2 * Copyright (c) 2008, 2009, 2010, 2011, 2012, 2013, 2014, 2015 Nicira, Inc.
4 * Licensed under the Apache License, Version 2.0 (the "License");
5 * you may not use this file except in compliance with the License.
6 * You may obtain a copy of the License at:
8 * http://www.apache.org/licenses/LICENSE-2.0
10 * Unless required by applicable law or agreed to in writing, software
11 * distributed under the License is distributed on an "AS IS" BASIS,
12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13 * See the License for the specific language governing permissions and
14 * limitations under the License.
21 #include <sys/types.h>
22 #include <netinet/in.h>
27 #include "openvswitch/types.h"
28 #include "odp-netlink.h"
31 #include "tun-metadata.h"
32 #include "unaligned.h"
38 /* Tunnel information used in flow key and metadata. */
50 uint8_t pad1[5]; /* Pad to 64 bits. */
51 struct tun_metadata metadata;
54 /* Some flags are exposed through OpenFlow while others are used only
58 #define FLOW_TNL_F_OAM (1 << 0)
60 #define FLOW_TNL_PUB_F_MASK ((1 << 1) - 1)
63 #define FLOW_TNL_F_DONT_FRAGMENT (1 << 1)
64 #define FLOW_TNL_F_CSUM (1 << 2)
65 #define FLOW_TNL_F_KEY (1 << 3)
67 #define FLOW_TNL_F_MASK ((1 << 4) - 1)
69 /* Purely internal to OVS userspace. These flags should never be exposed to
70 * the outside world and so aren't included in the flags mask. */
72 /* Tunnel information is in userspace datapath format. */
73 #define FLOW_TNL_F_UDPIF (1 << 4)
75 /* Returns an offset to 'src' covering all the meaningful fields in 'src'. */
77 flow_tnl_size(const struct flow_tnl *src)
80 /* Covers ip_dst only. */
81 return offsetof(struct flow_tnl, ip_src);
83 if (src->flags & FLOW_TNL_F_UDPIF) {
84 /* Datapath format, cover all options we have. */
85 return offsetof(struct flow_tnl, metadata.opts)
86 + src->metadata.present.len;
88 if (!src->metadata.present.map) {
89 /* No TLVs, opts is irrelevant. */
90 return offsetof(struct flow_tnl, metadata.opts);
92 /* Have decoded TLVs, opts is relevant. */
96 /* Copy flow_tnl, but avoid copying unused portions of tun_metadata. Unused
97 * data in 'dst' is NOT cleared, so this must not be used in cases where the
98 * uninitialized portion may be hashed over. */
100 flow_tnl_copy__(struct flow_tnl *dst, const struct flow_tnl *src)
102 memcpy(dst, src, flow_tnl_size(src));
106 flow_tnl_equal(const struct flow_tnl *a, const struct flow_tnl *b)
108 size_t a_size = flow_tnl_size(a);
110 return a_size == flow_tnl_size(b) && !memcmp(a, b, a_size);
113 /* Unfortunately, a "struct flow" sometimes has to handle OpenFlow port
114 * numbers and other times datapath (dpif) port numbers. This union allows
121 /* Datapath packet metadata */
122 struct pkt_metadata {
123 uint32_t recirc_id; /* Recirculation id carried with the
124 recirculating packets. 0 for packets
125 received from the wire. */
126 uint32_t dp_hash; /* hash value computed by the recirculation
128 uint32_t skb_priority; /* Packet priority for QoS. */
129 uint32_t pkt_mark; /* Packet mark. */
130 uint16_t ct_state; /* Connection state. */
131 uint16_t ct_zone; /* Connection zone. */
132 uint32_t ct_mark; /* Connection mark. */
133 union flow_in_port in_port; /* Input port. */
134 struct flow_tnl tunnel; /* Encapsulating tunnel parameters. Note that
135 * if 'ip_dst' == 0, the rest of the fields may
136 * be uninitialized. */
140 pkt_metadata_init(struct pkt_metadata *md, odp_port_t port)
142 /* It can be expensive to zero out all of the tunnel metadata. However,
143 * we can just zero out ip_dst and the rest of the data will never be
145 memset(md, 0, offsetof(struct pkt_metadata, in_port));
146 md->tunnel.ip_dst = 0;
148 md->in_port.odp_port = port;
151 bool dpid_from_string(const char *s, uint64_t *dpidp);
153 #define ETH_ADDR_LEN 6
155 static const struct eth_addr eth_addr_broadcast OVS_UNUSED
156 = { { { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff } } };
158 static const struct eth_addr eth_addr_exact OVS_UNUSED
159 = { { { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff } } };
161 static const struct eth_addr eth_addr_zero OVS_UNUSED
162 = { { { 0x00, 0x00, 0x00, 0x00, 0x00, 0x00 } } };
164 static const struct eth_addr eth_addr_stp OVS_UNUSED
165 = { { { 0x01, 0x80, 0xC2, 0x00, 0x00, 0x00 } } };
167 static const struct eth_addr eth_addr_lacp OVS_UNUSED
168 = { { { 0x01, 0x80, 0xC2, 0x00, 0x00, 0x02 } } };
170 static const struct eth_addr eth_addr_bfd OVS_UNUSED
171 = { { { 0x00, 0x23, 0x20, 0x00, 0x00, 0x01 } } };
173 static inline bool eth_addr_is_broadcast(const struct eth_addr a)
175 return (a.be16[0] & a.be16[1] & a.be16[2]) == htons(0xffff);
178 static inline bool eth_addr_is_multicast(const struct eth_addr a)
183 static inline bool eth_addr_is_local(const struct eth_addr a)
185 /* Local if it is either a locally administered address or a Nicira random
188 || (a.be16[0] == htons(0x0023)
189 && (a.be16[1] & htons(0xff80)) == htons(0x2080));
191 static inline bool eth_addr_is_zero(const struct eth_addr a)
193 return !(a.be16[0] | a.be16[1] | a.be16[2]);
196 static inline int eth_mask_is_exact(const struct eth_addr a)
198 return (a.be16[0] & a.be16[1] & a.be16[2]) == htons(0xffff);
201 static inline int eth_addr_compare_3way(const struct eth_addr a,
202 const struct eth_addr b)
204 return memcmp(&a, &b, sizeof a);
207 static inline bool eth_addr_equals(const struct eth_addr a,
208 const struct eth_addr b)
210 return !eth_addr_compare_3way(a, b);
213 static inline bool eth_addr_equal_except(const struct eth_addr a,
214 const struct eth_addr b,
215 const struct eth_addr mask)
217 return !(((a.be16[0] ^ b.be16[0]) & mask.be16[0])
218 || ((a.be16[1] ^ b.be16[1]) & mask.be16[1])
219 || ((a.be16[2] ^ b.be16[2]) & mask.be16[2]));
222 static inline uint64_t eth_addr_to_uint64(const struct eth_addr ea)
224 return (((uint64_t) ntohs(ea.be16[0]) << 32)
225 | ((uint64_t) ntohs(ea.be16[1]) << 16)
226 | ntohs(ea.be16[2]));
229 static inline uint64_t eth_addr_vlan_to_uint64(const struct eth_addr ea,
232 return (((uint64_t)vlan << 48) | eth_addr_to_uint64(ea));
235 static inline void eth_addr_from_uint64(uint64_t x, struct eth_addr *ea)
237 ea->be16[0] = htons(x >> 32);
238 ea->be16[1] = htons((x & 0xFFFF0000) >> 16);
239 ea->be16[2] = htons(x & 0xFFFF);
242 static inline struct eth_addr eth_addr_invert(const struct eth_addr src)
246 for (int i = 0; i < ARRAY_SIZE(src.be16); i++) {
247 dst.be16[i] = ~src.be16[i];
253 static inline void eth_addr_mark_random(struct eth_addr *ea)
255 ea->ea[0] &= ~1; /* Unicast. */
256 ea->ea[0] |= 2; /* Private. */
259 static inline void eth_addr_random(struct eth_addr *ea)
261 random_bytes((uint8_t *)ea, sizeof *ea);
262 eth_addr_mark_random(ea);
265 static inline void eth_addr_nicira_random(struct eth_addr *ea)
269 /* Set the OUI to the Nicira one. */
274 /* Set the top bit to indicate random Nicira address. */
277 static inline uint32_t hash_mac(const struct eth_addr ea,
278 const uint16_t vlan, const uint32_t basis)
280 return hash_uint64_basis(eth_addr_vlan_to_uint64(ea, vlan), basis);
283 bool eth_addr_is_reserved(const struct eth_addr);
284 bool eth_addr_from_string(const char *, struct eth_addr *);
286 void compose_rarp(struct dp_packet *, const struct eth_addr);
288 void eth_push_vlan(struct dp_packet *, ovs_be16 tpid, ovs_be16 tci);
289 void eth_pop_vlan(struct dp_packet *);
291 const char *eth_from_hex(const char *hex, struct dp_packet **packetp);
292 void eth_format_masked(const struct eth_addr ea,
293 const struct eth_addr *mask, struct ds *s);
295 void set_mpls_lse(struct dp_packet *, ovs_be32 label);
296 void push_mpls(struct dp_packet *packet, ovs_be16 ethtype, ovs_be32 lse);
297 void pop_mpls(struct dp_packet *, ovs_be16 ethtype);
299 void set_mpls_lse_ttl(ovs_be32 *lse, uint8_t ttl);
300 void set_mpls_lse_tc(ovs_be32 *lse, uint8_t tc);
301 void set_mpls_lse_label(ovs_be32 *lse, ovs_be32 label);
302 void set_mpls_lse_bos(ovs_be32 *lse, uint8_t bos);
303 ovs_be32 set_mpls_lse_values(uint8_t ttl, uint8_t tc, uint8_t bos,
308 * struct eth_addr mac;
310 * printf("The Ethernet address is "ETH_ADDR_FMT"\n", ETH_ADDR_ARGS(mac));
313 #define ETH_ADDR_FMT \
314 "%02"PRIx8":%02"PRIx8":%02"PRIx8":%02"PRIx8":%02"PRIx8":%02"PRIx8
315 #define ETH_ADDR_ARGS(EA) ETH_ADDR_BYTES_ARGS((EA).ea)
316 #define ETH_ADDR_BYTES_ARGS(EAB) \
317 (EAB)[0], (EAB)[1], (EAB)[2], (EAB)[3], (EAB)[4], (EAB)[5]
321 * char *string = "1 00:11:22:33:44:55 2";
322 * struct eth_addr mac;
325 * if (ovs_scan(string, "%d"ETH_ADDR_SCAN_FMT"%d",
326 * &a, ETH_ADDR_SCAN_ARGS(mac), &b)) {
330 #define ETH_ADDR_SCAN_FMT "%"SCNx8":%"SCNx8":%"SCNx8":%"SCNx8":%"SCNx8":%"SCNx8
331 #define ETH_ADDR_SCAN_ARGS(EA) \
332 &(EA).ea[0], &(EA).ea[1], &(EA).ea[2], &(EA).ea[3], &(EA).ea[4], &(EA).ea[5]
334 #define ETH_TYPE_IP 0x0800
335 #define ETH_TYPE_ARP 0x0806
336 #define ETH_TYPE_TEB 0x6558
337 #define ETH_TYPE_VLAN_8021Q 0x8100
338 #define ETH_TYPE_VLAN ETH_TYPE_VLAN_8021Q
339 #define ETH_TYPE_VLAN_8021AD 0x88a8
340 #define ETH_TYPE_IPV6 0x86dd
341 #define ETH_TYPE_LACP 0x8809
342 #define ETH_TYPE_RARP 0x8035
343 #define ETH_TYPE_MPLS 0x8847
344 #define ETH_TYPE_MPLS_MCAST 0x8848
346 static inline bool eth_type_mpls(ovs_be16 eth_type)
348 return eth_type == htons(ETH_TYPE_MPLS) ||
349 eth_type == htons(ETH_TYPE_MPLS_MCAST);
352 static inline bool eth_type_vlan(ovs_be16 eth_type)
354 return eth_type == htons(ETH_TYPE_VLAN_8021Q) ||
355 eth_type == htons(ETH_TYPE_VLAN_8021AD);
359 /* Minimum value for an Ethernet type. Values below this are IEEE 802.2 frame
361 #define ETH_TYPE_MIN 0x600
363 #define ETH_HEADER_LEN 14
364 #define ETH_PAYLOAD_MIN 46
365 #define ETH_PAYLOAD_MAX 1500
366 #define ETH_TOTAL_MIN (ETH_HEADER_LEN + ETH_PAYLOAD_MIN)
367 #define ETH_TOTAL_MAX (ETH_HEADER_LEN + ETH_PAYLOAD_MAX)
368 #define ETH_VLAN_TOTAL_MAX (ETH_HEADER_LEN + VLAN_HEADER_LEN + ETH_PAYLOAD_MAX)
371 struct eth_addr eth_dst;
372 struct eth_addr eth_src;
375 BUILD_ASSERT_DECL(ETH_HEADER_LEN == sizeof(struct eth_header));
377 #define LLC_DSAP_SNAP 0xaa
378 #define LLC_SSAP_SNAP 0xaa
379 #define LLC_CNTL_SNAP 3
381 #define LLC_HEADER_LEN 3
388 BUILD_ASSERT_DECL(LLC_HEADER_LEN == sizeof(struct llc_header));
390 /* LLC field values used for STP frames. */
391 #define STP_LLC_SSAP 0x42
392 #define STP_LLC_DSAP 0x42
393 #define STP_LLC_CNTL 0x03
395 #define SNAP_ORG_ETHERNET "\0\0" /* The compiler adds a null byte, so
396 sizeof(SNAP_ORG_ETHERNET) == 3. */
397 #define SNAP_HEADER_LEN 5
403 BUILD_ASSERT_DECL(SNAP_HEADER_LEN == sizeof(struct snap_header));
405 #define LLC_SNAP_HEADER_LEN (LLC_HEADER_LEN + SNAP_HEADER_LEN)
407 struct llc_snap_header {
408 struct llc_header llc;
409 struct snap_header snap;
411 BUILD_ASSERT_DECL(LLC_SNAP_HEADER_LEN == sizeof(struct llc_snap_header));
413 #define VLAN_VID_MASK 0x0fff
414 #define VLAN_VID_SHIFT 0
416 #define VLAN_PCP_MASK 0xe000
417 #define VLAN_PCP_SHIFT 13
419 #define VLAN_CFI 0x1000
420 #define VLAN_CFI_SHIFT 12
422 /* Given the vlan_tci field from an 802.1Q header, in network byte order,
423 * returns the VLAN ID in host byte order. */
424 static inline uint16_t
425 vlan_tci_to_vid(ovs_be16 vlan_tci)
427 return (ntohs(vlan_tci) & VLAN_VID_MASK) >> VLAN_VID_SHIFT;
430 /* Given the vlan_tci field from an 802.1Q header, in network byte order,
431 * returns the priority code point (PCP) in host byte order. */
433 vlan_tci_to_pcp(ovs_be16 vlan_tci)
435 return (ntohs(vlan_tci) & VLAN_PCP_MASK) >> VLAN_PCP_SHIFT;
438 /* Given the vlan_tci field from an 802.1Q header, in network byte order,
439 * returns the Canonical Format Indicator (CFI). */
441 vlan_tci_to_cfi(ovs_be16 vlan_tci)
443 return (vlan_tci & htons(VLAN_CFI)) != 0;
446 #define VLAN_HEADER_LEN 4
448 ovs_be16 vlan_tci; /* Lowest 12 bits are VLAN ID. */
449 ovs_be16 vlan_next_type;
451 BUILD_ASSERT_DECL(VLAN_HEADER_LEN == sizeof(struct vlan_header));
453 #define VLAN_ETH_HEADER_LEN (ETH_HEADER_LEN + VLAN_HEADER_LEN)
455 struct vlan_eth_header {
456 struct eth_addr veth_dst;
457 struct eth_addr veth_src;
458 ovs_be16 veth_type; /* Always htons(ETH_TYPE_VLAN). */
459 ovs_be16 veth_tci; /* Lowest 12 bits are VLAN ID. */
460 ovs_be16 veth_next_type;
462 BUILD_ASSERT_DECL(VLAN_ETH_HEADER_LEN == sizeof(struct vlan_eth_header));
464 /* MPLS related definitions */
465 #define MPLS_TTL_MASK 0x000000ff
466 #define MPLS_TTL_SHIFT 0
468 #define MPLS_BOS_MASK 0x00000100
469 #define MPLS_BOS_SHIFT 8
471 #define MPLS_TC_MASK 0x00000e00
472 #define MPLS_TC_SHIFT 9
474 #define MPLS_LABEL_MASK 0xfffff000
475 #define MPLS_LABEL_SHIFT 12
480 ovs_16aligned_be32 mpls_lse;
482 BUILD_ASSERT_DECL(MPLS_HLEN == sizeof(struct mpls_hdr));
484 /* Given a mpls label stack entry in network byte order
485 * return mpls label in host byte order */
486 static inline uint32_t
487 mpls_lse_to_label(ovs_be32 mpls_lse)
489 return (ntohl(mpls_lse) & MPLS_LABEL_MASK) >> MPLS_LABEL_SHIFT;
492 /* Given a mpls label stack entry in network byte order
494 static inline uint8_t
495 mpls_lse_to_tc(ovs_be32 mpls_lse)
497 return (ntohl(mpls_lse) & MPLS_TC_MASK) >> MPLS_TC_SHIFT;
500 /* Given a mpls label stack entry in network byte order
502 static inline uint8_t
503 mpls_lse_to_ttl(ovs_be32 mpls_lse)
505 return (ntohl(mpls_lse) & MPLS_TTL_MASK) >> MPLS_TTL_SHIFT;
508 /* Set TTL in mpls lse. */
510 flow_set_mpls_lse_ttl(ovs_be32 *mpls_lse, uint8_t ttl)
512 *mpls_lse &= ~htonl(MPLS_TTL_MASK);
513 *mpls_lse |= htonl(ttl << MPLS_TTL_SHIFT);
516 /* Given a mpls label stack entry in network byte order
517 * return mpls BoS bit */
518 static inline uint8_t
519 mpls_lse_to_bos(ovs_be32 mpls_lse)
521 return (mpls_lse & htonl(MPLS_BOS_MASK)) != 0;
524 #define IP_FMT "%"PRIu32".%"PRIu32".%"PRIu32".%"PRIu32
525 #define IP_ARGS(ip) \
527 (ntohl(ip) >> 16) & 0xff, \
528 (ntohl(ip) >> 8) & 0xff, \
533 * char *string = "1 33.44.55.66 2";
537 * if (ovs_scan(string, "%d"IP_SCAN_FMT"%d", &a, IP_SCAN_ARGS(&ip), &b)) {
541 #define IP_SCAN_FMT "%"SCNu8".%"SCNu8".%"SCNu8".%"SCNu8
542 #define IP_SCAN_ARGS(ip) \
543 ((void) (ovs_be32) *(ip), &((uint8_t *) ip)[0]), \
544 &((uint8_t *) ip)[1], \
545 &((uint8_t *) ip)[2], \
548 /* Returns true if 'netmask' is a CIDR netmask, that is, if it consists of N
549 * high-order 1-bits and 32-N low-order 0-bits. */
551 ip_is_cidr(ovs_be32 netmask)
553 uint32_t x = ~ntohl(netmask);
554 return !(x & (x + 1));
557 ip_is_multicast(ovs_be32 ip)
559 return (ip & htonl(0xf0000000)) == htonl(0xe0000000);
562 ip_is_local_multicast(ovs_be32 ip)
564 return (ip & htonl(0xffffff00)) == htonl(0xe0000000);
566 int ip_count_cidr_bits(ovs_be32 netmask);
567 void ip_format_masked(ovs_be32 ip, ovs_be32 mask, struct ds *);
569 #define IP_VER(ip_ihl_ver) ((ip_ihl_ver) >> 4)
570 #define IP_IHL(ip_ihl_ver) ((ip_ihl_ver) & 15)
571 #define IP_IHL_VER(ihl, ver) (((ver) << 4) | (ihl))
574 #define IPPROTO_SCTP 132
578 #define IP_ECN_NOT_ECT 0x0
579 #define IP_ECN_ECT_1 0x01
580 #define IP_ECN_ECT_0 0x02
581 #define IP_ECN_CE 0x03
582 #define IP_ECN_MASK 0x03
583 #define IP_DSCP_MASK 0xfc
587 #define IP_DONT_FRAGMENT 0x4000 /* Don't fragment. */
588 #define IP_MORE_FRAGMENTS 0x2000 /* More fragments. */
589 #define IP_FRAG_OFF_MASK 0x1fff /* Fragment offset. */
590 #define IP_IS_FRAGMENT(ip_frag_off) \
591 ((ip_frag_off) & htons(IP_MORE_FRAGMENTS | IP_FRAG_OFF_MASK))
593 #define IP_HEADER_LEN 20
599 ovs_be16 ip_frag_off;
603 ovs_16aligned_be32 ip_src;
604 ovs_16aligned_be32 ip_dst;
607 BUILD_ASSERT_DECL(IP_HEADER_LEN == sizeof(struct ip_header));
609 #define ICMP_HEADER_LEN 8
623 ovs_16aligned_be32 gateway;
626 BUILD_ASSERT_DECL(ICMP_HEADER_LEN == sizeof(struct icmp_header));
628 #define IGMP_HEADER_LEN 8
633 ovs_16aligned_be32 group;
635 BUILD_ASSERT_DECL(IGMP_HEADER_LEN == sizeof(struct igmp_header));
637 #define IGMPV3_HEADER_LEN 8
638 struct igmpv3_header {
645 BUILD_ASSERT_DECL(IGMPV3_HEADER_LEN == sizeof(struct igmpv3_header));
647 #define IGMPV3_RECORD_LEN 8
648 struct igmpv3_record {
652 ovs_16aligned_be32 maddr;
654 BUILD_ASSERT_DECL(IGMPV3_RECORD_LEN == sizeof(struct igmpv3_record));
656 #define IGMP_HOST_MEMBERSHIP_QUERY 0x11 /* From RFC1112 */
657 #define IGMP_HOST_MEMBERSHIP_REPORT 0x12 /* Ditto */
658 #define IGMPV2_HOST_MEMBERSHIP_REPORT 0x16 /* V2 version of 0x12 */
659 #define IGMP_HOST_LEAVE_MESSAGE 0x17
660 #define IGMPV3_HOST_MEMBERSHIP_REPORT 0x22 /* V3 version of 0x12 */
663 * IGMPv3 and MLDv2 use the same codes.
665 #define IGMPV3_MODE_IS_INCLUDE 1
666 #define IGMPV3_MODE_IS_EXCLUDE 2
667 #define IGMPV3_CHANGE_TO_INCLUDE_MODE 3
668 #define IGMPV3_CHANGE_TO_EXCLUDE_MODE 4
669 #define IGMPV3_ALLOW_NEW_SOURCES 5
670 #define IGMPV3_BLOCK_OLD_SOURCES 6
672 #define SCTP_HEADER_LEN 12
676 ovs_16aligned_be32 sctp_vtag;
677 ovs_16aligned_be32 sctp_csum;
679 BUILD_ASSERT_DECL(SCTP_HEADER_LEN == sizeof(struct sctp_header));
681 #define UDP_HEADER_LEN 8
688 BUILD_ASSERT_DECL(UDP_HEADER_LEN == sizeof(struct udp_header));
690 #define TCP_FIN 0x001
691 #define TCP_SYN 0x002
692 #define TCP_RST 0x004
693 #define TCP_PSH 0x008
694 #define TCP_ACK 0x010
695 #define TCP_URG 0x020
696 #define TCP_ECE 0x040
697 #define TCP_CWR 0x080
700 #define TCP_CTL(flags, offset) (htons((flags) | ((offset) << 12)))
701 #define TCP_FLAGS(tcp_ctl) (ntohs(tcp_ctl) & 0x0fff)
702 #define TCP_FLAGS_BE16(tcp_ctl) ((tcp_ctl) & htons(0x0fff))
703 #define TCP_OFFSET(tcp_ctl) (ntohs(tcp_ctl) >> 12)
705 #define TCP_HEADER_LEN 20
709 ovs_16aligned_be32 tcp_seq;
710 ovs_16aligned_be32 tcp_ack;
716 BUILD_ASSERT_DECL(TCP_HEADER_LEN == sizeof(struct tcp_header));
718 /* Connection states */
720 #define CS_ESTABLISHED 0x02
721 #define CS_RELATED 0x04
722 #define CS_INVALID 0x20
723 #define CS_REPLY_DIR 0x40
724 #define CS_TRACKED 0x80
726 /* Undefined connection state bits. */
727 #define CS_SUPPORTED_MASK (CS_NEW | CS_ESTABLISHED | CS_RELATED \
728 | CS_INVALID | CS_REPLY_DIR | CS_TRACKED)
729 #define CS_UNSUPPORTED_MASK (~(uint32_t)CS_SUPPORTED_MASK)
731 #define ARP_HRD_ETHERNET 1
732 #define ARP_PRO_IP 0x0800
733 #define ARP_OP_REQUEST 1
734 #define ARP_OP_REPLY 2
735 #define ARP_OP_RARP 3
737 #define ARP_ETH_HEADER_LEN 28
738 struct arp_eth_header {
739 /* Generic members. */
740 ovs_be16 ar_hrd; /* Hardware type. */
741 ovs_be16 ar_pro; /* Protocol type. */
742 uint8_t ar_hln; /* Hardware address length. */
743 uint8_t ar_pln; /* Protocol address length. */
744 ovs_be16 ar_op; /* Opcode. */
746 /* Ethernet+IPv4 specific members. */
747 struct eth_addr ar_sha; /* Sender hardware address. */
748 ovs_16aligned_be32 ar_spa; /* Sender protocol address. */
749 struct eth_addr ar_tha; /* Target hardware address. */
750 ovs_16aligned_be32 ar_tpa; /* Target protocol address. */
752 BUILD_ASSERT_DECL(ARP_ETH_HEADER_LEN == sizeof(struct arp_eth_header));
754 /* Like struct in6_addr, but whereas that struct requires 32-bit alignment on
755 * most implementations, this one only requires 16-bit alignment. */
756 union ovs_16aligned_in6_addr {
758 ovs_16aligned_be32 be32[4];
761 /* Like struct in6_hdr, but whereas that struct requires 32-bit alignment, this
762 * one only requires 16-bit alignment. */
763 struct ovs_16aligned_ip6_hdr {
765 struct ovs_16aligned_ip6_hdrctl {
766 ovs_16aligned_be32 ip6_un1_flow;
767 ovs_be16 ip6_un1_plen;
769 uint8_t ip6_un1_hlim;
773 union ovs_16aligned_in6_addr ip6_src;
774 union ovs_16aligned_in6_addr ip6_dst;
777 /* Like struct in6_frag, but whereas that struct requires 32-bit alignment,
778 * this one only requires 16-bit alignment. */
779 struct ovs_16aligned_ip6_frag {
781 uint8_t ip6f_reserved;
783 ovs_16aligned_be32 ip6f_ident;
786 #define ICMP6_HEADER_LEN 4
787 struct icmp6_header {
790 ovs_be16 icmp6_cksum;
792 BUILD_ASSERT_DECL(ICMP6_HEADER_LEN == sizeof(struct icmp6_header));
794 /* Neighbor Discovery option field.
795 * ND options are always a multiple of 8 bytes in size. */
798 uint8_t nd_opt_type; /* Values defined in icmp6.h */
799 uint8_t nd_opt_len; /* in units of 8 octets (the size of this struct) */
800 struct eth_addr nd_opt_mac; /* Ethernet address in the case of SLL or TLL options */
802 BUILD_ASSERT_DECL(ND_OPT_LEN == sizeof(struct ovs_nd_opt));
804 /* Like struct nd_msg (from ndisc.h), but whereas that struct requires 32-bit
805 * alignment, this one only requires 16-bit alignment. */
806 #define ND_MSG_LEN 24
808 struct icmp6_header icmph;
809 ovs_16aligned_be32 rco_flags;
810 union ovs_16aligned_in6_addr target;
811 struct ovs_nd_opt options[0];
813 BUILD_ASSERT_DECL(ND_MSG_LEN == sizeof(struct ovs_nd_msg));
816 * Use the same struct for MLD and MLD2, naming members as the defined fields in
817 * in the corresponding version of the protocol, though they are reserved in the
820 #define MLD_HEADER_LEN 8
828 BUILD_ASSERT_DECL(MLD_HEADER_LEN == sizeof(struct mld_header));
830 #define MLD2_RECORD_LEN 20
835 union ovs_16aligned_in6_addr maddr;
837 BUILD_ASSERT_DECL(MLD2_RECORD_LEN == sizeof(struct mld2_record));
839 #define MLD_QUERY 130
840 #define MLD_REPORT 131
842 #define MLD2_REPORT 143
844 /* The IPv6 flow label is in the lower 20 bits of the first 32-bit word. */
845 #define IPV6_LABEL_MASK 0x000fffff
849 * char *string = "1 ::1 2";
850 * char ipv6_s[IPV6_SCAN_LEN + 1];
851 * struct in6_addr ipv6;
853 * if (ovs_scan(string, "%d"IPV6_SCAN_FMT"%d", &a, ipv6_s, &b)
854 * && inet_pton(AF_INET6, ipv6_s, &ipv6) == 1) {
858 #define IPV6_SCAN_FMT "%46[0123456789abcdefABCDEF:.]"
859 #define IPV6_SCAN_LEN 46
861 extern const struct in6_addr in6addr_exact;
862 #define IN6ADDR_EXACT_INIT { { { 0xff,0xff,0xff,0xff,0xff,0xff,0xff,0xff, \
863 0xff,0xff,0xff,0xff,0xff,0xff,0xff,0xff } } }
865 extern const struct in6_addr in6addr_all_hosts;
866 #define IN6ADDR_ALL_HOSTS_INIT { { { 0xff,0x02,0x00,0x00,0x00,0x00,0x00,0x00, \
867 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x01 } } }
869 static inline bool ipv6_addr_equals(const struct in6_addr *a,
870 const struct in6_addr *b)
872 #ifdef IN6_ARE_ADDR_EQUAL
873 return IN6_ARE_ADDR_EQUAL(a, b);
875 return !memcmp(a, b, sizeof(*a));
879 static inline bool ipv6_mask_is_any(const struct in6_addr *mask) {
880 return ipv6_addr_equals(mask, &in6addr_any);
883 static inline bool ipv6_mask_is_exact(const struct in6_addr *mask) {
884 return ipv6_addr_equals(mask, &in6addr_exact);
887 static inline bool ipv6_is_all_hosts(const struct in6_addr *addr) {
888 return ipv6_addr_equals(addr, &in6addr_all_hosts);
891 static inline bool ipv6_addr_is_set(const struct in6_addr *addr) {
892 return !ipv6_addr_equals(addr, &in6addr_any);
895 static inline bool ipv6_addr_is_multicast(const struct in6_addr *ip) {
896 return ip->s6_addr[0] == 0xff;
900 in6_addr_set_mapped_ipv4(struct in6_addr *addr, ovs_be32 ip4)
902 union ovs_16aligned_in6_addr *taddr = (void *) addr;
903 memset(taddr->be16, 0, sizeof(taddr->be16));
904 taddr->be16[5] = OVS_BE16_MAX;
905 put_16aligned_be32(&taddr->be32[3], ip4);
908 static inline ovs_be32
909 in6_addr_get_mapped_ipv4(const struct in6_addr *addr)
911 union ovs_16aligned_in6_addr *taddr = (void *) addr;
912 if (IN6_IS_ADDR_V4MAPPED(addr)) {
913 return get_16aligned_be32(&taddr->be32[3]);
919 static inline bool dl_type_is_ip_any(ovs_be16 dl_type)
921 return dl_type == htons(ETH_TYPE_IP)
922 || dl_type == htons(ETH_TYPE_IPV6);
927 /* GRE protocol header */
928 struct gre_base_hdr {
933 #define GRE_CSUM 0x8000
934 #define GRE_ROUTING 0x4000
935 #define GRE_KEY 0x2000
936 #define GRE_SEQ 0x1000
937 #define GRE_STRICT 0x0800
938 #define GRE_REC 0x0700
939 #define GRE_FLAGS 0x00F8
940 #define GRE_VERSION 0x0007
942 /* VXLAN protocol header */
944 ovs_16aligned_be32 vx_flags;
945 ovs_16aligned_be32 vx_vni;
948 #define VXLAN_FLAGS 0x08000000 /* struct vxlanhdr.vx_flags required value. */
950 void format_ipv6_addr(char *addr_str, const struct in6_addr *addr);
951 void print_ipv6_addr(struct ds *string, const struct in6_addr *addr);
952 void print_ipv6_mapped(struct ds *string, const struct in6_addr *addr);
953 void print_ipv6_masked(struct ds *string, const struct in6_addr *addr,
954 const struct in6_addr *mask);
955 struct in6_addr ipv6_addr_bitand(const struct in6_addr *src,
956 const struct in6_addr *mask);
957 struct in6_addr ipv6_create_mask(int mask);
958 int ipv6_count_cidr_bits(const struct in6_addr *netmask);
959 bool ipv6_is_cidr(const struct in6_addr *netmask);
961 void *eth_compose(struct dp_packet *, const struct eth_addr eth_dst,
962 const struct eth_addr eth_src, uint16_t eth_type,
964 void *snap_compose(struct dp_packet *, const struct eth_addr eth_dst,
965 const struct eth_addr eth_src,
966 unsigned int oui, uint16_t snap_type, size_t size);
967 void packet_set_ipv4(struct dp_packet *, ovs_be32 src, ovs_be32 dst, uint8_t tos,
969 void packet_set_ipv6(struct dp_packet *, uint8_t proto, const ovs_be32 src[4],
970 const ovs_be32 dst[4], uint8_t tc,
971 ovs_be32 fl, uint8_t hlmit);
972 void packet_set_tcp_port(struct dp_packet *, ovs_be16 src, ovs_be16 dst);
973 void packet_set_udp_port(struct dp_packet *, ovs_be16 src, ovs_be16 dst);
974 void packet_set_sctp_port(struct dp_packet *, ovs_be16 src, ovs_be16 dst);
975 void packet_set_nd(struct dp_packet *, const ovs_be32 target[4],
976 const struct eth_addr sll, const struct eth_addr tll);
978 void packet_format_tcp_flags(struct ds *, uint16_t);
979 const char *packet_tcp_flag_to_string(uint32_t flag);
980 void compose_arp(struct dp_packet *, uint16_t arp_op,
981 const struct eth_addr arp_sha,
982 const struct eth_addr arp_tha, bool broadcast,
983 ovs_be32 arp_spa, ovs_be32 arp_tpa);
984 uint32_t packet_csum_pseudoheader(const struct ip_header *);
986 #endif /* packets.h */