X-Git-Url: http://git.onelab.eu/?a=blobdiff_plain;f=lib%2Fflow.c;h=e7ed2a97efe1b33e58d72765653289cea61573c6;hb=ea763e0e2828c35660310f8b7791781b17b30cbd;hp=e55f4fe3e2dff47204ca06e4deab150ae478a1f7;hpb=064af42167bf4fc9aaea2702d80ce08074b889c0;p=sliver-openvswitch.git diff --git a/lib/flow.c b/lib/flow.c index e55f4fe3e..e7ed2a97e 100644 --- a/lib/flow.c +++ b/lib/flow.c @@ -1,17 +1,17 @@ /* - * Copyright (c) 2008, 2009 Nicira Networks. + * Copyright (c) 2008, 2009, 2010 Nicira Networks. * - * Permission to use, copy, modify, and/or distribute this software for any - * purpose with or without fee is hereby granted, provided that the above - * copyright notice and this permission notice appear in all copies. + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at: * - * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES - * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF - * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR - * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES - * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN - * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF - * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. */ #include #include @@ -20,16 +20,27 @@ #include #include #include +#include "byte-order.h" #include "coverage.h" #include "dynamic-string.h" #include "hash.h" +#include "ofp-util.h" #include "ofpbuf.h" #include "openflow/openflow.h" #include "openvswitch/datapath-protocol.h" #include "packets.h" - +#include "unaligned.h" #include "vlog.h" -#define THIS_MODULE VLM_flow + +VLOG_DEFINE_THIS_MODULE(flow); + +COVERAGE_DEFINE(flow_extract); + +static struct arp_eth_header * +pull_arp(struct ofpbuf *packet) +{ + return ofpbuf_try_pull(packet, ARP_ETH_HEADER_LEN); +} static struct ip_header * pull_ip(struct ofpbuf *packet) @@ -45,7 +56,7 @@ pull_ip(struct ofpbuf *packet) } static struct tcp_header * -pull_tcp(struct ofpbuf *packet) +pull_tcp(struct ofpbuf *packet) { if (packet->size >= TCP_HEADER_LEN) { struct tcp_header *tcp = packet->data; @@ -58,32 +69,77 @@ pull_tcp(struct ofpbuf *packet) } static struct udp_header * -pull_udp(struct ofpbuf *packet) +pull_udp(struct ofpbuf *packet) { return ofpbuf_try_pull(packet, UDP_HEADER_LEN); } static struct icmp_header * -pull_icmp(struct ofpbuf *packet) +pull_icmp(struct ofpbuf *packet) { return ofpbuf_try_pull(packet, ICMP_HEADER_LEN); } -static struct eth_header * -pull_eth(struct ofpbuf *packet) +static void +parse_vlan(struct ofpbuf *b, struct flow *flow) { - return ofpbuf_try_pull(packet, ETH_HEADER_LEN); + struct qtag_prefix { + ovs_be16 eth_type; /* ETH_TYPE_VLAN */ + ovs_be16 tci; + }; + + if (b->size >= sizeof(struct qtag_prefix) + sizeof(ovs_be16)) { + struct qtag_prefix *qp = ofpbuf_pull(b, sizeof *qp); + flow->vlan_tci = qp->tci | htons(VLAN_CFI); + } } -static struct vlan_header * -pull_vlan(struct ofpbuf *packet) +static ovs_be16 +parse_ethertype(struct ofpbuf *b) { - return ofpbuf_try_pull(packet, VLAN_HEADER_LEN); + struct llc_snap_header *llc; + ovs_be16 proto; + + proto = *(ovs_be16 *) ofpbuf_pull(b, sizeof proto); + if (ntohs(proto) >= ODP_DL_TYPE_ETH2_CUTOFF) { + return proto; + } + + if (b->size < sizeof *llc) { + return htons(ODP_DL_TYPE_NOT_ETH_TYPE); + } + + llc = b->data; + if (llc->llc.llc_dsap != LLC_DSAP_SNAP + || llc->llc.llc_ssap != LLC_SSAP_SNAP + || llc->llc.llc_cntl != LLC_CNTL_SNAP + || memcmp(llc->snap.snap_org, SNAP_ORG_ETHERNET, + sizeof llc->snap.snap_org)) { + return htons(ODP_DL_TYPE_NOT_ETH_TYPE); + } + + ofpbuf_pull(b, sizeof *llc); + return llc->snap.snap_type; } -/* Returns 1 if 'packet' is an IP fragment, 0 otherwise. */ +/* Initializes 'flow' members from 'packet', 'tun_id', and 'in_port. + * Initializes 'packet' header pointers as follows: + * + * - packet->l2 to the start of the Ethernet 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. + * + * - 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 + * present and has a correct length, and otherwise NULL. + */ int -flow_extract(struct ofpbuf *packet, uint16_t in_port, flow_t *flow) +flow_extract(struct ofpbuf *packet, ovs_be64 tun_id, uint16_t in_port, + struct flow *flow) { struct ofpbuf b = *packet; struct eth_header *eth; @@ -92,7 +148,7 @@ flow_extract(struct ofpbuf *packet, uint16_t in_port, flow_t *flow) COVERAGE_INC(flow_extract); memset(flow, 0, sizeof *flow); - flow->dl_vlan = htons(OFP_VLAN_NONE); + flow->tun_id = tun_id; flow->in_port = in_port; packet->l2 = b.data; @@ -100,90 +156,74 @@ flow_extract(struct ofpbuf *packet, uint16_t in_port, flow_t *flow) packet->l4 = NULL; packet->l7 = NULL; - eth = pull_eth(&b); - if (eth) { - if (ntohs(eth->eth_type) >= OFP_DL_TYPE_ETH2_CUTOFF) { - /* This is an Ethernet II frame */ - flow->dl_type = eth->eth_type; - } else { - /* This is an 802.2 frame */ - struct llc_header *llc = ofpbuf_at(&b, 0, sizeof *llc); - struct snap_header *snap = ofpbuf_at(&b, sizeof *llc, - sizeof *snap); - if (llc == NULL) { - return 0; - } - if (snap - && llc->llc_dsap == LLC_DSAP_SNAP - && llc->llc_ssap == LLC_SSAP_SNAP - && llc->llc_cntl == LLC_CNTL_SNAP - && !memcmp(snap->snap_org, SNAP_ORG_ETHERNET, - sizeof snap->snap_org)) { - flow->dl_type = snap->snap_type; - ofpbuf_pull(&b, LLC_SNAP_HEADER_LEN); + if (b.size < sizeof *eth) { + return 0; + } + + /* Link layer. */ + eth = b.data; + memcpy(flow->dl_src, eth->eth_src, ETH_ADDR_LEN); + memcpy(flow->dl_dst, eth->eth_dst, ETH_ADDR_LEN); + + /* dl_type, vlan_tci. */ + ofpbuf_pull(&b, ETH_ADDR_LEN * 2); + if (eth->eth_type == htons(ETH_TYPE_VLAN)) { + parse_vlan(&b, flow); + } + flow->dl_type = parse_ethertype(&b); + + /* Network layer. */ + packet->l3 = b.data; + if (flow->dl_type == htons(ETH_TYPE_IP)) { + const struct ip_header *nh = pull_ip(&b); + if (nh) { + 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)) { + if (flow->nw_proto == IP_TYPE_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 == IP_TYPE_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 == IP_TYPE_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); + packet->l7 = b.data; + } + } } else { - flow->dl_type = htons(OFP_DL_TYPE_NOT_ETH_TYPE); - ofpbuf_pull(&b, sizeof(struct llc_header)); + retval = 1; } } - - /* Check for a VLAN tag */ - if (flow->dl_type == htons(ETH_TYPE_VLAN)) { - struct vlan_header *vh = pull_vlan(&b); - if (vh) { - flow->dl_type = vh->vlan_next_type; - flow->dl_vlan = vh->vlan_tci & htons(VLAN_VID_MASK); + } else if (flow->dl_type == htons(ETH_TYPE_ARP)) { + const struct arp_eth_header *arp = pull_arp(&b); + if (arp && arp->ar_hrd == htons(1) + && arp->ar_pro == htons(ETH_TYPE_IP) + && arp->ar_hln == ETH_ADDR_LEN + && arp->ar_pln == 4) { + /* We only match on the lower 8 bits of the opcode. */ + if (ntohs(arp->ar_op) <= 0xff) { + flow->nw_proto = ntohs(arp->ar_op); } - } - memcpy(flow->dl_src, eth->eth_src, ETH_ADDR_LEN); - memcpy(flow->dl_dst, eth->eth_dst, ETH_ADDR_LEN); - - packet->l3 = b.data; - if (flow->dl_type == htons(ETH_TYPE_IP)) { - const struct ip_header *nh = pull_ip(&b); - if (nh) { - flow->nw_src = nh->ip_src; - flow->nw_dst = nh->ip_dst; - flow->nw_proto = nh->ip_proto; - packet->l4 = b.data; - if (!IP_IS_FRAGMENT(nh->ip_frag_off)) { - if (flow->nw_proto == IP_TYPE_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 { - /* Avoid tricking other code into thinking that - * this packet has an L4 header. */ - flow->nw_proto = 0; - } - } else if (flow->nw_proto == IP_TYPE_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 { - /* Avoid tricking other code into thinking that - * this packet has an L4 header. */ - flow->nw_proto = 0; - } - } else if (flow->nw_proto == IP_TYPE_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); - packet->l7 = b.data; - } else { - /* Avoid tricking other code into thinking that - * this packet has an L4 header. */ - flow->nw_proto = 0; - } - } - } else { - retval = 1; - } + + 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; } } } @@ -194,14 +234,12 @@ flow_extract(struct ofpbuf *packet, uint16_t in_port, flow_t *flow) * arguments must have been initialized through a call to flow_extract(). */ void -flow_extract_stats(const flow_t *flow, struct ofpbuf *packet, +flow_extract_stats(const struct flow *flow, struct ofpbuf *packet, struct odp_flow_stats *stats) { memset(stats, '\0', sizeof(*stats)); if ((flow->dl_type == htons(ETH_TYPE_IP)) && packet->l4) { - struct ip_header *ip = packet->l3; - stats->ip_tos = ip->ip_tos; if ((flow->nw_proto == IP_TYPE_TCP) && packet->l7) { struct tcp_header *tcp = packet->l4; stats->tcp_flags = TCP_FLAGS(tcp->tcp_ctl); @@ -212,69 +250,208 @@ flow_extract_stats(const flow_t *flow, struct ofpbuf *packet, stats->n_packets = 1; } -void -flow_to_match(const flow_t *flow, uint32_t wildcards, struct ofp_match *match) +char * +flow_to_string(const struct flow *flow) { - match->wildcards = htonl(wildcards); - match->in_port = htons(flow->in_port == ODPP_LOCAL ? OFPP_LOCAL - : flow->in_port); - match->dl_vlan = flow->dl_vlan; - memcpy(match->dl_src, flow->dl_src, ETH_ADDR_LEN); - memcpy(match->dl_dst, flow->dl_dst, ETH_ADDR_LEN); - match->dl_type = flow->dl_type; - match->nw_src = flow->nw_src; - match->nw_dst = flow->nw_dst; - match->nw_proto = flow->nw_proto; - match->tp_src = flow->tp_src; - match->tp_dst = flow->tp_dst; - match->pad = 0; + struct ds ds = DS_EMPTY_INITIALIZER; + flow_format(&ds, flow); + return ds_cstr(&ds); } void -flow_from_match(flow_t *flow, uint32_t *wildcards, - const struct ofp_match *match) +flow_format(struct ds *ds, const struct flow *flow) { - if (wildcards) { - *wildcards = ntohl(match->wildcards); + ds_put_format(ds, "tunnel%#"PRIx64":in_port%04"PRIx16":tci(", + flow->tun_id, flow->in_port); + 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'); } - flow->nw_src = match->nw_src; - flow->nw_dst = match->nw_dst; - flow->in_port = (match->in_port == htons(OFPP_LOCAL) ? ODPP_LOCAL - : ntohs(match->in_port)); - flow->dl_vlan = match->dl_vlan; - flow->dl_type = match->dl_type; - flow->tp_src = match->tp_src; - flow->tp_dst = match->tp_dst; - memcpy(flow->dl_src, match->dl_src, ETH_ADDR_LEN); - memcpy(flow->dl_dst, match->dl_dst, ETH_ADDR_LEN); - flow->nw_proto = match->nw_proto; - flow->reserved = 0; + ds_put_format(ds, ") mac"ETH_ADDR_FMT"->"ETH_ADDR_FMT + " type%04"PRIx16 + " proto%"PRIu8 + " tos%"PRIu8 + " ip"IP_FMT"->"IP_FMT + " port%"PRIu16"->%"PRIu16, + ETH_ADDR_ARGS(flow->dl_src), + ETH_ADDR_ARGS(flow->dl_dst), + ntohs(flow->dl_type), + flow->nw_proto, + flow->nw_tos, + IP_ARGS(&flow->nw_src), + IP_ARGS(&flow->nw_dst), + ntohs(flow->tp_src), + ntohs(flow->tp_dst)); } -char * -flow_to_string(const flow_t *flow) +void +flow_print(FILE *stream, const struct flow *flow) { - struct ds ds = DS_EMPTY_INITIALIZER; - flow_format(&ds, flow); - return ds_cstr(&ds); + char *s = flow_to_string(flow); + fputs(s, stream); + free(s); } + +/* flow_wildcards functions. */ +/* Initializes 'wc' as a set of wildcards that matches every packet. */ void -flow_format(struct ds *ds, const flow_t *flow) +flow_wildcards_init_catchall(struct flow_wildcards *wc) { - ds_put_format(ds, "port%04x:vlan%d mac"ETH_ADDR_FMT"->"ETH_ADDR_FMT" " - "type%04x proto%"PRId8" ip"IP_FMT"->"IP_FMT" port%d->%d", - flow->in_port, ntohs(flow->dl_vlan), - ETH_ADDR_ARGS(flow->dl_src), ETH_ADDR_ARGS(flow->dl_dst), - ntohs(flow->dl_type), flow->nw_proto, - IP_ARGS(&flow->nw_src), IP_ARGS(&flow->nw_dst), - ntohs(flow->tp_src), ntohs(flow->tp_dst)); + wc->wildcards = FWW_ALL; + wc->nw_src_mask = htonl(0); + wc->nw_dst_mask = htonl(0); + memset(wc->reg_masks, 0, sizeof wc->reg_masks); + wc->vlan_tci_mask = htons(0); + wc->zero = 0; } +/* Initializes 'wc' as an exact-match set of wildcards; that is, 'wc' does not + * wildcard any bits or fields. */ void -flow_print(FILE *stream, const flow_t *flow) +flow_wildcards_init_exact(struct flow_wildcards *wc) { - char *s = flow_to_string(flow); - fputs(s, stream); - free(s); + wc->wildcards = 0; + wc->nw_src_mask = htonl(UINT32_MAX); + wc->nw_dst_mask = htonl(UINT32_MAX); + memset(wc->reg_masks, 0xff, sizeof wc->reg_masks); + wc->vlan_tci_mask = htons(UINT16_MAX); + wc->zero = 0; +} + +/* Returns true if 'wc' is exact-match, false if 'wc' wildcards any bits or + * fields. */ +bool +flow_wildcards_is_exact(const struct flow_wildcards *wc) +{ + int i; + + if (wc->wildcards + || wc->nw_src_mask != htonl(UINT32_MAX) + || wc->nw_dst_mask != htonl(UINT32_MAX) + || wc->vlan_tci_mask != htons(UINT16_MAX)) { + return false; + } + + for (i = 0; i < FLOW_N_REGS; i++) { + if (wc->reg_masks[i] != htonl(UINT32_MAX)) { + return false; + } + } + + return true; +} + +/* 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. */ +void +flow_wildcards_combine(struct flow_wildcards *dst, + const struct flow_wildcards *src1, + const struct flow_wildcards *src2) +{ + int i; + + dst->wildcards = src1->wildcards | src2->wildcards; + dst->nw_src_mask = src1->nw_src_mask & src2->nw_src_mask; + dst->nw_dst_mask = src1->nw_dst_mask & src2->nw_dst_mask; + for (i = 0; i < FLOW_N_REGS; i++) { + dst->reg_masks[i] = src1->reg_masks[i] & src2->reg_masks[i]; + } + dst->vlan_tci_mask = src1->vlan_tci_mask & src2->vlan_tci_mask; +} + +/* Returns a hash of the wildcards in 'wc'. */ +uint32_t +flow_wildcards_hash(const struct flow_wildcards *wc) +{ + /* 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 == 16 + FLOW_N_REGS * 4); + return hash_bytes(wc, sizeof *wc, 0); +} + +/* Returns true if 'a' and 'b' represent the same wildcards, false if they are + * different. */ +bool +flow_wildcards_equal(const struct flow_wildcards *a, + const struct flow_wildcards *b) +{ + int i; + + if (a->wildcards != b->wildcards + || a->nw_src_mask != b->nw_src_mask + || a->nw_dst_mask != b->nw_dst_mask + || a->vlan_tci_mask != b->vlan_tci_mask) { + return false; + } + + for (i = 0; i < FLOW_N_REGS; i++) { + if (a->reg_masks[i] != b->reg_masks[i]) { + return false; + } + } + + return true; +} + +/* Returns true if at least one bit or field is wildcarded in 'a' but not in + * 'b', false otherwise. */ +bool +flow_wildcards_has_extra(const struct flow_wildcards *a, + const struct flow_wildcards *b) +{ + int i; + + for (i = 0; i < FLOW_N_REGS; i++) { + if ((a->reg_masks[i] & b->reg_masks[i]) != b->reg_masks[i]) { + return true; + } + } + + return (a->wildcards & ~b->wildcards + || (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); +} + +/* Sets the wildcard mask for register 'idx' in 'wc' to 'mask'. + * (A 0-bit indicates a wildcard bit.) */ +void +flow_wildcards_set_reg_mask(struct flow_wildcards *wc, int idx, uint32_t mask) +{ + wc->reg_masks[idx] = mask; }