2 * Copyright (c) 2008, 2009, 2010, 2011, 2012, 2013, 2014, 2015, 2016, 2017 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>
25 #include "openvswitch/geneve.h"
26 #include "openvswitch/packets.h"
27 #include "openvswitch/types.h"
28 #include "openvswitch/nsh.h"
29 #include "odp-netlink.h"
32 #include "tun-metadata.h"
33 #include "unaligned.h"
41 /* Purely internal to OVS userspace. These flags should never be exposed to
42 * the outside world and so aren't included in the flags mask. */
44 /* Tunnel information is in userspace datapath format. */
45 #define FLOW_TNL_F_UDPIF (1 << 4)
47 static inline bool ipv6_addr_is_set(const struct in6_addr
*addr
);
50 flow_tnl_dst_is_set(const struct flow_tnl
*tnl
)
52 return tnl
->ip_dst
|| ipv6_addr_is_set(&tnl
->ipv6_dst
);
56 flow_tnl_src_is_set(const struct flow_tnl
*tnl
)
58 return tnl
->ip_src
|| ipv6_addr_is_set(&tnl
->ipv6_src
);
61 struct in6_addr
flow_tnl_dst(const struct flow_tnl
*tnl
);
62 struct in6_addr
flow_tnl_src(const struct flow_tnl
*tnl
);
64 /* Returns an offset to 'src' covering all the meaningful fields in 'src'. */
66 flow_tnl_size(const struct flow_tnl
*src
)
68 if (!flow_tnl_dst_is_set(src
)) {
69 /* Covers ip_dst and ipv6_dst only. */
70 return offsetof(struct flow_tnl
, ip_src
);
72 if (src
->flags
& FLOW_TNL_F_UDPIF
) {
73 /* Datapath format, cover all options we have. */
74 return offsetof(struct flow_tnl
, metadata
.opts
)
75 + src
->metadata
.present
.len
;
77 if (!src
->metadata
.present
.map
) {
78 /* No TLVs, opts is irrelevant. */
79 return offsetof(struct flow_tnl
, metadata
.opts
);
81 /* Have decoded TLVs, opts is relevant. */
85 /* Copy flow_tnl, but avoid copying unused portions of tun_metadata. Unused
86 * data in 'dst' is NOT cleared, so this must not be used in cases where the
87 * uninitialized portion may be hashed over. */
89 flow_tnl_copy__(struct flow_tnl
*dst
, const struct flow_tnl
*src
)
91 memcpy(dst
, src
, flow_tnl_size(src
));
95 flow_tnl_equal(const struct flow_tnl
*a
, const struct flow_tnl
*b
)
97 size_t a_size
= flow_tnl_size(a
);
99 return a_size
== flow_tnl_size(b
) && !memcmp(a
, b
, a_size
);
102 /* Datapath packet metadata */
103 struct pkt_metadata
{
104 PADDED_MEMBERS_CACHELINE_MARKER(CACHE_LINE_SIZE
, cacheline0
,
105 uint32_t recirc_id
; /* Recirculation id carried with the
106 recirculating packets. 0 for packets
107 received from the wire. */
108 uint32_t dp_hash
; /* hash value computed by the recirculation
110 uint32_t skb_priority
; /* Packet priority for QoS. */
111 uint32_t pkt_mark
; /* Packet mark. */
112 uint8_t ct_state
; /* Connection state. */
113 bool ct_orig_tuple_ipv6
;
114 uint16_t ct_zone
; /* Connection zone. */
115 uint32_t ct_mark
; /* Connection mark. */
116 ovs_u128 ct_label
; /* Connection label. */
117 union flow_in_port in_port
; /* Input port. */
118 struct conn
*conn
; /* Cached conntrack connection. */
119 bool reply
; /* True if reply direction. */
120 bool icmp_related
; /* True if ICMP related. */
123 PADDED_MEMBERS_CACHELINE_MARKER(CACHE_LINE_SIZE
, cacheline1
,
124 union { /* Populated only for non-zero 'ct_state'. */
125 struct ovs_key_ct_tuple_ipv4 ipv4
;
126 struct ovs_key_ct_tuple_ipv6 ipv6
; /* Used only if */
127 } ct_orig_tuple
; /* 'ct_orig_tuple_ipv6' is set */
130 PADDED_MEMBERS_CACHELINE_MARKER(CACHE_LINE_SIZE
, cacheline2
,
131 struct flow_tnl tunnel
; /* Encapsulating tunnel parameters. Note that
132 * if 'ip_dst' == 0, the rest of the fields may
133 * be uninitialized. */
137 BUILD_ASSERT_DECL(offsetof(struct pkt_metadata
, cacheline0
) == 0);
138 BUILD_ASSERT_DECL(offsetof(struct pkt_metadata
, cacheline1
) ==
140 BUILD_ASSERT_DECL(offsetof(struct pkt_metadata
, cacheline2
) ==
141 2 * CACHE_LINE_SIZE
);
144 pkt_metadata_init_tnl(struct pkt_metadata
*md
)
146 /* Zero up through the tunnel metadata options. The length and table
147 * are before this and as long as they are empty, the options won't
149 memset(md
, 0, offsetof(struct pkt_metadata
, tunnel
.metadata
.opts
));
153 pkt_metadata_init_conn(struct pkt_metadata
*md
)
159 pkt_metadata_init(struct pkt_metadata
*md
, odp_port_t port
)
161 /* This is called for every packet in userspace datapath and affects
162 * performance if all the metadata is initialized. Hence, fields should
163 * only be zeroed out when necessary.
165 * Initialize only till ct_state. Once the ct_state is zeroed out rest
166 * of ct fields will not be looked at unless ct_state != 0.
168 memset(md
, 0, offsetof(struct pkt_metadata
, ct_orig_tuple_ipv6
));
170 /* It can be expensive to zero out all of the tunnel metadata. However,
171 * we can just zero out ip_dst and the rest of the data will never be
173 md
->tunnel
.ip_dst
= 0;
174 md
->tunnel
.ipv6_dst
= in6addr_any
;
175 md
->in_port
.odp_port
= port
;
179 /* This function prefetches the cachelines touched by pkt_metadata_init()
180 * and pkt_metadata_init_tnl(). For performance reasons the two functions
181 * should be kept in sync. */
183 pkt_metadata_prefetch_init(struct pkt_metadata
*md
)
185 /* Prefetch cacheline0 as members till ct_state and odp_port will
186 * be initialized later in pkt_metadata_init(). */
187 OVS_PREFETCH(md
->cacheline0
);
189 /* Prefetch cacheline1 as members of this cacheline will be zeroed out
190 * in pkt_metadata_init_tnl(). */
191 OVS_PREFETCH(md
->cacheline1
);
193 /* Prefetch cachline2 as ip_dst & ipv6_dst fields will be initialized. */
194 OVS_PREFETCH(md
->cacheline2
);
197 bool dpid_from_string(const char *s
, uint64_t *dpidp
);
199 #define ETH_ADDR_LEN 6
201 static const struct eth_addr eth_addr_broadcast OVS_UNUSED
202 = ETH_ADDR_C(ff
,ff
,ff
,ff
,ff
,ff
);
204 static const struct eth_addr eth_addr_exact OVS_UNUSED
205 = ETH_ADDR_C(ff
,ff
,ff
,ff
,ff
,ff
);
207 static const struct eth_addr eth_addr_zero OVS_UNUSED
208 = ETH_ADDR_C(00,00,00,00,00,00);
209 static const struct eth_addr64 eth_addr64_zero OVS_UNUSED
210 = ETH_ADDR64_C(00,00,00,00,00,00,00,00);
212 static const struct eth_addr eth_addr_stp OVS_UNUSED
213 = ETH_ADDR_C(01,80,c2
,00,00,00);
215 static const struct eth_addr eth_addr_lacp OVS_UNUSED
216 = ETH_ADDR_C(01,80,c2
,00,00,02);
218 static const struct eth_addr eth_addr_bfd OVS_UNUSED
219 = ETH_ADDR_C(00,23,20,00,00,01);
221 static inline bool eth_addr_is_broadcast(const struct eth_addr a
)
223 return (a
.be16
[0] & a
.be16
[1] & a
.be16
[2]) == htons(0xffff);
226 static inline bool eth_addr_is_multicast(const struct eth_addr a
)
231 static inline bool eth_addr_is_local(const struct eth_addr a
)
233 /* Local if it is either a locally administered address or a Nicira random
236 || (a
.be16
[0] == htons(0x0023)
237 && (a
.be16
[1] & htons(0xff80)) == htons(0x2080));
239 static inline bool eth_addr_is_zero(const struct eth_addr a
)
241 return !(a
.be16
[0] | a
.be16
[1] | a
.be16
[2]);
243 static inline bool eth_addr64_is_zero(const struct eth_addr64 a
)
245 return !(a
.be16
[0] | a
.be16
[1] | a
.be16
[2] | a
.be16
[3]);
248 static inline int eth_mask_is_exact(const struct eth_addr a
)
250 return (a
.be16
[0] & a
.be16
[1] & a
.be16
[2]) == htons(0xffff);
253 static inline int eth_addr_compare_3way(const struct eth_addr a
,
254 const struct eth_addr b
)
256 return memcmp(&a
, &b
, sizeof a
);
258 static inline int eth_addr64_compare_3way(const struct eth_addr64 a
,
259 const struct eth_addr64 b
)
261 return memcmp(&a
, &b
, sizeof a
);
264 static inline bool eth_addr_equals(const struct eth_addr a
,
265 const struct eth_addr b
)
267 return !eth_addr_compare_3way(a
, b
);
269 static inline bool eth_addr64_equals(const struct eth_addr64 a
,
270 const struct eth_addr64 b
)
272 return !eth_addr64_compare_3way(a
, b
);
275 static inline bool eth_addr_equal_except(const struct eth_addr a
,
276 const struct eth_addr b
,
277 const struct eth_addr mask
)
279 return !(((a
.be16
[0] ^ b
.be16
[0]) & mask
.be16
[0])
280 || ((a
.be16
[1] ^ b
.be16
[1]) & mask
.be16
[1])
281 || ((a
.be16
[2] ^ b
.be16
[2]) & mask
.be16
[2]));
284 uint64_t eth_addr_to_uint64(const struct eth_addr ea
);
286 static inline uint64_t eth_addr_vlan_to_uint64(const struct eth_addr ea
,
289 return (((uint64_t)vlan
<< 48) | eth_addr_to_uint64(ea
));
292 void eth_addr_from_uint64(uint64_t x
, struct eth_addr
*ea
);
294 static inline struct eth_addr
eth_addr_invert(const struct eth_addr src
)
298 for (int i
= 0; i
< ARRAY_SIZE(src
.be16
); i
++) {
299 dst
.be16
[i
] = ~src
.be16
[i
];
305 void eth_addr_mark_random(struct eth_addr
*ea
);
307 static inline void eth_addr_random(struct eth_addr
*ea
)
309 random_bytes((uint8_t *)ea
, sizeof *ea
);
310 eth_addr_mark_random(ea
);
313 static inline void eth_addr_nicira_random(struct eth_addr
*ea
)
317 /* Set the OUI to the Nicira one. */
322 /* Set the top bit to indicate random Nicira address. */
325 static inline uint32_t hash_mac(const struct eth_addr ea
,
326 const uint16_t vlan
, const uint32_t basis
)
328 return hash_uint64_basis(eth_addr_vlan_to_uint64(ea
, vlan
), basis
);
331 bool eth_addr_is_reserved(const struct eth_addr
);
332 bool eth_addr_from_string(const char *, struct eth_addr
*);
334 void compose_rarp(struct dp_packet
*, const struct eth_addr
);
336 void eth_push_vlan(struct dp_packet
*, ovs_be16 tpid
, ovs_be16 tci
);
337 void eth_pop_vlan(struct dp_packet
*);
339 const char *eth_from_hex(const char *hex
, struct dp_packet
**packetp
);
340 void eth_format_masked(const struct eth_addr ea
,
341 const struct eth_addr
*mask
, struct ds
*s
);
343 void set_mpls_lse(struct dp_packet
*, ovs_be32 label
);
344 void push_mpls(struct dp_packet
*packet
, ovs_be16 ethtype
, ovs_be32 lse
);
345 void pop_mpls(struct dp_packet
*, ovs_be16 ethtype
);
347 void set_mpls_lse_ttl(ovs_be32
*lse
, uint8_t ttl
);
348 void set_mpls_lse_tc(ovs_be32
*lse
, uint8_t tc
);
349 void set_mpls_lse_label(ovs_be32
*lse
, ovs_be32 label
);
350 void set_mpls_lse_bos(ovs_be32
*lse
, uint8_t bos
);
351 ovs_be32
set_mpls_lse_values(uint8_t ttl
, uint8_t tc
, uint8_t bos
,
356 * struct eth_addr mac;
358 * printf("The Ethernet address is "ETH_ADDR_FMT"\n", ETH_ADDR_ARGS(mac));
361 #define ETH_ADDR_FMT \
362 "%02"PRIx8":%02"PRIx8":%02"PRIx8":%02"PRIx8":%02"PRIx8":%02"PRIx8
363 #define ETH_ADDR_ARGS(EA) ETH_ADDR_BYTES_ARGS((EA).ea)
364 #define ETH_ADDR_BYTES_ARGS(EAB) \
365 (EAB)[0], (EAB)[1], (EAB)[2], (EAB)[3], (EAB)[4], (EAB)[5]
366 #define ETH_ADDR_STRLEN 17
370 * struct eth_addr64 eui64;
372 * printf("The EUI-64 address is "ETH_ADDR64_FMT"\n", ETH_ADDR64_ARGS(mac));
375 #define ETH_ADDR64_FMT \
376 "%02"PRIx8":%02"PRIx8":%02"PRIx8":%02"PRIx8":" \
377 "%02"PRIx8":%02"PRIx8":%02"PRIx8":%02"PRIx8
378 #define ETH_ADDR64_ARGS(EA) ETH_ADDR64_BYTES_ARGS((EA).ea64)
379 #define ETH_ADDR64_BYTES_ARGS(EAB) \
380 (EAB)[0], (EAB)[1], (EAB)[2], (EAB)[3], \
381 (EAB)[4], (EAB)[5], (EAB)[6], (EAB)[7]
382 #define ETH_ADDR64_STRLEN 23
386 * char *string = "1 00:11:22:33:44:55 2";
387 * struct eth_addr mac;
390 * if (ovs_scan(string, "%d"ETH_ADDR_SCAN_FMT"%d",
391 * &a, ETH_ADDR_SCAN_ARGS(mac), &b)) {
395 #define ETH_ADDR_SCAN_FMT "%"SCNx8":%"SCNx8":%"SCNx8":%"SCNx8":%"SCNx8":%"SCNx8
396 #define ETH_ADDR_SCAN_ARGS(EA) \
397 &(EA).ea[0], &(EA).ea[1], &(EA).ea[2], &(EA).ea[3], &(EA).ea[4], &(EA).ea[5]
399 #define ETH_TYPE_IP 0x0800
400 #define ETH_TYPE_ARP 0x0806
401 #define ETH_TYPE_TEB 0x6558
402 #define ETH_TYPE_VLAN_8021Q 0x8100
403 #define ETH_TYPE_VLAN ETH_TYPE_VLAN_8021Q
404 #define ETH_TYPE_VLAN_8021AD 0x88a8
405 #define ETH_TYPE_IPV6 0x86dd
406 #define ETH_TYPE_LACP 0x8809
407 #define ETH_TYPE_RARP 0x8035
408 #define ETH_TYPE_MPLS 0x8847
409 #define ETH_TYPE_MPLS_MCAST 0x8848
410 #define ETH_TYPE_NSH 0x894f
411 #define ETH_TYPE_ERSPAN1 0x88be /* version 1 type II */
412 #define ETH_TYPE_ERSPAN2 0x22eb /* version 2 type III */
414 static inline bool eth_type_mpls(ovs_be16 eth_type
)
416 return eth_type
== htons(ETH_TYPE_MPLS
) ||
417 eth_type
== htons(ETH_TYPE_MPLS_MCAST
);
420 static inline bool eth_type_vlan(ovs_be16 eth_type
)
422 return eth_type
== htons(ETH_TYPE_VLAN_8021Q
) ||
423 eth_type
== htons(ETH_TYPE_VLAN_8021AD
);
427 /* Minimum value for an Ethernet type. Values below this are IEEE 802.2 frame
429 #define ETH_TYPE_MIN 0x600
431 #define ETH_HEADER_LEN 14
432 #define ETH_PAYLOAD_MIN 46
433 #define ETH_PAYLOAD_MAX 1500
434 #define ETH_TOTAL_MIN (ETH_HEADER_LEN + ETH_PAYLOAD_MIN)
435 #define ETH_TOTAL_MAX (ETH_HEADER_LEN + ETH_PAYLOAD_MAX)
436 #define ETH_VLAN_TOTAL_MAX (ETH_HEADER_LEN + VLAN_HEADER_LEN + ETH_PAYLOAD_MAX)
438 struct eth_addr eth_dst
;
439 struct eth_addr eth_src
;
442 BUILD_ASSERT_DECL(ETH_HEADER_LEN
== sizeof(struct eth_header
));
444 void push_eth(struct dp_packet
*packet
, const struct eth_addr
*dst
,
445 const struct eth_addr
*src
);
446 void pop_eth(struct dp_packet
*packet
);
448 void push_nsh(struct dp_packet
*packet
, const struct nsh_hdr
*nsh_hdr_src
);
449 bool pop_nsh(struct dp_packet
*packet
);
451 #define LLC_DSAP_SNAP 0xaa
452 #define LLC_SSAP_SNAP 0xaa
453 #define LLC_CNTL_SNAP 3
455 #define LLC_HEADER_LEN 3
461 BUILD_ASSERT_DECL(LLC_HEADER_LEN
== sizeof(struct llc_header
));
463 /* LLC field values used for STP frames. */
464 #define STP_LLC_SSAP 0x42
465 #define STP_LLC_DSAP 0x42
466 #define STP_LLC_CNTL 0x03
468 #define SNAP_ORG_ETHERNET "\0\0" /* The compiler adds a null byte, so
469 sizeof(SNAP_ORG_ETHERNET) == 3. */
470 #define SNAP_HEADER_LEN 5
476 BUILD_ASSERT_DECL(SNAP_HEADER_LEN
== sizeof(struct snap_header
));
478 #define LLC_SNAP_HEADER_LEN (LLC_HEADER_LEN + SNAP_HEADER_LEN)
480 struct llc_snap_header
{
481 struct llc_header llc
;
482 struct snap_header snap
;
484 BUILD_ASSERT_DECL(LLC_SNAP_HEADER_LEN
== sizeof(struct llc_snap_header
));
486 #define VLAN_VID_MASK 0x0fff
487 #define VLAN_VID_SHIFT 0
489 #define VLAN_PCP_MASK 0xe000
490 #define VLAN_PCP_SHIFT 13
492 #define VLAN_CFI 0x1000
493 #define VLAN_CFI_SHIFT 12
495 /* Given the vlan_tci field from an 802.1Q header, in network byte order,
496 * returns the VLAN ID in host byte order. */
497 static inline uint16_t
498 vlan_tci_to_vid(ovs_be16 vlan_tci
)
500 return (ntohs(vlan_tci
) & VLAN_VID_MASK
) >> VLAN_VID_SHIFT
;
503 /* Given the vlan_tci field from an 802.1Q header, in network byte order,
504 * returns the priority code point (PCP) in host byte order. */
506 vlan_tci_to_pcp(ovs_be16 vlan_tci
)
508 return (ntohs(vlan_tci
) & VLAN_PCP_MASK
) >> VLAN_PCP_SHIFT
;
511 /* Given the vlan_tci field from an 802.1Q header, in network byte order,
512 * returns the Canonical Format Indicator (CFI). */
514 vlan_tci_to_cfi(ovs_be16 vlan_tci
)
516 return (vlan_tci
& htons(VLAN_CFI
)) != 0;
519 #define VLAN_HEADER_LEN 4
521 ovs_be16 vlan_tci
; /* Lowest 12 bits are VLAN ID. */
522 ovs_be16 vlan_next_type
;
524 BUILD_ASSERT_DECL(VLAN_HEADER_LEN
== sizeof(struct vlan_header
));
526 #define VLAN_ETH_HEADER_LEN (ETH_HEADER_LEN + VLAN_HEADER_LEN)
527 struct vlan_eth_header
{
528 struct eth_addr veth_dst
;
529 struct eth_addr veth_src
;
530 ovs_be16 veth_type
; /* Always htons(ETH_TYPE_VLAN). */
531 ovs_be16 veth_tci
; /* Lowest 12 bits are VLAN ID. */
532 ovs_be16 veth_next_type
;
534 BUILD_ASSERT_DECL(VLAN_ETH_HEADER_LEN
== sizeof(struct vlan_eth_header
));
536 /* MPLS related definitions */
537 #define MPLS_TTL_MASK 0x000000ff
538 #define MPLS_TTL_SHIFT 0
540 #define MPLS_BOS_MASK 0x00000100
541 #define MPLS_BOS_SHIFT 8
543 #define MPLS_TC_MASK 0x00000e00
544 #define MPLS_TC_SHIFT 9
546 #define MPLS_LABEL_MASK 0xfffff000
547 #define MPLS_LABEL_SHIFT 12
552 ovs_16aligned_be32 mpls_lse
;
554 BUILD_ASSERT_DECL(MPLS_HLEN
== sizeof(struct mpls_hdr
));
556 /* Given a mpls label stack entry in network byte order
557 * return mpls label in host byte order */
558 static inline uint32_t
559 mpls_lse_to_label(ovs_be32 mpls_lse
)
561 return (ntohl(mpls_lse
) & MPLS_LABEL_MASK
) >> MPLS_LABEL_SHIFT
;
564 /* Given a mpls label stack entry in network byte order
566 static inline uint8_t
567 mpls_lse_to_tc(ovs_be32 mpls_lse
)
569 return (ntohl(mpls_lse
) & MPLS_TC_MASK
) >> MPLS_TC_SHIFT
;
572 /* Given a mpls label stack entry in network byte order
574 static inline uint8_t
575 mpls_lse_to_ttl(ovs_be32 mpls_lse
)
577 return (ntohl(mpls_lse
) & MPLS_TTL_MASK
) >> MPLS_TTL_SHIFT
;
580 /* Set label in mpls lse. */
582 flow_set_mpls_lse_label(ovs_be32
*mpls_lse
, uint32_t label
)
584 *mpls_lse
&= ~htonl(MPLS_LABEL_MASK
);
585 *mpls_lse
|= htonl(label
<< MPLS_LABEL_SHIFT
);
588 /* Set TC in mpls lse. */
590 flow_set_mpls_lse_tc(ovs_be32
*mpls_lse
, uint8_t tc
)
592 *mpls_lse
&= ~htonl(MPLS_TC_MASK
);
593 *mpls_lse
|= htonl((tc
& 0x7) << MPLS_TC_SHIFT
);
596 /* Set BOS in mpls lse. */
598 flow_set_mpls_lse_bos(ovs_be32
*mpls_lse
, uint8_t bos
)
600 *mpls_lse
&= ~htonl(MPLS_BOS_MASK
);
601 *mpls_lse
|= htonl((bos
& 0x1) << MPLS_BOS_SHIFT
);
604 /* Set TTL in mpls lse. */
606 flow_set_mpls_lse_ttl(ovs_be32
*mpls_lse
, uint8_t ttl
)
608 *mpls_lse
&= ~htonl(MPLS_TTL_MASK
);
609 *mpls_lse
|= htonl(ttl
<< MPLS_TTL_SHIFT
);
612 /* Given a mpls label stack entry in network byte order
613 * return mpls BoS bit */
614 static inline uint8_t
615 mpls_lse_to_bos(ovs_be32 mpls_lse
)
617 return (mpls_lse
& htonl(MPLS_BOS_MASK
)) != 0;
620 #define IP_FMT "%"PRIu32".%"PRIu32".%"PRIu32".%"PRIu32
621 #define IP_ARGS(ip) \
623 (ntohl(ip) >> 16) & 0xff, \
624 (ntohl(ip) >> 8) & 0xff, \
629 * char *string = "1 33.44.55.66 2";
633 * if (ovs_scan(string, "%d"IP_SCAN_FMT"%d", &a, IP_SCAN_ARGS(&ip), &b)) {
637 #define IP_SCAN_FMT "%"SCNu8".%"SCNu8".%"SCNu8".%"SCNu8
638 #define IP_SCAN_ARGS(ip) \
639 ((void) (ovs_be32) *(ip), &((uint8_t *) ip)[0]), \
640 &((uint8_t *) ip)[1], \
641 &((uint8_t *) ip)[2], \
644 #define IP_PORT_SCAN_FMT "%"SCNu8".%"SCNu8".%"SCNu8".%"SCNu8":%"SCNu16
645 #define IP_PORT_SCAN_ARGS(ip, port) \
646 ((void) (ovs_be32) *(ip), &((uint8_t *) ip)[0]), \
647 &((uint8_t *) ip)[1], \
648 &((uint8_t *) ip)[2], \
649 &((uint8_t *) ip)[3], \
650 ((void) (ovs_be16) *(port), (uint16_t *) port)
652 /* Returns true if 'netmask' is a CIDR netmask, that is, if it consists of N
653 * high-order 1-bits and 32-N low-order 0-bits. */
655 ip_is_cidr(ovs_be32 netmask
)
657 uint32_t x
= ~ntohl(netmask
);
658 return !(x
& (x
+ 1));
661 ip_is_multicast(ovs_be32 ip
)
663 return (ip
& htonl(0xf0000000)) == htonl(0xe0000000);
666 ip_is_local_multicast(ovs_be32 ip
)
668 return (ip
& htonl(0xffffff00)) == htonl(0xe0000000);
670 int ip_count_cidr_bits(ovs_be32 netmask
);
671 void ip_format_masked(ovs_be32 ip
, ovs_be32 mask
, struct ds
*);
672 bool ip_parse(const char *s
, ovs_be32
*ip
);
673 char *ip_parse_port(const char *s
, ovs_be32
*ip
, ovs_be16
*port
)
674 OVS_WARN_UNUSED_RESULT
;
675 char *ip_parse_masked(const char *s
, ovs_be32
*ip
, ovs_be32
*mask
)
676 OVS_WARN_UNUSED_RESULT
;
677 char *ip_parse_cidr(const char *s
, ovs_be32
*ip
, unsigned int *plen
)
678 OVS_WARN_UNUSED_RESULT
;
679 char *ip_parse_masked_len(const char *s
, int *n
, ovs_be32
*ip
, ovs_be32
*mask
)
680 OVS_WARN_UNUSED_RESULT
;
681 char *ip_parse_cidr_len(const char *s
, int *n
, ovs_be32
*ip
,
683 OVS_WARN_UNUSED_RESULT
;
685 #define IP_VER(ip_ihl_ver) ((ip_ihl_ver) >> 4)
686 #define IP_IHL(ip_ihl_ver) ((ip_ihl_ver) & 15)
687 #define IP_IHL_VER(ihl, ver) (((ver) << 4) | (ihl))
690 #define IPPROTO_SCTP 132
694 #define IPPROTO_DCCP 33
698 #define IPPROTO_IGMP 2
701 #ifndef IPPROTO_UDPLITE
702 #define IPPROTO_UDPLITE 136
706 #define IP_ECN_NOT_ECT 0x0
707 #define IP_ECN_ECT_1 0x01
708 #define IP_ECN_ECT_0 0x02
709 #define IP_ECN_CE 0x03
710 #define IP_ECN_MASK 0x03
711 #define IP_DSCP_CS6 0xc0
712 #define IP_DSCP_MASK 0xfc
715 IP_ECN_is_ce(uint8_t dsfield
)
717 return (dsfield
& IP_ECN_MASK
) == IP_ECN_CE
;
722 #define IP_DONT_FRAGMENT 0x4000 /* Don't fragment. */
723 #define IP_MORE_FRAGMENTS 0x2000 /* More fragments. */
724 #define IP_FRAG_OFF_MASK 0x1fff /* Fragment offset. */
725 #define IP_IS_FRAGMENT(ip_frag_off) \
726 ((ip_frag_off) & htons(IP_MORE_FRAGMENTS | IP_FRAG_OFF_MASK))
728 #define IP_HEADER_LEN 20
734 ovs_be16 ip_frag_off
;
738 ovs_16aligned_be32 ip_src
;
739 ovs_16aligned_be32 ip_dst
;
741 BUILD_ASSERT_DECL(IP_HEADER_LEN
== sizeof(struct ip_header
));
744 #define ICMP4_ECHO_REPLY 0
745 #define ICMP4_DST_UNREACH 3
746 #define ICMP4_SOURCEQUENCH 4
747 #define ICMP4_REDIRECT 5
748 #define ICMP4_ECHO_REQUEST 8
749 #define ICMP4_TIME_EXCEEDED 11
750 #define ICMP4_PARAM_PROB 12
751 #define ICMP4_TIMESTAMP 13
752 #define ICMP4_TIMESTAMPREPLY 14
753 #define ICMP4_INFOREQUEST 15
754 #define ICMP4_INFOREPLY 16
756 #define ICMP_HEADER_LEN 8
770 ovs_16aligned_be32 gateway
;
773 BUILD_ASSERT_DECL(ICMP_HEADER_LEN
== sizeof(struct icmp_header
));
776 #define ICMP_ERROR_DATA_L4_LEN 8
778 #define IGMP_HEADER_LEN 8
783 ovs_16aligned_be32 group
;
785 BUILD_ASSERT_DECL(IGMP_HEADER_LEN
== sizeof(struct igmp_header
));
787 #define IGMPV3_HEADER_LEN 8
788 struct igmpv3_header
{
795 BUILD_ASSERT_DECL(IGMPV3_HEADER_LEN
== sizeof(struct igmpv3_header
));
797 #define IGMPV3_QUERY_HEADER_LEN 12
798 struct igmpv3_query_header
{
802 ovs_16aligned_be32 group
;
808 IGMPV3_QUERY_HEADER_LEN
== sizeof(struct igmpv3_query_header
811 #define IGMPV3_RECORD_LEN 8
812 struct igmpv3_record
{
816 ovs_16aligned_be32 maddr
;
818 BUILD_ASSERT_DECL(IGMPV3_RECORD_LEN
== sizeof(struct igmpv3_record
));
820 #define IGMP_HOST_MEMBERSHIP_QUERY 0x11 /* From RFC1112 */
821 #define IGMP_HOST_MEMBERSHIP_REPORT 0x12 /* Ditto */
822 #define IGMPV2_HOST_MEMBERSHIP_REPORT 0x16 /* V2 version of 0x12 */
823 #define IGMP_HOST_LEAVE_MESSAGE 0x17
824 #define IGMPV3_HOST_MEMBERSHIP_REPORT 0x22 /* V3 version of 0x12 */
827 * IGMPv3 and MLDv2 use the same codes.
829 #define IGMPV3_MODE_IS_INCLUDE 1
830 #define IGMPV3_MODE_IS_EXCLUDE 2
831 #define IGMPV3_CHANGE_TO_INCLUDE_MODE 3
832 #define IGMPV3_CHANGE_TO_EXCLUDE_MODE 4
833 #define IGMPV3_ALLOW_NEW_SOURCES 5
834 #define IGMPV3_BLOCK_OLD_SOURCES 6
836 #define SCTP_HEADER_LEN 12
840 ovs_16aligned_be32 sctp_vtag
;
841 ovs_16aligned_be32 sctp_csum
;
843 BUILD_ASSERT_DECL(SCTP_HEADER_LEN
== sizeof(struct sctp_header
));
845 #define UDP_HEADER_LEN 8
852 BUILD_ASSERT_DECL(UDP_HEADER_LEN
== sizeof(struct udp_header
));
854 #define ESP_HEADER_LEN 8
859 BUILD_ASSERT_DECL(ESP_HEADER_LEN
== sizeof(struct esp_header
));
861 #define ESP_TRAILER_LEN 2
866 BUILD_ASSERT_DECL(ESP_TRAILER_LEN
== sizeof(struct esp_trailer
));
868 #define TCP_FIN 0x001
869 #define TCP_SYN 0x002
870 #define TCP_RST 0x004
871 #define TCP_PSH 0x008
872 #define TCP_ACK 0x010
873 #define TCP_URG 0x020
874 #define TCP_ECE 0x040
875 #define TCP_CWR 0x080
878 #define TCP_CTL(flags, offset) (htons((flags) | ((offset) << 12)))
879 #define TCP_FLAGS(tcp_ctl) (ntohs(tcp_ctl) & 0x0fff)
880 #define TCP_FLAGS_BE16(tcp_ctl) ((tcp_ctl) & htons(0x0fff))
881 #define TCP_OFFSET(tcp_ctl) (ntohs(tcp_ctl) >> 12)
883 #define TCP_HEADER_LEN 20
887 ovs_16aligned_be32 tcp_seq
;
888 ovs_16aligned_be32 tcp_ack
;
894 BUILD_ASSERT_DECL(TCP_HEADER_LEN
== sizeof(struct tcp_header
));
896 /* Connection states.
898 * Names like CS_RELATED are bit values, e.g. 1 << 2.
899 * Names like CS_RELATED_BIT are bit indexes, e.g. 2. */
901 CS_STATE(NEW, 0, "new") \
902 CS_STATE(ESTABLISHED, 1, "est") \
903 CS_STATE(RELATED, 2, "rel") \
904 CS_STATE(REPLY_DIR, 3, "rpl") \
905 CS_STATE(INVALID, 4, "inv") \
906 CS_STATE(TRACKED, 5, "trk") \
907 CS_STATE(SRC_NAT, 6, "snat") \
908 CS_STATE(DST_NAT, 7, "dnat")
911 #define CS_STATE(ENUM, INDEX, NAME) \
912 CS_##ENUM = 1 << INDEX, \
913 CS_##ENUM##_BIT = INDEX,
918 /* Undefined connection state bits. */
920 #define CS_STATE(ENUM, INDEX, NAME) +CS_##ENUM
921 CS_SUPPORTED_MASK
= CS_STATES
924 #define CS_UNSUPPORTED_MASK (~(uint32_t)CS_SUPPORTED_MASK)
926 #define ARP_HRD_ETHERNET 1
927 #define ARP_PRO_IP 0x0800
928 #define ARP_OP_REQUEST 1
929 #define ARP_OP_REPLY 2
930 #define ARP_OP_RARP 3
932 #define ARP_ETH_HEADER_LEN 28
933 struct arp_eth_header
{
934 /* Generic members. */
935 ovs_be16 ar_hrd
; /* Hardware type. */
936 ovs_be16 ar_pro
; /* Protocol type. */
937 uint8_t ar_hln
; /* Hardware address length. */
938 uint8_t ar_pln
; /* Protocol address length. */
939 ovs_be16 ar_op
; /* Opcode. */
941 /* Ethernet+IPv4 specific members. */
942 struct eth_addr ar_sha
; /* Sender hardware address. */
943 ovs_16aligned_be32 ar_spa
; /* Sender protocol address. */
944 struct eth_addr ar_tha
; /* Target hardware address. */
945 ovs_16aligned_be32 ar_tpa
; /* Target protocol address. */
947 BUILD_ASSERT_DECL(ARP_ETH_HEADER_LEN
== sizeof(struct arp_eth_header
));
949 #define IPV6_HEADER_LEN 40
951 /* Like struct in6_addr, but whereas that struct requires 32-bit alignment on
952 * most implementations, this one only requires 16-bit alignment. */
953 union ovs_16aligned_in6_addr
{
955 ovs_16aligned_be32 be32
[4];
958 /* Like struct ip6_hdr, but whereas that struct requires 32-bit alignment, this
959 * one only requires 16-bit alignment. */
960 struct ovs_16aligned_ip6_hdr
{
962 struct ovs_16aligned_ip6_hdrctl
{
963 ovs_16aligned_be32 ip6_un1_flow
;
964 ovs_be16 ip6_un1_plen
;
966 uint8_t ip6_un1_hlim
;
970 union ovs_16aligned_in6_addr ip6_src
;
971 union ovs_16aligned_in6_addr ip6_dst
;
974 /* Like struct in6_frag, but whereas that struct requires 32-bit alignment,
975 * this one only requires 16-bit alignment. */
976 struct ovs_16aligned_ip6_frag
{
978 uint8_t ip6f_reserved
;
980 ovs_16aligned_be32 ip6f_ident
;
983 #define ICMP6_HEADER_LEN 4
984 struct icmp6_header
{
987 ovs_be16 icmp6_cksum
;
989 BUILD_ASSERT_DECL(ICMP6_HEADER_LEN
== sizeof(struct icmp6_header
));
991 #define ICMP6_DATA_HEADER_LEN 8
992 struct icmp6_data_header
{
993 struct icmp6_header icmp6_base
;
995 ovs_16aligned_be32 be32
[1];
1000 BUILD_ASSERT_DECL(ICMP6_DATA_HEADER_LEN
== sizeof(struct icmp6_data_header
));
1002 uint32_t packet_csum_pseudoheader6(const struct ovs_16aligned_ip6_hdr
*);
1003 ovs_be16
packet_csum_upperlayer6(const struct ovs_16aligned_ip6_hdr
*,
1004 const void *, uint8_t, uint16_t);
1006 /* Neighbor Discovery option field.
1007 * ND options are always a multiple of 8 bytes in size. */
1008 #define ND_LLA_OPT_LEN 8
1009 struct ovs_nd_lla_opt
{
1010 uint8_t type
; /* One of ND_OPT_*_LINKADDR. */
1012 struct eth_addr mac
;
1014 BUILD_ASSERT_DECL(ND_LLA_OPT_LEN
== sizeof(struct ovs_nd_lla_opt
));
1016 /* Neighbor Discovery option: Prefix Information. */
1017 #define ND_PREFIX_OPT_LEN 32
1018 struct ovs_nd_prefix_opt
{
1019 uint8_t type
; /* ND_OPT_PREFIX_INFORMATION. */
1020 uint8_t len
; /* Always 4. */
1022 uint8_t la_flags
; /* ND_PREFIX_* flags. */
1023 ovs_16aligned_be32 valid_lifetime
;
1024 ovs_16aligned_be32 preferred_lifetime
;
1025 ovs_16aligned_be32 reserved
; /* Always 0. */
1026 union ovs_16aligned_in6_addr prefix
;
1028 BUILD_ASSERT_DECL(ND_PREFIX_OPT_LEN
== sizeof(struct ovs_nd_prefix_opt
));
1030 /* Neighbor Discovery option: MTU. */
1031 #define ND_MTU_OPT_LEN 8
1032 #define ND_MTU_DEFAULT 0
1033 struct ovs_nd_mtu_opt
{
1034 uint8_t type
; /* ND_OPT_MTU */
1035 uint8_t len
; /* Always 1. */
1036 ovs_be16 reserved
; /* Always 0. */
1037 ovs_16aligned_be32 mtu
;
1039 BUILD_ASSERT_DECL(ND_MTU_OPT_LEN
== sizeof(struct ovs_nd_mtu_opt
));
1041 /* Like struct nd_msg (from ndisc.h), but whereas that struct requires 32-bit
1042 * alignment, this one only requires 16-bit alignment. */
1043 #define ND_MSG_LEN 24
1045 struct icmp6_header icmph
;
1046 ovs_16aligned_be32 rso_flags
;
1047 union ovs_16aligned_in6_addr target
;
1048 struct ovs_nd_lla_opt options
[0];
1050 BUILD_ASSERT_DECL(ND_MSG_LEN
== sizeof(struct ovs_nd_msg
));
1052 /* Neighbor Discovery packet flags. */
1053 #define ND_RSO_ROUTER 0x80000000
1054 #define ND_RSO_SOLICITED 0x40000000
1055 #define ND_RSO_OVERRIDE 0x20000000
1057 #define RA_MSG_LEN 16
1059 struct icmp6_header icmph
;
1060 uint8_t cur_hop_limit
;
1061 uint8_t mo_flags
; /* ND_RA_MANAGED_ADDRESS and ND_RA_OTHER_CONFIG flags. */
1062 ovs_be16 router_lifetime
;
1063 ovs_be32 reachable_time
;
1064 ovs_be32 retrans_timer
;
1065 struct ovs_nd_lla_opt options
[0];
1067 BUILD_ASSERT_DECL(RA_MSG_LEN
== sizeof(struct ovs_ra_msg
));
1069 #define ND_RA_MANAGED_ADDRESS 0x80
1070 #define ND_RA_OTHER_CONFIG 0x40
1072 /* Defaults based on MaxRtrInterval and MinRtrInterval from RFC 4861 section
1075 #define ND_RA_MAX_INTERVAL_DEFAULT 600
1078 nd_ra_min_interval_default(int max
)
1080 return max
>= 9 ? max
/ 3 : max
* 3 / 4;
1084 * Use the same struct for MLD and MLD2, naming members as the defined fields in
1085 * in the corresponding version of the protocol, though they are reserved in the
1088 #define MLD_HEADER_LEN 8
1096 BUILD_ASSERT_DECL(MLD_HEADER_LEN
== sizeof(struct mld_header
));
1098 #define MLD2_RECORD_LEN 20
1099 struct mld2_record
{
1103 union ovs_16aligned_in6_addr maddr
;
1105 BUILD_ASSERT_DECL(MLD2_RECORD_LEN
== sizeof(struct mld2_record
));
1107 #define MLD_QUERY 130
1108 #define MLD_REPORT 131
1109 #define MLD_DONE 132
1110 #define MLD2_REPORT 143
1112 /* The IPv6 flow label is in the lower 20 bits of the first 32-bit word. */
1113 #define IPV6_LABEL_MASK 0x000fffff
1117 * char *string = "1 ::1 2";
1118 * char ipv6_s[IPV6_SCAN_LEN + 1];
1119 * struct in6_addr ipv6;
1121 * if (ovs_scan(string, "%d"IPV6_SCAN_FMT"%d", &a, ipv6_s, &b)
1122 * && inet_pton(AF_INET6, ipv6_s, &ipv6) == 1) {
1126 #define IPV6_SCAN_FMT "%46[0123456789abcdefABCDEF:.]"
1127 #define IPV6_SCAN_LEN 46
1129 extern const struct in6_addr in6addr_exact
;
1130 #define IN6ADDR_EXACT_INIT { { { 0xff,0xff,0xff,0xff,0xff,0xff,0xff,0xff, \
1131 0xff,0xff,0xff,0xff,0xff,0xff,0xff,0xff } } }
1133 extern const struct in6_addr in6addr_all_hosts
;
1134 #define IN6ADDR_ALL_HOSTS_INIT { { { 0xff,0x02,0x00,0x00,0x00,0x00,0x00,0x00, \
1135 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x01 } } }
1137 extern const struct in6_addr in6addr_all_routers
;
1138 #define IN6ADDR_ALL_ROUTERS_INIT { { { 0xff,0x02,0x00,0x00,0x00,0x00,0x00,0x00, \
1139 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x02 } } }
1141 static inline bool ipv6_addr_equals(const struct in6_addr
*a
,
1142 const struct in6_addr
*b
)
1144 #ifdef IN6_ARE_ADDR_EQUAL
1145 return IN6_ARE_ADDR_EQUAL(a
, b
);
1147 return !memcmp(a
, b
, sizeof(*a
));
1151 /* Checks the IPv6 address in 'mask' for all zeroes. */
1152 static inline bool ipv6_mask_is_any(const struct in6_addr
*mask
) {
1153 return ipv6_addr_equals(mask
, &in6addr_any
);
1156 static inline bool ipv6_mask_is_exact(const struct in6_addr
*mask
) {
1157 return ipv6_addr_equals(mask
, &in6addr_exact
);
1160 static inline bool ipv6_is_all_hosts(const struct in6_addr
*addr
) {
1161 return ipv6_addr_equals(addr
, &in6addr_all_hosts
);
1164 static inline bool ipv6_addr_is_set(const struct in6_addr
*addr
) {
1165 return !ipv6_addr_equals(addr
, &in6addr_any
);
1168 static inline bool ipv6_addr_is_multicast(const struct in6_addr
*ip
) {
1169 return ip
->s6_addr
[0] == 0xff;
1172 static inline struct in6_addr
1173 in6_addr_mapped_ipv4(ovs_be32 ip4
)
1175 struct in6_addr ip6
;
1176 memset(&ip6
, 0, sizeof(ip6
));
1177 ip6
.s6_addr
[10] = 0xff, ip6
.s6_addr
[11] = 0xff;
1178 memcpy(&ip6
.s6_addr
[12], &ip4
, 4);
1183 in6_addr_set_mapped_ipv4(struct in6_addr
*ip6
, ovs_be32 ip4
)
1185 *ip6
= in6_addr_mapped_ipv4(ip4
);
1188 static inline ovs_be32
1189 in6_addr_get_mapped_ipv4(const struct in6_addr
*addr
)
1191 union ovs_16aligned_in6_addr
*taddr
=
1192 (union ovs_16aligned_in6_addr
*) addr
;
1193 if (IN6_IS_ADDR_V4MAPPED(addr
)) {
1194 return get_16aligned_be32(&taddr
->be32
[3]);
1200 void in6_addr_solicited_node(struct in6_addr
*addr
,
1201 const struct in6_addr
*ip6
);
1203 void in6_generate_eui64(struct eth_addr ea
, const struct in6_addr
*prefix
,
1204 struct in6_addr
*lla
);
1206 void in6_generate_lla(struct eth_addr ea
, struct in6_addr
*lla
);
1208 /* Returns true if 'addr' is a link local address. Otherwise, false. */
1209 bool in6_is_lla(struct in6_addr
*addr
);
1211 void ipv6_multicast_to_ethernet(struct eth_addr
*eth
,
1212 const struct in6_addr
*ip6
);
1214 static inline bool dl_type_is_ip_any(ovs_be16 dl_type
)
1216 return dl_type
== htons(ETH_TYPE_IP
)
1217 || dl_type
== htons(ETH_TYPE_IPV6
);
1222 /* GRE protocol header */
1223 struct gre_base_hdr
{
1228 #define GRE_CSUM 0x8000
1229 #define GRE_ROUTING 0x4000
1230 #define GRE_KEY 0x2000
1231 #define GRE_SEQ 0x1000
1232 #define GRE_STRICT 0x0800
1233 #define GRE_REC 0x0700
1234 #define GRE_FLAGS 0x00F8
1235 #define GRE_VERSION 0x0007
1238 * ERSPAN protocol header and metadata
1240 * Version 1 (Type II) header (8 octets [42:49])
1242 * 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
1243 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1244 * | Ver | VLAN | COS | En|T| Session ID |
1245 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1246 * | Reserved | Index |
1247 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1250 * ERSPAN Version 2 (Type III) header (12 octets [42:49])
1252 * 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
1253 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1254 * | Ver | VLAN | COS |BSO|T| Session ID |
1255 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1257 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1258 * | SGT |P| FT | Hw ID |D|Gra|O|
1259 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1263 /* ERSPAN has fixed 8-byte GRE header */
1264 #define ERSPAN_GREHDR_LEN 8
1265 #define ERSPAN_HDR(gre_base_hdr) \
1266 ((struct erspan_base_hdr *)((char *)gre_base_hdr + ERSPAN_GREHDR_LEN))
1268 #define ERSPAN_V1_MDSIZE 4
1269 #define ERSPAN_V2_MDSIZE 8
1271 #define ERSPAN_SID_MASK 0x03ff /* 10-bit Session ID. */
1272 #define ERSPAN_IDX_MASK 0xfffff /* v1 Index */
1273 #define ERSPAN_HWID_MASK 0x03f0
1274 #define ERSPAN_DIR_MASK 0x0008
1276 struct erspan_base_hdr
{
1277 #ifdef WORDS_BIGENDIAN
1285 uint8_t session_id
:8;
1287 uint8_t vlan_upper
:4,
1290 uint8_t session_id_upper
:2,
1294 uint8_t session_id
:8;
1299 ovs_16aligned_be32 timestamp
;
1301 #ifdef WORDS_BIGENDIAN
1310 uint8_t hwid_upper
:2,
1320 struct erspan_metadata
{
1323 ovs_be32 index
; /* Version 1 (type II)*/
1324 struct erspan_md2 md2
; /* Version 2 (type III) */
1328 static inline uint16_t get_sid(const struct erspan_base_hdr
*ershdr
)
1330 return (ershdr
->session_id_upper
<< 8) + ershdr
->session_id
;
1333 static inline void set_sid(struct erspan_base_hdr
*ershdr
, uint16_t id
)
1335 ershdr
->session_id
= id
& 0xff;
1336 ershdr
->session_id_upper
= (id
>> 8) &0x3;
1339 static inline uint8_t get_hwid(const struct erspan_md2
*md2
)
1341 return (md2
->hwid_upper
<< 4) + md2
->hwid
;
1344 static inline void set_hwid(struct erspan_md2
*md2
, uint8_t hwid
)
1346 md2
->hwid
= hwid
& 0xf;
1347 md2
->hwid_upper
= (hwid
>> 4) & 0x3;
1350 /* ERSPAN timestamp granularity
1351 * 00b --> granularity = 100 microseconds
1352 * 01b --> granularity = 100 nanoseconds
1353 * 10b --> granularity = IEEE 1588
1354 * Here we only support 100 microseconds.
1356 enum erspan_ts_gra
{
1362 static inline ovs_be32
get_erspan_ts(enum erspan_ts_gra gra
)
1368 ts
= htonl((uint32_t)(time_wall_usec() / 100));
1372 case ERSPAN_IEEE1588
:
1382 * GTP-U protocol header and metadata
1384 * User Plane Protocol and Architectural Analysis on 3GPP 5G System
1385 * draft-hmm-dmm-5g-uplane-analysis-00
1388 * 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
1389 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1390 * | Ver |P|R|E|S|N| Message Type| Length |
1391 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1392 * | Tunnel Endpoint Identifier |
1393 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1394 * | Sequence Number | N-PDU Number | Next-Ext-Hdr |
1395 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1398 * P: Protocol Type (Set to '1')
1399 * R: Reserved Bit (Set to '0')
1400 * E: Extension Header Flag (Set to '1' if extension header exists)
1401 * S: Sequence Number Flag (Set to '1' if sequence number exists)
1402 * N: N-PDU Number Flag (Set to '1' if N-PDU number exists)
1404 * GTP-U Message Type:
1405 * Indicates the type of GTP-U message.
1408 * Indicates the length in octets of the payload.
1410 * User payload is transmitted in G-PDU packets.
1413 #define GTPU_VER_MASK 0xe0
1414 #define GTPU_P_MASK 0x10
1415 #define GTPU_E_MASK 0x04
1416 #define GTPU_S_MASK 0x02
1418 /* GTP-U UDP port. */
1419 #define GTPU_DST_PORT 2152
1421 /* Default GTP-U flags: Ver = 1 and P = 1. */
1422 #define GTPU_FLAGS_DEFAULT 0x30
1424 /* GTP-U message type for normal user plane PDU. */
1425 #define GTPU_MSGTYPE_REQ 1 /* Echo Request. */
1426 #define GTPU_MSGTYPE_REPL 2 /* Echo Reply. */
1427 #define GTPU_MSGTYPE_GPDU 255 /* User Payload. */
1429 struct gtpu_metadata
{
1433 BUILD_ASSERT_DECL(sizeof(struct gtpu_metadata
) == 2);
1436 struct gtpu_metadata md
;
1438 ovs_16aligned_be32 teid
;
1440 BUILD_ASSERT_DECL(sizeof(struct gtpuhdr
) == 8);
1442 struct gtpuhdr_opt
{
1445 uint8_t next_ext_type
;
1447 BUILD_ASSERT_DECL(sizeof(struct gtpuhdr_opt
) == 4);
1449 /* VXLAN protocol header */
1452 ovs_16aligned_be32 vx_flags
; /* VXLAN flags. */
1454 uint8_t flags
; /* VXLAN GPE flags. */
1455 uint8_t reserved
[2]; /* 16 bits reserved. */
1456 uint8_t next_protocol
; /* Next Protocol field for VXLAN GPE. */
1459 ovs_16aligned_be32 vx_vni
;
1461 BUILD_ASSERT_DECL(sizeof(struct vxlanhdr
) == 8);
1463 #define VXLAN_FLAGS 0x08000000 /* struct vxlanhdr.vx_flags required value. */
1466 * VXLAN Generic Protocol Extension (VXLAN_F_GPE):
1467 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1468 * |R|R|Ver|I|P|R|O| Reserved |Next Protocol |
1469 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1470 * | VXLAN Network Identifier (VNI) | Reserved |
1471 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1473 * Ver = Version. Indicates VXLAN GPE protocol version.
1475 * P = Next Protocol Bit. The P bit is set to indicate that the
1476 * Next Protocol field is present.
1478 * O = OAM Flag Bit. The O bit is set to indicate that the packet
1481 * Next Protocol = This 8 bit field indicates the protocol header
1482 * immediately following the VXLAN GPE header.
1484 * https://tools.ietf.org/html/draft-ietf-nvo3-vxlan-gpe-01
1487 /* Fields in struct vxlanhdr.vx_gpe.flags */
1488 #define VXLAN_GPE_FLAGS_VER 0x30 /* Version. */
1489 #define VLXAN_GPE_FLAGS_P 0x04 /* Next Protocol Bit. */
1490 #define VXLAN_GPE_FLAGS_O 0x01 /* OAM Bit. */
1492 /* VXLAN-GPE header flags. */
1493 #define VXLAN_HF_VER ((1U <<29) | (1U <<28))
1494 #define VXLAN_HF_NP (1U <<26)
1495 #define VXLAN_HF_OAM (1U <<24)
1497 #define VXLAN_GPE_USED_BITS (VXLAN_HF_VER | VXLAN_HF_NP | VXLAN_HF_OAM | \
1500 /* VXLAN-GPE header Next Protocol. */
1501 #define VXLAN_GPE_NP_IPV4 0x01
1502 #define VXLAN_GPE_NP_IPV6 0x02
1503 #define VXLAN_GPE_NP_ETHERNET 0x03
1504 #define VXLAN_GPE_NP_NSH 0x04
1506 #define VXLAN_F_GPE 0x4000
1507 #define VXLAN_HF_GPE 0x04000000
1509 /* Input values for PACKET_TYPE macros have to be in host byte order.
1510 * The _BE postfix indicates result is in network byte order. Otherwise result
1511 * is in host byte order. */
1512 #define PACKET_TYPE(NS, NS_TYPE) ((uint32_t) ((NS) << 16 | (NS_TYPE)))
1513 #define PACKET_TYPE_BE(NS, NS_TYPE) (htonl((NS) << 16 | (NS_TYPE)))
1515 /* Returns the host byte ordered namespace of 'packet type'. */
1516 static inline uint16_t
1517 pt_ns(ovs_be32 packet_type
)
1519 return ntohl(packet_type
) >> 16;
1522 /* Returns the network byte ordered namespace type of 'packet type'. */
1523 static inline ovs_be16
1524 pt_ns_type_be(ovs_be32 packet_type
)
1526 return be32_to_be16(packet_type
);
1529 /* Returns the host byte ordered namespace type of 'packet type'. */
1530 static inline uint16_t
1531 pt_ns_type(ovs_be32 packet_type
)
1533 return ntohs(pt_ns_type_be(packet_type
));
1536 /* Well-known packet_type field values. */
1538 PT_ETH
= PACKET_TYPE(OFPHTN_ONF
, 0x0000), /* Default PT: Ethernet */
1539 PT_USE_NEXT_PROTO
= PACKET_TYPE(OFPHTN_ONF
, 0xfffe), /* Pseudo PT for decap. */
1540 PT_IPV4
= PACKET_TYPE(OFPHTN_ETHERTYPE
, ETH_TYPE_IP
),
1541 PT_IPV6
= PACKET_TYPE(OFPHTN_ETHERTYPE
, ETH_TYPE_IPV6
),
1542 PT_MPLS
= PACKET_TYPE(OFPHTN_ETHERTYPE
, ETH_TYPE_MPLS
),
1543 PT_MPLS_MC
= PACKET_TYPE(OFPHTN_ETHERTYPE
, ETH_TYPE_MPLS_MCAST
),
1544 PT_NSH
= PACKET_TYPE(OFPHTN_ETHERTYPE
, ETH_TYPE_NSH
),
1545 PT_UNKNOWN
= PACKET_TYPE(0xffff, 0xffff), /* Unknown packet type. */
1549 void ipv6_format_addr(const struct in6_addr
*addr
, struct ds
*);
1550 void ipv6_format_addr_bracket(const struct in6_addr
*addr
, struct ds
*,
1552 void ipv6_format_mapped(const struct in6_addr
*addr
, struct ds
*);
1553 void ipv6_format_masked(const struct in6_addr
*addr
,
1554 const struct in6_addr
*mask
, struct ds
*);
1555 const char * ipv6_string_mapped(char *addr_str
, const struct in6_addr
*addr
);
1556 struct in6_addr
ipv6_addr_bitand(const struct in6_addr
*src
,
1557 const struct in6_addr
*mask
);
1558 struct in6_addr
ipv6_addr_bitxor(const struct in6_addr
*a
,
1559 const struct in6_addr
*b
);
1560 bool ipv6_is_zero(const struct in6_addr
*a
);
1561 struct in6_addr
ipv6_create_mask(int mask
);
1562 int ipv6_count_cidr_bits(const struct in6_addr
*netmask
);
1563 bool ipv6_is_cidr(const struct in6_addr
*netmask
);
1565 bool ipv6_parse(const char *s
, struct in6_addr
*ip
);
1566 char *ipv6_parse_masked(const char *s
, struct in6_addr
*ipv6
,
1567 struct in6_addr
*mask
);
1568 char *ipv6_parse_cidr(const char *s
, struct in6_addr
*ip
, unsigned int *plen
)
1569 OVS_WARN_UNUSED_RESULT
;
1570 char *ipv6_parse_masked_len(const char *s
, int *n
, struct in6_addr
*ipv6
,
1571 struct in6_addr
*mask
);
1572 char *ipv6_parse_cidr_len(const char *s
, int *n
, struct in6_addr
*ip
,
1574 OVS_WARN_UNUSED_RESULT
;
1576 void *eth_compose(struct dp_packet
*, const struct eth_addr eth_dst
,
1577 const struct eth_addr eth_src
, uint16_t eth_type
,
1579 void *snap_compose(struct dp_packet
*, const struct eth_addr eth_dst
,
1580 const struct eth_addr eth_src
,
1581 unsigned int oui
, uint16_t snap_type
, size_t size
);
1582 void packet_set_ipv4(struct dp_packet
*, ovs_be32 src
, ovs_be32 dst
, uint8_t tos
,
1584 void packet_set_ipv4_addr(struct dp_packet
*packet
, ovs_16aligned_be32
*addr
,
1586 void packet_set_ipv6(struct dp_packet
*, const struct in6_addr
*src
,
1587 const struct in6_addr
*dst
, uint8_t tc
,
1588 ovs_be32 fl
, uint8_t hlmit
);
1589 void packet_set_ipv6_addr(struct dp_packet
*packet
, uint8_t proto
,
1590 ovs_16aligned_be32 addr
[4],
1591 const struct in6_addr
*new_addr
,
1592 bool recalculate_csum
);
1593 void packet_set_tcp_port(struct dp_packet
*, ovs_be16 src
, ovs_be16 dst
);
1594 void packet_set_udp_port(struct dp_packet
*, ovs_be16 src
, ovs_be16 dst
);
1595 void packet_set_sctp_port(struct dp_packet
*, ovs_be16 src
, ovs_be16 dst
);
1596 void packet_set_icmp(struct dp_packet
*, uint8_t type
, uint8_t code
);
1597 void packet_set_nd(struct dp_packet
*, const struct in6_addr
*target
,
1598 const struct eth_addr sll
, const struct eth_addr tll
);
1599 void packet_set_nd_ext(struct dp_packet
*packet
,
1600 const ovs_16aligned_be32 rso_flags
,
1601 const uint8_t opt_type
);
1602 void packet_set_igmp3_query(struct dp_packet
*, uint8_t max_resp
,
1603 ovs_be32 group
, bool srs
, uint8_t qrv
,
1605 void packet_format_tcp_flags(struct ds
*, uint16_t);
1606 const char *packet_tcp_flag_to_string(uint32_t flag
);
1607 void *compose_ipv6(struct dp_packet
*packet
, uint8_t proto
,
1608 const struct in6_addr
*src
, const struct in6_addr
*dst
,
1609 uint8_t key_tc
, ovs_be32 key_fl
, uint8_t key_hl
, int size
);
1610 void compose_arp__(struct dp_packet
*);
1611 void compose_arp(struct dp_packet
*, uint16_t arp_op
,
1612 const struct eth_addr arp_sha
,
1613 const struct eth_addr arp_tha
, bool broadcast
,
1614 ovs_be32 arp_spa
, ovs_be32 arp_tpa
);
1615 void compose_nd_ns(struct dp_packet
*, const struct eth_addr eth_src
,
1616 const struct in6_addr
*ipv6_src
,
1617 const struct in6_addr
*ipv6_dst
);
1618 void compose_nd_na(struct dp_packet
*, const struct eth_addr eth_src
,
1619 const struct eth_addr eth_dst
,
1620 const struct in6_addr
*ipv6_src
,
1621 const struct in6_addr
*ipv6_dst
,
1622 ovs_be32 rso_flags
);
1623 void compose_nd_ra(struct dp_packet
*,
1624 const struct eth_addr eth_src
,
1625 const struct eth_addr eth_dst
,
1626 const struct in6_addr
*ipv6_src
,
1627 const struct in6_addr
*ipv6_dst
,
1628 uint8_t cur_hop_limit
, uint8_t mo_flags
,
1629 ovs_be16 router_lt
, ovs_be32 reachable_time
,
1630 ovs_be32 retrans_timer
, uint32_t mtu
);
1631 void packet_put_ra_prefix_opt(struct dp_packet
*,
1632 uint8_t plen
, uint8_t la_flags
,
1633 ovs_be32 valid_lifetime
,
1634 ovs_be32 preferred_lifetime
,
1635 const ovs_be128 router_prefix
);
1636 uint32_t packet_csum_pseudoheader(const struct ip_header
*);
1637 void IP_ECN_set_ce(struct dp_packet
*pkt
, bool is_ipv6
);
1639 #define DNS_HEADER_LEN 12
1642 uint8_t lo_flag
; /* QR (1), OPCODE (4), AA (1), TC (1) and RD (1) */
1643 uint8_t hi_flag
; /* RA (1), Z (3) and RCODE (4) */
1644 ovs_be16 qdcount
; /* Num of entries in the question section. */
1645 ovs_be16 ancount
; /* Num of resource records in the answer section. */
1647 /* Num of name server records in the authority record section. */
1650 /* Num of resource records in the additional records section. */
1654 BUILD_ASSERT_DECL(DNS_HEADER_LEN
== sizeof(struct dns_header
));
1656 #define DNS_QUERY_TYPE_A 0x01
1657 #define DNS_QUERY_TYPE_AAAA 0x1c
1658 #define DNS_QUERY_TYPE_ANY 0xff
1660 #define DNS_CLASS_IN 0x01
1661 #define DNS_DEFAULT_RR_TTL 3600
1663 #endif /* packets.h */