_mask(prefix); + } else if (ovs_scan(s, "/"IPV6_SCAN_FMT"%n", ipv6_s, &n) + && !s[n] + && ipv6_parse(ipv6_s, mask)) { + /* OK. */ + } else { + return xasprintf("%s: syntax error expecting IPv6 prefix length " + "or mask", s); + } + return NULL; + } + return xasprintf("%s: invalid IPv6 address", s); +} + +/* Similar to ipv6_parse_masked(), but the mask, if present, must be a CIDR + * mask and is returned as a prefix length in '*plen'. */ +char * OVS_WARN_UNUSED_RESULT +ipv6_parse_cidr(const char *s, struct in6_addr *ip, unsigned int *plen) +{ + struct in6_addr mask; + char *error; + + error = ipv6_parse_masked(s, ip, &mask); + if (error) { + return error; + } + + if (!ipv6_is_cidr(&mask)) { + return xasprintf("%s: IPv6 CIDR network required", s); + } + *plen = ipv6_count_cidr_bits(&mask); + return NULL; +} /* Stores the string representation of the IPv6 address 'addr' into the * character array 'addr_str', which must be at least INET6_ADDRSTRLEN * bytes long. */ void -format_ipv6_addr(char *addr_str, const struct in6_addr *addr) +ipv6_format_addr(const struct in6_addr *addr, struct ds *s) { - inet_ntop(AF_INET6, addr, addr_str, INET6_ADDRSTRLEN); + char *dst; + + ds_reserve(s, s->length + INET6_ADDRSTRLEN); + + dst = s->string + s->length; + inet_ntop(AF_INET6, addr, dst, INET6_ADDRSTRLEN); + s->length += strlen(dst); } +/* Same as print_ipv6_addr, but optionally encloses the address in square + * brackets. */ void -print_ipv6_addr(struct ds *string, const struct in6_addr *addr) +ipv6_format_addr_bracket(const struct in6_addr *addr, struct ds *s, + bool bracket) { - char *dst; - - ds_reserve(string, string->length + INET6_ADDRSTRLEN); + if (bracket) { + ds_put_char(s, '['); + } + ipv6_format_addr(addr, s); + if (bracket) { + ds_put_char(s, ']'); + } +} - dst = string->string + string->length; - format_ipv6_addr(dst, addr); - string->length += strlen(dst); +void +ipv6_format_mapped(const struct in6_addr *addr, struct ds *s) +{ + if (IN6_IS_ADDR_V4MAPPED(addr)) { + ds_put_format(s, IP_FMT, addr->s6_addr[12], addr->s6_addr[13], + addr->s6_addr[14], addr->s6_addr[15]); + } else { + ipv6_format_addr(addr, s); + } } void -print_ipv6_masked(struct ds *s, const struct in6_addr *addr, - const struct in6_addr *mask) +ipv6_format_masked(const struct in6_addr *addr, const struct in6_addr *mask, + struct ds *s) { - print_ipv6_addr(s, addr); + ipv6_format_addr(addr, s); if (mask && !ipv6_mask_is_exact(mask)) { if (ipv6_is_cidr(mask)) { int cidr_bits = ipv6_count_cidr_bits(mask); ds_put_format(s, "/%d", cidr_bits); } else { ds_put_char(s, '/'); - print_ipv6_addr(s, mask); + ipv6_format_addr(mask, s); } } } +/* Stores the string representation of the IPv6 address 'addr' into the + * character array 'addr_str', which must be at least INET6_ADDRSTRLEN + * bytes long. If addr is IPv4-mapped, store an IPv4 dotted-decimal string. */ +const char * +ipv6_string_mapped(char *addr_str, const struct in6_addr *addr) +{ + ovs_be32 ip; + ip = in6_addr_get_mapped_ipv4(addr); + if (ip) { + return inet_ntop(AF_INET, &ip, addr_str, INET6_ADDRSTRLEN); + } else { + return inet_ntop(AF_INET6, addr, addr_str, INET6_ADDRSTRLEN); + } +} + struct in6_addr ipv6_addr_bitand(const struct in6_addr *a, const struct in6_addr *b) { @@ -605,52 +710,53 @@ ipv6_is_cidr(const struct in6_addr *netmask) /* Populates 'b' with an Ethernet II packet headed with the given 'eth_dst', * 'eth_src' and 'eth_type' parameters. A payload of 'size' bytes is allocated * in 'b' and returned. This payload may be populated with appropriate - * information by the caller. Sets 'b''s 'l2' and 'l3' pointers to the - * Ethernet header and payload respectively. Aligns b->l3 on a 32-bit + * information by the caller. Sets 'b''s 'frame' pointer and 'l3' offset to + * the Ethernet header and payload respectively. Aligns b->l3 on a 32-bit * boundary. * * The returned packet has enough headroom to insert an 802.1Q VLAN header if * desired. */ void * -eth_compose(struct ofpbuf *b, const uint8_t eth_dst[ETH_ADDR_LEN], - const uint8_t eth_src[ETH_ADDR_LEN], uint16_t eth_type, +eth_compose(struct dp_packet *b, const struct eth_addr eth_dst, + const struct eth_addr eth_src, uint16_t eth_type, size_t size) { void *data; struct eth_header *eth; - ofpbuf_clear(b); + dp_packet_clear(b); /* The magic 2 here ensures that the L3 header (when it is added later) * will be 32-bit aligned. */ - ofpbuf_prealloc_tailroom(b, 2 + ETH_HEADER_LEN + VLAN_HEADER_LEN + size); - ofpbuf_reserve(b, 2 + VLAN_HEADER_LEN); - eth = ofpbuf_put_uninit(b, ETH_HEADER_LEN); - data = ofpbuf_put_uninit(b, size); + dp_packet_prealloc_tailroom(b, 2 + ETH_HEADER_LEN + VLAN_HEADER_LEN + size); + dp_packet_reserve(b, 2 + VLAN_HEADER_LEN); + eth = dp_packet_put_uninit(b, ETH_HEADER_LEN); + data = dp_packet_put_uninit(b, size); - memcpy(eth->eth_dst, eth_dst, ETH_ADDR_LEN); - memcpy(eth->eth_src, eth_src, ETH_ADDR_LEN); + eth->eth_dst = eth_dst; + eth->eth_src = eth_src; eth->eth_type = htons(eth_type); - b->l2 = eth; - b->l3 = data; + dp_packet_reset_offsets(b); + dp_packet_set_l3(b, data); return data; } static void -packet_set_ipv4_addr(struct ofpbuf *packet, +packet_set_ipv4_addr(struct dp_packet *packet, ovs_16aligned_be32 *addr, ovs_be32 new_addr) { - struct ip_header *nh = packet->l3; + struct ip_header *nh = dp_packet_l3(packet); ovs_be32 old_addr = get_16aligned_be32(addr); + size_t l4_size = dp_packet_l4_size(packet); - if (nh->ip_proto == IPPROTO_TCP && packet->l7) { - struct tcp_header *th = packet->l4; + if (nh->ip_proto == IPPROTO_TCP && l4_size >= TCP_HEADER_LEN) { + struct tcp_header *th = dp_packet_l4(packet); th->tcp_csum = recalc_csum32(th->tcp_csum, old_addr, new_addr); - } else if (nh->ip_proto == IPPROTO_UDP && packet->l7) { - struct udp_header *uh = packet->l4; + } else if (nh->ip_proto == IPPROTO_UDP && l4_size >= UDP_HEADER_LEN ) { + struct udp_header *uh = dp_packet_l4(packet); if (uh->udp_csum) { uh->udp_csum = recalc_csum32(uh->udp_csum, old_addr, new_addr); @@ -666,17 +772,17 @@ packet_set_ipv4_addr(struct ofpbuf *packet, /* Returns true, if packet contains at least one routing header where * segements_left > 0. * - * This function assumes that L3 and L4 markers are set in the packet. */ + * This function assumes that L3 and L4 offsets are set in the packet. */ static bool -packet_rh_present(struct ofpbuf *packet) +packet_rh_present(struct dp_packet *packet) { const struct ovs_16aligned_ip6_hdr *nh; int nexthdr; size_t len; size_t remaining; - uint8_t *data = packet->l3; + uint8_t *data = dp_packet_l3(packet); - remaining = (uint8_t *)packet->l4 - (uint8_t *)packet->l3; + remaining = packet->l4_ofs - packet->l3_ofs; if (remaining < sizeof *nh) { return false; @@ -749,15 +855,17 @@ packet_rh_present(struct ofpbuf *packet) } static void -packet_update_csum128(struct ofpbuf *packet, uint8_t proto, +packet_update_csum128(struct dp_packet *packet, uint8_t proto, ovs_16aligned_be32 addr[4], const ovs_be32 new_addr[4]) { - if (proto == IPPROTO_TCP && packet->l7) { - struct tcp_header *th = packet->l4; + size_t l4_size = dp_packet_l4_size(packet); + + if (proto == IPPROTO_TCP && l4_size >= TCP_HEADER_LEN) { + struct tcp_header *th = dp_packet_l4(packet); th->tcp_csum = recalc_csum128(th->tcp_csum, addr, new_addr); - } else if (proto == IPPROTO_UDP && packet->l7) { - struct udp_header *uh = packet->l4; + } else if (proto == IPPROTO_UDP && l4_size >= UDP_HEADER_LEN) { + struct udp_header *uh = dp_packet_l4(packet); if (uh->udp_csum) { uh->udp_csum = recalc_csum128(uh->udp_csum, addr, new_addr); @@ -765,18 +873,23 @@ packet_update_csum128(struct ofpbuf *packet, uint8_t proto, uh->udp_csum = htons(0xffff); } } + } else if (proto == IPPROTO_ICMPV6 && + l4_size >= sizeof(struct icmp6_header)) { + struct icmp6_header *icmp = dp_packet_l4(packet); + + icmp->icmp6_cksum = recalc_csum128(icmp->icmp6_cksum, addr, new_addr); } } static void -packet_set_ipv6_addr(struct ofpbuf *packet, uint8_t proto, - ovs_16aligned_be32 *addr, const ovs_be32 new_addr[4], +packet_set_ipv6_addr(struct dp_packet *packet, uint8_t proto, + ovs_16aligned_be32 addr[4], const ovs_be32 new_addr[4], bool recalculate_csum) { if (recalculate_csum) { packet_update_csum128(packet, proto, addr, new_addr); } - memcpy(addr, new_addr, sizeof(*addr)); + memcpy(addr, new_addr, sizeof(ovs_be32[4])); } static void @@ -800,10 +913,10 @@ packet_set_ipv6_tc(ovs_16aligned_be32 *flow_label, uint8_t tc) * '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, +packet_set_ipv4(struct dp_packet *packet, ovs_be32 src, ovs_be32 dst, uint8_t tos, uint8_t ttl) { - struct ip_header *nh = packet->l3; + struct ip_header *nh = dp_packet_l3(packet); if (get_16aligned_be32(&nh->ip_src) != src) { packet_set_ipv4_addr(packet, &nh->ip_src, src); @@ -833,13 +946,13 @@ packet_set_ipv4(struct ofpbuf *packet, ovs_be32 src, ovs_be32 dst, /* Modifies the IPv6 header fields of 'packet' to be consistent with 'src', * 'dst', 'traffic class', and 'next hop'. Updates 'packet''s L4 checksums as * appropriate. 'packet' must contain a valid IPv6 packet with correctly - * populated l[347] markers. */ + * populated l[34] offsets. */ void -packet_set_ipv6(struct ofpbuf *packet, uint8_t proto, const ovs_be32 src[4], +packet_set_ipv6(struct dp_packet *packet, uint8_t proto, const ovs_be32 src[4], const ovs_be32 dst[4], uint8_t key_tc, ovs_be32 key_fl, uint8_t key_hl) { - struct ovs_16aligned_ip6_hdr *nh = packet->l3; + struct ovs_16aligned_ip6_hdr *nh = dp_packet_l3(packet); if (memcmp(&nh->ip6_src, src, sizeof(ovs_be32[4]))) { packet_set_ipv6_addr(packet, proto, nh->ip6_src.be32, src, true); @@ -868,11 +981,11 @@ packet_set_port(ovs_be16 *port, ovs_be16 new_port, ovs_be16 *csum) /* 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. */ + * with its l4 offset properly populated. */ void -packet_set_tcp_port(struct ofpbuf *packet, ovs_be16 src, ovs_be16 dst) +packet_set_tcp_port(struct dp_packet *packet, ovs_be16 src, ovs_be16 dst) { - struct tcp_header *th = packet->l4; + struct tcp_header *th = dp_packet_l4(packet); packet_set_port(&th->tcp_src, src, &th->tcp_csum); packet_set_port(&th->tcp_dst, dst, &th->tcp_csum); @@ -880,11 +993,11 @@ packet_set_tcp_port(struct ofpbuf *packet, ovs_be16 src, ovs_be16 dst) /* 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. */ + * with its l4 offset properly populated. */ void -packet_set_udp_port(struct ofpbuf *packet, ovs_be16 src, ovs_be16 dst) +packet_set_udp_port(struct dp_packet *packet, ovs_be16 src, ovs_be16 dst) { - struct udp_header *uh = packet->l4; + struct udp_header *uh = dp_packet_l4(packet); if (uh->udp_csum) { packet_set_port(&uh->udp_src, src, &uh->udp_csum); @@ -901,46 +1014,132 @@ packet_set_udp_port(struct ofpbuf *packet, ovs_be16 src, ovs_be16 dst) /* Sets the SCTP source and destination port ('src' and 'dst' respectively) of * the SCTP header contained in 'packet'. 'packet' must be a valid SCTP packet - * with its l4 marker properly populated. */ + * with its l4 offset properly populated. */ void -packet_set_sctp_port(struct ofpbuf *packet, ovs_be16 src, ovs_be16 dst) +packet_set_sctp_port(struct dp_packet *packet, ovs_be16 src, ovs_be16 dst) { - struct sctp_header *sh = packet->l4; + struct sctp_header *sh = dp_packet_l4(packet); ovs_be32 old_csum, old_correct_csum, new_csum; - uint16_t tp_len = packet->size - ((uint8_t*)sh - (uint8_t*)packet->data); + uint16_t tp_len = dp_packet_l4_size(packet); - old_csum = sh->sctp_csum; - sh->sctp_csum = 0; - old_correct_csum = crc32c(packet->l4, tp_len); + old_csum = get_16aligned_be32(&sh->sctp_csum); + put_16aligned_be32(&sh->sctp_csum, 0); + old_correct_csum = crc32c((void *)sh, tp_len); sh->sctp_src = src; sh->sctp_dst = dst; - new_csum = crc32c(packet->l4, tp_len); - sh->sctp_csum = old_csum ^ old_correct_csum ^ new_csum; + new_csum = crc32c((void *)sh, tp_len); + put_16aligned_be32(&sh->sctp_csum, old_csum ^ old_correct_csum ^ new_csum); } -/* If 'packet' is a TCP packet, returns the TCP flags. Otherwise, returns 0. - * - * '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) -{ - if (dl_type_is_ip_any(flow->dl_type) && - flow->nw_proto == IPPROTO_TCP && packet->l7) { - const struct tcp_header *tcp = packet->l4; - return TCP_FLAGS(tcp->tcp_ctl); - } else { - return 0; +/* Sets the ICMP type and code of the ICMP header contained in 'packet'. + * 'packet' must be a valid ICMP packet with its l4 offset properly + * populated. */ +void +packet_set_icmp(struct dp_packet *packet, uint8_t type, uint8_t code) +{ + struct icmp_header *ih = dp_packet_l4(packet); + ovs_be16 orig_tc = htons(ih->icmp_type << 8 | ih->icmp_code); + ovs_be16 new_tc = htons(type << 8 | code); + + if (orig_tc != new_tc) { + ih->icmp_type = type; + ih->icmp_code = code; + + ih->icmp_csum = recalc_csum16(ih->icmp_csum, orig_tc, new_tc); + } +} + +void +packet_set_nd(struct dp_packet *packet, const ovs_be32 target[4], + const struct eth_addr sll, const struct eth_addr tll) { + struct ovs_nd_msg *ns; + struct ovs_nd_opt *nd_opt; + int bytes_remain = dp_packet_l4_size(packet); + + if (OVS_UNLIKELY(bytes_remain < sizeof(*ns))) { + return; + } + + ns = dp_packet_l4(packet); + nd_opt = &ns->options[0]; + bytes_remain -= sizeof(*ns); + + if (memcmp(&ns->target, target, sizeof(ovs_be32[4]))) { + packet_set_ipv6_addr(packet, IPPROTO_ICMPV6, + ns->target.be32, + target, true); + } + + while (bytes_remain >= ND_OPT_LEN && nd_opt->nd_opt_len != 0) { + if (nd_opt->nd_opt_type == ND_OPT_SOURCE_LINKADDR + && nd_opt->nd_opt_len == 1) { + if (!eth_addr_equals(nd_opt->nd_opt_mac, sll)) { + ovs_be16 *csum = &(ns->icmph.icmp6_cksum); + + *csum = recalc_csum48(*csum, nd_opt->nd_opt_mac, sll); + nd_opt->nd_opt_mac = sll; + } + + /* A packet can only contain one SLL or TLL option */ + break; + } else if (nd_opt->nd_opt_type == ND_OPT_TARGET_LINKADDR + && nd_opt->nd_opt_len == 1) { + if (!eth_addr_equals(nd_opt->nd_opt_mac, tll)) { + ovs_be16 *csum = &(ns->icmph.icmp6_cksum); + + *csum = recalc_csum48(*csum, nd_opt->nd_opt_mac, tll); + nd_opt->nd_opt_mac = tll; + } + + /* A packet can only contain one SLL or TLL option */ + break; + } + + nd_opt += nd_opt->nd_opt_len; + bytes_remain -= nd_opt->nd_opt_len * ND_OPT_LEN; + } +} + +const char * +packet_tcp_flag_to_string(uint32_t flag) +{ + switch (flag) { + case TCP_FIN: + return "fin"; + case TCP_SYN: + return "syn"; + case TCP_RST: + return "rst"; + case TCP_PSH: + return "psh"; + case TCP_ACK: + return "ack"; + case TCP_URG: + return "urg"; + case TCP_ECE: + return "ece"; + case TCP_CWR: + return "cwr"; + case TCP_NS: + return "ns"; + case 0x200: + return "[200]"; + case 0x400: + return "[400]"; + case 0x800: + return "[800]"; + default: + return NULL; } } /* 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 + * (e.g. from struct flow.tcp_flags or obtained via TCP_FLAGS) to 's', in the * format used by tcpdump. */ void -packet_format_tcp_flags(struct ds *s, uint8_t tcp_flags) +packet_format_tcp_flags(struct ds *s, uint16_t tcp_flags) { if (!tcp_flags) { ds_put_cstr(s, "none"); @@ -965,10 +1164,131 @@ packet_format_tcp_flags(struct ds *s, uint8_t tcp_flags) if (tcp_flags & TCP_ACK) { ds_put_char(s, '.'); } - if (tcp_flags & 0x40) { - ds_put_cstr(s, "[40]"); + if (tcp_flags & TCP_ECE) { + ds_put_cstr(s, "E"); + } + if (tcp_flags & TCP_CWR) { + ds_put_cstr(s, "C"); } - if (tcp_flags & 0x80) { - ds_put_cstr(s, "[80]"); + if (tcp_flags & TCP_NS) { + ds_put_cstr(s, "N"); + } + if (tcp_flags & 0x200) { + ds_put_cstr(s, "[200]"); + } + if (tcp_flags & 0x400) { + ds_put_cstr(s, "[400]"); + } + if (tcp_flags & 0x800) { + ds_put_cstr(s, "[800]"); } } + +#define ARP_PACKET_SIZE (2 + ETH_HEADER_LEN + VLAN_HEADER_LEN + \ + ARP_ETH_HEADER_LEN) + +/* Clears 'b' and replaces its contents by an ARP frame with the specified + * 'arp_op', 'arp_sha', 'arp_tha', 'arp_spa', and 'arp_tpa'. The outer + * Ethernet frame is initialized with Ethernet source 'arp_sha' and destination + * 'arp_tha', except that destination ff:ff:ff:ff:ff:ff is used instead if + * 'broadcast' is true. */ +void +compose_arp(struct dp_packet *b, uint16_t arp_op, + const struct eth_addr arp_sha, const struct eth_addr arp_tha, + bool broadcast, ovs_be32 arp_spa, ovs_be32 arp_tpa) +{ + struct eth_header *eth; + struct arp_eth_header *arp; + + dp_packet_clear(b); + dp_packet_prealloc_tailroom(b, ARP_PACKET_SIZE); + dp_packet_reserve(b, 2 + VLAN_HEADER_LEN); + + eth = dp_packet_put_uninit(b, sizeof *eth); + eth->eth_dst = broadcast ? eth_addr_broadcast : arp_tha; + eth->eth_src = arp_sha; + eth->eth_type = htons(ETH_TYPE_ARP); + + arp = dp_packet_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); + arp->ar_sha = arp_sha; + arp->ar_tha = arp_tha; + + put_16aligned_be32(&arp->ar_spa, arp_spa); + put_16aligned_be32(&arp->ar_tpa, arp_tpa); + + dp_packet_reset_offsets(b); + dp_packet_set_l3(b, arp); +} + +void +compose_nd(struct dp_packet *b, const struct eth_addr eth_src, + struct in6_addr * ipv6_src, struct in6_addr * ipv6_dst) +{ + struct in6_addr sn_addr; + struct eth_addr eth_dst; + struct ovs_nd_msg *ns; + struct ovs_nd_opt *nd_opt; + + in6_addr_solicited_node(&sn_addr, ipv6_dst); + ipv6_multicast_to_ethernet(ð_dst, &sn_addr); + + eth_compose(b, eth_dst, eth_src, ETH_TYPE_IPV6, + IPV6_HEADER_LEN + ICMP6_HEADER_LEN + ND_OPT_LEN); + packet_set_ipv6(b, IPPROTO_ICMPV6, + ALIGNED_CAST(ovs_be32 *, ipv6_src->s6_addr), + ALIGNED_CAST(ovs_be32 *, sn_addr.s6_addr), + 0, 0, 255); + + ns = dp_packet_l4(b); + nd_opt = &ns->options[0]; + + ns->icmph.icmp6_type = ND_NEIGHBOR_SOLICIT; + ns->icmph.icmp6_code = 0; + + nd_opt->nd_opt_type = ND_OPT_SOURCE_LINKADDR; + packet_set_nd(b, ALIGNED_CAST(ovs_be32 *, ipv6_dst->s6_addr), + eth_src, eth_addr_zero); +} + +uint32_t +packet_csum_pseudoheader(const struct ip_header *ip) +{ + uint32_t partial = 0; + + partial = csum_add32(partial, get_16aligned_be32(&ip->ip_src)); + partial = csum_add32(partial, get_16aligned_be32(&ip->ip_dst)); + partial = csum_add16(partial, htons(ip->ip_proto)); + partial = csum_add16(partial, htons(ntohs(ip->ip_tot_len) - + IP_IHL(ip->ip_ihl_ver) * 4)); + + return partial; +} + +#ifndef __CHECKER__ +uint32_t +packet_csum_pseudoheader6(const struct ovs_16aligned_ip6_hdr *ip6) +{ + uint32_t partial = 0; + + partial = csum_add32(partial, get_16aligned_be32(&(ip6->ip6_src.be32[0]))); + partial = csum_add32(partial, get_16aligned_be32(&(ip6->ip6_src.be32[1]))); + partial = csum_add32(partial, get_16aligned_be32(&(ip6->ip6_src.be32[2]))); + partial = csum_add32(partial, get_16aligned_be32(&(ip6->ip6_src.be32[3]))); + partial = csum_add32(partial, get_16aligned_be32(&(ip6->ip6_dst.be32[0]))); + partial = csum_add32(partial, get_16aligned_be32(&(ip6->ip6_dst.be32[1]))); + partial = csum_add32(partial, get_16aligned_be32(&(ip6->ip6_dst.be32[2]))); + partial = csum_add32(partial, get_16aligned_be32(&(ip6->ip6_dst.be32[3]))); + + partial = csum_add16(partial, 0); + partial = csum_add16(partial, ip6->ip6_plen); + partial = csum_add16(partial, 0); + partial = csum_add16(partial, ip6->ip6_nxt); + + return partial; +} +#endif