X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=lib%2Fflow.c;h=9ad1898f72f299a8a7979ef8d63a99abf40de3ea;hb=be812f2d2b0c948e5f52017c4df79645809fcb90;hp=879e462f0d059b487182d57456e6c67f39a45a36;hpb=685a51a5b89750cead1b2934c2079d2bb9c52a4a;p=openvswitch diff --git a/lib/flow.c b/lib/flow.c index 879e462f..9ad1898f 100644 --- a/lib/flow.c +++ b/lib/flow.c @@ -1,5 +1,5 @@ /* - * Copyright (c) 2008, 2009, 2010, 2011 Nicira Networks. + * Copyright (c) 2008, 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. @@ -16,6 +16,7 @@ #include #include #include "flow.h" +#include #include #include #include @@ -25,12 +26,10 @@ #include #include "byte-order.h" #include "coverage.h" -#include "dpif.h" #include "dynamic-string.h" #include "hash.h" #include "ofpbuf.h" #include "openflow/openflow.h" -#include "openvswitch/datapath-protocol.h" #include "packets.h" #include "unaligned.h" #include "vlog.h" @@ -134,36 +133,26 @@ parse_ethertype(struct ofpbuf *b) static int parse_ipv6(struct ofpbuf *packet, struct flow *flow) { - struct ip6_hdr *nh; - int nh_len = sizeof(struct ip6_hdr); - int payload_len; + const struct ip6_hdr *nh; ovs_be32 tc_flow; int nexthdr; - if (packet->size < sizeof *nh) { - return -EINVAL; + nh = ofpbuf_try_pull(packet, sizeof *nh); + if (!nh) { + return EINVAL; } - nh = packet->data; nexthdr = nh->ip6_nxt; - payload_len = ntohs(nh->ip6_plen); flow->ipv6_src = nh->ip6_src; flow->ipv6_dst = nh->ip6_dst; tc_flow = get_unaligned_be32(&nh->ip6_flow); - flow->nw_tos = (ntohl(tc_flow) >> 4) & IP_DSCP_MASK; + flow->nw_tos = ntohl(tc_flow) >> 20; + flow->ipv6_label = tc_flow & htonl(IPV6_LABEL_MASK); + flow->nw_ttl = nh->ip6_hlim; flow->nw_proto = IPPROTO_NONE; - /* We don't process jumbograms. */ - if (!payload_len) { - return -EINVAL; - } - - if (packet->size < sizeof *nh + payload_len) { - return -EINVAL; - } - while (1) { if ((nexthdr != IPPROTO_HOPOPTS) && (nexthdr != IPPROTO_ROUTING) @@ -179,9 +168,10 @@ parse_ipv6(struct ofpbuf *packet, struct flow *flow) /* We only verify that at least 8 bytes of the next header are * available, but many of these headers are longer. Ensure that * accesses within the extension header are within those first 8 + * bytes. All extension headers are required to be at least 8 * bytes. */ - if (packet->size < nh_len + 8) { - return -EINVAL; + if (packet->size < 8) { + return EINVAL; } if ((nexthdr == IPPROTO_HOPOPTS) @@ -189,55 +179,70 @@ parse_ipv6(struct ofpbuf *packet, struct flow *flow) || (nexthdr == IPPROTO_DSTOPTS)) { /* These headers, while different, have the fields we care about * in the same location and with the same interpretation. */ - struct ip6_ext *ext_hdr; - - ext_hdr = (struct ip6_ext *)((char *)packet->data + nh_len); + const struct ip6_ext *ext_hdr = (struct ip6_ext *)packet->data; nexthdr = ext_hdr->ip6e_nxt; - nh_len += (ext_hdr->ip6e_len + 1) * 8; + if (!ofpbuf_try_pull(packet, (ext_hdr->ip6e_len + 1) * 8)) { + return EINVAL; + } } else if (nexthdr == IPPROTO_AH) { /* A standard AH definition isn't available, but the fields * we care about are in the same location as the generic * option header--only the header length is calculated * differently. */ - struct ip6_ext *ext_hdr; - - ext_hdr = (struct ip6_ext *)((char *)packet->data + nh_len); + const struct ip6_ext *ext_hdr = (struct ip6_ext *)packet->data; nexthdr = ext_hdr->ip6e_nxt; - nh_len += (ext_hdr->ip6e_len + 2) * 4; + if (!ofpbuf_try_pull(packet, (ext_hdr->ip6e_len + 2) * 4)) { + return EINVAL; + } } else if (nexthdr == IPPROTO_FRAGMENT) { - struct ip6_frag *frag_hdr; - - frag_hdr = (struct ip6_frag *)((char *)packet->data + nh_len); + const struct ip6_frag *frag_hdr = (struct ip6_frag *)packet->data; nexthdr = frag_hdr->ip6f_nxt; - nh_len += sizeof *frag_hdr; + if (!ofpbuf_try_pull(packet, sizeof *frag_hdr)) { + return EINVAL; + } /* We only process the first fragment. */ - if ((frag_hdr->ip6f_offlg & IP6F_OFF_MASK) != htons(0)) { - nexthdr = IPPROTO_FRAGMENT; - break; + if (frag_hdr->ip6f_offlg != htons(0)) { + if ((frag_hdr->ip6f_offlg & IP6F_OFF_MASK) == htons(0)) { + flow->nw_frag = FLOW_NW_FRAG_ANY; + } else { + flow->nw_frag |= FLOW_NW_FRAG_LATER; + nexthdr = IPPROTO_FRAGMENT; + break; + } } } } - /* The payload length claims to be smaller than the size of the - * headers we've already processed. */ - if (payload_len < nh_len - sizeof *nh) { - return -EINVAL; - } - flow->nw_proto = nexthdr; - return nh_len; + return 0; +} + +static void +parse_tcp(struct ofpbuf *packet, struct ofpbuf *b, struct flow *flow) +{ + const struct tcp_header *tcp = pull_tcp(b); + if (tcp) { + flow->tp_src = tcp->tcp_src; + flow->tp_dst = tcp->tcp_dst; + packet->l7 = b->data; + } } -/* Neighbor Discovery Solicitation and Advertisement messages are - * identical in structure, so we'll just use one of them. To be safe, - * we'll assert that they're still identical. */ -BUILD_ASSERT_DECL(sizeof(struct nd_neighbor_solicit) - == sizeof(struct nd_neighbor_advert)); +static void +parse_udp(struct ofpbuf *packet, struct ofpbuf *b, struct flow *flow) +{ + const struct udp_header *udp = pull_udp(b); + if (udp) { + flow->tp_src = udp->udp_src; + flow->tp_dst = udp->udp_dst; + packet->l7 = b->data; + } +} static bool -parse_icmpv6(struct ofpbuf *b, struct flow *flow, int icmp_len) +parse_icmpv6(struct ofpbuf *b, struct flow *flow) { const struct icmp6_hdr *icmp = pull_icmpv6(b); @@ -247,37 +252,29 @@ parse_icmpv6(struct ofpbuf *b, struct flow *flow, int icmp_len) /* The ICMPv6 type and code fields use the 16-bit transport port * fields, so we need to store them in 16-bit network byte order. */ - flow->icmp_type = htons(icmp->icmp6_type); - flow->icmp_code = htons(icmp->icmp6_code); - - if (!icmp->icmp6_code - && ((icmp->icmp6_type == ND_NEIGHBOR_SOLICIT) - || (icmp->icmp6_type == ND_NEIGHBOR_ADVERT))) { - struct nd_neighbor_solicit *nd_ns; /* Identical to ND advert */ - - /* In order to process neighbor discovery options, we need the - * entire packet. */ - if ((icmp_len < sizeof *nd_ns) - || (!ofpbuf_try_pull(b, sizeof *nd_ns - sizeof *icmp))) { + flow->tp_src = htons(icmp->icmp6_type); + flow->tp_dst = htons(icmp->icmp6_code); + + if (icmp->icmp6_code == 0 && + (icmp->icmp6_type == ND_NEIGHBOR_SOLICIT || + icmp->icmp6_type == ND_NEIGHBOR_ADVERT)) { + const struct in6_addr *nd_target; + + nd_target = ofpbuf_try_pull(b, sizeof *nd_target); + if (!nd_target) { return false; } - nd_ns = (struct nd_neighbor_solicit *)icmp; - flow->nd_target = nd_ns->nd_ns_target; - - icmp_len -= sizeof(*nd_ns); - while (icmp_len >= 8) { - struct nd_opt_hdr *nd_opt; - int opt_len; - const uint8_t *data; + flow->nd_target = *nd_target; + while (b->size >= 8) { /* The minimum size of an option is 8 bytes, which also is * the size of Ethernet link-layer options. */ - nd_opt = ofpbuf_pull(b, 8); - if (!nd_opt->nd_opt_len || nd_opt->nd_opt_len * 8 > icmp_len) { + const struct nd_opt_hdr *nd_opt = b->data; + int opt_len = nd_opt->nd_opt_len * 8; + + if (!opt_len || opt_len > b->size) { goto invalid; } - opt_len = nd_opt->nd_opt_len * 8; - data = (const uint8_t *)(nd_opt + 1); /* Store the link layer address if the appropriate option is * provided. It is considered an error if the same link @@ -285,40 +282,39 @@ parse_icmpv6(struct ofpbuf *b, struct flow *flow, int icmp_len) if (nd_opt->nd_opt_type == ND_OPT_SOURCE_LINKADDR && opt_len == 8) { if (eth_addr_is_zero(flow->arp_sha)) { - memcpy(flow->arp_sha, data, ETH_ADDR_LEN); + memcpy(flow->arp_sha, nd_opt + 1, ETH_ADDR_LEN); } else { goto invalid; } } else if (nd_opt->nd_opt_type == ND_OPT_TARGET_LINKADDR && opt_len == 8) { if (eth_addr_is_zero(flow->arp_tha)) { - memcpy(flow->arp_tha, data, ETH_ADDR_LEN); + memcpy(flow->arp_tha, nd_opt + 1, ETH_ADDR_LEN); } else { goto invalid; } } - /* Pull the rest of this option. */ - if (!ofpbuf_try_pull(b, opt_len - 8)) { + if (!ofpbuf_try_pull(b, opt_len)) { goto invalid; } - - icmp_len -= opt_len; } } return true; invalid: - memset(&flow->nd_target, '\0', sizeof(flow->nd_target)); - memset(flow->arp_sha, '\0', sizeof(flow->arp_sha)); - memset(flow->arp_tha, '\0', sizeof(flow->arp_tha)); + memset(&flow->nd_target, 0, sizeof(flow->nd_target)); + memset(flow->arp_sha, 0, sizeof(flow->arp_sha)); + memset(flow->arp_tha, 0, sizeof(flow->arp_tha)); return false; } -/* Initializes 'flow' members from 'packet', 'tun_id', and 'in_port. +/* Initializes 'flow' members from 'packet', 'skb_priority', 'tun_id', and + * 'ofp_in_port'. + * * Initializes 'packet' header pointers as follows: * * - packet->l2 to the start of the Ethernet header. @@ -333,19 +329,19 @@ invalid: * - packet->l7 to just past the TCP or UDP or ICMP header, if one is * present and has a correct length, and otherwise NULL. */ -int -flow_extract(struct ofpbuf *packet, ovs_be64 tun_id, uint16_t in_port, - struct flow *flow) +void +flow_extract(struct ofpbuf *packet, uint32_t skb_priority, ovs_be64 tun_id, + uint16_t ofp_in_port, struct flow *flow) { struct ofpbuf b = *packet; struct eth_header *eth; - int retval = 0; COVERAGE_INC(flow_extract); memset(flow, 0, sizeof *flow); flow->tun_id = tun_id; - flow->in_port = in_port; + flow->in_port = ofp_in_port; + flow->skb_priority = skb_priority; packet->l2 = b.data; packet->l3 = NULL; @@ -353,7 +349,7 @@ flow_extract(struct ofpbuf *packet, ovs_be64 tun_id, uint16_t in_port, packet->l7 = NULL; if (b.size < sizeof *eth) { - return 0; + return; } /* Link layer. */ @@ -373,69 +369,49 @@ flow_extract(struct ofpbuf *packet, ovs_be64 tun_id, uint16_t in_port, if (flow->dl_type == htons(ETH_TYPE_IP)) { const struct ip_header *nh = pull_ip(&b); if (nh) { + packet->l4 = b.data; + flow->nw_src = get_unaligned_be32(&nh->ip_src); flow->nw_dst = get_unaligned_be32(&nh->ip_dst); - flow->nw_tos = nh->ip_tos & IP_DSCP_MASK; flow->nw_proto = nh->ip_proto; - packet->l4 = b.data; - if (!IP_IS_FRAGMENT(nh->ip_frag_off)) { + + flow->nw_tos = nh->ip_tos; + if (IP_IS_FRAGMENT(nh->ip_frag_off)) { + flow->nw_frag = FLOW_NW_FRAG_ANY; + if (nh->ip_frag_off & htons(IP_FRAG_OFF_MASK)) { + flow->nw_frag |= FLOW_NW_FRAG_LATER; + } + } + flow->nw_ttl = nh->ip_ttl; + + if (!(nh->ip_frag_off & htons(IP_FRAG_OFF_MASK))) { if (flow->nw_proto == IPPROTO_TCP) { - const struct tcp_header *tcp = pull_tcp(&b); - if (tcp) { - flow->tp_src = tcp->tcp_src; - flow->tp_dst = tcp->tcp_dst; - packet->l7 = b.data; - } + parse_tcp(packet, &b, flow); } else if (flow->nw_proto == IPPROTO_UDP) { - const struct udp_header *udp = pull_udp(&b); - if (udp) { - flow->tp_src = udp->udp_src; - flow->tp_dst = udp->udp_dst; - packet->l7 = b.data; - } + parse_udp(packet, &b, flow); } else if (flow->nw_proto == IPPROTO_ICMP) { const struct icmp_header *icmp = pull_icmp(&b); if (icmp) { - flow->icmp_type = htons(icmp->icmp_type); - flow->icmp_code = htons(icmp->icmp_code); + flow->tp_src = htons(icmp->icmp_type); + flow->tp_dst = htons(icmp->icmp_code); packet->l7 = b.data; } } - } else { - retval = 1; } } } else if (flow->dl_type == htons(ETH_TYPE_IPV6)) { - int nh_len; - const struct ip6_hdr *nh; - - nh_len = parse_ipv6(&b, flow); - if (nh_len < 0) { - return 0; + if (parse_ipv6(&b, flow)) { + return; } - nh = ofpbuf_pull(&b, nh_len); - if (nh) { - packet->l4 = b.data; - if (flow->nw_proto == IPPROTO_TCP) { - const struct tcp_header *tcp = pull_tcp(&b); - if (tcp) { - flow->tp_src = tcp->tcp_src; - flow->tp_dst = tcp->tcp_dst; - packet->l7 = b.data; - } - } else if (flow->nw_proto == IPPROTO_UDP) { - const struct udp_header *udp = pull_udp(&b); - if (udp) { - flow->tp_src = udp->udp_src; - flow->tp_dst = udp->udp_dst; - packet->l7 = b.data; - } - } else if (flow->nw_proto == IPPROTO_ICMPV6) { - int icmp_len = ntohs(nh->ip6_plen) + sizeof *nh - nh_len; - if (parse_icmpv6(&b, flow, icmp_len)) { - packet->l7 = b.data; - } + packet->l4 = b.data; + if (flow->nw_proto == IPPROTO_TCP) { + parse_tcp(packet, &b, flow); + } else if (flow->nw_proto == IPPROTO_UDP) { + parse_udp(packet, &b, flow); + } else if (flow->nw_proto == IPPROTO_ICMPV6) { + if (parse_icmpv6(&b, flow)) { + packet->l7 = b.data; } } } else if (flow->dl_type == htons(ETH_TYPE_ARP)) { @@ -458,28 +434,83 @@ flow_extract(struct ofpbuf *packet, ovs_be64 tun_id, uint16_t in_port, } } } - - return retval; } -/* Extracts the flow stats for a packet. The 'flow' and 'packet' - * arguments must have been initialized through a call to flow_extract(). - */ +/* For every bit of a field that is wildcarded in 'wildcards', sets the + * corresponding bit in 'flow' to zero. */ void -flow_extract_stats(const struct flow *flow, struct ofpbuf *packet, - struct dpif_flow_stats *stats) +flow_zero_wildcards(struct flow *flow, const struct flow_wildcards *wildcards) { - memset(stats, 0, sizeof(*stats)); + const flow_wildcards_t wc = wildcards->wildcards; + int i; - if ((flow->dl_type == htons(ETH_TYPE_IP)) && packet->l4) { - if ((flow->nw_proto == IPPROTO_TCP) && packet->l7) { - struct tcp_header *tcp = packet->l4; - stats->tcp_flags = TCP_FLAGS(tcp->tcp_ctl); - } + BUILD_ASSERT_DECL(FLOW_WC_SEQ == 12); + + for (i = 0; i < FLOW_N_REGS; i++) { + flow->regs[i] &= wildcards->reg_masks[i]; + } + flow->tun_id &= wildcards->tun_id_mask; + flow->metadata &= wildcards->metadata_mask; + flow->nw_src &= wildcards->nw_src_mask; + flow->nw_dst &= wildcards->nw_dst_mask; + if (wc & FWW_IN_PORT) { + flow->in_port = 0; + } + flow->vlan_tci &= wildcards->vlan_tci_mask; + if (wc & FWW_DL_TYPE) { + flow->dl_type = htons(0); + } + flow->tp_src &= wildcards->tp_src_mask; + flow->tp_dst &= wildcards->tp_dst_mask; + eth_addr_bitand(flow->dl_src, wildcards->dl_src_mask, flow->dl_src); + eth_addr_bitand(flow->dl_dst, wildcards->dl_dst_mask, flow->dl_dst); + if (wc & FWW_NW_PROTO) { + flow->nw_proto = 0; + } + if (wc & FWW_IPV6_LABEL) { + flow->ipv6_label = htonl(0); + } + if (wc & FWW_NW_DSCP) { + flow->nw_tos &= ~IP_DSCP_MASK; } + if (wc & FWW_NW_ECN) { + flow->nw_tos &= ~IP_ECN_MASK; + } + if (wc & FWW_NW_TTL) { + flow->nw_ttl = 0; + } + flow->nw_frag &= wildcards->nw_frag_mask; + if (wc & FWW_ARP_SHA) { + memset(flow->arp_sha, 0, sizeof flow->arp_sha); + } + if (wc & FWW_ARP_THA) { + memset(flow->arp_tha, 0, sizeof flow->arp_tha); + } + flow->ipv6_src = ipv6_addr_bitand(&flow->ipv6_src, + &wildcards->ipv6_src_mask); + flow->ipv6_dst = ipv6_addr_bitand(&flow->ipv6_dst, + &wildcards->ipv6_dst_mask); + flow->nd_target = ipv6_addr_bitand(&flow->nd_target, + &wildcards->nd_target_mask); + flow->skb_priority = 0; +} + +/* Initializes 'fmd' with the metadata found in 'flow'. */ +void +flow_get_metadata(const struct flow *flow, struct flow_metadata *fmd) +{ + BUILD_ASSERT_DECL(FLOW_WC_SEQ == 12); + + fmd->tun_id = flow->tun_id; + fmd->tun_id_mask = htonll(UINT64_MAX); - stats->n_bytes = packet->size; - stats->n_packets = 1; + fmd->metadata = flow->metadata; + fmd->metadata_mask = htonll(UINT64_MAX); + + memcpy(fmd->regs, flow->regs, sizeof fmd->regs); + memset(fmd->reg_masks, 0xff, sizeof fmd->reg_masks); + + fmd->in_port = flow->in_port; } char * @@ -493,43 +524,56 @@ flow_to_string(const struct flow *flow) void flow_format(struct ds *ds, const struct flow *flow) { - ds_put_format(ds, "tunnel%#"PRIx64":in_port%04"PRIx16":tci(", - flow->tun_id, flow->in_port); + ds_put_format(ds, "priority:%"PRIu32 + ",tunnel:%#"PRIx64 + ",metadata:%#"PRIx64 + ",in_port:%04"PRIx16, + flow->skb_priority, + ntohll(flow->tun_id), + ntohll(flow->metadata), + flow->in_port); + + ds_put_format(ds, ",tci("); if (flow->vlan_tci) { - ds_put_format(ds, "vlan%"PRIu16",pcp%d", + ds_put_format(ds, "vlan:%"PRIu16",pcp:%d", vlan_tci_to_vid(flow->vlan_tci), vlan_tci_to_pcp(flow->vlan_tci)); } else { ds_put_char(ds, '0'); } - ds_put_format(ds, ") mac"ETH_ADDR_FMT"->"ETH_ADDR_FMT - " type%04"PRIx16, + ds_put_format(ds, ") mac("ETH_ADDR_FMT"->"ETH_ADDR_FMT + ") type:%04"PRIx16, ETH_ADDR_ARGS(flow->dl_src), ETH_ADDR_ARGS(flow->dl_dst), ntohs(flow->dl_type)); if (flow->dl_type == htons(ETH_TYPE_IPV6)) { - ds_put_format(ds, " proto%"PRIu8" tos%"PRIu8" ipv6", - flow->nw_proto, flow->nw_tos); + ds_put_format(ds, " label:%#"PRIx32" proto:%"PRIu8" tos:%#"PRIx8 + " ttl:%"PRIu8" ipv6(", + ntohl(flow->ipv6_label), flow->nw_proto, + flow->nw_tos, flow->nw_ttl); print_ipv6_addr(ds, &flow->ipv6_src); ds_put_cstr(ds, "->"); print_ipv6_addr(ds, &flow->ipv6_dst); - + ds_put_char(ds, ')'); } else { - ds_put_format(ds, " proto%"PRIu8 - " tos%"PRIu8 - " ip"IP_FMT"->"IP_FMT, - flow->nw_proto, - flow->nw_tos, - IP_ARGS(&flow->nw_src), - IP_ARGS(&flow->nw_dst)); + ds_put_format(ds, " proto:%"PRIu8" tos:%#"PRIx8" ttl:%"PRIu8 + " ip("IP_FMT"->"IP_FMT")", + flow->nw_proto, flow->nw_tos, flow->nw_ttl, + IP_ARGS(&flow->nw_src), IP_ARGS(&flow->nw_dst)); + } + if (flow->nw_frag) { + ds_put_format(ds, " frag(%s)", + flow->nw_frag == FLOW_NW_FRAG_ANY ? "first" + : flow->nw_frag == (FLOW_NW_FRAG_ANY | FLOW_NW_FRAG_LATER) + ? "later" : ""); } if (flow->tp_src || flow->tp_dst) { - ds_put_format(ds, " port%"PRIu16"->%"PRIu16, + ds_put_format(ds, " port(%"PRIu16"->%"PRIu16")", ntohs(flow->tp_src), ntohs(flow->tp_dst)); } if (!eth_addr_is_zero(flow->arp_sha) || !eth_addr_is_zero(flow->arp_tha)) { - ds_put_format(ds, " arp_ha"ETH_ADDR_FMT"->"ETH_ADDR_FMT, + ds_put_format(ds, " arp_ha("ETH_ADDR_FMT"->"ETH_ADDR_FMT")", ETH_ADDR_ARGS(flow->arp_sha), ETH_ADDR_ARGS(flow->arp_tha)); } @@ -549,15 +593,24 @@ flow_print(FILE *stream, const struct flow *flow) void flow_wildcards_init_catchall(struct flow_wildcards *wc) { + BUILD_ASSERT_DECL(FLOW_WC_SEQ == 12); + wc->wildcards = FWW_ALL; wc->tun_id_mask = htonll(0); wc->nw_src_mask = htonl(0); wc->nw_dst_mask = htonl(0); wc->ipv6_src_mask = in6addr_any; wc->ipv6_dst_mask = in6addr_any; + wc->nd_target_mask = in6addr_any; memset(wc->reg_masks, 0, sizeof wc->reg_masks); + wc->metadata_mask = htonll(0); wc->vlan_tci_mask = htons(0); - wc->zero = 0; + wc->nw_frag_mask = 0; + wc->tp_src_mask = htons(0); + wc->tp_dst_mask = htons(0); + memset(wc->dl_src_mask, 0, ETH_ADDR_LEN); + memset(wc->dl_dst_mask, 0, ETH_ADDR_LEN); + memset(wc->zeros, 0, sizeof wc->zeros); } /* Initializes 'wc' as an exact-match set of wildcards; that is, 'wc' does not @@ -565,15 +618,24 @@ flow_wildcards_init_catchall(struct flow_wildcards *wc) void flow_wildcards_init_exact(struct flow_wildcards *wc) { + BUILD_ASSERT_DECL(FLOW_WC_SEQ == 12); + wc->wildcards = 0; wc->tun_id_mask = htonll(UINT64_MAX); wc->nw_src_mask = htonl(UINT32_MAX); wc->nw_dst_mask = htonl(UINT32_MAX); wc->ipv6_src_mask = in6addr_exact; wc->ipv6_dst_mask = in6addr_exact; + wc->nd_target_mask = in6addr_exact; memset(wc->reg_masks, 0xff, sizeof wc->reg_masks); + wc->metadata_mask = htonll(UINT64_MAX); wc->vlan_tci_mask = htons(UINT16_MAX); - wc->zero = 0; + wc->nw_frag_mask = UINT8_MAX; + wc->tp_src_mask = htons(UINT16_MAX); + wc->tp_dst_mask = htons(UINT16_MAX); + memset(wc->dl_src_mask, 0xff, ETH_ADDR_LEN); + memset(wc->dl_dst_mask, 0xff, ETH_ADDR_LEN); + memset(wc->zeros, 0, sizeof wc->zeros); } /* Returns true if 'wc' is exact-match, false if 'wc' wildcards any bits or @@ -583,18 +645,62 @@ flow_wildcards_is_exact(const struct flow_wildcards *wc) { int i; + BUILD_ASSERT_DECL(FLOW_WC_SEQ == 12); + if (wc->wildcards || wc->tun_id_mask != htonll(UINT64_MAX) || wc->nw_src_mask != htonl(UINT32_MAX) || wc->nw_dst_mask != htonl(UINT32_MAX) + || wc->tp_src_mask != htons(UINT16_MAX) + || wc->tp_dst_mask != htons(UINT16_MAX) || wc->vlan_tci_mask != htons(UINT16_MAX) + || wc->metadata_mask != htonll(UINT64_MAX) + || !eth_mask_is_exact(wc->dl_src_mask) + || !eth_mask_is_exact(wc->dl_dst_mask) || !ipv6_mask_is_exact(&wc->ipv6_src_mask) - || !ipv6_mask_is_exact(&wc->ipv6_dst_mask)) { + || !ipv6_mask_is_exact(&wc->ipv6_dst_mask) + || !ipv6_mask_is_exact(&wc->nd_target_mask) + || wc->nw_frag_mask != UINT8_MAX) { + return false; + } + + for (i = 0; i < FLOW_N_REGS; i++) { + if (wc->reg_masks[i] != UINT32_MAX) { + return false; + } + } + + return true; +} + +/* Returns true if 'wc' matches every packet, false if 'wc' fixes any bits or + * fields. */ +bool +flow_wildcards_is_catchall(const struct flow_wildcards *wc) +{ + int i; + + BUILD_ASSERT_DECL(FLOW_WC_SEQ == 12); + + if (wc->wildcards != FWW_ALL + || wc->tun_id_mask != htonll(0) + || wc->nw_src_mask != htonl(0) + || wc->nw_dst_mask != htonl(0) + || wc->tp_src_mask != htons(0) + || wc->tp_dst_mask != htons(0) + || wc->vlan_tci_mask != htons(0) + || wc->metadata_mask != htonll(0) + || !eth_addr_is_zero(wc->dl_src_mask) + || !eth_addr_is_zero(wc->dl_dst_mask) + || !ipv6_mask_is_any(&wc->ipv6_src_mask) + || !ipv6_mask_is_any(&wc->ipv6_dst_mask) + || !ipv6_mask_is_any(&wc->nd_target_mask) + || wc->nw_frag_mask != 0) { return false; } for (i = 0; i < FLOW_N_REGS; i++) { - if (wc->reg_masks[i] != htonl(UINT32_MAX)) { + if (wc->reg_masks[i] != 0) { return false; } } @@ -612,6 +718,8 @@ flow_wildcards_combine(struct flow_wildcards *dst, { int i; + BUILD_ASSERT_DECL(FLOW_WC_SEQ == 12); + dst->wildcards = src1->wildcards | src2->wildcards; dst->tun_id_mask = src1->tun_id_mask & src2->tun_id_mask; dst->nw_src_mask = src1->nw_src_mask & src2->nw_src_mask; @@ -620,21 +728,28 @@ flow_wildcards_combine(struct flow_wildcards *dst, &src2->ipv6_src_mask); dst->ipv6_dst_mask = ipv6_addr_bitand(&src1->ipv6_dst_mask, &src2->ipv6_dst_mask); + dst->nd_target_mask = ipv6_addr_bitand(&src1->nd_target_mask, + &src2->nd_target_mask); for (i = 0; i < FLOW_N_REGS; i++) { dst->reg_masks[i] = src1->reg_masks[i] & src2->reg_masks[i]; } + dst->metadata_mask = src1->metadata_mask & src2->metadata_mask; dst->vlan_tci_mask = src1->vlan_tci_mask & src2->vlan_tci_mask; + dst->tp_src_mask = src1->tp_src_mask & src2->tp_src_mask; + dst->tp_dst_mask = src1->tp_dst_mask & src2->tp_dst_mask; + eth_addr_bitand(src1->dl_src_mask, src2->dl_src_mask, dst->dl_src_mask); + eth_addr_bitand(src1->dl_dst_mask, src2->dl_dst_mask, dst->dl_dst_mask); } /* Returns a hash of the wildcards in 'wc'. */ uint32_t -flow_wildcards_hash(const struct flow_wildcards *wc) +flow_wildcards_hash(const struct flow_wildcards *wc, uint32_t basis) { /* If you change struct flow_wildcards and thereby trigger this * assertion, please check that the new struct flow_wildcards has no holes * in it before you update the assertion. */ - BUILD_ASSERT_DECL(sizeof *wc == 56 + FLOW_N_REGS * 4); - return hash_bytes(wc, sizeof *wc, 0); + BUILD_ASSERT_DECL(sizeof *wc == 96 + FLOW_N_REGS * 4); + return hash_bytes(wc, sizeof *wc, basis); } /* Returns true if 'a' and 'b' represent the same wildcards, false if they are @@ -645,13 +760,21 @@ flow_wildcards_equal(const struct flow_wildcards *a, { int i; + BUILD_ASSERT_DECL(FLOW_WC_SEQ == 12); + if (a->wildcards != b->wildcards || a->tun_id_mask != b->tun_id_mask || a->nw_src_mask != b->nw_src_mask || a->nw_dst_mask != b->nw_dst_mask - || a->vlan_tci_mask != b->vlan_tci_mask + || a->vlan_tci_mask != b->vlan_tci_mask + || a->metadata_mask != b->metadata_mask || !ipv6_addr_equals(&a->ipv6_src_mask, &b->ipv6_src_mask) - || !ipv6_addr_equals(&a->ipv6_dst_mask, &b->ipv6_dst_mask)) { + || !ipv6_addr_equals(&a->ipv6_dst_mask, &b->ipv6_dst_mask) + || !ipv6_addr_equals(&a->nd_target_mask, &b->nd_target_mask) + || a->tp_src_mask != b->tp_src_mask + || a->tp_dst_mask != b->tp_dst_mask + || !eth_addr_equals(a->dl_src_mask, b->dl_src_mask) + || !eth_addr_equals(a->dl_dst_mask, b->dl_dst_mask)) { return false; } @@ -671,14 +794,27 @@ flow_wildcards_has_extra(const struct flow_wildcards *a, const struct flow_wildcards *b) { int i; + uint8_t eth_masked[ETH_ADDR_LEN]; struct in6_addr ipv6_masked; + BUILD_ASSERT_DECL(FLOW_WC_SEQ == 12); + for (i = 0; i < FLOW_N_REGS; i++) { if ((a->reg_masks[i] & b->reg_masks[i]) != b->reg_masks[i]) { return true; } } + eth_addr_bitand(a->dl_src_mask, b->dl_src_mask, eth_masked); + if (!eth_addr_equals(eth_masked, b->dl_src_mask)) { + return true; + } + + eth_addr_bitand(a->dl_dst_mask, b->dl_dst_mask, eth_masked); + if (!eth_addr_equals(eth_masked, b->dl_dst_mask)) { + return true; + } + ipv6_masked = ipv6_addr_bitand(&a->ipv6_src_mask, &b->ipv6_src_mask); if (!ipv6_addr_equals(&ipv6_masked, &b->ipv6_src_mask)) { return true; @@ -689,71 +825,19 @@ flow_wildcards_has_extra(const struct flow_wildcards *a, return true; } + ipv6_masked = ipv6_addr_bitand(&a->nd_target_mask, &b->nd_target_mask); + if (!ipv6_addr_equals(&ipv6_masked, &b->nd_target_mask)) { + return true; + } + return (a->wildcards & ~b->wildcards || (a->tun_id_mask & b->tun_id_mask) != b->tun_id_mask || (a->nw_src_mask & b->nw_src_mask) != b->nw_src_mask || (a->nw_dst_mask & b->nw_dst_mask) != b->nw_dst_mask - || (a->vlan_tci_mask & b->vlan_tci_mask) != b->vlan_tci_mask); -} - -static bool -set_nw_mask(ovs_be32 *maskp, ovs_be32 mask) -{ - if (ip_is_cidr(mask)) { - *maskp = mask; - return true; - } else { - return false; - } -} - -/* Sets the IP (or ARP) source wildcard mask to CIDR 'mask' (consisting of N - * high-order 1-bit and 32-N low-order 0-bits). Returns true if successful, - * false if 'mask' is not a CIDR mask. */ -bool -flow_wildcards_set_nw_src_mask(struct flow_wildcards *wc, ovs_be32 mask) -{ - return set_nw_mask(&wc->nw_src_mask, mask); -} - -/* Sets the IP (or ARP) destination wildcard mask to CIDR 'mask' (consisting of - * N high-order 1-bit and 32-N low-order 0-bits). Returns true if successful, - * false if 'mask' is not a CIDR mask. */ -bool -flow_wildcards_set_nw_dst_mask(struct flow_wildcards *wc, ovs_be32 mask) -{ - return set_nw_mask(&wc->nw_dst_mask, mask); -} - -static bool -set_ipv6_mask(struct in6_addr *maskp, const struct in6_addr *mask) -{ - if (ipv6_is_cidr(mask)) { - *maskp = *mask; - return true; - } else { - return false; - } -} - -/* Sets the IPv6 source wildcard mask to CIDR 'mask' (consisting of N - * high-order 1-bit and 128-N low-order 0-bits). Returns true if successful, - * false if 'mask' is not a CIDR mask. */ -bool -flow_wildcards_set_ipv6_src_mask(struct flow_wildcards *wc, - const struct in6_addr *mask) -{ - return set_ipv6_mask(&wc->ipv6_src_mask, mask); -} - -/* Sets the IPv6 destination wildcard mask to CIDR 'mask' (consisting of - * N high-order 1-bit and 128-N low-order 0-bits). Returns true if - * successful, false if 'mask' is not a CIDR mask. */ -bool -flow_wildcards_set_ipv6_dst_mask(struct flow_wildcards *wc, - const struct in6_addr *mask) -{ - return set_ipv6_mask(&wc->ipv6_dst_mask, mask); + || (a->vlan_tci_mask & b->vlan_tci_mask) != b->vlan_tci_mask + || (a->metadata_mask & b->metadata_mask) != b->metadata_mask + || (a->tp_src_mask & b->tp_src_mask) != b->tp_src_mask + || (a->tp_dst_mask & b->tp_dst_mask) != b->tp_dst_mask); } /* Sets the wildcard mask for register 'idx' in 'wc' to 'mask'. @@ -775,7 +859,7 @@ flow_hash_symmetric_l4(const struct flow *flow, uint32_t basis) }; ovs_be16 eth_type; ovs_be16 vlan_tci; - ovs_be16 tp_addr; + ovs_be16 tp_port; uint8_t eth_addr[ETH_ADDR_LEN]; uint8_t ip_proto; } fields; @@ -788,11 +872,14 @@ flow_hash_symmetric_l4(const struct flow *flow, uint32_t basis) } fields.vlan_tci = flow->vlan_tci & htons(VLAN_VID_MASK); fields.eth_type = flow->dl_type; + + /* UDP source and destination port are not taken into account because they + * will not necessarily be symmetric in a bidirectional flow. */ if (fields.eth_type == htons(ETH_TYPE_IP)) { fields.ipv4_addr = flow->nw_src ^ flow->nw_dst; fields.ip_proto = flow->nw_proto; - if (fields.ip_proto == IPPROTO_TCP || fields.ip_proto == IPPROTO_UDP) { - fields.tp_addr = flow->tp_src ^ flow->tp_dst; + if (fields.ip_proto == IPPROTO_TCP) { + fields.tp_port = flow->tp_src ^ flow->tp_dst; } } else if (fields.eth_type == htons(ETH_TYPE_IPV6)) { const uint8_t *a = &flow->ipv6_src.s6_addr[0]; @@ -803,9 +890,167 @@ flow_hash_symmetric_l4(const struct flow *flow, uint32_t basis) ipv6_addr[i] = a[i] ^ b[i]; } fields.ip_proto = flow->nw_proto; - if (fields.ip_proto == IPPROTO_TCP || fields.ip_proto == IPPROTO_UDP) { - fields.tp_addr = flow->tp_src ^ flow->tp_dst; + if (fields.ip_proto == IPPROTO_TCP) { + fields.tp_port = flow->tp_src ^ flow->tp_dst; } } return hash_bytes(&fields, sizeof fields, basis); } + +/* Hashes the portions of 'flow' designated by 'fields'. */ +uint32_t +flow_hash_fields(const struct flow *flow, enum nx_hash_fields fields, + uint16_t basis) +{ + switch (fields) { + + case NX_HASH_FIELDS_ETH_SRC: + return hash_bytes(flow->dl_src, sizeof flow->dl_src, basis); + + case NX_HASH_FIELDS_SYMMETRIC_L4: + return flow_hash_symmetric_l4(flow, basis); + } + + NOT_REACHED(); +} + +/* Returns a string representation of 'fields'. */ +const char * +flow_hash_fields_to_str(enum nx_hash_fields fields) +{ + switch (fields) { + case NX_HASH_FIELDS_ETH_SRC: return "eth_src"; + case NX_HASH_FIELDS_SYMMETRIC_L4: return "symmetric_l4"; + default: return ""; + } +} + +/* Returns true if the value of 'fields' is supported. Otherwise false. */ +bool +flow_hash_fields_valid(enum nx_hash_fields fields) +{ + return fields == NX_HASH_FIELDS_ETH_SRC + || fields == NX_HASH_FIELDS_SYMMETRIC_L4; +} + +/* Sets the VLAN VID that 'flow' matches to 'vid', which is interpreted as an + * OpenFlow 1.0 "dl_vlan" value: + * + * - If it is in the range 0...4095, 'flow->vlan_tci' is set to match + * that VLAN. Any existing PCP match is unchanged (it becomes 0 if + * 'flow' previously matched packets without a VLAN header). + * + * - If it is OFP_VLAN_NONE, 'flow->vlan_tci' is set to match a packet + * without a VLAN tag. + * + * - Other values of 'vid' should not be used. */ +void +flow_set_vlan_vid(struct flow *flow, ovs_be16 vid) +{ + if (vid == htons(OFP10_VLAN_NONE)) { + flow->vlan_tci = htons(0); + } else { + vid &= htons(VLAN_VID_MASK); + flow->vlan_tci &= ~htons(VLAN_VID_MASK); + flow->vlan_tci |= htons(VLAN_CFI) | vid; + } +} + +/* Sets the VLAN PCP that 'flow' matches to 'pcp', which should be in the + * range 0...7. + * + * This function has no effect on the VLAN ID that 'flow' matches. + * + * After calling this function, 'flow' will not match packets without a VLAN + * header. */ +void +flow_set_vlan_pcp(struct flow *flow, uint8_t pcp) +{ + pcp &= 0x07; + flow->vlan_tci &= ~htons(VLAN_PCP_MASK); + flow->vlan_tci |= htons((pcp << VLAN_PCP_SHIFT) | VLAN_CFI); +} + +/* Puts into 'b' a packet that flow_extract() would parse as having the given + * 'flow'. + * + * (This is useful only for testing, obviously, and the packet isn't really + * valid. It hasn't got any checksums filled in, for one, and lots of fields + * are just zeroed.) */ +void +flow_compose(struct ofpbuf *b, const struct flow *flow) +{ + eth_compose(b, flow->dl_dst, flow->dl_src, ntohs(flow->dl_type), 0); + if (flow->dl_type == htons(FLOW_DL_TYPE_NONE)) { + struct eth_header *eth = b->l2; + eth->eth_type = htons(b->size); + return; + } + + if (flow->vlan_tci & htons(VLAN_CFI)) { + eth_push_vlan(b, flow->vlan_tci); + } + + if (flow->dl_type == htons(ETH_TYPE_IP)) { + struct ip_header *ip; + + b->l3 = ip = ofpbuf_put_zeros(b, sizeof *ip); + ip->ip_ihl_ver = IP_IHL_VER(5, 4); + ip->ip_tos = flow->nw_tos; + ip->ip_proto = flow->nw_proto; + ip->ip_src = flow->nw_src; + ip->ip_dst = flow->nw_dst; + + if (flow->nw_frag & FLOW_NW_FRAG_ANY) { + ip->ip_frag_off |= htons(IP_MORE_FRAGMENTS); + if (flow->nw_frag & FLOW_NW_FRAG_LATER) { + ip->ip_frag_off |= htons(100); + } + } + if (!(flow->nw_frag & FLOW_NW_FRAG_ANY) + || !(flow->nw_frag & FLOW_NW_FRAG_LATER)) { + if (flow->nw_proto == IPPROTO_TCP) { + struct tcp_header *tcp; + + b->l4 = tcp = ofpbuf_put_zeros(b, sizeof *tcp); + tcp->tcp_src = flow->tp_src; + tcp->tcp_dst = flow->tp_dst; + tcp->tcp_ctl = TCP_CTL(0, 5); + } else if (flow->nw_proto == IPPROTO_UDP) { + struct udp_header *udp; + + b->l4 = udp = ofpbuf_put_zeros(b, sizeof *udp); + udp->udp_src = flow->tp_src; + udp->udp_dst = flow->tp_dst; + } else if (flow->nw_proto == IPPROTO_ICMP) { + struct icmp_header *icmp; + + b->l4 = icmp = ofpbuf_put_zeros(b, sizeof *icmp); + icmp->icmp_type = ntohs(flow->tp_src); + icmp->icmp_code = ntohs(flow->tp_dst); + } + } + + ip->ip_tot_len = htons((uint8_t *) b->data + b->size + - (uint8_t *) b->l3); + } else if (flow->dl_type == htons(ETH_TYPE_IPV6)) { + /* XXX */ + } else if (flow->dl_type == htons(ETH_TYPE_ARP)) { + struct arp_eth_header *arp; + + b->l3 = arp = ofpbuf_put_zeros(b, sizeof *arp); + arp->ar_hrd = htons(1); + arp->ar_pro = htons(ETH_TYPE_IP); + arp->ar_hln = ETH_ADDR_LEN; + arp->ar_pln = 4; + arp->ar_op = htons(flow->nw_proto); + + if (flow->nw_proto == ARP_OP_REQUEST || + flow->nw_proto == ARP_OP_REPLY) { + arp->ar_spa = flow->nw_src; + arp->ar_tpa = flow->nw_dst; + memcpy(arp->ar_sha, flow->arp_sha, ETH_ADDR_LEN); + memcpy(arp->ar_tha, flow->arp_tha, ETH_ADDR_LEN); + } + } +}