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 union flow_in_port in_port
; /* Input port. */
133 struct flow_tnl tunnel
; /* Encapsulating tunnel parameters. Note that
134 * if 'ip_dst' == 0, the rest of the fields may
135 * be uninitialized. */
139 pkt_metadata_init(struct pkt_metadata
*md
, odp_port_t port
)
141 /* It can be expensive to zero out all of the tunnel metadata. However,
142 * we can just zero out ip_dst and the rest of the data will never be
144 memset(md
, 0, offsetof(struct pkt_metadata
, in_port
));
145 md
->tunnel
.ip_dst
= 0;
147 md
->in_port
.odp_port
= port
;
150 bool dpid_from_string(const char *s
, uint64_t *dpidp
);
152 #define ETH_ADDR_LEN 6
154 static const struct eth_addr eth_addr_broadcast OVS_UNUSED
155 = { { { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff } } };
157 static const struct eth_addr eth_addr_exact OVS_UNUSED
158 = { { { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff } } };
160 static const struct eth_addr eth_addr_zero OVS_UNUSED
161 = { { { 0x00, 0x00, 0x00, 0x00, 0x00, 0x00 } } };
163 static const struct eth_addr eth_addr_stp OVS_UNUSED
164 = { { { 0x01, 0x80, 0xC2, 0x00, 0x00, 0x00 } } };
166 static const struct eth_addr eth_addr_lacp OVS_UNUSED
167 = { { { 0x01, 0x80, 0xC2, 0x00, 0x00, 0x02 } } };
169 static const struct eth_addr eth_addr_bfd OVS_UNUSED
170 = { { { 0x00, 0x23, 0x20, 0x00, 0x00, 0x01 } } };
172 static inline bool eth_addr_is_broadcast(const struct eth_addr a
)
174 return (a
.be16
[0] & a
.be16
[1] & a
.be16
[2]) == htons(0xffff);
177 static inline bool eth_addr_is_multicast(const struct eth_addr a
)
182 static inline bool eth_addr_is_local(const struct eth_addr a
)
184 /* Local if it is either a locally administered address or a Nicira random
187 || (a
.be16
[0] == htons(0x0023)
188 && (a
.be16
[1] & htons(0xff80)) == htons(0x2080));
190 static inline bool eth_addr_is_zero(const struct eth_addr a
)
192 return !(a
.be16
[0] | a
.be16
[1] | a
.be16
[2]);
195 static inline int eth_mask_is_exact(const struct eth_addr a
)
197 return (a
.be16
[0] & a
.be16
[1] & a
.be16
[2]) == htons(0xffff);
200 static inline int eth_addr_compare_3way(const struct eth_addr a
,
201 const struct eth_addr b
)
203 return memcmp(&a
, &b
, sizeof a
);
206 static inline bool eth_addr_equals(const struct eth_addr a
,
207 const struct eth_addr b
)
209 return !eth_addr_compare_3way(a
, b
);
212 static inline bool eth_addr_equal_except(const struct eth_addr a
,
213 const struct eth_addr b
,
214 const struct eth_addr mask
)
216 return !(((a
.be16
[0] ^ b
.be16
[0]) & mask
.be16
[0])
217 || ((a
.be16
[1] ^ b
.be16
[1]) & mask
.be16
[1])
218 || ((a
.be16
[2] ^ b
.be16
[2]) & mask
.be16
[2]));
221 static inline uint64_t eth_addr_to_uint64(const struct eth_addr ea
)
223 return (((uint64_t) ntohs(ea
.be16
[0]) << 32)
224 | ((uint64_t) ntohs(ea
.be16
[1]) << 16)
225 | ntohs(ea
.be16
[2]));
228 static inline uint64_t eth_addr_vlan_to_uint64(const struct eth_addr ea
,
231 return (((uint64_t)vlan
<< 48) | eth_addr_to_uint64(ea
));
234 static inline void eth_addr_from_uint64(uint64_t x
, struct eth_addr
*ea
)
236 ea
->be16
[0] = htons(x
>> 32);
237 ea
->be16
[1] = htons((x
& 0xFFFF0000) >> 16);
238 ea
->be16
[2] = htons(x
& 0xFFFF);
241 static inline struct eth_addr
eth_addr_invert(const struct eth_addr src
)
245 for (int i
= 0; i
< ARRAY_SIZE(src
.be16
); i
++) {
246 dst
.be16
[i
] = ~src
.be16
[i
];
252 static inline void eth_addr_mark_random(struct eth_addr
*ea
)
254 ea
->ea
[0] &= ~1; /* Unicast. */
255 ea
->ea
[0] |= 2; /* Private. */
258 static inline void eth_addr_random(struct eth_addr
*ea
)
260 random_bytes((uint8_t *)ea
, sizeof *ea
);
261 eth_addr_mark_random(ea
);
264 static inline void eth_addr_nicira_random(struct eth_addr
*ea
)
268 /* Set the OUI to the Nicira one. */
273 /* Set the top bit to indicate random Nicira address. */
276 static inline uint32_t hash_mac(const struct eth_addr ea
,
277 const uint16_t vlan
, const uint32_t basis
)
279 return hash_uint64_basis(eth_addr_vlan_to_uint64(ea
, vlan
), basis
);
282 bool eth_addr_is_reserved(const struct eth_addr
);
283 bool eth_addr_from_string(const char *, struct eth_addr
*);
285 void compose_rarp(struct dp_packet
*, const struct eth_addr
);
287 void eth_push_vlan(struct dp_packet
*, ovs_be16 tpid
, ovs_be16 tci
);
288 void eth_pop_vlan(struct dp_packet
*);
290 const char *eth_from_hex(const char *hex
, struct dp_packet
**packetp
);
291 void eth_format_masked(const struct eth_addr ea
,
292 const struct eth_addr
*mask
, struct ds
*s
);
294 void set_mpls_lse(struct dp_packet
*, ovs_be32 label
);
295 void push_mpls(struct dp_packet
*packet
, ovs_be16 ethtype
, ovs_be32 lse
);
296 void pop_mpls(struct dp_packet
*, ovs_be16 ethtype
);
298 void set_mpls_lse_ttl(ovs_be32
*lse
, uint8_t ttl
);
299 void set_mpls_lse_tc(ovs_be32
*lse
, uint8_t tc
);
300 void set_mpls_lse_label(ovs_be32
*lse
, ovs_be32 label
);
301 void set_mpls_lse_bos(ovs_be32
*lse
, uint8_t bos
);
302 ovs_be32
set_mpls_lse_values(uint8_t ttl
, uint8_t tc
, uint8_t bos
,
307 * struct eth_addr mac;
309 * printf("The Ethernet address is "ETH_ADDR_FMT"\n", ETH_ADDR_ARGS(mac));
312 #define ETH_ADDR_FMT \
313 "%02"PRIx8":%02"PRIx8":%02"PRIx8":%02"PRIx8":%02"PRIx8":%02"PRIx8
314 #define ETH_ADDR_ARGS(EA) ETH_ADDR_BYTES_ARGS((EA).ea)
315 #define ETH_ADDR_BYTES_ARGS(EAB) \
316 (EAB)[0], (EAB)[1], (EAB)[2], (EAB)[3], (EAB)[4], (EAB)[5]
320 * char *string = "1 00:11:22:33:44:55 2";
321 * struct eth_addr mac;
324 * if (ovs_scan(string, "%d"ETH_ADDR_SCAN_FMT"%d",
325 * &a, ETH_ADDR_SCAN_ARGS(mac), &b)) {
329 #define ETH_ADDR_SCAN_FMT "%"SCNx8":%"SCNx8":%"SCNx8":%"SCNx8":%"SCNx8":%"SCNx8
330 #define ETH_ADDR_SCAN_ARGS(EA) \
331 &(EA).ea[0], &(EA).ea[1], &(EA).ea[2], &(EA).ea[3], &(EA).ea[4], &(EA).ea[5]
333 #define ETH_TYPE_IP 0x0800
334 #define ETH_TYPE_ARP 0x0806
335 #define ETH_TYPE_TEB 0x6558
336 #define ETH_TYPE_VLAN_8021Q 0x8100
337 #define ETH_TYPE_VLAN ETH_TYPE_VLAN_8021Q
338 #define ETH_TYPE_VLAN_8021AD 0x88a8
339 #define ETH_TYPE_IPV6 0x86dd
340 #define ETH_TYPE_LACP 0x8809
341 #define ETH_TYPE_RARP 0x8035
342 #define ETH_TYPE_MPLS 0x8847
343 #define ETH_TYPE_MPLS_MCAST 0x8848
345 static inline bool eth_type_mpls(ovs_be16 eth_type
)
347 return eth_type
== htons(ETH_TYPE_MPLS
) ||
348 eth_type
== htons(ETH_TYPE_MPLS_MCAST
);
351 static inline bool eth_type_vlan(ovs_be16 eth_type
)
353 return eth_type
== htons(ETH_TYPE_VLAN_8021Q
) ||
354 eth_type
== htons(ETH_TYPE_VLAN_8021AD
);
358 /* Minimum value for an Ethernet type. Values below this are IEEE 802.2 frame
360 #define ETH_TYPE_MIN 0x600
362 #define ETH_HEADER_LEN 14
363 #define ETH_PAYLOAD_MIN 46
364 #define ETH_PAYLOAD_MAX 1500
365 #define ETH_TOTAL_MIN (ETH_HEADER_LEN + ETH_PAYLOAD_MIN)
366 #define ETH_TOTAL_MAX (ETH_HEADER_LEN + ETH_PAYLOAD_MAX)
367 #define ETH_VLAN_TOTAL_MAX (ETH_HEADER_LEN + VLAN_HEADER_LEN + ETH_PAYLOAD_MAX)
370 struct eth_addr eth_dst
;
371 struct eth_addr eth_src
;
374 BUILD_ASSERT_DECL(ETH_HEADER_LEN
== sizeof(struct eth_header
));
376 #define LLC_DSAP_SNAP 0xaa
377 #define LLC_SSAP_SNAP 0xaa
378 #define LLC_CNTL_SNAP 3
380 #define LLC_HEADER_LEN 3
387 BUILD_ASSERT_DECL(LLC_HEADER_LEN
== sizeof(struct llc_header
));
389 /* LLC field values used for STP frames. */
390 #define STP_LLC_SSAP 0x42
391 #define STP_LLC_DSAP 0x42
392 #define STP_LLC_CNTL 0x03
394 #define SNAP_ORG_ETHERNET "\0\0" /* The compiler adds a null byte, so
395 sizeof(SNAP_ORG_ETHERNET) == 3. */
396 #define SNAP_HEADER_LEN 5
402 BUILD_ASSERT_DECL(SNAP_HEADER_LEN
== sizeof(struct snap_header
));
404 #define LLC_SNAP_HEADER_LEN (LLC_HEADER_LEN + SNAP_HEADER_LEN)
406 struct llc_snap_header
{
407 struct llc_header llc
;
408 struct snap_header snap
;
410 BUILD_ASSERT_DECL(LLC_SNAP_HEADER_LEN
== sizeof(struct llc_snap_header
));
412 #define VLAN_VID_MASK 0x0fff
413 #define VLAN_VID_SHIFT 0
415 #define VLAN_PCP_MASK 0xe000
416 #define VLAN_PCP_SHIFT 13
418 #define VLAN_CFI 0x1000
419 #define VLAN_CFI_SHIFT 12
421 /* Given the vlan_tci field from an 802.1Q header, in network byte order,
422 * returns the VLAN ID in host byte order. */
423 static inline uint16_t
424 vlan_tci_to_vid(ovs_be16 vlan_tci
)
426 return (ntohs(vlan_tci
) & VLAN_VID_MASK
) >> VLAN_VID_SHIFT
;
429 /* Given the vlan_tci field from an 802.1Q header, in network byte order,
430 * returns the priority code point (PCP) in host byte order. */
432 vlan_tci_to_pcp(ovs_be16 vlan_tci
)
434 return (ntohs(vlan_tci
) & VLAN_PCP_MASK
) >> VLAN_PCP_SHIFT
;
437 /* Given the vlan_tci field from an 802.1Q header, in network byte order,
438 * returns the Canonical Format Indicator (CFI). */
440 vlan_tci_to_cfi(ovs_be16 vlan_tci
)
442 return (vlan_tci
& htons(VLAN_CFI
)) != 0;
445 #define VLAN_HEADER_LEN 4
447 ovs_be16 vlan_tci
; /* Lowest 12 bits are VLAN ID. */
448 ovs_be16 vlan_next_type
;
450 BUILD_ASSERT_DECL(VLAN_HEADER_LEN
== sizeof(struct vlan_header
));
452 #define VLAN_ETH_HEADER_LEN (ETH_HEADER_LEN + VLAN_HEADER_LEN)
454 struct vlan_eth_header
{
455 struct eth_addr veth_dst
;
456 struct eth_addr veth_src
;
457 ovs_be16 veth_type
; /* Always htons(ETH_TYPE_VLAN). */
458 ovs_be16 veth_tci
; /* Lowest 12 bits are VLAN ID. */
459 ovs_be16 veth_next_type
;
461 BUILD_ASSERT_DECL(VLAN_ETH_HEADER_LEN
== sizeof(struct vlan_eth_header
));
463 /* MPLS related definitions */
464 #define MPLS_TTL_MASK 0x000000ff
465 #define MPLS_TTL_SHIFT 0
467 #define MPLS_BOS_MASK 0x00000100
468 #define MPLS_BOS_SHIFT 8
470 #define MPLS_TC_MASK 0x00000e00
471 #define MPLS_TC_SHIFT 9
473 #define MPLS_LABEL_MASK 0xfffff000
474 #define MPLS_LABEL_SHIFT 12
479 ovs_16aligned_be32 mpls_lse
;
481 BUILD_ASSERT_DECL(MPLS_HLEN
== sizeof(struct mpls_hdr
));
483 /* Given a mpls label stack entry in network byte order
484 * return mpls label in host byte order */
485 static inline uint32_t
486 mpls_lse_to_label(ovs_be32 mpls_lse
)
488 return (ntohl(mpls_lse
) & MPLS_LABEL_MASK
) >> MPLS_LABEL_SHIFT
;
491 /* Given a mpls label stack entry in network byte order
493 static inline uint8_t
494 mpls_lse_to_tc(ovs_be32 mpls_lse
)
496 return (ntohl(mpls_lse
) & MPLS_TC_MASK
) >> MPLS_TC_SHIFT
;
499 /* Given a mpls label stack entry in network byte order
501 static inline uint8_t
502 mpls_lse_to_ttl(ovs_be32 mpls_lse
)
504 return (ntohl(mpls_lse
) & MPLS_TTL_MASK
) >> MPLS_TTL_SHIFT
;
507 /* Set TTL in mpls lse. */
509 flow_set_mpls_lse_ttl(ovs_be32
*mpls_lse
, uint8_t ttl
)
511 *mpls_lse
&= ~htonl(MPLS_TTL_MASK
);
512 *mpls_lse
|= htonl(ttl
<< MPLS_TTL_SHIFT
);
515 /* Given a mpls label stack entry in network byte order
516 * return mpls BoS bit */
517 static inline uint8_t
518 mpls_lse_to_bos(ovs_be32 mpls_lse
)
520 return (mpls_lse
& htonl(MPLS_BOS_MASK
)) != 0;
523 #define IP_FMT "%"PRIu32".%"PRIu32".%"PRIu32".%"PRIu32
524 #define IP_ARGS(ip) \
526 (ntohl(ip) >> 16) & 0xff, \
527 (ntohl(ip) >> 8) & 0xff, \
532 * char *string = "1 33.44.55.66 2";
536 * if (ovs_scan(string, "%d"IP_SCAN_FMT"%d", &a, IP_SCAN_ARGS(&ip), &b)) {
540 #define IP_SCAN_FMT "%"SCNu8".%"SCNu8".%"SCNu8".%"SCNu8
541 #define IP_SCAN_ARGS(ip) \
542 ((void) (ovs_be32) *(ip), &((uint8_t *) ip)[0]), \
543 &((uint8_t *) ip)[1], \
544 &((uint8_t *) ip)[2], \
547 /* Returns true if 'netmask' is a CIDR netmask, that is, if it consists of N
548 * high-order 1-bits and 32-N low-order 0-bits. */
550 ip_is_cidr(ovs_be32 netmask
)
552 uint32_t x
= ~ntohl(netmask
);
553 return !(x
& (x
+ 1));
556 ip_is_multicast(ovs_be32 ip
)
558 return (ip
& htonl(0xf0000000)) == htonl(0xe0000000);
561 ip_is_local_multicast(ovs_be32 ip
)
563 return (ip
& htonl(0xffffff00)) == htonl(0xe0000000);
565 int ip_count_cidr_bits(ovs_be32 netmask
);
566 void ip_format_masked(ovs_be32 ip
, ovs_be32 mask
, struct ds
*);
568 #define IP_VER(ip_ihl_ver) ((ip_ihl_ver) >> 4)
569 #define IP_IHL(ip_ihl_ver) ((ip_ihl_ver) & 15)
570 #define IP_IHL_VER(ihl, ver) (((ver) << 4) | (ihl))
573 #define IPPROTO_SCTP 132
577 #define IP_ECN_NOT_ECT 0x0
578 #define IP_ECN_ECT_1 0x01
579 #define IP_ECN_ECT_0 0x02
580 #define IP_ECN_CE 0x03
581 #define IP_ECN_MASK 0x03
582 #define IP_DSCP_MASK 0xfc
586 #define IP_DONT_FRAGMENT 0x4000 /* Don't fragment. */
587 #define IP_MORE_FRAGMENTS 0x2000 /* More fragments. */
588 #define IP_FRAG_OFF_MASK 0x1fff /* Fragment offset. */
589 #define IP_IS_FRAGMENT(ip_frag_off) \
590 ((ip_frag_off) & htons(IP_MORE_FRAGMENTS | IP_FRAG_OFF_MASK))
592 #define IP_HEADER_LEN 20
598 ovs_be16 ip_frag_off
;
602 ovs_16aligned_be32 ip_src
;
603 ovs_16aligned_be32 ip_dst
;
606 BUILD_ASSERT_DECL(IP_HEADER_LEN
== sizeof(struct ip_header
));
608 #define ICMP_HEADER_LEN 8
622 ovs_16aligned_be32 gateway
;
625 BUILD_ASSERT_DECL(ICMP_HEADER_LEN
== sizeof(struct icmp_header
));
627 #define IGMP_HEADER_LEN 8
632 ovs_16aligned_be32 group
;
634 BUILD_ASSERT_DECL(IGMP_HEADER_LEN
== sizeof(struct igmp_header
));
636 #define IGMPV3_HEADER_LEN 8
637 struct igmpv3_header
{
644 BUILD_ASSERT_DECL(IGMPV3_HEADER_LEN
== sizeof(struct igmpv3_header
));
646 #define IGMPV3_RECORD_LEN 8
647 struct igmpv3_record
{
651 ovs_16aligned_be32 maddr
;
653 BUILD_ASSERT_DECL(IGMPV3_RECORD_LEN
== sizeof(struct igmpv3_record
));
655 #define IGMP_HOST_MEMBERSHIP_QUERY 0x11 /* From RFC1112 */
656 #define IGMP_HOST_MEMBERSHIP_REPORT 0x12 /* Ditto */
657 #define IGMPV2_HOST_MEMBERSHIP_REPORT 0x16 /* V2 version of 0x12 */
658 #define IGMP_HOST_LEAVE_MESSAGE 0x17
659 #define IGMPV3_HOST_MEMBERSHIP_REPORT 0x22 /* V3 version of 0x12 */
662 * IGMPv3 and MLDv2 use the same codes.
664 #define IGMPV3_MODE_IS_INCLUDE 1
665 #define IGMPV3_MODE_IS_EXCLUDE 2
666 #define IGMPV3_CHANGE_TO_INCLUDE_MODE 3
667 #define IGMPV3_CHANGE_TO_EXCLUDE_MODE 4
668 #define IGMPV3_ALLOW_NEW_SOURCES 5
669 #define IGMPV3_BLOCK_OLD_SOURCES 6
671 #define SCTP_HEADER_LEN 12
675 ovs_16aligned_be32 sctp_vtag
;
676 ovs_16aligned_be32 sctp_csum
;
678 BUILD_ASSERT_DECL(SCTP_HEADER_LEN
== sizeof(struct sctp_header
));
680 #define UDP_HEADER_LEN 8
687 BUILD_ASSERT_DECL(UDP_HEADER_LEN
== sizeof(struct udp_header
));
689 #define TCP_FIN 0x001
690 #define TCP_SYN 0x002
691 #define TCP_RST 0x004
692 #define TCP_PSH 0x008
693 #define TCP_ACK 0x010
694 #define TCP_URG 0x020
695 #define TCP_ECE 0x040
696 #define TCP_CWR 0x080
699 #define TCP_CTL(flags, offset) (htons((flags) | ((offset) << 12)))
700 #define TCP_FLAGS(tcp_ctl) (ntohs(tcp_ctl) & 0x0fff)
701 #define TCP_FLAGS_BE16(tcp_ctl) ((tcp_ctl) & htons(0x0fff))
702 #define TCP_OFFSET(tcp_ctl) (ntohs(tcp_ctl) >> 12)
704 #define TCP_HEADER_LEN 20
708 ovs_16aligned_be32 tcp_seq
;
709 ovs_16aligned_be32 tcp_ack
;
715 BUILD_ASSERT_DECL(TCP_HEADER_LEN
== sizeof(struct tcp_header
));
717 /* Connection states */
719 #define CS_ESTABLISHED 0x02
720 #define CS_RELATED 0x04
721 #define CS_INVALID 0x20
722 #define CS_REPLY_DIR 0x40
723 #define CS_TRACKED 0x80
725 /* Undefined connection state bits. */
726 #define CS_SUPPORTED_MASK (CS_NEW | CS_ESTABLISHED | CS_RELATED \
727 | CS_INVALID | CS_REPLY_DIR | CS_TRACKED)
728 #define CS_UNSUPPORTED_MASK (~(uint32_t)CS_SUPPORTED_MASK)
730 #define ARP_HRD_ETHERNET 1
731 #define ARP_PRO_IP 0x0800
732 #define ARP_OP_REQUEST 1
733 #define ARP_OP_REPLY 2
734 #define ARP_OP_RARP 3
736 #define ARP_ETH_HEADER_LEN 28
737 struct arp_eth_header
{
738 /* Generic members. */
739 ovs_be16 ar_hrd
; /* Hardware type. */
740 ovs_be16 ar_pro
; /* Protocol type. */
741 uint8_t ar_hln
; /* Hardware address length. */
742 uint8_t ar_pln
; /* Protocol address length. */
743 ovs_be16 ar_op
; /* Opcode. */
745 /* Ethernet+IPv4 specific members. */
746 struct eth_addr ar_sha
; /* Sender hardware address. */
747 ovs_16aligned_be32 ar_spa
; /* Sender protocol address. */
748 struct eth_addr ar_tha
; /* Target hardware address. */
749 ovs_16aligned_be32 ar_tpa
; /* Target protocol address. */
751 BUILD_ASSERT_DECL(ARP_ETH_HEADER_LEN
== sizeof(struct arp_eth_header
));
753 /* Like struct in6_addr, but whereas that struct requires 32-bit alignment on
754 * most implementations, this one only requires 16-bit alignment. */
755 union ovs_16aligned_in6_addr
{
757 ovs_16aligned_be32 be32
[4];
760 /* Like struct in6_hdr, but whereas that struct requires 32-bit alignment, this
761 * one only requires 16-bit alignment. */
762 struct ovs_16aligned_ip6_hdr
{
764 struct ovs_16aligned_ip6_hdrctl
{
765 ovs_16aligned_be32 ip6_un1_flow
;
766 ovs_be16 ip6_un1_plen
;
768 uint8_t ip6_un1_hlim
;
772 union ovs_16aligned_in6_addr ip6_src
;
773 union ovs_16aligned_in6_addr ip6_dst
;
776 /* Like struct in6_frag, but whereas that struct requires 32-bit alignment,
777 * this one only requires 16-bit alignment. */
778 struct ovs_16aligned_ip6_frag
{
780 uint8_t ip6f_reserved
;
782 ovs_16aligned_be32 ip6f_ident
;
785 #define ICMP6_HEADER_LEN 4
786 struct icmp6_header
{
789 ovs_be16 icmp6_cksum
;
791 BUILD_ASSERT_DECL(ICMP6_HEADER_LEN
== sizeof(struct icmp6_header
));
793 /* Neighbor Discovery option field.
794 * ND options are always a multiple of 8 bytes in size. */
797 uint8_t nd_opt_type
; /* Values defined in icmp6.h */
798 uint8_t nd_opt_len
; /* in units of 8 octets (the size of this struct) */
799 struct eth_addr nd_opt_mac
; /* Ethernet address in the case of SLL or TLL options */
801 BUILD_ASSERT_DECL(ND_OPT_LEN
== sizeof(struct ovs_nd_opt
));
803 /* Like struct nd_msg (from ndisc.h), but whereas that struct requires 32-bit
804 * alignment, this one only requires 16-bit alignment. */
805 #define ND_MSG_LEN 24
807 struct icmp6_header icmph
;
808 ovs_16aligned_be32 rco_flags
;
809 union ovs_16aligned_in6_addr target
;
810 struct ovs_nd_opt options
[0];
812 BUILD_ASSERT_DECL(ND_MSG_LEN
== sizeof(struct ovs_nd_msg
));
815 * Use the same struct for MLD and MLD2, naming members as the defined fields in
816 * in the corresponding version of the protocol, though they are reserved in the
819 #define MLD_HEADER_LEN 8
827 BUILD_ASSERT_DECL(MLD_HEADER_LEN
== sizeof(struct mld_header
));
829 #define MLD2_RECORD_LEN 20
834 union ovs_16aligned_in6_addr maddr
;
836 BUILD_ASSERT_DECL(MLD2_RECORD_LEN
== sizeof(struct mld2_record
));
838 #define MLD_QUERY 130
839 #define MLD_REPORT 131
841 #define MLD2_REPORT 143
843 /* The IPv6 flow label is in the lower 20 bits of the first 32-bit word. */
844 #define IPV6_LABEL_MASK 0x000fffff
848 * char *string = "1 ::1 2";
849 * char ipv6_s[IPV6_SCAN_LEN + 1];
850 * struct in6_addr ipv6;
852 * if (ovs_scan(string, "%d"IPV6_SCAN_FMT"%d", &a, ipv6_s, &b)
853 * && inet_pton(AF_INET6, ipv6_s, &ipv6) == 1) {
857 #define IPV6_SCAN_FMT "%46[0123456789abcdefABCDEF:.]"
858 #define IPV6_SCAN_LEN 46
860 extern const struct in6_addr in6addr_exact
;
861 #define IN6ADDR_EXACT_INIT { { { 0xff,0xff,0xff,0xff,0xff,0xff,0xff,0xff, \
862 0xff,0xff,0xff,0xff,0xff,0xff,0xff,0xff } } }
864 extern const struct in6_addr in6addr_all_hosts
;
865 #define IN6ADDR_ALL_HOSTS_INIT { { { 0xff,0x02,0x00,0x00,0x00,0x00,0x00,0x00, \
866 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x01 } } }
868 static inline bool ipv6_addr_equals(const struct in6_addr
*a
,
869 const struct in6_addr
*b
)
871 #ifdef IN6_ARE_ADDR_EQUAL
872 return IN6_ARE_ADDR_EQUAL(a
, b
);
874 return !memcmp(a
, b
, sizeof(*a
));
878 static inline bool ipv6_mask_is_any(const struct in6_addr
*mask
) {
879 return ipv6_addr_equals(mask
, &in6addr_any
);
882 static inline bool ipv6_mask_is_exact(const struct in6_addr
*mask
) {
883 return ipv6_addr_equals(mask
, &in6addr_exact
);
886 static inline bool ipv6_is_all_hosts(const struct in6_addr
*addr
) {
887 return ipv6_addr_equals(addr
, &in6addr_all_hosts
);
890 static inline bool ipv6_addr_is_set(const struct in6_addr
*addr
) {
891 return !ipv6_addr_equals(addr
, &in6addr_any
);
894 static inline bool ipv6_addr_is_multicast(const struct in6_addr
*ip
) {
895 return ip
->s6_addr
[0] == 0xff;
899 in6_addr_set_mapped_ipv4(struct in6_addr
*addr
, ovs_be32 ip4
)
901 union ovs_16aligned_in6_addr
*taddr
= (void *) addr
;
902 memset(taddr
->be16
, 0, sizeof(taddr
->be16
));
903 taddr
->be16
[5] = OVS_BE16_MAX
;
904 put_16aligned_be32(&taddr
->be32
[3], ip4
);
907 static inline ovs_be32
908 in6_addr_get_mapped_ipv4(const struct in6_addr
*addr
)
910 union ovs_16aligned_in6_addr
*taddr
= (void *) addr
;
911 if (IN6_IS_ADDR_V4MAPPED(addr
)) {
912 return get_16aligned_be32(&taddr
->be32
[3]);
918 static inline bool dl_type_is_ip_any(ovs_be16 dl_type
)
920 return dl_type
== htons(ETH_TYPE_IP
)
921 || dl_type
== htons(ETH_TYPE_IPV6
);
926 /* GRE protocol header */
927 struct gre_base_hdr
{
932 #define GRE_CSUM 0x8000
933 #define GRE_ROUTING 0x4000
934 #define GRE_KEY 0x2000
935 #define GRE_SEQ 0x1000
936 #define GRE_STRICT 0x0800
937 #define GRE_REC 0x0700
938 #define GRE_FLAGS 0x00F8
939 #define GRE_VERSION 0x0007
941 /* VXLAN protocol header */
943 ovs_16aligned_be32 vx_flags
;
944 ovs_16aligned_be32 vx_vni
;
947 #define VXLAN_FLAGS 0x08000000 /* struct vxlanhdr.vx_flags required value. */
949 void format_ipv6_addr(char *addr_str
, const struct in6_addr
*addr
);
950 void print_ipv6_addr(struct ds
*string
, const struct in6_addr
*addr
);
951 void print_ipv6_mapped(struct ds
*string
, const struct in6_addr
*addr
);
952 void print_ipv6_masked(struct ds
*string
, const struct in6_addr
*addr
,
953 const struct in6_addr
*mask
);
954 struct in6_addr
ipv6_addr_bitand(const struct in6_addr
*src
,
955 const struct in6_addr
*mask
);
956 struct in6_addr
ipv6_create_mask(int mask
);
957 int ipv6_count_cidr_bits(const struct in6_addr
*netmask
);
958 bool ipv6_is_cidr(const struct in6_addr
*netmask
);
960 void *eth_compose(struct dp_packet
*, const struct eth_addr eth_dst
,
961 const struct eth_addr eth_src
, uint16_t eth_type
,
963 void *snap_compose(struct dp_packet
*, const struct eth_addr eth_dst
,
964 const struct eth_addr eth_src
,
965 unsigned int oui
, uint16_t snap_type
, size_t size
);
966 void packet_set_ipv4(struct dp_packet
*, ovs_be32 src
, ovs_be32 dst
, uint8_t tos
,
968 void packet_set_ipv6(struct dp_packet
*, uint8_t proto
, const ovs_be32 src
[4],
969 const ovs_be32 dst
[4], uint8_t tc
,
970 ovs_be32 fl
, uint8_t hlmit
);
971 void packet_set_tcp_port(struct dp_packet
*, ovs_be16 src
, ovs_be16 dst
);
972 void packet_set_udp_port(struct dp_packet
*, ovs_be16 src
, ovs_be16 dst
);
973 void packet_set_sctp_port(struct dp_packet
*, ovs_be16 src
, ovs_be16 dst
);
974 void packet_set_nd(struct dp_packet
*, const ovs_be32 target
[4],
975 const struct eth_addr sll
, const struct eth_addr tll
);
977 void packet_format_tcp_flags(struct ds
*, uint16_t);
978 const char *packet_tcp_flag_to_string(uint32_t flag
);
979 void compose_arp(struct dp_packet
*, uint16_t arp_op
,
980 const struct eth_addr arp_sha
,
981 const struct eth_addr arp_tha
, bool broadcast
,
982 ovs_be32 arp_spa
, ovs_be32 arp_tpa
);
983 uint32_t packet_csum_pseudoheader(const struct ip_header
*);
985 #endif /* packets.h */