X-Git-Url: http://git.onelab.eu/?a=blobdiff_plain;f=lib%2Fflow.c;h=ac8d748c449a074a8c68d32000559ce90dcec0e8;hb=55c955bd8aecc90273ffa93114c1c38f4f9b78ec;hp=5ba3e10df435dfa569e6712d850c59456e3619a9;hpb=e878338bf1409f0dd6d23cb57822a2c4c1a69e0f;p=sliver-openvswitch.git diff --git a/lib/flow.c b/lib/flow.c index 5ba3e10df..ac8d748c4 100644 --- a/lib/flow.c +++ b/lib/flow.c @@ -1,5 +1,5 @@ /* - * Copyright (c) 2008, 2009, 2010, 2011, 2012 Nicira, Inc. + * Copyright (c) 2008, 2009, 2010, 2011, 2012, 2013, 2014 Nicira, Inc. * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. @@ -16,27 +16,39 @@ #include #include #include "flow.h" -#include #include #include +#include #include #include #include +#include #include #include #include "byte-order.h" #include "coverage.h" +#include "csum.h" #include "dynamic-string.h" #include "hash.h" +#include "jhash.h" +#include "match.h" #include "ofpbuf.h" #include "openflow/openflow.h" #include "packets.h" +#include "odp-util.h" +#include "random.h" #include "unaligned.h" -#include "vlog.h" - -VLOG_DEFINE_THIS_MODULE(flow); COVERAGE_DEFINE(flow_extract); +COVERAGE_DEFINE(miniflow_malloc); + +/* U32 indices for segmented flow classification. */ +const uint8_t flow_segment_u32s[4] = { + FLOW_SEGMENT_1_ENDS_AT / 4, + FLOW_SEGMENT_2_ENDS_AT / 4, + FLOW_SEGMENT_3_ENDS_AT / 4, + FLOW_U32S +}; static struct arp_eth_header * pull_arp(struct ofpbuf *packet) @@ -76,6 +88,12 @@ pull_udp(struct ofpbuf *packet) return ofpbuf_try_pull(packet, UDP_HEADER_LEN); } +static struct sctp_header * +pull_sctp(struct ofpbuf *packet) +{ + return ofpbuf_try_pull(packet, SCTP_HEADER_LEN); +} + static struct icmp_header * pull_icmp(struct ofpbuf *packet) { @@ -88,6 +106,22 @@ pull_icmpv6(struct ofpbuf *packet) return ofpbuf_try_pull(packet, sizeof(struct icmp6_hdr)); } +static void +parse_mpls(struct ofpbuf *b, struct flow *flow) +{ + struct mpls_hdr *mh; + int idx = 0; + + while ((mh = ofpbuf_try_pull(b, sizeof *mh))) { + if (idx < FLOW_MAX_MPLS_LABELS) { + flow->mpls_lse[idx++] = mh->mpls_lse; + } + if (mh->mpls_lse & htonl(MPLS_BOS_MASK)) { + break; + } + } +} + static void parse_vlan(struct ofpbuf *b, struct flow *flow) { @@ -127,13 +161,18 @@ parse_ethertype(struct ofpbuf *b) } ofpbuf_pull(b, sizeof *llc); - return llc->snap.snap_type; + + if (ntohs(llc->snap.snap_type) >= ETH_TYPE_MIN) { + return llc->snap.snap_type; + } + + return htons(FLOW_DL_TYPE_NONE); } static int parse_ipv6(struct ofpbuf *packet, struct flow *flow) { - const struct ip6_hdr *nh; + const struct ovs_16aligned_ip6_hdr *nh; ovs_be32 tc_flow; int nexthdr; @@ -144,10 +183,10 @@ parse_ipv6(struct ofpbuf *packet, struct flow *flow) nexthdr = nh->ip6_nxt; - flow->ipv6_src = nh->ip6_src; - flow->ipv6_dst = nh->ip6_dst; + memcpy(&flow->ipv6_src, &nh->ip6_src, sizeof flow->ipv6_src); + memcpy(&flow->ipv6_dst, &nh->ip6_dst, sizeof flow->ipv6_dst); - tc_flow = get_unaligned_be32(&nh->ip6_flow); + tc_flow = get_16aligned_be32(&nh->ip6_flow); flow->nw_tos = ntohl(tc_flow) >> 20; flow->ipv6_label = tc_flow & htonl(IPV6_LABEL_MASK); flow->nw_ttl = nh->ip6_hlim; @@ -179,7 +218,7 @@ 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. */ - const struct ip6_ext *ext_hdr = (struct ip6_ext *)packet->data; + const struct ip6_ext *ext_hdr = packet->data; nexthdr = ext_hdr->ip6e_nxt; if (!ofpbuf_try_pull(packet, (ext_hdr->ip6e_len + 1) * 8)) { return EINVAL; @@ -189,13 +228,13 @@ parse_ipv6(struct ofpbuf *packet, struct flow *flow) * we care about are in the same location as the generic * option header--only the header length is calculated * differently. */ - const struct ip6_ext *ext_hdr = (struct ip6_ext *)packet->data; + const struct ip6_ext *ext_hdr = packet->data; nexthdr = ext_hdr->ip6e_nxt; if (!ofpbuf_try_pull(packet, (ext_hdr->ip6e_len + 2) * 4)) { return EINVAL; } } else if (nexthdr == IPPROTO_FRAGMENT) { - const struct ip6_frag *frag_hdr = (struct ip6_frag *)packet->data; + const struct ovs_16aligned_ip6_frag *frag_hdr = packet->data; nexthdr = frag_hdr->ip6f_nxt; if (!ofpbuf_try_pull(packet, sizeof *frag_hdr)) { @@ -204,9 +243,8 @@ parse_ipv6(struct ofpbuf *packet, struct flow *flow) /* We only process the first fragment. */ 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_ANY; + if ((frag_hdr->ip6f_offlg & IP6F_OFF_MASK) != htons(0)) { flow->nw_frag |= FLOW_NW_FRAG_LATER; nexthdr = IPPROTO_FRAGMENT; break; @@ -226,6 +264,7 @@ parse_tcp(struct ofpbuf *packet, struct ofpbuf *b, struct flow *flow) if (tcp) { flow->tp_src = tcp->tcp_src; flow->tp_dst = tcp->tcp_dst; + flow->tcp_flags = tcp->tcp_ctl & htons(0x0fff); packet->l7 = b->data; } } @@ -241,6 +280,17 @@ parse_udp(struct ofpbuf *packet, struct ofpbuf *b, struct flow *flow) } } +static void +parse_sctp(struct ofpbuf *packet, struct ofpbuf *b, struct flow *flow) +{ + const struct sctp_header *sctp = pull_sctp(b); + if (sctp) { + flow->tp_src = sctp->sctp_src; + flow->tp_dst = sctp->sctp_dst; + packet->l7 = b->data; + } +} + static bool parse_icmpv6(struct ofpbuf *b, struct flow *flow) { @@ -312,13 +362,14 @@ invalid: } -/* Initializes 'flow' members from 'packet', 'skb_priority', 'tun_id', and - * 'ofp_in_port'. +/* Initializes 'flow' members from 'packet' and 'md' * * Initializes 'packet' header pointers as follows: * * - packet->l2 to the start of the Ethernet header. * + * - packet->l2_5 to the start of the MPLS shim header. + * * - packet->l3 to just past the Ethernet header, or just past the * vlan_header if one is present, to the first byte of the payload of the * Ethernet frame. @@ -326,12 +377,12 @@ invalid: * - packet->l4 to just past the IPv4 header, if one is present and has a * correct length, and otherwise NULL. * - * - packet->l7 to just past the TCP or UDP or ICMP header, if one is + * - packet->l7 to just past the TCP/UDP/SCTP/ICMP header, if one is * present and has a correct length, and otherwise NULL. */ void -flow_extract(struct ofpbuf *packet, uint32_t skb_priority, ovs_be64 tun_id, - uint16_t ofp_in_port, struct flow *flow) +flow_extract(struct ofpbuf *packet, const struct pkt_metadata *md, + struct flow *flow) { struct ofpbuf b = *packet; struct eth_header *eth; @@ -339,14 +390,19 @@ flow_extract(struct ofpbuf *packet, uint32_t skb_priority, ovs_be64 tun_id, COVERAGE_INC(flow_extract); memset(flow, 0, sizeof *flow); - flow->tun_id = tun_id; - flow->in_port = ofp_in_port; - flow->skb_priority = skb_priority; - packet->l2 = b.data; - packet->l3 = NULL; - packet->l4 = NULL; - packet->l7 = NULL; + if (md) { + flow->tunnel = md->tunnel; + flow->in_port = md->in_port; + flow->skb_priority = md->skb_priority; + flow->pkt_mark = md->pkt_mark; + } + + packet->l2 = b.data; + packet->l2_5 = NULL; + packet->l3 = NULL; + packet->l4 = NULL; + packet->l7 = NULL; if (b.size < sizeof *eth) { return; @@ -364,6 +420,12 @@ flow_extract(struct ofpbuf *packet, uint32_t skb_priority, ovs_be64 tun_id, } flow->dl_type = parse_ethertype(&b); + /* Parse mpls, copy l3 ttl. */ + if (eth_type_mpls(flow->dl_type)) { + packet->l2_5 = b.data; + parse_mpls(&b, flow); + } + /* Network layer. */ packet->l3 = b.data; if (flow->dl_type == htons(ETH_TYPE_IP)) { @@ -371,8 +433,8 @@ flow_extract(struct ofpbuf *packet, uint32_t skb_priority, ovs_be64 tun_id, 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_src = get_16aligned_be32(&nh->ip_src); + flow->nw_dst = get_16aligned_be32(&nh->ip_dst); flow->nw_proto = nh->ip_proto; flow->nw_tos = nh->ip_tos; @@ -389,6 +451,8 @@ flow_extract(struct ofpbuf *packet, uint32_t skb_priority, ovs_be64 tun_id, parse_tcp(packet, &b, flow); } else if (flow->nw_proto == IPPROTO_UDP) { parse_udp(packet, &b, flow); + } else if (flow->nw_proto == IPPROTO_SCTP) { + parse_sctp(packet, &b, flow); } else if (flow->nw_proto == IPPROTO_ICMP) { const struct icmp_header *icmp = pull_icmp(&b); if (icmp) { @@ -409,12 +473,15 @@ flow_extract(struct ofpbuf *packet, uint32_t skb_priority, ovs_be64 tun_id, parse_tcp(packet, &b, flow); } else if (flow->nw_proto == IPPROTO_UDP) { parse_udp(packet, &b, flow); + } else if (flow->nw_proto == IPPROTO_SCTP) { + parse_sctp(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)) { + } else if (flow->dl_type == htons(ETH_TYPE_ARP) || + flow->dl_type == htons(ETH_TYPE_RARP)) { const struct arp_eth_header *arp = pull_arp(&b); if (arp && arp->ar_hrd == htons(1) && arp->ar_pro == htons(ETH_TYPE_IP) @@ -425,13 +492,10 @@ flow_extract(struct ofpbuf *packet, uint32_t skb_priority, ovs_be64 tun_id, flow->nw_proto = ntohs(arp->ar_op); } - if ((flow->nw_proto == ARP_OP_REQUEST) - || (flow->nw_proto == ARP_OP_REPLY)) { - flow->nw_src = arp->ar_spa; - flow->nw_dst = arp->ar_tpa; - memcpy(flow->arp_sha, arp->ar_sha, ETH_ADDR_LEN); - memcpy(flow->arp_tha, arp->ar_tha, ETH_ADDR_LEN); - } + flow->nw_src = get_16aligned_be32(&arp->ar_spa); + flow->nw_dst = get_16aligned_be32(&arp->ar_tpa); + memcpy(flow->arp_sha, arp->ar_sha, ETH_ADDR_LEN); + memcpy(flow->arp_tha, arp->ar_tha, ETH_ADDR_LEN); } } } @@ -441,70 +505,40 @@ flow_extract(struct ofpbuf *packet, uint32_t skb_priority, ovs_be64 tun_id, void flow_zero_wildcards(struct flow *flow, const struct flow_wildcards *wildcards) { - const flow_wildcards_t wc = wildcards->wildcards; - int i; - - BUILD_ASSERT_DECL(FLOW_WC_SEQ == 14); + uint32_t *flow_u32 = (uint32_t *) flow; + const uint32_t *wc_u32 = (const uint32_t *) &wildcards->masks; + size_t i; - 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); + for (i = 0; i < FLOW_U32S; i++) { + flow_u32[i] &= wc_u32[i]; } - 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; - } - flow->ipv6_label &= wildcards->ipv6_label_mask; - 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; +} + +void +flow_unwildcard_tp_ports(const struct flow *flow, struct flow_wildcards *wc) +{ + if (flow->nw_proto != IPPROTO_ICMP) { + memset(&wc->masks.tp_src, 0xff, sizeof wc->masks.tp_src); + memset(&wc->masks.tp_dst, 0xff, sizeof wc->masks.tp_dst); + } else { + wc->masks.tp_src = htons(0xff); + wc->masks.tp_dst = htons(0xff); } - flow->nw_frag &= wildcards->nw_frag_mask; - eth_addr_bitand(flow->arp_sha, wildcards->arp_sha_mask, flow->arp_sha); - eth_addr_bitand(flow->arp_tha, wildcards->arp_tha_mask, 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 == 14); - - fmd->tun_id = flow->tun_id; - fmd->tun_id_mask = htonll(UINT64_MAX); + BUILD_ASSERT_DECL(FLOW_WC_SEQ == 24); + fmd->tun_id = flow->tunnel.tun_id; + fmd->tun_src = flow->tunnel.ip_src; + fmd->tun_dst = flow->tunnel.ip_dst; 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; + fmd->pkt_mark = flow->pkt_mark; + fmd->in_port = flow->in_port.ofp_port; } char * @@ -515,64 +549,77 @@ flow_to_string(const struct flow *flow) return ds_cstr(&ds); } +const char * +flow_tun_flag_to_string(uint32_t flags) +{ + switch (flags) { + case FLOW_TNL_F_DONT_FRAGMENT: + return "df"; + case FLOW_TNL_F_CSUM: + return "csum"; + case FLOW_TNL_F_KEY: + return "key"; + default: + return NULL; + } +} + void -flow_format(struct ds *ds, const struct flow *flow) +format_flags(struct ds *ds, const char *(*bit_to_string)(uint32_t), + uint32_t flags, char del) { - 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", - 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, - 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, " 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:%#"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)); + uint32_t bad = 0; + + if (!flags) { + return; } - 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" : ""); + while (flags) { + uint32_t bit = rightmost_1bit(flags); + const char *s; + + s = bit_to_string(bit); + if (s) { + ds_put_format(ds, "%s%c", s, del); + } else { + bad |= bit; + } + + flags &= ~bit; } - if (flow->tp_src || flow->tp_dst) { - ds_put_format(ds, " port(%"PRIu16"->%"PRIu16")", - ntohs(flow->tp_src), ntohs(flow->tp_dst)); + + if (bad) { + ds_put_format(ds, "0x%"PRIx32"%c", bad, del); + } + ds_chomp(ds, del); +} + +void +format_flags_masked(struct ds *ds, const char *name, + const char *(*bit_to_string)(uint32_t), uint32_t flags, + uint32_t mask) +{ + if (name) { + ds_put_format(ds, "%s=", name); } - 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")", - ETH_ADDR_ARGS(flow->arp_sha), - ETH_ADDR_ARGS(flow->arp_tha)); + while (mask) { + uint32_t bit = rightmost_1bit(mask); + const char *s = bit_to_string(bit); + + ds_put_format(ds, "%s%s", (flags & bit) ? "+" : "-", + s ? s : "[Unknown]"); + mask &= ~bit; } } +void +flow_format(struct ds *ds, const struct flow *flow) +{ + struct match match; + + match_wc_init(&match, flow); + match_format(&match, ds, OFP_DEFAULT_PRIORITY); +} + void flow_print(FILE *stream, const struct flow *flow) { @@ -587,178 +634,134 @@ flow_print(FILE *stream, const struct flow *flow) void flow_wildcards_init_catchall(struct flow_wildcards *wc) { - BUILD_ASSERT_DECL(FLOW_WC_SEQ == 14); - - 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->ipv6_label_mask = htonl(0); - 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->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->arp_sha_mask, 0, ETH_ADDR_LEN); - memset(wc->arp_tha_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 - * wildcard any bits or fields. */ + memset(&wc->masks, 0, sizeof wc->masks); +} + +/* Clear the metadata and register wildcard masks. They are not packet + * header fields. */ void -flow_wildcards_init_exact(struct flow_wildcards *wc) -{ - BUILD_ASSERT_DECL(FLOW_WC_SEQ == 14); - - 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->ipv6_label_mask = htonl(UINT32_MAX); - 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->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->arp_sha_mask, 0xff, ETH_ADDR_LEN); - memset(wc->arp_tha_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 +flow_wildcards_clear_non_packet_fields(struct flow_wildcards *wc) +{ + memset(&wc->masks.metadata, 0, sizeof wc->masks.metadata); + memset(&wc->masks.regs, 0, sizeof wc->masks.regs); +} + +/* Returns true if 'wc' matches every packet, false if 'wc' fixes any bits or * fields. */ bool -flow_wildcards_is_exact(const struct flow_wildcards *wc) +flow_wildcards_is_catchall(const struct flow_wildcards *wc) { - int i; - - BUILD_ASSERT_DECL(FLOW_WC_SEQ == 14); - - 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) - || !eth_mask_is_exact(wc->arp_sha_mask) - || !eth_mask_is_exact(wc->arp_tha_mask) - || !ipv6_mask_is_exact(&wc->ipv6_src_mask) - || !ipv6_mask_is_exact(&wc->ipv6_dst_mask) - || wc->ipv6_label_mask != htonl(UINT32_MAX) - || !ipv6_mask_is_exact(&wc->nd_target_mask) - || wc->nw_frag_mask != UINT8_MAX) { - return false; - } + const uint32_t *wc_u32 = (const uint32_t *) &wc->masks; + size_t i; - for (i = 0; i < FLOW_N_REGS; i++) { - if (wc->reg_masks[i] != UINT32_MAX) { + for (i = 0; i < FLOW_U32S; i++) { + if (wc_u32[i]) { 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) +/* Sets 'dst' as the bitwise AND of wildcards in 'src1' and 'src2'. + * That is, a bit or a field is wildcarded in 'dst' if it is wildcarded + * in 'src1' or 'src2' or both. */ +void +flow_wildcards_and(struct flow_wildcards *dst, + const struct flow_wildcards *src1, + const struct flow_wildcards *src2) { - int i; + uint32_t *dst_u32 = (uint32_t *) &dst->masks; + const uint32_t *src1_u32 = (const uint32_t *) &src1->masks; + const uint32_t *src2_u32 = (const uint32_t *) &src2->masks; + size_t i; - BUILD_ASSERT_DECL(FLOW_WC_SEQ == 14); - - 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) - || !eth_addr_is_zero(wc->arp_sha_mask) - || !eth_addr_is_zero(wc->arp_tha_mask) - || !ipv6_mask_is_any(&wc->ipv6_src_mask) - || !ipv6_mask_is_any(&wc->ipv6_dst_mask) - || wc->ipv6_label_mask != htonl(0) - || !ipv6_mask_is_any(&wc->nd_target_mask) - || wc->nw_frag_mask != 0) { - return false; + for (i = 0; i < FLOW_U32S; i++) { + dst_u32[i] = src1_u32[i] & src2_u32[i]; } +} - for (i = 0; i < FLOW_N_REGS; i++) { - if (wc->reg_masks[i] != 0) { - return false; - } +/* Sets 'dst' as the bitwise OR of wildcards in 'src1' and 'src2'. That + * is, a bit or a field is wildcarded in 'dst' if it is neither + * wildcarded in 'src1' nor 'src2'. */ +void +flow_wildcards_or(struct flow_wildcards *dst, + const struct flow_wildcards *src1, + const struct flow_wildcards *src2) +{ + uint32_t *dst_u32 = (uint32_t *) &dst->masks; + const uint32_t *src1_u32 = (const uint32_t *) &src1->masks; + const uint32_t *src2_u32 = (const uint32_t *) &src2->masks; + size_t i; + + for (i = 0; i < FLOW_U32S; i++) { + dst_u32[i] = src1_u32[i] | src2_u32[i]; } +} - return true; +/* Perform a bitwise OR of miniflow 'src' flow data with the equivalent + * fields in 'dst', storing the result in 'dst'. */ +static void +flow_union_with_miniflow(struct flow *dst, const struct miniflow *src) +{ + uint32_t *dst_u32 = (uint32_t *) dst; + const uint32_t *p = src->values; + uint64_t map; + + for (map = src->map; map; map = zero_rightmost_1bit(map)) { + dst_u32[raw_ctz(map)] |= *p++; + } } -/* Initializes 'dst' as the combination of wildcards in 'src1' and 'src2'. - * That is, a bit or a field is wildcarded in 'dst' if it is wildcarded in - * 'src1' or 'src2' or both. */ +/* Fold minimask 'mask''s wildcard mask into 'wc's wildcard mask. */ void -flow_wildcards_combine(struct flow_wildcards *dst, - const struct flow_wildcards *src1, - const struct flow_wildcards *src2) +flow_wildcards_fold_minimask(struct flow_wildcards *wc, + const struct minimask *mask) { - int i; + flow_union_with_miniflow(&wc->masks, &mask->masks); +} + +uint64_t +miniflow_get_map_in_range(const struct miniflow *miniflow, + uint8_t start, uint8_t end, unsigned int *offset) +{ + uint64_t map = miniflow->map; + *offset = 0; - BUILD_ASSERT_DECL(FLOW_WC_SEQ == 14); - - 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; - dst->nw_dst_mask = src1->nw_dst_mask & src2->nw_dst_mask; - dst->ipv6_src_mask = ipv6_addr_bitand(&src1->ipv6_src_mask, - &src2->ipv6_src_mask); - dst->ipv6_dst_mask = ipv6_addr_bitand(&src1->ipv6_dst_mask, - &src2->ipv6_dst_mask); - dst->ipv6_label_mask = src1->ipv6_label_mask & src2->ipv6_label_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); - eth_addr_bitand(src1->arp_sha_mask, src2->arp_sha_mask, dst->arp_sha_mask); - eth_addr_bitand(src1->arp_tha_mask, src2->arp_tha_mask, dst->arp_tha_mask); + if (start > 0) { + uint64_t msk = (UINT64_C(1) << start) - 1; /* 'start' LSBs set */ + *offset = count_1bits(map & msk); + map &= ~msk; + } + if (end < FLOW_U32S) { + uint64_t msk = (UINT64_C(1) << end) - 1; /* 'end' LSBs set */ + map &= msk; + } + return map; +} + +/* Fold minimask 'mask''s wildcard mask into 'wc's wildcard mask + * in range [start, end). */ +void +flow_wildcards_fold_minimask_range(struct flow_wildcards *wc, + const struct minimask *mask, + uint8_t start, uint8_t end) +{ + uint32_t *dst_u32 = (uint32_t *)&wc->masks; + unsigned int offset; + uint64_t map = miniflow_get_map_in_range(&mask->masks, start, end, + &offset); + const uint32_t *p = mask->masks.values + offset; + + for (; map; map = zero_rightmost_1bit(map)) { + dst_u32[raw_ctz(map)] |= *p++; + } } /* Returns a hash of the wildcards in 'wc'. */ uint32_t 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 == 112 + FLOW_N_REGS * 4); - return hash_bytes(wc, sizeof *wc, basis); + return flow_hash(&wc->masks, basis); } /* Returns true if 'a' and 'b' represent the same wildcards, false if they are @@ -767,36 +770,7 @@ bool flow_wildcards_equal(const struct flow_wildcards *a, const struct flow_wildcards *b) { - int i; - - BUILD_ASSERT_DECL(FLOW_WC_SEQ == 14); - - 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->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) - || a->ipv6_label_mask != b->ipv6_label_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) - || !eth_addr_equals(a->arp_sha_mask, b->arp_sha_mask) - || !eth_addr_equals(a->arp_tha_mask, b->arp_tha_mask)) { - return false; - } - - for (i = 0; i < FLOW_N_REGS; i++) { - if (a->reg_masks[i] != b->reg_masks[i]) { - return false; - } - } - - return true; + return flow_equal(&a->masks, &b->masks); } /* Returns true if at least one bit or field is wildcarded in 'a' but not in @@ -805,62 +779,35 @@ bool 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; + const uint32_t *a_u32 = (const uint32_t *) &a->masks; + const uint32_t *b_u32 = (const uint32_t *) &b->masks; + size_t i; - BUILD_ASSERT_DECL(FLOW_WC_SEQ == 14); - - for (i = 0; i < FLOW_N_REGS; i++) { - if ((a->reg_masks[i] & b->reg_masks[i]) != b->reg_masks[i]) { + for (i = 0; i < FLOW_U32S; i++) { + if ((a_u32[i] & b_u32[i]) != b_u32[i]) { return true; } } + return false; +} - 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; - } - - eth_addr_bitand(a->arp_sha_mask, b->arp_sha_mask, eth_masked); - if (!eth_addr_equals(eth_masked, b->arp_sha_mask)) { - return true; - } - - eth_addr_bitand(a->arp_tha_mask, b->arp_tha_mask, eth_masked); - if (!eth_addr_equals(eth_masked, b->arp_tha_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; - } - - ipv6_masked = ipv6_addr_bitand(&a->ipv6_dst_mask, &b->ipv6_dst_mask); - if (!ipv6_addr_equals(&ipv6_masked, &b->ipv6_dst_mask)) { - return true; - } +/* Returns true if 'a' and 'b' are equal, except that 0-bits (wildcarded bits) + * in 'wc' do not need to be equal in 'a' and 'b'. */ +bool +flow_equal_except(const struct flow *a, const struct flow *b, + const struct flow_wildcards *wc) +{ + const uint32_t *a_u32 = (const uint32_t *) a; + const uint32_t *b_u32 = (const uint32_t *) b; + const uint32_t *wc_u32 = (const uint32_t *) &wc->masks; + size_t i; - 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; + for (i = 0; i < FLOW_U32S; i++) { + if ((a_u32[i] ^ b_u32[i]) & wc_u32[i]) { + return false; + } } - - 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->ipv6_label_mask & b->ipv6_label_mask) != b->ipv6_label_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); + return true; } /* Sets the wildcard mask for register 'idx' in 'wc' to 'mask'. @@ -868,7 +815,26 @@ flow_wildcards_has_extra(const struct flow_wildcards *a, void flow_wildcards_set_reg_mask(struct flow_wildcards *wc, int idx, uint32_t mask) { - wc->reg_masks[idx] = mask; + wc->masks.regs[idx] = mask; +} + +/* Calculates the 5-tuple hash from the given flow. */ +uint32_t +flow_hash_5tuple(const struct flow *flow, uint32_t basis) +{ + uint32_t hash = 0; + + if (!flow) { + return 0; + } + + hash = mhash_add(basis, (OVS_FORCE uint32_t) flow->nw_src); + hash = mhash_add(hash, (OVS_FORCE uint32_t) flow->nw_dst); + hash = mhash_add(hash, ((OVS_FORCE uint32_t) flow->tp_src << 16) + | (OVS_FORCE uint32_t) flow->tp_dst); + hash = mhash_add(hash, flow->nw_proto); + + return mhash_finish(hash, 13); } /* Hashes 'flow' based on its L2 through L4 protocol information. */ @@ -901,7 +867,7 @@ flow_hash_symmetric_l4(const struct flow *flow, uint32_t basis) 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) { + if (fields.ip_proto == IPPROTO_TCP || fields.ip_proto == IPPROTO_SCTP) { fields.tp_port = flow->tp_src ^ flow->tp_dst; } } else if (fields.eth_type == htons(ETH_TYPE_IPV6)) { @@ -913,11 +879,83 @@ 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) { + if (fields.ip_proto == IPPROTO_TCP || fields.ip_proto == IPPROTO_SCTP) { fields.tp_port = flow->tp_src ^ flow->tp_dst; } } - return hash_bytes(&fields, sizeof fields, basis); + return jhash_bytes(&fields, sizeof fields, basis); +} + +/* Initialize a flow with random fields that matter for nx_hash_fields. */ +void +flow_random_hash_fields(struct flow *flow) +{ + uint16_t rnd = random_uint16(); + + /* Initialize to all zeros. */ + memset(flow, 0, sizeof *flow); + + eth_addr_random(flow->dl_src); + eth_addr_random(flow->dl_dst); + + flow->vlan_tci = (OVS_FORCE ovs_be16) (random_uint16() & VLAN_VID_MASK); + + /* Make most of the random flows IPv4, some IPv6, and rest random. */ + flow->dl_type = rnd < 0x8000 ? htons(ETH_TYPE_IP) : + rnd < 0xc000 ? htons(ETH_TYPE_IPV6) : (OVS_FORCE ovs_be16)rnd; + + if (dl_type_is_ip_any(flow->dl_type)) { + if (flow->dl_type == htons(ETH_TYPE_IP)) { + flow->nw_src = (OVS_FORCE ovs_be32)random_uint32(); + flow->nw_dst = (OVS_FORCE ovs_be32)random_uint32(); + } else { + random_bytes(&flow->ipv6_src, sizeof flow->ipv6_src); + random_bytes(&flow->ipv6_dst, sizeof flow->ipv6_dst); + } + /* Make most of IP flows TCP, some UDP or SCTP, and rest random. */ + rnd = random_uint16(); + flow->nw_proto = rnd < 0x8000 ? IPPROTO_TCP : + rnd < 0xc000 ? IPPROTO_UDP : + rnd < 0xd000 ? IPPROTO_SCTP : (uint8_t)rnd; + if (flow->nw_proto == IPPROTO_TCP || + flow->nw_proto == IPPROTO_UDP || + flow->nw_proto == IPPROTO_SCTP) { + flow->tp_src = (OVS_FORCE ovs_be16)random_uint16(); + flow->tp_dst = (OVS_FORCE ovs_be16)random_uint16(); + } + } +} + +/* Masks the fields in 'wc' that are used by the flow hash 'fields'. */ +void +flow_mask_hash_fields(const struct flow *flow, struct flow_wildcards *wc, + enum nx_hash_fields fields) +{ + switch (fields) { + case NX_HASH_FIELDS_ETH_SRC: + memset(&wc->masks.dl_src, 0xff, sizeof wc->masks.dl_src); + break; + + case NX_HASH_FIELDS_SYMMETRIC_L4: + memset(&wc->masks.dl_src, 0xff, sizeof wc->masks.dl_src); + memset(&wc->masks.dl_dst, 0xff, sizeof wc->masks.dl_dst); + if (flow->dl_type == htons(ETH_TYPE_IP)) { + memset(&wc->masks.nw_src, 0xff, sizeof wc->masks.nw_src); + memset(&wc->masks.nw_dst, 0xff, sizeof wc->masks.nw_dst); + } else if (flow->dl_type == htons(ETH_TYPE_IPV6)) { + memset(&wc->masks.ipv6_src, 0xff, sizeof wc->masks.ipv6_src); + memset(&wc->masks.ipv6_dst, 0xff, sizeof wc->masks.ipv6_dst); + } + if (is_ip_any(flow)) { + memset(&wc->masks.nw_proto, 0xff, sizeof wc->masks.nw_proto); + flow_unwildcard_tp_ports(flow, wc); + } + wc->masks.vlan_tci |= htons(VLAN_VID_MASK | VLAN_CFI); + break; + + default: + OVS_NOT_REACHED(); + } } /* Hashes the portions of 'flow' designated by 'fields'. */ @@ -928,13 +966,13 @@ flow_hash_fields(const struct flow *flow, enum nx_hash_fields fields, switch (fields) { case NX_HASH_FIELDS_ETH_SRC: - return hash_bytes(flow->dl_src, sizeof flow->dl_src, basis); + return jhash_bytes(flow->dl_src, sizeof flow->dl_src, basis); case NX_HASH_FIELDS_SYMMETRIC_L4: return flow_hash_symmetric_l4(flow, basis); } - NOT_REACHED(); + OVS_NOT_REACHED(); } /* Returns a string representation of 'fields'. */ @@ -956,6 +994,24 @@ flow_hash_fields_valid(enum nx_hash_fields fields) || fields == NX_HASH_FIELDS_SYMMETRIC_L4; } +/* Returns a hash value for the bits of 'flow' that are active based on + * 'wc', given 'basis'. */ +uint32_t +flow_hash_in_wildcards(const struct flow *flow, + const struct flow_wildcards *wc, uint32_t basis) +{ + const uint32_t *wc_u32 = (const uint32_t *) &wc->masks; + const uint32_t *flow_u32 = (const uint32_t *) flow; + uint32_t hash; + size_t i; + + hash = basis; + for (i = 0; i < FLOW_U32S; i++) { + hash = mhash_add(hash, flow_u32[i] & wc_u32[i]); + } + return mhash_finish(hash, 4 * FLOW_U32S); +} + /* Sets the VLAN VID that 'flow' matches to 'vid', which is interpreted as an * OpenFlow 1.0 "dl_vlan" value: * @@ -968,7 +1024,7 @@ flow_hash_fields_valid(enum nx_hash_fields fields) * * - Other values of 'vid' should not be used. */ void -flow_set_vlan_vid(struct flow *flow, ovs_be16 vid) +flow_set_dl_vlan(struct flow *flow, ovs_be16 vid) { if (vid == htons(OFP10_VLAN_NONE)) { flow->vlan_tci = htons(0); @@ -979,6 +1035,17 @@ flow_set_vlan_vid(struct flow *flow, ovs_be16 vid) } } +/* Sets the VLAN VID that 'flow' matches to 'vid', which is interpreted as an + * OpenFlow 1.2 "vlan_vid" value, that is, the low 13 bits of 'vlan_tci' (VID + * plus CFI). */ +void +flow_set_vlan_vid(struct flow *flow, ovs_be16 vid) +{ + ovs_be16 mask = htons(VLAN_VID_MASK | VLAN_CFI); + flow->vlan_tci &= ~mask; + flow->vlan_tci |= vid & mask; +} + /* Sets the VLAN PCP that 'flow' matches to 'pcp', which should be in the * range 0...7. * @@ -994,15 +1061,294 @@ flow_set_vlan_pcp(struct flow *flow, uint8_t pcp) flow->vlan_tci |= htons((pcp << VLAN_PCP_SHIFT) | VLAN_CFI); } +/* Returns the number of MPLS LSEs present in 'flow' + * + * Returns 0 if the 'dl_type' of 'flow' is not an MPLS ethernet type. + * Otherwise traverses 'flow''s MPLS label stack stopping at the + * first entry that has the BoS bit set. If no such entry exists then + * the maximum number of LSEs that can be stored in 'flow' is returned. + */ +int +flow_count_mpls_labels(const struct flow *flow, struct flow_wildcards *wc) +{ + if (wc) { + wc->masks.dl_type = OVS_BE16_MAX; + } + if (eth_type_mpls(flow->dl_type)) { + int i; + int len = FLOW_MAX_MPLS_LABELS; + + for (i = 0; i < len; i++) { + if (wc) { + wc->masks.mpls_lse[i] |= htonl(MPLS_BOS_MASK); + } + if (flow->mpls_lse[i] & htonl(MPLS_BOS_MASK)) { + return i + 1; + } + } + + return len; + } else { + return 0; + } +} + +/* Returns the number consecutive of MPLS LSEs, starting at the + * innermost LSE, that are common in 'a' and 'b'. + * + * 'an' must be flow_count_mpls_labels(a). + * 'bn' must be flow_count_mpls_labels(b). + */ +int +flow_count_common_mpls_labels(const struct flow *a, int an, + const struct flow *b, int bn, + struct flow_wildcards *wc) +{ + int min_n = MIN(an, bn); + if (min_n == 0) { + return 0; + } else { + int common_n = 0; + int a_last = an - 1; + int b_last = bn - 1; + int i; + + for (i = 0; i < min_n; i++) { + if (wc) { + wc->masks.mpls_lse[a_last - i] = OVS_BE32_MAX; + wc->masks.mpls_lse[b_last - i] = OVS_BE32_MAX; + } + if (a->mpls_lse[a_last - i] != b->mpls_lse[b_last - i]) { + break; + } else { + common_n++; + } + } + + return common_n; + } +} + +/* Adds a new outermost MPLS label to 'flow' and changes 'flow''s Ethernet type + * to 'mpls_eth_type', which must be an MPLS Ethertype. + * + * If the new label is the first MPLS label in 'flow', it is generated as; + * + * - label: 2, if 'flow' is IPv6, otherwise 0. + * + * - TTL: IPv4 or IPv6 TTL, if present and nonzero, otherwise 64. + * + * - TC: IPv4 or IPv6 TOS, if present, otherwise 0. + * + * - BoS: 1. + * + * If the new label is the second or label MPLS label in 'flow', it is + * generated as; + * + * - label: Copied from outer label. + * + * - TTL: Copied from outer label. + * + * - TC: Copied from outer label. + * + * - BoS: 0. + * + * 'n' must be flow_count_mpls_labels(flow). 'n' must be less than + * FLOW_MAX_MPLS_LABELS (because otherwise flow->mpls_lse[] would overflow). + */ +void +flow_push_mpls(struct flow *flow, int n, ovs_be16 mpls_eth_type, + struct flow_wildcards *wc) +{ + ovs_assert(eth_type_mpls(mpls_eth_type)); + ovs_assert(n < FLOW_MAX_MPLS_LABELS); + + memset(wc->masks.mpls_lse, 0xff, sizeof wc->masks.mpls_lse); + if (n) { + int i; + + for (i = n; i >= 1; i--) { + flow->mpls_lse[i] = flow->mpls_lse[i - 1]; + } + flow->mpls_lse[0] = (flow->mpls_lse[1] + & htonl(~MPLS_BOS_MASK)); + } else { + int label = 0; /* IPv4 Explicit Null. */ + int tc = 0; + int ttl = 64; + + if (flow->dl_type == htons(ETH_TYPE_IPV6)) { + label = 2; + } + + if (is_ip_any(flow)) { + tc = (flow->nw_tos & IP_DSCP_MASK) >> 2; + wc->masks.nw_tos |= IP_DSCP_MASK; + + if (flow->nw_ttl) { + ttl = flow->nw_ttl; + } + wc->masks.nw_ttl = 0xff; + } + + flow->mpls_lse[0] = set_mpls_lse_values(ttl, tc, 1, htonl(label)); + + /* Clear all L3 and L4 fields. */ + BUILD_ASSERT(FLOW_WC_SEQ == 24); + memset((char *) flow + FLOW_SEGMENT_2_ENDS_AT, 0, + sizeof(struct flow) - FLOW_SEGMENT_2_ENDS_AT); + } + flow->dl_type = mpls_eth_type; +} + +/* Tries to remove the outermost MPLS label from 'flow'. Returns true if + * successful, false otherwise. On success, sets 'flow''s Ethernet type to + * 'eth_type'. + * + * 'n' must be flow_count_mpls_labels(flow). */ +bool +flow_pop_mpls(struct flow *flow, int n, ovs_be16 eth_type, + struct flow_wildcards *wc) +{ + int i; + + if (n == 0) { + /* Nothing to pop. */ + return false; + } else if (n == FLOW_MAX_MPLS_LABELS + && !(flow->mpls_lse[n - 1] & htonl(MPLS_BOS_MASK))) { + /* Can't pop because we don't know what to fill in mpls_lse[n - 1]. */ + return false; + } + + memset(wc->masks.mpls_lse, 0xff, sizeof wc->masks.mpls_lse); + for (i = 1; i < n; i++) { + flow->mpls_lse[i - 1] = flow->mpls_lse[i]; + } + flow->mpls_lse[n - 1] = 0; + flow->dl_type = eth_type; + return true; +} + +/* Sets the MPLS Label that 'flow' matches to 'label', which is interpreted + * as an OpenFlow 1.1 "mpls_label" value. */ +void +flow_set_mpls_label(struct flow *flow, int idx, ovs_be32 label) +{ + set_mpls_lse_label(&flow->mpls_lse[idx], label); +} + +/* Sets the MPLS TTL that 'flow' matches to 'ttl', which should be in the + * range 0...255. */ +void +flow_set_mpls_ttl(struct flow *flow, int idx, uint8_t ttl) +{ + set_mpls_lse_ttl(&flow->mpls_lse[idx], ttl); +} + +/* Sets the MPLS TC that 'flow' matches to 'tc', which should be in the + * range 0...7. */ +void +flow_set_mpls_tc(struct flow *flow, int idx, uint8_t tc) +{ + set_mpls_lse_tc(&flow->mpls_lse[idx], tc); +} + +/* Sets the MPLS BOS bit that 'flow' matches to which should be 0 or 1. */ +void +flow_set_mpls_bos(struct flow *flow, int idx, uint8_t bos) +{ + set_mpls_lse_bos(&flow->mpls_lse[idx], bos); +} + +/* Sets the entire MPLS LSE. */ +void +flow_set_mpls_lse(struct flow *flow, int idx, ovs_be32 lse) +{ + flow->mpls_lse[idx] = lse; +} + +static void +flow_compose_l4(struct ofpbuf *b, const struct flow *flow) +{ + 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; + + tcp = ofpbuf_put_zeros(b, sizeof *tcp); + tcp->tcp_src = flow->tp_src; + tcp->tcp_dst = flow->tp_dst; + tcp->tcp_ctl = TCP_CTL(ntohs(flow->tcp_flags), 5); + b->l7 = ofpbuf_tail(b); + } else if (flow->nw_proto == IPPROTO_UDP) { + struct udp_header *udp; + + udp = ofpbuf_put_zeros(b, sizeof *udp); + udp->udp_src = flow->tp_src; + udp->udp_dst = flow->tp_dst; + b->l7 = ofpbuf_tail(b); + } else if (flow->nw_proto == IPPROTO_SCTP) { + struct sctp_header *sctp; + + sctp = ofpbuf_put_zeros(b, sizeof *sctp); + sctp->sctp_src = flow->tp_src; + sctp->sctp_dst = flow->tp_dst; + b->l7 = ofpbuf_tail(b); + } else if (flow->nw_proto == IPPROTO_ICMP) { + struct icmp_header *icmp; + + icmp = ofpbuf_put_zeros(b, sizeof *icmp); + icmp->icmp_type = ntohs(flow->tp_src); + icmp->icmp_code = ntohs(flow->tp_dst); + icmp->icmp_csum = csum(icmp, ICMP_HEADER_LEN); + b->l7 = ofpbuf_tail(b); + } else if (flow->nw_proto == IPPROTO_ICMPV6) { + struct icmp6_hdr *icmp; + + icmp = ofpbuf_put_zeros(b, sizeof *icmp); + icmp->icmp6_type = ntohs(flow->tp_src); + icmp->icmp6_code = ntohs(flow->tp_dst); + + if (icmp->icmp6_code == 0 && + (icmp->icmp6_type == ND_NEIGHBOR_SOLICIT || + icmp->icmp6_type == ND_NEIGHBOR_ADVERT)) { + struct in6_addr *nd_target; + struct nd_opt_hdr *nd_opt; + + nd_target = ofpbuf_put_zeros(b, sizeof *nd_target); + *nd_target = flow->nd_target; + + if (!eth_addr_is_zero(flow->arp_sha)) { + nd_opt = ofpbuf_put_zeros(b, 8); + nd_opt->nd_opt_len = 1; + nd_opt->nd_opt_type = ND_OPT_SOURCE_LINKADDR; + memcpy(nd_opt + 1, flow->arp_sha, ETH_ADDR_LEN); + } + if (!eth_addr_is_zero(flow->arp_tha)) { + nd_opt = ofpbuf_put_zeros(b, 8); + nd_opt->nd_opt_len = 1; + nd_opt->nd_opt_type = ND_OPT_TARGET_LINKADDR; + memcpy(nd_opt + 1, flow->arp_tha, ETH_ADDR_LEN); + } + } + icmp->icmp6_cksum = (OVS_FORCE uint16_t) + csum(icmp, (char *)ofpbuf_tail(b) - (char *)icmp); + b->l7 = ofpbuf_tail(b); + } + } +} + /* 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 + * valid. It hasn't got some checksums filled in, for one, and lots of fields * are just zeroed.) */ void flow_compose(struct ofpbuf *b, const struct flow *flow) { + /* eth_compose() sets l3 pointer and makes sure it is 32-bit aligned. */ 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; @@ -1011,18 +1357,19 @@ flow_compose(struct ofpbuf *b, const struct flow *flow) } if (flow->vlan_tci & htons(VLAN_CFI)) { - eth_push_vlan(b, flow->vlan_tci); + eth_push_vlan(b, htons(ETH_TYPE_VLAN), 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 = ofpbuf_put_zeros(b, sizeof *ip); ip->ip_ihl_ver = IP_IHL_VER(5, 4); ip->ip_tos = flow->nw_tos; + ip->ip_ttl = flow->nw_ttl; ip->ip_proto = flow->nw_proto; - ip->ip_src = flow->nw_src; - ip->ip_dst = flow->nw_dst; + put_16aligned_be32(&ip->ip_src, flow->nw_src); + put_16aligned_be32(&ip->ip_dst, flow->nw_dst); if (flow->nw_frag & FLOW_NW_FRAG_ANY) { ip->ip_frag_off |= htons(IP_MORE_FRAGMENTS); @@ -1030,35 +1377,34 @@ flow_compose(struct ofpbuf *b, const struct flow *flow) 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); - } - } + + b->l4 = ofpbuf_tail(b); + + flow_compose_l4(b, flow); ip->ip_tot_len = htons((uint8_t *) b->data + b->size - (uint8_t *) b->l3); + ip->ip_csum = csum(ip, sizeof *ip); } else if (flow->dl_type == htons(ETH_TYPE_IPV6)) { - /* XXX */ - } else if (flow->dl_type == htons(ETH_TYPE_ARP)) { + struct ovs_16aligned_ip6_hdr *nh; + + nh = ofpbuf_put_zeros(b, sizeof *nh); + put_16aligned_be32(&nh->ip6_flow, htonl(6 << 28) | + htonl(flow->nw_tos << 20) | flow->ipv6_label); + nh->ip6_hlim = flow->nw_ttl; + nh->ip6_nxt = flow->nw_proto; + + memcpy(&nh->ip6_src, &flow->ipv6_src, sizeof(nh->ip6_src)); + memcpy(&nh->ip6_dst, &flow->ipv6_dst, sizeof(nh->ip6_dst)); + + b->l4 = ofpbuf_tail(b); + + flow_compose_l4(b, flow); + + nh->ip6_plen = + b->l7 ? htons((uint8_t *) b->l7 - (uint8_t *) b->l4) : htons(0); + } else if (flow->dl_type == htons(ETH_TYPE_ARP) || + flow->dl_type == htons(ETH_TYPE_RARP)) { struct arp_eth_header *arp; b->l3 = arp = ofpbuf_put_zeros(b, sizeof *arp); @@ -1070,10 +1416,491 @@ flow_compose(struct ofpbuf *b, const struct flow *flow) 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; + put_16aligned_be32(&arp->ar_spa, flow->nw_src); + put_16aligned_be32(&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); } } + + if (eth_type_mpls(flow->dl_type)) { + int n; + + b->l2_5 = b->l3; + for (n = 1; n < FLOW_MAX_MPLS_LABELS; n++) { + if (flow->mpls_lse[n - 1] & htonl(MPLS_BOS_MASK)) { + break; + } + } + while (n > 0) { + push_mpls(b, flow->dl_type, flow->mpls_lse[--n]); + } + } +} + +/* Compressed flow. */ + +static int +miniflow_n_values(const struct miniflow *flow) +{ + return count_1bits(flow->map); +} + +static uint32_t * +miniflow_alloc_values(struct miniflow *flow, int n) +{ + if (n <= MINI_N_INLINE) { + return flow->inline_values; + } else { + COVERAGE_INC(miniflow_malloc); + return xmalloc(n * sizeof *flow->values); + } +} + +/* Completes an initialization of 'dst' as a miniflow copy of 'src' begun by + * the caller. The caller must have already initialized 'dst->map' properly + * to indicate the significant uint32_t elements of 'src'. 'n' must be the + * number of 1-bits in 'dst->map'. + * + * Normally the significant elements are the ones that are non-zero. However, + * when a miniflow is initialized from a (mini)mask, the values can be zeroes, + * so that the flow and mask always have the same maps. + * + * This function initializes 'dst->values' (either inline if possible or with + * malloc() otherwise) and copies the uint32_t elements of 'src' indicated by + * 'dst->map' into it. */ +static void +miniflow_init__(struct miniflow *dst, const struct flow *src, int n) +{ + const uint32_t *src_u32 = (const uint32_t *) src; + unsigned int ofs; + uint64_t map; + + dst->values = miniflow_alloc_values(dst, n); + ofs = 0; + for (map = dst->map; map; map = zero_rightmost_1bit(map)) { + dst->values[ofs++] = src_u32[raw_ctz(map)]; + } +} + +/* Initializes 'dst' as a copy of 'src'. The caller must eventually free 'dst' + * with miniflow_destroy(). */ +void +miniflow_init(struct miniflow *dst, const struct flow *src) +{ + const uint32_t *src_u32 = (const uint32_t *) src; + unsigned int i; + int n; + + /* Initialize dst->map, counting the number of nonzero elements. */ + n = 0; + dst->map = 0; + + for (i = 0; i < FLOW_U32S; i++) { + if (src_u32[i]) { + dst->map |= UINT64_C(1) << i; + n++; + } + } + + miniflow_init__(dst, src, n); +} + +/* Initializes 'dst' as a copy of 'src', using 'mask->map' as 'dst''s map. The + * caller must eventually free 'dst' with miniflow_destroy(). */ +void +miniflow_init_with_minimask(struct miniflow *dst, const struct flow *src, + const struct minimask *mask) +{ + dst->map = mask->masks.map; + miniflow_init__(dst, src, miniflow_n_values(dst)); +} + +/* Initializes 'dst' as a copy of 'src'. The caller must eventually free 'dst' + * with miniflow_destroy(). */ +void +miniflow_clone(struct miniflow *dst, const struct miniflow *src) +{ + int n = miniflow_n_values(src); + dst->map = src->map; + dst->values = miniflow_alloc_values(dst, n); + memcpy(dst->values, src->values, n * sizeof *dst->values); +} + +/* Initializes 'dst' with the data in 'src', destroying 'src'. + * The caller must eventually free 'dst' with miniflow_destroy(). */ +void +miniflow_move(struct miniflow *dst, struct miniflow *src) +{ + if (src->values == src->inline_values) { + dst->values = dst->inline_values; + memcpy(dst->values, src->values, + miniflow_n_values(src) * sizeof *dst->values); + } else { + dst->values = src->values; + } + dst->map = src->map; +} + +/* Frees any memory owned by 'flow'. Does not free the storage in which 'flow' + * itself resides; the caller is responsible for that. */ +void +miniflow_destroy(struct miniflow *flow) +{ + if (flow->values != flow->inline_values) { + free(flow->values); + } +} + +/* Initializes 'dst' as a copy of 'src'. */ +void +miniflow_expand(const struct miniflow *src, struct flow *dst) +{ + memset(dst, 0, sizeof *dst); + flow_union_with_miniflow(dst, src); +} + +static const uint32_t * +miniflow_get__(const struct miniflow *flow, unsigned int u32_ofs) +{ + if (!(flow->map & (UINT64_C(1) << u32_ofs))) { + static const uint32_t zero = 0; + return &zero; + } + return flow->values + + count_1bits(flow->map & ((UINT64_C(1) << u32_ofs) - 1)); +} + +/* Returns the uint32_t that would be at byte offset '4 * u32_ofs' if 'flow' + * were expanded into a "struct flow". */ +uint32_t +miniflow_get(const struct miniflow *flow, unsigned int u32_ofs) +{ + return *miniflow_get__(flow, u32_ofs); +} + +/* Returns the ovs_be16 that would be at byte offset 'u8_ofs' if 'flow' were + * expanded into a "struct flow". */ +static ovs_be16 +miniflow_get_be16(const struct miniflow *flow, unsigned int u8_ofs) +{ + const uint32_t *u32p = miniflow_get__(flow, u8_ofs / 4); + const ovs_be16 *be16p = (const ovs_be16 *) u32p; + return be16p[u8_ofs % 4 != 0]; +} + +/* Returns the VID within the vlan_tci member of the "struct flow" represented + * by 'flow'. */ +uint16_t +miniflow_get_vid(const struct miniflow *flow) +{ + ovs_be16 tci = miniflow_get_be16(flow, offsetof(struct flow, vlan_tci)); + return vlan_tci_to_vid(tci); +} + +/* Returns true if 'a' and 'b' are the same flow, false otherwise. */ +bool +miniflow_equal(const struct miniflow *a, const struct miniflow *b) +{ + const uint32_t *ap = a->values; + const uint32_t *bp = b->values; + const uint64_t a_map = a->map; + const uint64_t b_map = b->map; + uint64_t map; + + if (a_map == b_map) { + for (map = a_map; map; map = zero_rightmost_1bit(map)) { + if (*ap++ != *bp++) { + return false; + } + } + } else { + for (map = a_map | b_map; map; map = zero_rightmost_1bit(map)) { + uint64_t bit = rightmost_1bit(map); + uint64_t a_value = a_map & bit ? *ap++ : 0; + uint64_t b_value = b_map & bit ? *bp++ : 0; + + if (a_value != b_value) { + return false; + } + } + } + + return true; +} + +/* Returns true if 'a' and 'b' are equal at the places where there are 1-bits + * in 'mask', false if they differ. */ +bool +miniflow_equal_in_minimask(const struct miniflow *a, const struct miniflow *b, + const struct minimask *mask) +{ + const uint32_t *p; + uint64_t map; + + p = mask->masks.values; + + for (map = mask->masks.map; map; map = zero_rightmost_1bit(map)) { + int ofs = raw_ctz(map); + + if ((miniflow_get(a, ofs) ^ miniflow_get(b, ofs)) & *p) { + return false; + } + p++; + } + + return true; +} + +/* Returns true if 'a' and 'b' are equal at the places where there are 1-bits + * in 'mask', false if they differ. */ +bool +miniflow_equal_flow_in_minimask(const struct miniflow *a, const struct flow *b, + const struct minimask *mask) +{ + const uint32_t *b_u32 = (const uint32_t *) b; + const uint32_t *p; + uint64_t map; + + p = mask->masks.values; + + for (map = mask->masks.map; map; map = zero_rightmost_1bit(map)) { + int ofs = raw_ctz(map); + + if ((miniflow_get(a, ofs) ^ b_u32[ofs]) & *p) { + return false; + } + p++; + } + + return true; +} + +/* Returns a hash value for 'flow', given 'basis'. */ +uint32_t +miniflow_hash(const struct miniflow *flow, uint32_t basis) +{ + const uint32_t *p = flow->values; + uint32_t hash = basis; + uint64_t hash_map = 0; + uint64_t map; + + for (map = flow->map; map; map = zero_rightmost_1bit(map)) { + if (*p) { + hash = mhash_add(hash, *p); + hash_map |= rightmost_1bit(map); + } + p++; + } + hash = mhash_add(hash, hash_map); + hash = mhash_add(hash, hash_map >> 32); + + return mhash_finish(hash, p - flow->values); +} + +/* Returns a hash value for the bits of 'flow' where there are 1-bits in + * 'mask', given 'basis'. + * + * The hash values returned by this function are the same as those returned by + * flow_hash_in_minimask(), only the form of the arguments differ. */ +uint32_t +miniflow_hash_in_minimask(const struct miniflow *flow, + const struct minimask *mask, uint32_t basis) +{ + const uint32_t *p = mask->masks.values; + uint32_t hash; + uint64_t map; + + hash = basis; + + for (map = mask->masks.map; map; map = zero_rightmost_1bit(map)) { + hash = mhash_add(hash, miniflow_get(flow, raw_ctz(map)) & *p++); + } + + return mhash_finish(hash, (p - mask->masks.values) * 4); +} + +/* Returns a hash value for the bits of 'flow' where there are 1-bits in + * 'mask', given 'basis'. + * + * The hash values returned by this function are the same as those returned by + * miniflow_hash_in_minimask(), only the form of the arguments differ. */ +uint32_t +flow_hash_in_minimask(const struct flow *flow, const struct minimask *mask, + uint32_t basis) +{ + const uint32_t *flow_u32 = (const uint32_t *)flow; + const uint32_t *p = mask->masks.values; + uint32_t hash; + uint64_t map; + + hash = basis; + for (map = mask->masks.map; map; map = zero_rightmost_1bit(map)) { + hash = mhash_add(hash, flow_u32[raw_ctz(map)] & *p++); + } + + return mhash_finish(hash, (p - mask->masks.values) * 4); +} + +/* Returns a hash value for the bits of range [start, end) in 'flow', + * where there are 1-bits in 'mask', given 'hash'. + * + * The hash values returned by this function are the same as those returned by + * minimatch_hash_range(), only the form of the arguments differ. */ +uint32_t +flow_hash_in_minimask_range(const struct flow *flow, + const struct minimask *mask, + uint8_t start, uint8_t end, uint32_t *basis) +{ + const uint32_t *flow_u32 = (const uint32_t *)flow; + unsigned int offset; + uint64_t map = miniflow_get_map_in_range(&mask->masks, start, end, + &offset); + const uint32_t *p = mask->masks.values + offset; + uint32_t hash = *basis; + + for (; map; map = zero_rightmost_1bit(map)) { + hash = mhash_add(hash, flow_u32[raw_ctz(map)] & *p++); + } + + *basis = hash; /* Allow continuation from the unfinished value. */ + return mhash_finish(hash, (p - mask->masks.values) * 4); +} + + +/* Initializes 'dst' as a copy of 'src'. The caller must eventually free 'dst' + * with minimask_destroy(). */ +void +minimask_init(struct minimask *mask, const struct flow_wildcards *wc) +{ + miniflow_init(&mask->masks, &wc->masks); +} + +/* Initializes 'dst' as a copy of 'src'. The caller must eventually free 'dst' + * with minimask_destroy(). */ +void +minimask_clone(struct minimask *dst, const struct minimask *src) +{ + miniflow_clone(&dst->masks, &src->masks); +} + +/* Initializes 'dst' with the data in 'src', destroying 'src'. + * The caller must eventually free 'dst' with minimask_destroy(). */ +void +minimask_move(struct minimask *dst, struct minimask *src) +{ + miniflow_move(&dst->masks, &src->masks); +} + +/* Initializes 'dst_' as the bit-wise "and" of 'a_' and 'b_'. + * + * The caller must provide room for FLOW_U32S "uint32_t"s in 'storage', for use + * by 'dst_'. The caller must *not* free 'dst_' with minimask_destroy(). */ +void +minimask_combine(struct minimask *dst_, + const struct minimask *a_, const struct minimask *b_, + uint32_t storage[FLOW_U32S]) +{ + struct miniflow *dst = &dst_->masks; + const struct miniflow *a = &a_->masks; + const struct miniflow *b = &b_->masks; + uint64_t map; + int n = 0; + + dst->values = storage; + + dst->map = 0; + for (map = a->map & b->map; map; map = zero_rightmost_1bit(map)) { + int ofs = raw_ctz(map); + uint32_t mask = miniflow_get(a, ofs) & miniflow_get(b, ofs); + + if (mask) { + dst->map |= rightmost_1bit(map); + dst->values[n++] = mask; + } + } +} + +/* Frees any memory owned by 'mask'. Does not free the storage in which 'mask' + * itself resides; the caller is responsible for that. */ +void +minimask_destroy(struct minimask *mask) +{ + miniflow_destroy(&mask->masks); +} + +/* Initializes 'dst' as a copy of 'src'. */ +void +minimask_expand(const struct minimask *mask, struct flow_wildcards *wc) +{ + miniflow_expand(&mask->masks, &wc->masks); +} + +/* Returns the uint32_t that would be at byte offset '4 * u32_ofs' if 'mask' + * were expanded into a "struct flow_wildcards". */ +uint32_t +minimask_get(const struct minimask *mask, unsigned int u32_ofs) +{ + return miniflow_get(&mask->masks, u32_ofs); +} + +/* Returns the VID mask within the vlan_tci member of the "struct + * flow_wildcards" represented by 'mask'. */ +uint16_t +minimask_get_vid_mask(const struct minimask *mask) +{ + return miniflow_get_vid(&mask->masks); +} + +/* Returns true if 'a' and 'b' are the same flow mask, false otherwise. */ +bool +minimask_equal(const struct minimask *a, const struct minimask *b) +{ + return miniflow_equal(&a->masks, &b->masks); +} + +/* Returns a hash value for 'mask', given 'basis'. */ +uint32_t +minimask_hash(const struct minimask *mask, uint32_t basis) +{ + return miniflow_hash(&mask->masks, basis); +} + +/* Returns true if at least one bit is wildcarded in 'a_' but not in 'b_', + * false otherwise. */ +bool +minimask_has_extra(const struct minimask *a_, const struct minimask *b_) +{ + const struct miniflow *a = &a_->masks; + const struct miniflow *b = &b_->masks; + uint64_t map; + + for (map = a->map | b->map; map; map = zero_rightmost_1bit(map)) { + int ofs = raw_ctz(map); + uint32_t a_u32 = miniflow_get(a, ofs); + uint32_t b_u32 = miniflow_get(b, ofs); + + if ((a_u32 & b_u32) != b_u32) { + return true; + } + } + + return false; +} + +/* Returns true if 'mask' matches every packet, false if 'mask' fixes any bits + * or fields. */ +bool +minimask_is_catchall(const struct minimask *mask_) +{ + const struct miniflow *mask = &mask_->masks; + const uint32_t *p = mask->values; + uint64_t map; + + for (map = mask->map; map; map = zero_rightmost_1bit(map)) { + if (*p++) { + return false; + } + } + return true; }