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
);
55 struct in6_addr
flow_tnl_dst(const struct flow_tnl
*tnl
);
56 struct in6_addr
flow_tnl_src(const struct flow_tnl
*tnl
);
58 /* Returns an offset to 'src' covering all the meaningful fields in 'src'. */
60 flow_tnl_size(const struct flow_tnl
*src
)
62 if (!flow_tnl_dst_is_set(src
)) {
63 /* Covers ip_dst and ipv6_dst only. */
64 return offsetof(struct flow_tnl
, ip_src
);
66 if (src
->flags
& FLOW_TNL_F_UDPIF
) {
67 /* Datapath format, cover all options we have. */
68 return offsetof(struct flow_tnl
, metadata
.opts
)
69 + src
->metadata
.present
.len
;
71 if (!src
->metadata
.present
.map
) {
72 /* No TLVs, opts is irrelevant. */
73 return offsetof(struct flow_tnl
, metadata
.opts
);
75 /* Have decoded TLVs, opts is relevant. */
79 /* Copy flow_tnl, but avoid copying unused portions of tun_metadata. Unused
80 * data in 'dst' is NOT cleared, so this must not be used in cases where the
81 * uninitialized portion may be hashed over. */
83 flow_tnl_copy__(struct flow_tnl
*dst
, const struct flow_tnl
*src
)
85 memcpy(dst
, src
, flow_tnl_size(src
));
89 flow_tnl_equal(const struct flow_tnl
*a
, const struct flow_tnl
*b
)
91 size_t a_size
= flow_tnl_size(a
);
93 return a_size
== flow_tnl_size(b
) && !memcmp(a
, b
, a_size
);
96 /* Datapath packet metadata */
98 PADDED_MEMBERS_CACHELINE_MARKER(CACHE_LINE_SIZE
, cacheline0
,
99 uint32_t recirc_id
; /* Recirculation id carried with the
100 recirculating packets. 0 for packets
101 received from the wire. */
102 uint32_t dp_hash
; /* hash value computed by the recirculation
104 uint32_t skb_priority
; /* Packet priority for QoS. */
105 uint32_t pkt_mark
; /* Packet mark. */
106 uint8_t ct_state
; /* Connection state. */
107 bool ct_orig_tuple_ipv6
;
108 uint16_t ct_zone
; /* Connection zone. */
109 uint32_t ct_mark
; /* Connection mark. */
110 ovs_u128 ct_label
; /* Connection label. */
111 union flow_in_port in_port
; /* Input port. */
112 struct conn
*conn
; /* Cached conntrack connection. */
113 bool reply
; /* True if reply direction. */
114 bool icmp_related
; /* True if ICMP related. */
117 PADDED_MEMBERS_CACHELINE_MARKER(CACHE_LINE_SIZE
, cacheline1
,
118 union { /* Populated only for non-zero 'ct_state'. */
119 struct ovs_key_ct_tuple_ipv4 ipv4
;
120 struct ovs_key_ct_tuple_ipv6 ipv6
; /* Used only if */
121 } ct_orig_tuple
; /* 'ct_orig_tuple_ipv6' is set */
124 PADDED_MEMBERS_CACHELINE_MARKER(CACHE_LINE_SIZE
, cacheline2
,
125 struct flow_tnl tunnel
; /* Encapsulating tunnel parameters. Note that
126 * if 'ip_dst' == 0, the rest of the fields may
127 * be uninitialized. */
131 BUILD_ASSERT_DECL(offsetof(struct pkt_metadata
, cacheline0
) == 0);
132 BUILD_ASSERT_DECL(offsetof(struct pkt_metadata
, cacheline1
) ==
134 BUILD_ASSERT_DECL(offsetof(struct pkt_metadata
, cacheline2
) ==
135 2 * CACHE_LINE_SIZE
);
138 pkt_metadata_init_tnl(struct pkt_metadata
*md
)
140 /* Zero up through the tunnel metadata options. The length and table
141 * are before this and as long as they are empty, the options won't
143 memset(md
, 0, offsetof(struct pkt_metadata
, tunnel
.metadata
.opts
));
147 pkt_metadata_init_conn(struct pkt_metadata
*md
)
153 pkt_metadata_init(struct pkt_metadata
*md
, odp_port_t port
)
155 /* This is called for every packet in userspace datapath and affects
156 * performance if all the metadata is initialized. Hence, fields should
157 * only be zeroed out when necessary.
159 * Initialize only till ct_state. Once the ct_state is zeroed out rest
160 * of ct fields will not be looked at unless ct_state != 0.
162 memset(md
, 0, offsetof(struct pkt_metadata
, ct_orig_tuple_ipv6
));
164 /* It can be expensive to zero out all of the tunnel metadata. However,
165 * we can just zero out ip_dst and the rest of the data will never be
167 md
->tunnel
.ip_dst
= 0;
168 md
->tunnel
.ipv6_dst
= in6addr_any
;
169 md
->in_port
.odp_port
= port
;
173 /* This function prefetches the cachelines touched by pkt_metadata_init()
174 * and pkt_metadata_init_tnl(). For performance reasons the two functions
175 * should be kept in sync. */
177 pkt_metadata_prefetch_init(struct pkt_metadata
*md
)
179 /* Prefetch cacheline0 as members till ct_state and odp_port will
180 * be initialized later in pkt_metadata_init(). */
181 OVS_PREFETCH(md
->cacheline0
);
183 /* Prefetch cacheline1 as members of this cacheline will be zeroed out
184 * in pkt_metadata_init_tnl(). */
185 OVS_PREFETCH(md
->cacheline1
);
187 /* Prefetch cachline2 as ip_dst & ipv6_dst fields will be initialized. */
188 OVS_PREFETCH(md
->cacheline2
);
191 bool dpid_from_string(const char *s
, uint64_t *dpidp
);
193 #define ETH_ADDR_LEN 6
195 static const struct eth_addr eth_addr_broadcast OVS_UNUSED
196 = ETH_ADDR_C(ff
,ff
,ff
,ff
,ff
,ff
);
198 static const struct eth_addr eth_addr_exact OVS_UNUSED
199 = ETH_ADDR_C(ff
,ff
,ff
,ff
,ff
,ff
);
201 static const struct eth_addr eth_addr_zero OVS_UNUSED
202 = ETH_ADDR_C(00,00,00,00,00,00);
203 static const struct eth_addr64 eth_addr64_zero OVS_UNUSED
204 = ETH_ADDR64_C(00,00,00,00,00,00,00,00);
206 static const struct eth_addr eth_addr_stp OVS_UNUSED
207 = ETH_ADDR_C(01,80,c2
,00,00,00);
209 static const struct eth_addr eth_addr_lacp OVS_UNUSED
210 = ETH_ADDR_C(01,80,c2
,00,00,02);
212 static const struct eth_addr eth_addr_bfd OVS_UNUSED
213 = ETH_ADDR_C(00,23,20,00,00,01);
215 static inline bool eth_addr_is_broadcast(const struct eth_addr a
)
217 return (a
.be16
[0] & a
.be16
[1] & a
.be16
[2]) == htons(0xffff);
220 static inline bool eth_addr_is_multicast(const struct eth_addr a
)
225 static inline bool eth_addr_is_local(const struct eth_addr a
)
227 /* Local if it is either a locally administered address or a Nicira random
230 || (a
.be16
[0] == htons(0x0023)
231 && (a
.be16
[1] & htons(0xff80)) == htons(0x2080));
233 static inline bool eth_addr_is_zero(const struct eth_addr a
)
235 return !(a
.be16
[0] | a
.be16
[1] | a
.be16
[2]);
237 static inline bool eth_addr64_is_zero(const struct eth_addr64 a
)
239 return !(a
.be16
[0] | a
.be16
[1] | a
.be16
[2] | a
.be16
[3]);
242 static inline int eth_mask_is_exact(const struct eth_addr a
)
244 return (a
.be16
[0] & a
.be16
[1] & a
.be16
[2]) == htons(0xffff);
247 static inline int eth_addr_compare_3way(const struct eth_addr a
,
248 const struct eth_addr b
)
250 return memcmp(&a
, &b
, sizeof a
);
252 static inline int eth_addr64_compare_3way(const struct eth_addr64 a
,
253 const struct eth_addr64 b
)
255 return memcmp(&a
, &b
, sizeof a
);
258 static inline bool eth_addr_equals(const struct eth_addr a
,
259 const struct eth_addr b
)
261 return !eth_addr_compare_3way(a
, b
);
263 static inline bool eth_addr64_equals(const struct eth_addr64 a
,
264 const struct eth_addr64 b
)
266 return !eth_addr64_compare_3way(a
, b
);
269 static inline bool eth_addr_equal_except(const struct eth_addr a
,
270 const struct eth_addr b
,
271 const struct eth_addr mask
)
273 return !(((a
.be16
[0] ^ b
.be16
[0]) & mask
.be16
[0])
274 || ((a
.be16
[1] ^ b
.be16
[1]) & mask
.be16
[1])
275 || ((a
.be16
[2] ^ b
.be16
[2]) & mask
.be16
[2]));
278 static inline uint64_t eth_addr_to_uint64(const struct eth_addr ea
)
280 return (((uint64_t) ntohs(ea
.be16
[0]) << 32)
281 | ((uint64_t) ntohs(ea
.be16
[1]) << 16)
282 | ntohs(ea
.be16
[2]));
285 static inline uint64_t eth_addr_vlan_to_uint64(const struct eth_addr ea
,
288 return (((uint64_t)vlan
<< 48) | eth_addr_to_uint64(ea
));
291 static inline void eth_addr_from_uint64(uint64_t x
, struct eth_addr
*ea
)
293 ea
->be16
[0] = htons(x
>> 32);
294 ea
->be16
[1] = htons((x
& 0xFFFF0000) >> 16);
295 ea
->be16
[2] = htons(x
& 0xFFFF);
298 static inline struct eth_addr
eth_addr_invert(const struct eth_addr src
)
302 for (int i
= 0; i
< ARRAY_SIZE(src
.be16
); i
++) {
303 dst
.be16
[i
] = ~src
.be16
[i
];
309 static inline void eth_addr_mark_random(struct eth_addr
*ea
)
311 ea
->ea
[0] &= ~1; /* Unicast. */
312 ea
->ea
[0] |= 2; /* Private. */
315 static inline void eth_addr_random(struct eth_addr
*ea
)
317 random_bytes((uint8_t *)ea
, sizeof *ea
);
318 eth_addr_mark_random(ea
);
321 static inline void eth_addr_nicira_random(struct eth_addr
*ea
)
325 /* Set the OUI to the Nicira one. */
330 /* Set the top bit to indicate random Nicira address. */
333 static inline uint32_t hash_mac(const struct eth_addr ea
,
334 const uint16_t vlan
, const uint32_t basis
)
336 return hash_uint64_basis(eth_addr_vlan_to_uint64(ea
, vlan
), basis
);
339 bool eth_addr_is_reserved(const struct eth_addr
);
340 bool eth_addr_from_string(const char *, struct eth_addr
*);
342 void compose_rarp(struct dp_packet
*, const struct eth_addr
);
344 void eth_push_vlan(struct dp_packet
*, ovs_be16 tpid
, ovs_be16 tci
);
345 void eth_pop_vlan(struct dp_packet
*);
347 const char *eth_from_hex(const char *hex
, struct dp_packet
**packetp
);
348 void eth_format_masked(const struct eth_addr ea
,
349 const struct eth_addr
*mask
, struct ds
*s
);
351 void set_mpls_lse(struct dp_packet
*, ovs_be32 label
);
352 void push_mpls(struct dp_packet
*packet
, ovs_be16 ethtype
, ovs_be32 lse
);
353 void pop_mpls(struct dp_packet
*, ovs_be16 ethtype
);
355 void set_mpls_lse_ttl(ovs_be32
*lse
, uint8_t ttl
);
356 void set_mpls_lse_tc(ovs_be32
*lse
, uint8_t tc
);
357 void set_mpls_lse_label(ovs_be32
*lse
, ovs_be32 label
);
358 void set_mpls_lse_bos(ovs_be32
*lse
, uint8_t bos
);
359 ovs_be32
set_mpls_lse_values(uint8_t ttl
, uint8_t tc
, uint8_t bos
,
364 * struct eth_addr mac;
366 * printf("The Ethernet address is "ETH_ADDR_FMT"\n", ETH_ADDR_ARGS(mac));
369 #define ETH_ADDR_FMT \
370 "%02"PRIx8":%02"PRIx8":%02"PRIx8":%02"PRIx8":%02"PRIx8":%02"PRIx8
371 #define ETH_ADDR_ARGS(EA) ETH_ADDR_BYTES_ARGS((EA).ea)
372 #define ETH_ADDR_BYTES_ARGS(EAB) \
373 (EAB)[0], (EAB)[1], (EAB)[2], (EAB)[3], (EAB)[4], (EAB)[5]
374 #define ETH_ADDR_STRLEN 17
378 * struct eth_addr64 eui64;
380 * printf("The EUI-64 address is "ETH_ADDR64_FMT"\n", ETH_ADDR64_ARGS(mac));
383 #define ETH_ADDR64_FMT \
384 "%02"PRIx8":%02"PRIx8":%02"PRIx8":%02"PRIx8":" \
385 "%02"PRIx8":%02"PRIx8":%02"PRIx8":%02"PRIx8
386 #define ETH_ADDR64_ARGS(EA) ETH_ADDR64_BYTES_ARGS((EA).ea64)
387 #define ETH_ADDR64_BYTES_ARGS(EAB) \
388 (EAB)[0], (EAB)[1], (EAB)[2], (EAB)[3], \
389 (EAB)[4], (EAB)[5], (EAB)[6], (EAB)[7]
390 #define ETH_ADDR64_STRLEN 23
394 * char *string = "1 00:11:22:33:44:55 2";
395 * struct eth_addr mac;
398 * if (ovs_scan(string, "%d"ETH_ADDR_SCAN_FMT"%d",
399 * &a, ETH_ADDR_SCAN_ARGS(mac), &b)) {
403 #define ETH_ADDR_SCAN_FMT "%"SCNx8":%"SCNx8":%"SCNx8":%"SCNx8":%"SCNx8":%"SCNx8
404 #define ETH_ADDR_SCAN_ARGS(EA) \
405 &(EA).ea[0], &(EA).ea[1], &(EA).ea[2], &(EA).ea[3], &(EA).ea[4], &(EA).ea[5]
407 #define ETH_TYPE_IP 0x0800
408 #define ETH_TYPE_ARP 0x0806
409 #define ETH_TYPE_TEB 0x6558
410 #define ETH_TYPE_VLAN_8021Q 0x8100
411 #define ETH_TYPE_VLAN ETH_TYPE_VLAN_8021Q
412 #define ETH_TYPE_VLAN_8021AD 0x88a8
413 #define ETH_TYPE_IPV6 0x86dd
414 #define ETH_TYPE_LACP 0x8809
415 #define ETH_TYPE_RARP 0x8035
416 #define ETH_TYPE_MPLS 0x8847
417 #define ETH_TYPE_MPLS_MCAST 0x8848
418 #define ETH_TYPE_NSH 0x894f
419 #define ETH_TYPE_ERSPAN1 0x88be /* version 1 type II */
420 #define ETH_TYPE_ERSPAN2 0x22eb /* version 2 type III */
422 static inline bool eth_type_mpls(ovs_be16 eth_type
)
424 return eth_type
== htons(ETH_TYPE_MPLS
) ||
425 eth_type
== htons(ETH_TYPE_MPLS_MCAST
);
428 static inline bool eth_type_vlan(ovs_be16 eth_type
)
430 return eth_type
== htons(ETH_TYPE_VLAN_8021Q
) ||
431 eth_type
== htons(ETH_TYPE_VLAN_8021AD
);
435 /* Minimum value for an Ethernet type. Values below this are IEEE 802.2 frame
437 #define ETH_TYPE_MIN 0x600
439 #define ETH_HEADER_LEN 14
440 #define ETH_PAYLOAD_MIN 46
441 #define ETH_PAYLOAD_MAX 1500
442 #define ETH_TOTAL_MIN (ETH_HEADER_LEN + ETH_PAYLOAD_MIN)
443 #define ETH_TOTAL_MAX (ETH_HEADER_LEN + ETH_PAYLOAD_MAX)
444 #define ETH_VLAN_TOTAL_MAX (ETH_HEADER_LEN + VLAN_HEADER_LEN + ETH_PAYLOAD_MAX)
446 struct eth_addr eth_dst
;
447 struct eth_addr eth_src
;
450 BUILD_ASSERT_DECL(ETH_HEADER_LEN
== sizeof(struct eth_header
));
452 void push_eth(struct dp_packet
*packet
, const struct eth_addr
*dst
,
453 const struct eth_addr
*src
);
454 void pop_eth(struct dp_packet
*packet
);
456 void push_nsh(struct dp_packet
*packet
, const struct nsh_hdr
*nsh_hdr_src
);
457 bool pop_nsh(struct dp_packet
*packet
);
459 #define LLC_DSAP_SNAP 0xaa
460 #define LLC_SSAP_SNAP 0xaa
461 #define LLC_CNTL_SNAP 3
463 #define LLC_HEADER_LEN 3
469 BUILD_ASSERT_DECL(LLC_HEADER_LEN
== sizeof(struct llc_header
));
471 /* LLC field values used for STP frames. */
472 #define STP_LLC_SSAP 0x42
473 #define STP_LLC_DSAP 0x42
474 #define STP_LLC_CNTL 0x03
476 #define SNAP_ORG_ETHERNET "\0\0" /* The compiler adds a null byte, so
477 sizeof(SNAP_ORG_ETHERNET) == 3. */
478 #define SNAP_HEADER_LEN 5
484 BUILD_ASSERT_DECL(SNAP_HEADER_LEN
== sizeof(struct snap_header
));
486 #define LLC_SNAP_HEADER_LEN (LLC_HEADER_LEN + SNAP_HEADER_LEN)
488 struct llc_snap_header
{
489 struct llc_header llc
;
490 struct snap_header snap
;
492 BUILD_ASSERT_DECL(LLC_SNAP_HEADER_LEN
== sizeof(struct llc_snap_header
));
494 #define VLAN_VID_MASK 0x0fff
495 #define VLAN_VID_SHIFT 0
497 #define VLAN_PCP_MASK 0xe000
498 #define VLAN_PCP_SHIFT 13
500 #define VLAN_CFI 0x1000
501 #define VLAN_CFI_SHIFT 12
503 /* Given the vlan_tci field from an 802.1Q header, in network byte order,
504 * returns the VLAN ID in host byte order. */
505 static inline uint16_t
506 vlan_tci_to_vid(ovs_be16 vlan_tci
)
508 return (ntohs(vlan_tci
) & VLAN_VID_MASK
) >> VLAN_VID_SHIFT
;
511 /* Given the vlan_tci field from an 802.1Q header, in network byte order,
512 * returns the priority code point (PCP) in host byte order. */
514 vlan_tci_to_pcp(ovs_be16 vlan_tci
)
516 return (ntohs(vlan_tci
) & VLAN_PCP_MASK
) >> VLAN_PCP_SHIFT
;
519 /* Given the vlan_tci field from an 802.1Q header, in network byte order,
520 * returns the Canonical Format Indicator (CFI). */
522 vlan_tci_to_cfi(ovs_be16 vlan_tci
)
524 return (vlan_tci
& htons(VLAN_CFI
)) != 0;
527 #define VLAN_HEADER_LEN 4
529 ovs_be16 vlan_tci
; /* Lowest 12 bits are VLAN ID. */
530 ovs_be16 vlan_next_type
;
532 BUILD_ASSERT_DECL(VLAN_HEADER_LEN
== sizeof(struct vlan_header
));
534 #define VLAN_ETH_HEADER_LEN (ETH_HEADER_LEN + VLAN_HEADER_LEN)
535 struct vlan_eth_header
{
536 struct eth_addr veth_dst
;
537 struct eth_addr veth_src
;
538 ovs_be16 veth_type
; /* Always htons(ETH_TYPE_VLAN). */
539 ovs_be16 veth_tci
; /* Lowest 12 bits are VLAN ID. */
540 ovs_be16 veth_next_type
;
542 BUILD_ASSERT_DECL(VLAN_ETH_HEADER_LEN
== sizeof(struct vlan_eth_header
));
544 /* MPLS related definitions */
545 #define MPLS_TTL_MASK 0x000000ff
546 #define MPLS_TTL_SHIFT 0
548 #define MPLS_BOS_MASK 0x00000100
549 #define MPLS_BOS_SHIFT 8
551 #define MPLS_TC_MASK 0x00000e00
552 #define MPLS_TC_SHIFT 9
554 #define MPLS_LABEL_MASK 0xfffff000
555 #define MPLS_LABEL_SHIFT 12
560 ovs_16aligned_be32 mpls_lse
;
562 BUILD_ASSERT_DECL(MPLS_HLEN
== sizeof(struct mpls_hdr
));
564 /* Given a mpls label stack entry in network byte order
565 * return mpls label in host byte order */
566 static inline uint32_t
567 mpls_lse_to_label(ovs_be32 mpls_lse
)
569 return (ntohl(mpls_lse
) & MPLS_LABEL_MASK
) >> MPLS_LABEL_SHIFT
;
572 /* Given a mpls label stack entry in network byte order
574 static inline uint8_t
575 mpls_lse_to_tc(ovs_be32 mpls_lse
)
577 return (ntohl(mpls_lse
) & MPLS_TC_MASK
) >> MPLS_TC_SHIFT
;
580 /* Given a mpls label stack entry in network byte order
582 static inline uint8_t
583 mpls_lse_to_ttl(ovs_be32 mpls_lse
)
585 return (ntohl(mpls_lse
) & MPLS_TTL_MASK
) >> MPLS_TTL_SHIFT
;
588 /* Set label in mpls lse. */
590 flow_set_mpls_lse_label(ovs_be32
*mpls_lse
, uint32_t label
)
592 *mpls_lse
&= ~htonl(MPLS_LABEL_MASK
);
593 *mpls_lse
|= htonl(label
<< MPLS_LABEL_SHIFT
);
596 /* Set TC in mpls lse. */
598 flow_set_mpls_lse_tc(ovs_be32
*mpls_lse
, uint8_t tc
)
600 *mpls_lse
&= ~htonl(MPLS_TC_MASK
);
601 *mpls_lse
|= htonl((tc
& 0x7) << MPLS_TC_SHIFT
);
604 /* Set BOS in mpls lse. */
606 flow_set_mpls_lse_bos(ovs_be32
*mpls_lse
, uint8_t bos
)
608 *mpls_lse
&= ~htonl(MPLS_BOS_MASK
);
609 *mpls_lse
|= htonl((bos
& 0x1) << MPLS_BOS_SHIFT
);
612 /* Set TTL in mpls lse. */
614 flow_set_mpls_lse_ttl(ovs_be32
*mpls_lse
, uint8_t ttl
)
616 *mpls_lse
&= ~htonl(MPLS_TTL_MASK
);
617 *mpls_lse
|= htonl(ttl
<< MPLS_TTL_SHIFT
);
620 /* Given a mpls label stack entry in network byte order
621 * return mpls BoS bit */
622 static inline uint8_t
623 mpls_lse_to_bos(ovs_be32 mpls_lse
)
625 return (mpls_lse
& htonl(MPLS_BOS_MASK
)) != 0;
628 #define IP_FMT "%"PRIu32".%"PRIu32".%"PRIu32".%"PRIu32
629 #define IP_ARGS(ip) \
631 (ntohl(ip) >> 16) & 0xff, \
632 (ntohl(ip) >> 8) & 0xff, \
637 * char *string = "1 33.44.55.66 2";
641 * if (ovs_scan(string, "%d"IP_SCAN_FMT"%d", &a, IP_SCAN_ARGS(&ip), &b)) {
645 #define IP_SCAN_FMT "%"SCNu8".%"SCNu8".%"SCNu8".%"SCNu8
646 #define IP_SCAN_ARGS(ip) \
647 ((void) (ovs_be32) *(ip), &((uint8_t *) ip)[0]), \
648 &((uint8_t *) ip)[1], \
649 &((uint8_t *) ip)[2], \
652 #define IP_PORT_SCAN_FMT "%"SCNu8".%"SCNu8".%"SCNu8".%"SCNu8":%"SCNu16
653 #define IP_PORT_SCAN_ARGS(ip, port) \
654 ((void) (ovs_be32) *(ip), &((uint8_t *) ip)[0]), \
655 &((uint8_t *) ip)[1], \
656 &((uint8_t *) ip)[2], \
657 &((uint8_t *) ip)[3], \
658 ((void) (ovs_be16) *(port), (uint16_t *) port)
660 /* Returns true if 'netmask' is a CIDR netmask, that is, if it consists of N
661 * high-order 1-bits and 32-N low-order 0-bits. */
663 ip_is_cidr(ovs_be32 netmask
)
665 uint32_t x
= ~ntohl(netmask
);
666 return !(x
& (x
+ 1));
669 ip_is_multicast(ovs_be32 ip
)
671 return (ip
& htonl(0xf0000000)) == htonl(0xe0000000);
674 ip_is_local_multicast(ovs_be32 ip
)
676 return (ip
& htonl(0xffffff00)) == htonl(0xe0000000);
678 int ip_count_cidr_bits(ovs_be32 netmask
);
679 void ip_format_masked(ovs_be32 ip
, ovs_be32 mask
, struct ds
*);
680 bool ip_parse(const char *s
, ovs_be32
*ip
);
681 char *ip_parse_port(const char *s
, ovs_be32
*ip
, ovs_be16
*port
)
682 OVS_WARN_UNUSED_RESULT
;
683 char *ip_parse_masked(const char *s
, ovs_be32
*ip
, ovs_be32
*mask
)
684 OVS_WARN_UNUSED_RESULT
;
685 char *ip_parse_cidr(const char *s
, ovs_be32
*ip
, unsigned int *plen
)
686 OVS_WARN_UNUSED_RESULT
;
687 char *ip_parse_masked_len(const char *s
, int *n
, ovs_be32
*ip
, ovs_be32
*mask
)
688 OVS_WARN_UNUSED_RESULT
;
689 char *ip_parse_cidr_len(const char *s
, int *n
, ovs_be32
*ip
,
691 OVS_WARN_UNUSED_RESULT
;
693 #define IP_VER(ip_ihl_ver) ((ip_ihl_ver) >> 4)
694 #define IP_IHL(ip_ihl_ver) ((ip_ihl_ver) & 15)
695 #define IP_IHL_VER(ihl, ver) (((ver) << 4) | (ihl))
698 #define IPPROTO_SCTP 132
702 #define IPPROTO_DCCP 33
706 #define IPPROTO_IGMP 2
709 #ifndef IPPROTO_UDPLITE
710 #define IPPROTO_UDPLITE 136
714 #define IP_ECN_NOT_ECT 0x0
715 #define IP_ECN_ECT_1 0x01
716 #define IP_ECN_ECT_0 0x02
717 #define IP_ECN_CE 0x03
718 #define IP_ECN_MASK 0x03
719 #define IP_DSCP_CS6 0xc0
720 #define IP_DSCP_MASK 0xfc
723 IP_ECN_is_ce(uint8_t dsfield
)
725 return (dsfield
& IP_ECN_MASK
) == IP_ECN_CE
;
730 #define IP_DONT_FRAGMENT 0x4000 /* Don't fragment. */
731 #define IP_MORE_FRAGMENTS 0x2000 /* More fragments. */
732 #define IP_FRAG_OFF_MASK 0x1fff /* Fragment offset. */
733 #define IP_IS_FRAGMENT(ip_frag_off) \
734 ((ip_frag_off) & htons(IP_MORE_FRAGMENTS | IP_FRAG_OFF_MASK))
736 #define IP_HEADER_LEN 20
742 ovs_be16 ip_frag_off
;
746 ovs_16aligned_be32 ip_src
;
747 ovs_16aligned_be32 ip_dst
;
749 BUILD_ASSERT_DECL(IP_HEADER_LEN
== sizeof(struct ip_header
));
752 #define ICMP4_ECHO_REPLY 0
753 #define ICMP4_DST_UNREACH 3
754 #define ICMP4_SOURCEQUENCH 4
755 #define ICMP4_REDIRECT 5
756 #define ICMP4_ECHO_REQUEST 8
757 #define ICMP4_TIME_EXCEEDED 11
758 #define ICMP4_PARAM_PROB 12
759 #define ICMP4_TIMESTAMP 13
760 #define ICMP4_TIMESTAMPREPLY 14
761 #define ICMP4_INFOREQUEST 15
762 #define ICMP4_INFOREPLY 16
764 #define ICMP_HEADER_LEN 8
778 ovs_16aligned_be32 gateway
;
781 BUILD_ASSERT_DECL(ICMP_HEADER_LEN
== sizeof(struct icmp_header
));
784 #define ICMP_ERROR_DATA_L4_LEN 8
786 #define IGMP_HEADER_LEN 8
791 ovs_16aligned_be32 group
;
793 BUILD_ASSERT_DECL(IGMP_HEADER_LEN
== sizeof(struct igmp_header
));
795 #define IGMPV3_HEADER_LEN 8
796 struct igmpv3_header
{
803 BUILD_ASSERT_DECL(IGMPV3_HEADER_LEN
== sizeof(struct igmpv3_header
));
805 #define IGMPV3_QUERY_HEADER_LEN 12
806 struct igmpv3_query_header
{
810 ovs_16aligned_be32 group
;
816 IGMPV3_QUERY_HEADER_LEN
== sizeof(struct igmpv3_query_header
819 #define IGMPV3_RECORD_LEN 8
820 struct igmpv3_record
{
824 ovs_16aligned_be32 maddr
;
826 BUILD_ASSERT_DECL(IGMPV3_RECORD_LEN
== sizeof(struct igmpv3_record
));
828 #define IGMP_HOST_MEMBERSHIP_QUERY 0x11 /* From RFC1112 */
829 #define IGMP_HOST_MEMBERSHIP_REPORT 0x12 /* Ditto */
830 #define IGMPV2_HOST_MEMBERSHIP_REPORT 0x16 /* V2 version of 0x12 */
831 #define IGMP_HOST_LEAVE_MESSAGE 0x17
832 #define IGMPV3_HOST_MEMBERSHIP_REPORT 0x22 /* V3 version of 0x12 */
835 * IGMPv3 and MLDv2 use the same codes.
837 #define IGMPV3_MODE_IS_INCLUDE 1
838 #define IGMPV3_MODE_IS_EXCLUDE 2
839 #define IGMPV3_CHANGE_TO_INCLUDE_MODE 3
840 #define IGMPV3_CHANGE_TO_EXCLUDE_MODE 4
841 #define IGMPV3_ALLOW_NEW_SOURCES 5
842 #define IGMPV3_BLOCK_OLD_SOURCES 6
844 #define SCTP_HEADER_LEN 12
848 ovs_16aligned_be32 sctp_vtag
;
849 ovs_16aligned_be32 sctp_csum
;
851 BUILD_ASSERT_DECL(SCTP_HEADER_LEN
== sizeof(struct sctp_header
));
853 #define UDP_HEADER_LEN 8
860 BUILD_ASSERT_DECL(UDP_HEADER_LEN
== sizeof(struct udp_header
));
862 #define ESP_HEADER_LEN 8
867 BUILD_ASSERT_DECL(ESP_HEADER_LEN
== sizeof(struct esp_header
));
869 #define ESP_TRAILER_LEN 2
874 BUILD_ASSERT_DECL(ESP_TRAILER_LEN
== sizeof(struct esp_trailer
));
876 #define TCP_FIN 0x001
877 #define TCP_SYN 0x002
878 #define TCP_RST 0x004
879 #define TCP_PSH 0x008
880 #define TCP_ACK 0x010
881 #define TCP_URG 0x020
882 #define TCP_ECE 0x040
883 #define TCP_CWR 0x080
886 #define TCP_CTL(flags, offset) (htons((flags) | ((offset) << 12)))
887 #define TCP_FLAGS(tcp_ctl) (ntohs(tcp_ctl) & 0x0fff)
888 #define TCP_FLAGS_BE16(tcp_ctl) ((tcp_ctl) & htons(0x0fff))
889 #define TCP_OFFSET(tcp_ctl) (ntohs(tcp_ctl) >> 12)
891 #define TCP_HEADER_LEN 20
895 ovs_16aligned_be32 tcp_seq
;
896 ovs_16aligned_be32 tcp_ack
;
902 BUILD_ASSERT_DECL(TCP_HEADER_LEN
== sizeof(struct tcp_header
));
904 /* Connection states.
906 * Names like CS_RELATED are bit values, e.g. 1 << 2.
907 * Names like CS_RELATED_BIT are bit indexes, e.g. 2. */
909 CS_STATE(NEW, 0, "new") \
910 CS_STATE(ESTABLISHED, 1, "est") \
911 CS_STATE(RELATED, 2, "rel") \
912 CS_STATE(REPLY_DIR, 3, "rpl") \
913 CS_STATE(INVALID, 4, "inv") \
914 CS_STATE(TRACKED, 5, "trk") \
915 CS_STATE(SRC_NAT, 6, "snat") \
916 CS_STATE(DST_NAT, 7, "dnat")
919 #define CS_STATE(ENUM, INDEX, NAME) \
920 CS_##ENUM = 1 << INDEX, \
921 CS_##ENUM##_BIT = INDEX,
926 /* Undefined connection state bits. */
928 #define CS_STATE(ENUM, INDEX, NAME) +CS_##ENUM
929 CS_SUPPORTED_MASK
= CS_STATES
932 #define CS_UNSUPPORTED_MASK (~(uint32_t)CS_SUPPORTED_MASK)
934 #define ARP_HRD_ETHERNET 1
935 #define ARP_PRO_IP 0x0800
936 #define ARP_OP_REQUEST 1
937 #define ARP_OP_REPLY 2
938 #define ARP_OP_RARP 3
940 #define ARP_ETH_HEADER_LEN 28
941 struct arp_eth_header
{
942 /* Generic members. */
943 ovs_be16 ar_hrd
; /* Hardware type. */
944 ovs_be16 ar_pro
; /* Protocol type. */
945 uint8_t ar_hln
; /* Hardware address length. */
946 uint8_t ar_pln
; /* Protocol address length. */
947 ovs_be16 ar_op
; /* Opcode. */
949 /* Ethernet+IPv4 specific members. */
950 struct eth_addr ar_sha
; /* Sender hardware address. */
951 ovs_16aligned_be32 ar_spa
; /* Sender protocol address. */
952 struct eth_addr ar_tha
; /* Target hardware address. */
953 ovs_16aligned_be32 ar_tpa
; /* Target protocol address. */
955 BUILD_ASSERT_DECL(ARP_ETH_HEADER_LEN
== sizeof(struct arp_eth_header
));
957 #define IPV6_HEADER_LEN 40
959 /* Like struct in6_addr, but whereas that struct requires 32-bit alignment on
960 * most implementations, this one only requires 16-bit alignment. */
961 union ovs_16aligned_in6_addr
{
963 ovs_16aligned_be32 be32
[4];
966 /* Like struct in6_hdr, but whereas that struct requires 32-bit alignment, this
967 * one only requires 16-bit alignment. */
968 struct ovs_16aligned_ip6_hdr
{
970 struct ovs_16aligned_ip6_hdrctl
{
971 ovs_16aligned_be32 ip6_un1_flow
;
972 ovs_be16 ip6_un1_plen
;
974 uint8_t ip6_un1_hlim
;
978 union ovs_16aligned_in6_addr ip6_src
;
979 union ovs_16aligned_in6_addr ip6_dst
;
982 /* Like struct in6_frag, but whereas that struct requires 32-bit alignment,
983 * this one only requires 16-bit alignment. */
984 struct ovs_16aligned_ip6_frag
{
986 uint8_t ip6f_reserved
;
988 ovs_16aligned_be32 ip6f_ident
;
991 #define ICMP6_HEADER_LEN 4
992 struct icmp6_header
{
995 ovs_be16 icmp6_cksum
;
997 BUILD_ASSERT_DECL(ICMP6_HEADER_LEN
== sizeof(struct icmp6_header
));
999 #define ICMP6_DATA_HEADER_LEN 8
1000 struct icmp6_data_header
{
1001 struct icmp6_header icmp6_base
;
1003 ovs_16aligned_be32 be32
[1];
1008 BUILD_ASSERT_DECL(ICMP6_DATA_HEADER_LEN
== sizeof(struct icmp6_data_header
));
1010 uint32_t packet_csum_pseudoheader6(const struct ovs_16aligned_ip6_hdr
*);
1011 ovs_be16
packet_csum_upperlayer6(const struct ovs_16aligned_ip6_hdr
*,
1012 const void *, uint8_t, uint16_t);
1014 /* Neighbor Discovery option field.
1015 * ND options are always a multiple of 8 bytes in size. */
1016 #define ND_LLA_OPT_LEN 8
1017 struct ovs_nd_lla_opt
{
1018 uint8_t type
; /* One of ND_OPT_*_LINKADDR. */
1020 struct eth_addr mac
;
1022 BUILD_ASSERT_DECL(ND_LLA_OPT_LEN
== sizeof(struct ovs_nd_lla_opt
));
1024 /* Neighbor Discovery option: Prefix Information. */
1025 #define ND_PREFIX_OPT_LEN 32
1026 struct ovs_nd_prefix_opt
{
1027 uint8_t type
; /* ND_OPT_PREFIX_INFORMATION. */
1028 uint8_t len
; /* Always 4. */
1030 uint8_t la_flags
; /* ND_PREFIX_* flags. */
1031 ovs_16aligned_be32 valid_lifetime
;
1032 ovs_16aligned_be32 preferred_lifetime
;
1033 ovs_16aligned_be32 reserved
; /* Always 0. */
1034 union ovs_16aligned_in6_addr prefix
;
1036 BUILD_ASSERT_DECL(ND_PREFIX_OPT_LEN
== sizeof(struct ovs_nd_prefix_opt
));
1038 /* Neighbor Discovery option: MTU. */
1039 #define ND_MTU_OPT_LEN 8
1040 #define ND_MTU_DEFAULT 0
1041 struct ovs_nd_mtu_opt
{
1042 uint8_t type
; /* ND_OPT_MTU */
1043 uint8_t len
; /* Always 1. */
1044 ovs_be16 reserved
; /* Always 0. */
1045 ovs_16aligned_be32 mtu
;
1047 BUILD_ASSERT_DECL(ND_MTU_OPT_LEN
== sizeof(struct ovs_nd_mtu_opt
));
1049 /* Like struct nd_msg (from ndisc.h), but whereas that struct requires 32-bit
1050 * alignment, this one only requires 16-bit alignment. */
1051 #define ND_MSG_LEN 24
1053 struct icmp6_header icmph
;
1054 ovs_16aligned_be32 rso_flags
;
1055 union ovs_16aligned_in6_addr target
;
1056 struct ovs_nd_lla_opt options
[0];
1058 BUILD_ASSERT_DECL(ND_MSG_LEN
== sizeof(struct ovs_nd_msg
));
1060 /* Neighbor Discovery packet flags. */
1061 #define ND_RSO_ROUTER 0x80000000
1062 #define ND_RSO_SOLICITED 0x40000000
1063 #define ND_RSO_OVERRIDE 0x20000000
1065 #define RA_MSG_LEN 16
1067 struct icmp6_header icmph
;
1068 uint8_t cur_hop_limit
;
1069 uint8_t mo_flags
; /* ND_RA_MANAGED_ADDRESS and ND_RA_OTHER_CONFIG flags. */
1070 ovs_be16 router_lifetime
;
1071 ovs_be32 reachable_time
;
1072 ovs_be32 retrans_timer
;
1073 struct ovs_nd_lla_opt options
[0];
1075 BUILD_ASSERT_DECL(RA_MSG_LEN
== sizeof(struct ovs_ra_msg
));
1077 #define ND_RA_MANAGED_ADDRESS 0x80
1078 #define ND_RA_OTHER_CONFIG 0x40
1080 /* Defaults based on MaxRtrInterval and MinRtrInterval from RFC 4861 section
1083 #define ND_RA_MAX_INTERVAL_DEFAULT 600
1086 nd_ra_min_interval_default(int max
)
1088 return max
>= 9 ? max
/ 3 : max
* 3 / 4;
1092 * Use the same struct for MLD and MLD2, naming members as the defined fields in
1093 * in the corresponding version of the protocol, though they are reserved in the
1096 #define MLD_HEADER_LEN 8
1104 BUILD_ASSERT_DECL(MLD_HEADER_LEN
== sizeof(struct mld_header
));
1106 #define MLD2_RECORD_LEN 20
1107 struct mld2_record
{
1111 union ovs_16aligned_in6_addr maddr
;
1113 BUILD_ASSERT_DECL(MLD2_RECORD_LEN
== sizeof(struct mld2_record
));
1115 #define MLD_QUERY 130
1116 #define MLD_REPORT 131
1117 #define MLD_DONE 132
1118 #define MLD2_REPORT 143
1120 /* The IPv6 flow label is in the lower 20 bits of the first 32-bit word. */
1121 #define IPV6_LABEL_MASK 0x000fffff
1125 * char *string = "1 ::1 2";
1126 * char ipv6_s[IPV6_SCAN_LEN + 1];
1127 * struct in6_addr ipv6;
1129 * if (ovs_scan(string, "%d"IPV6_SCAN_FMT"%d", &a, ipv6_s, &b)
1130 * && inet_pton(AF_INET6, ipv6_s, &ipv6) == 1) {
1134 #define IPV6_SCAN_FMT "%46[0123456789abcdefABCDEF:.]"
1135 #define IPV6_SCAN_LEN 46
1137 extern const struct in6_addr in6addr_exact
;
1138 #define IN6ADDR_EXACT_INIT { { { 0xff,0xff,0xff,0xff,0xff,0xff,0xff,0xff, \
1139 0xff,0xff,0xff,0xff,0xff,0xff,0xff,0xff } } }
1141 extern const struct in6_addr in6addr_all_hosts
;
1142 #define IN6ADDR_ALL_HOSTS_INIT { { { 0xff,0x02,0x00,0x00,0x00,0x00,0x00,0x00, \
1143 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x01 } } }
1145 extern const struct in6_addr in6addr_all_routers
;
1146 #define IN6ADDR_ALL_ROUTERS_INIT { { { 0xff,0x02,0x00,0x00,0x00,0x00,0x00,0x00, \
1147 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x02 } } }
1149 static inline bool ipv6_addr_equals(const struct in6_addr
*a
,
1150 const struct in6_addr
*b
)
1152 #ifdef IN6_ARE_ADDR_EQUAL
1153 return IN6_ARE_ADDR_EQUAL(a
, b
);
1155 return !memcmp(a
, b
, sizeof(*a
));
1159 /* Checks the IPv6 address in 'mask' for all zeroes. */
1160 static inline bool ipv6_mask_is_any(const struct in6_addr
*mask
) {
1161 return ipv6_addr_equals(mask
, &in6addr_any
);
1164 static inline bool ipv6_mask_is_exact(const struct in6_addr
*mask
) {
1165 return ipv6_addr_equals(mask
, &in6addr_exact
);
1168 static inline bool ipv6_is_all_hosts(const struct in6_addr
*addr
) {
1169 return ipv6_addr_equals(addr
, &in6addr_all_hosts
);
1172 static inline bool ipv6_addr_is_set(const struct in6_addr
*addr
) {
1173 return !ipv6_addr_equals(addr
, &in6addr_any
);
1176 static inline bool ipv6_addr_is_multicast(const struct in6_addr
*ip
) {
1177 return ip
->s6_addr
[0] == 0xff;
1180 static inline struct in6_addr
1181 in6_addr_mapped_ipv4(ovs_be32 ip4
)
1183 struct in6_addr ip6
;
1184 memset(&ip6
, 0, sizeof(ip6
));
1185 ip6
.s6_addr
[10] = 0xff, ip6
.s6_addr
[11] = 0xff;
1186 memcpy(&ip6
.s6_addr
[12], &ip4
, 4);
1191 in6_addr_set_mapped_ipv4(struct in6_addr
*ip6
, ovs_be32 ip4
)
1193 *ip6
= in6_addr_mapped_ipv4(ip4
);
1196 static inline ovs_be32
1197 in6_addr_get_mapped_ipv4(const struct in6_addr
*addr
)
1199 union ovs_16aligned_in6_addr
*taddr
=
1200 (union ovs_16aligned_in6_addr
*) addr
;
1201 if (IN6_IS_ADDR_V4MAPPED(addr
)) {
1202 return get_16aligned_be32(&taddr
->be32
[3]);
1209 in6_addr_solicited_node(struct in6_addr
*addr
, const struct in6_addr
*ip6
)
1211 union ovs_16aligned_in6_addr
*taddr
=
1212 (union ovs_16aligned_in6_addr
*) addr
;
1213 memset(taddr
->be16
, 0, sizeof(taddr
->be16
));
1214 taddr
->be16
[0] = htons(0xff02);
1215 taddr
->be16
[5] = htons(0x1);
1216 taddr
->be16
[6] = htons(0xff00);
1217 memcpy(&addr
->s6_addr
[13], &ip6
->s6_addr
[13], 3);
1221 * Generates ipv6 EUI64 address from the given eth addr
1222 * and prefix and stores it in 'lla'
1225 in6_generate_eui64(struct eth_addr ea
, struct in6_addr
*prefix
,
1226 struct in6_addr
*lla
)
1228 union ovs_16aligned_in6_addr
*taddr
=
1229 (union ovs_16aligned_in6_addr
*) lla
;
1230 union ovs_16aligned_in6_addr
*prefix_taddr
=
1231 (union ovs_16aligned_in6_addr
*) prefix
;
1232 taddr
->be16
[0] = prefix_taddr
->be16
[0];
1233 taddr
->be16
[1] = prefix_taddr
->be16
[1];
1234 taddr
->be16
[2] = prefix_taddr
->be16
[2];
1235 taddr
->be16
[3] = prefix_taddr
->be16
[3];
1236 taddr
->be16
[4] = htons(((ea
.ea
[0] ^ 0x02) << 8) | ea
.ea
[1]);
1237 taddr
->be16
[5] = htons(ea
.ea
[2] << 8 | 0x00ff);
1238 taddr
->be16
[6] = htons(0xfe << 8 | ea
.ea
[3]);
1239 taddr
->be16
[7] = ea
.be16
[2];
1243 * Generates ipv6 link local address from the given eth addr
1244 * with prefix 'fe80::/64' and stores it in 'lla'
1247 in6_generate_lla(struct eth_addr ea
, struct in6_addr
*lla
)
1249 union ovs_16aligned_in6_addr
*taddr
=
1250 (union ovs_16aligned_in6_addr
*) lla
;
1251 memset(taddr
->be16
, 0, sizeof(taddr
->be16
));
1252 taddr
->be16
[0] = htons(0xfe80);
1253 taddr
->be16
[4] = htons(((ea
.ea
[0] ^ 0x02) << 8) | ea
.ea
[1]);
1254 taddr
->be16
[5] = htons(ea
.ea
[2] << 8 | 0x00ff);
1255 taddr
->be16
[6] = htons(0xfe << 8 | ea
.ea
[3]);
1256 taddr
->be16
[7] = ea
.be16
[2];
1259 /* Returns true if 'addr' is a link local address. Otherwise, false. */
1261 in6_is_lla(struct in6_addr
*addr
)
1264 return addr
->s6_addr32
[0] == htonl(0xfe800000) && !(addr
->s6_addr32
[1]);
1266 return addr
->s6_addr
[0] == 0xfe && addr
->s6_addr
[1] == 0x80 &&
1267 !(addr
->s6_addr
[2] | addr
->s6_addr
[3] | addr
->s6_addr
[4] |
1268 addr
->s6_addr
[5] | addr
->s6_addr
[6] | addr
->s6_addr
[7]);
1273 ipv6_multicast_to_ethernet(struct eth_addr
*eth
, const struct in6_addr
*ip6
)
1277 eth
->ea
[2] = ip6
->s6_addr
[12];
1278 eth
->ea
[3] = ip6
->s6_addr
[13];
1279 eth
->ea
[4] = ip6
->s6_addr
[14];
1280 eth
->ea
[5] = ip6
->s6_addr
[15];
1283 static inline bool dl_type_is_ip_any(ovs_be16 dl_type
)
1285 return dl_type
== htons(ETH_TYPE_IP
)
1286 || dl_type
== htons(ETH_TYPE_IPV6
);
1291 /* GRE protocol header */
1292 struct gre_base_hdr
{
1297 #define GRE_CSUM 0x8000
1298 #define GRE_ROUTING 0x4000
1299 #define GRE_KEY 0x2000
1300 #define GRE_SEQ 0x1000
1301 #define GRE_STRICT 0x0800
1302 #define GRE_REC 0x0700
1303 #define GRE_FLAGS 0x00F8
1304 #define GRE_VERSION 0x0007
1307 * ERSPAN protocol header and metadata
1309 * Version 1 (Type II) header (8 octets [42:49])
1311 * 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
1312 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1313 * | Ver | VLAN | COS | En|T| Session ID |
1314 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1315 * | Reserved | Index |
1316 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1319 * ERSPAN Version 2 (Type III) header (12 octets [42:49])
1321 * 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
1322 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1323 * | Ver | VLAN | COS |BSO|T| Session ID |
1324 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1326 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1327 * | SGT |P| FT | Hw ID |D|Gra|O|
1328 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1332 /* ERSPAN has fixed 8-byte GRE header */
1333 #define ERSPAN_GREHDR_LEN 8
1334 #define ERSPAN_HDR(gre_base_hdr) \
1335 ((struct erspan_base_hdr *)((char *)gre_base_hdr + ERSPAN_GREHDR_LEN))
1337 #define ERSPAN_V1_MDSIZE 4
1338 #define ERSPAN_V2_MDSIZE 8
1340 #define ERSPAN_SID_MASK 0x03ff /* 10-bit Session ID. */
1341 #define ERSPAN_IDX_MASK 0xfffff /* v1 Index */
1342 #define ERSPAN_HWID_MASK 0x03f0
1343 #define ERSPAN_DIR_MASK 0x0008
1345 struct erspan_base_hdr
{
1346 #ifdef WORDS_BIGENDIAN
1354 uint8_t session_id
:8;
1356 uint8_t vlan_upper
:4,
1359 uint8_t session_id_upper
:2,
1363 uint8_t session_id
:8;
1368 ovs_16aligned_be32 timestamp
;
1370 #ifdef WORDS_BIGENDIAN
1379 uint8_t hwid_upper
:2,
1389 struct erspan_metadata
{
1392 ovs_be32 index
; /* Version 1 (type II)*/
1393 struct erspan_md2 md2
; /* Version 2 (type III) */
1397 static inline uint16_t get_sid(const struct erspan_base_hdr
*ershdr
)
1399 return (ershdr
->session_id_upper
<< 8) + ershdr
->session_id
;
1402 static inline void set_sid(struct erspan_base_hdr
*ershdr
, uint16_t id
)
1404 ershdr
->session_id
= id
& 0xff;
1405 ershdr
->session_id_upper
= (id
>> 8) &0x3;
1408 static inline uint8_t get_hwid(const struct erspan_md2
*md2
)
1410 return (md2
->hwid_upper
<< 4) + md2
->hwid
;
1413 static inline void set_hwid(struct erspan_md2
*md2
, uint8_t hwid
)
1415 md2
->hwid
= hwid
& 0xf;
1416 md2
->hwid_upper
= (hwid
>> 4) & 0x3;
1419 /* ERSPAN timestamp granularity
1420 * 00b --> granularity = 100 microseconds
1421 * 01b --> granularity = 100 nanoseconds
1422 * 10b --> granularity = IEEE 1588
1423 * Here we only support 100 microseconds.
1425 enum erspan_ts_gra
{
1431 static inline ovs_be32
get_erspan_ts(enum erspan_ts_gra gra
)
1437 ts
= htonl((uint32_t)(time_wall_usec() / 100));
1441 case ERSPAN_IEEE1588
:
1450 /* VXLAN protocol header */
1453 ovs_16aligned_be32 vx_flags
; /* VXLAN flags. */
1455 uint8_t flags
; /* VXLAN GPE flags. */
1456 uint8_t reserved
[2]; /* 16 bits reserved. */
1457 uint8_t next_protocol
; /* Next Protocol field for VXLAN GPE. */
1460 ovs_16aligned_be32 vx_vni
;
1462 BUILD_ASSERT_DECL(sizeof(struct vxlanhdr
) == 8);
1464 #define VXLAN_FLAGS 0x08000000 /* struct vxlanhdr.vx_flags required value. */
1467 * VXLAN Generic Protocol Extension (VXLAN_F_GPE):
1468 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1469 * |R|R|Ver|I|P|R|O| Reserved |Next Protocol |
1470 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1471 * | VXLAN Network Identifier (VNI) | Reserved |
1472 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1474 * Ver = Version. Indicates VXLAN GPE protocol version.
1476 * P = Next Protocol Bit. The P bit is set to indicate that the
1477 * Next Protocol field is present.
1479 * O = OAM Flag Bit. The O bit is set to indicate that the packet
1482 * Next Protocol = This 8 bit field indicates the protocol header
1483 * immediately following the VXLAN GPE header.
1485 * https://tools.ietf.org/html/draft-ietf-nvo3-vxlan-gpe-01
1488 /* Fields in struct vxlanhdr.vx_gpe.flags */
1489 #define VXLAN_GPE_FLAGS_VER 0x30 /* Version. */
1490 #define VLXAN_GPE_FLAGS_P 0x04 /* Next Protocol Bit. */
1491 #define VXLAN_GPE_FLAGS_O 0x01 /* OAM Bit. */
1493 /* VXLAN-GPE header flags. */
1494 #define VXLAN_HF_VER ((1U <<29) | (1U <<28))
1495 #define VXLAN_HF_NP (1U <<26)
1496 #define VXLAN_HF_OAM (1U <<24)
1498 #define VXLAN_GPE_USED_BITS (VXLAN_HF_VER | VXLAN_HF_NP | VXLAN_HF_OAM | \
1501 /* VXLAN-GPE header Next Protocol. */
1502 #define VXLAN_GPE_NP_IPV4 0x01
1503 #define VXLAN_GPE_NP_IPV6 0x02
1504 #define VXLAN_GPE_NP_ETHERNET 0x03
1505 #define VXLAN_GPE_NP_NSH 0x04
1507 #define VXLAN_F_GPE 0x4000
1508 #define VXLAN_HF_GPE 0x04000000
1510 /* Input values for PACKET_TYPE macros have to be in host byte order.
1511 * The _BE postfix indicates result is in network byte order. Otherwise result
1512 * is in host byte order. */
1513 #define PACKET_TYPE(NS, NS_TYPE) ((uint32_t) ((NS) << 16 | (NS_TYPE)))
1514 #define PACKET_TYPE_BE(NS, NS_TYPE) (htonl((NS) << 16 | (NS_TYPE)))
1516 /* Returns the host byte ordered namespace of 'packet type'. */
1517 static inline uint16_t
1518 pt_ns(ovs_be32 packet_type
)
1520 return ntohl(packet_type
) >> 16;
1523 /* Returns the network byte ordered namespace type of 'packet type'. */
1524 static inline ovs_be16
1525 pt_ns_type_be(ovs_be32 packet_type
)
1527 return be32_to_be16(packet_type
);
1530 /* Returns the host byte ordered namespace type of 'packet type'. */
1531 static inline uint16_t
1532 pt_ns_type(ovs_be32 packet_type
)
1534 return ntohs(pt_ns_type_be(packet_type
));
1537 /* Well-known packet_type field values. */
1539 PT_ETH
= PACKET_TYPE(OFPHTN_ONF
, 0x0000), /* Default PT: Ethernet */
1540 PT_USE_NEXT_PROTO
= PACKET_TYPE(OFPHTN_ONF
, 0xfffe), /* Pseudo PT for decap. */
1541 PT_IPV4
= PACKET_TYPE(OFPHTN_ETHERTYPE
, ETH_TYPE_IP
),
1542 PT_IPV6
= PACKET_TYPE(OFPHTN_ETHERTYPE
, ETH_TYPE_IPV6
),
1543 PT_MPLS
= PACKET_TYPE(OFPHTN_ETHERTYPE
, ETH_TYPE_MPLS
),
1544 PT_MPLS_MC
= PACKET_TYPE(OFPHTN_ETHERTYPE
, ETH_TYPE_MPLS_MCAST
),
1545 PT_NSH
= PACKET_TYPE(OFPHTN_ETHERTYPE
, ETH_TYPE_NSH
),
1546 PT_UNKNOWN
= PACKET_TYPE(0xffff, 0xffff), /* Unknown packet type. */
1550 void ipv6_format_addr(const struct in6_addr
*addr
, struct ds
*);
1551 void ipv6_format_addr_bracket(const struct in6_addr
*addr
, struct ds
*,
1553 void ipv6_format_mapped(const struct in6_addr
*addr
, struct ds
*);
1554 void ipv6_format_masked(const struct in6_addr
*addr
,
1555 const struct in6_addr
*mask
, struct ds
*);
1556 const char * ipv6_string_mapped(char *addr_str
, const struct in6_addr
*addr
);
1557 struct in6_addr
ipv6_addr_bitand(const struct in6_addr
*src
,
1558 const struct in6_addr
*mask
);
1559 struct in6_addr
ipv6_addr_bitxor(const struct in6_addr
*a
,
1560 const struct in6_addr
*b
);
1561 bool ipv6_is_zero(const struct in6_addr
*a
);
1562 struct in6_addr
ipv6_create_mask(int mask
);
1563 int ipv6_count_cidr_bits(const struct in6_addr
*netmask
);
1564 bool ipv6_is_cidr(const struct in6_addr
*netmask
);
1566 bool ipv6_parse(const char *s
, struct in6_addr
*ip
);
1567 char *ipv6_parse_masked(const char *s
, struct in6_addr
*ipv6
,
1568 struct in6_addr
*mask
);
1569 char *ipv6_parse_cidr(const char *s
, struct in6_addr
*ip
, unsigned int *plen
)
1570 OVS_WARN_UNUSED_RESULT
;
1571 char *ipv6_parse_masked_len(const char *s
, int *n
, struct in6_addr
*ipv6
,
1572 struct in6_addr
*mask
);
1573 char *ipv6_parse_cidr_len(const char *s
, int *n
, struct in6_addr
*ip
,
1575 OVS_WARN_UNUSED_RESULT
;
1577 void *eth_compose(struct dp_packet
*, const struct eth_addr eth_dst
,
1578 const struct eth_addr eth_src
, uint16_t eth_type
,
1580 void *snap_compose(struct dp_packet
*, const struct eth_addr eth_dst
,
1581 const struct eth_addr eth_src
,
1582 unsigned int oui
, uint16_t snap_type
, size_t size
);
1583 void packet_set_ipv4(struct dp_packet
*, ovs_be32 src
, ovs_be32 dst
, uint8_t tos
,
1585 void packet_set_ipv4_addr(struct dp_packet
*packet
, ovs_16aligned_be32
*addr
,
1587 void packet_set_ipv6(struct dp_packet
*, const struct in6_addr
*src
,
1588 const struct in6_addr
*dst
, uint8_t tc
,
1589 ovs_be32 fl
, uint8_t hlmit
);
1590 void packet_set_ipv6_addr(struct dp_packet
*packet
, uint8_t proto
,
1591 ovs_16aligned_be32 addr
[4],
1592 const struct in6_addr
*new_addr
,
1593 bool recalculate_csum
);
1594 void packet_set_tcp_port(struct dp_packet
*, ovs_be16 src
, ovs_be16 dst
);
1595 void packet_set_udp_port(struct dp_packet
*, ovs_be16 src
, ovs_be16 dst
);
1596 void packet_set_sctp_port(struct dp_packet
*, ovs_be16 src
, ovs_be16 dst
);
1597 void packet_set_icmp(struct dp_packet
*, uint8_t type
, uint8_t code
);
1598 void packet_set_nd(struct dp_packet
*, const struct in6_addr
*target
,
1599 const struct eth_addr sll
, const struct eth_addr tll
);
1600 void packet_set_nd_ext(struct dp_packet
*packet
,
1601 const ovs_16aligned_be32 rso_flags
,
1602 const uint8_t opt_type
);
1603 void packet_set_igmp3_query(struct dp_packet
*, uint8_t max_resp
,
1604 ovs_be32 group
, bool srs
, uint8_t qrv
,
1606 void packet_format_tcp_flags(struct ds
*, uint16_t);
1607 const char *packet_tcp_flag_to_string(uint32_t flag
);
1608 void compose_arp__(struct dp_packet
*);
1609 void compose_arp(struct dp_packet
*, uint16_t arp_op
,
1610 const struct eth_addr arp_sha
,
1611 const struct eth_addr arp_tha
, bool broadcast
,
1612 ovs_be32 arp_spa
, ovs_be32 arp_tpa
);
1613 void compose_nd_ns(struct dp_packet
*, const struct eth_addr eth_src
,
1614 const struct in6_addr
*ipv6_src
,
1615 const struct in6_addr
*ipv6_dst
);
1616 void compose_nd_na(struct dp_packet
*, const struct eth_addr eth_src
,
1617 const struct eth_addr eth_dst
,
1618 const struct in6_addr
*ipv6_src
,
1619 const struct in6_addr
*ipv6_dst
,
1620 ovs_be32 rso_flags
);
1621 void compose_nd_ra(struct dp_packet
*,
1622 const struct eth_addr eth_src
,
1623 const struct eth_addr eth_dst
,
1624 const struct in6_addr
*ipv6_src
,
1625 const struct in6_addr
*ipv6_dst
,
1626 uint8_t cur_hop_limit
, uint8_t mo_flags
,
1627 ovs_be16 router_lt
, ovs_be32 reachable_time
,
1628 ovs_be32 retrans_timer
, uint32_t mtu
);
1629 void packet_put_ra_prefix_opt(struct dp_packet
*,
1630 uint8_t plen
, uint8_t la_flags
,
1631 ovs_be32 valid_lifetime
,
1632 ovs_be32 preferred_lifetime
,
1633 const ovs_be128 router_prefix
);
1634 uint32_t packet_csum_pseudoheader(const struct ip_header
*);
1635 void IP_ECN_set_ce(struct dp_packet
*pkt
, bool is_ipv6
);
1637 #define DNS_HEADER_LEN 12
1640 uint8_t lo_flag
; /* QR (1), OPCODE (4), AA (1), TC (1) and RD (1) */
1641 uint8_t hi_flag
; /* RA (1), Z (3) and RCODE (4) */
1642 ovs_be16 qdcount
; /* Num of entries in the question section. */
1643 ovs_be16 ancount
; /* Num of resource records in the answer section. */
1645 /* Num of name server records in the authority record section. */
1648 /* Num of resource records in the additional records section. */
1652 BUILD_ASSERT_DECL(DNS_HEADER_LEN
== sizeof(struct dns_header
));
1654 #define DNS_QUERY_TYPE_A 0x01
1655 #define DNS_QUERY_TYPE_AAAA 0x1c
1656 #define DNS_QUERY_TYPE_ANY 0xff
1658 #define DNS_CLASS_IN 0x01
1659 #define DNS_DEFAULT_RR_TTL 3600
1661 #endif /* packets.h */