lib/packets.h

   1 /*
   2  * Copyright (c) 2008, 2009, 2010, 2011, 2012, 2013, 2014, 2015, 2016, 2017 Nicira, Inc.
   3  *
   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:
   7  *
   8  *     http://www.apache.org/licenses/LICENSE-2.0
   9  *
  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.
  15  */
  16
  17 #ifndef PACKETS_H
  18 #define PACKETS_H 1
  19
  20 #include <inttypes.h>
  21 #include <sys/types.h>
  22 #include <stdint.h>
  23 #include <string.h>
  24 #include "compiler.h"
  25 #include "openvswitch/geneve.h"
  26 #include "openvswitch/packets.h"
  27 #include "openvswitch/types.h"
  28 #include "odp-netlink.h"
  29 #include "random.h"
  30 #include "hash.h"
  31 #include "tun-metadata.h"
  32 #include "unaligned.h"
  33 #include "util.h"
  34
  35 struct dp_packet;
  36 struct ds;
  37
  38 /* Purely internal to OVS userspace. These flags should never be exposed to
  39  * the outside world and so aren't included in the flags mask. */
  40
  41 /* Tunnel information is in userspace datapath format. */
  42 #define FLOW_TNL_F_UDPIF (1 << 4)
  43
  44 static inline bool ipv6_addr_is_set(const struct in6_addr *addr);
  45
  46 static inline bool
  47 flow_tnl_dst_is_set(const struct flow_tnl *tnl)
  48 {
  49     return tnl->ip_dst || ipv6_addr_is_set(&tnl->ipv6_dst);
  50 }
  51
  52 struct in6_addr flow_tnl_dst(const struct flow_tnl *tnl);
  53 struct in6_addr flow_tnl_src(const struct flow_tnl *tnl);
  54
  55 /* Returns an offset to 'src' covering all the meaningful fields in 'src'. */
  56 static inline size_t
  57 flow_tnl_size(const struct flow_tnl *src)
  58 {
  59     if (!flow_tnl_dst_is_set(src)) {
  60         /* Covers ip_dst and ipv6_dst only. */
  61         return offsetof(struct flow_tnl, ip_src);
  62     }
  63     if (src->flags & FLOW_TNL_F_UDPIF) {
  64         /* Datapath format, cover all options we have. */
  65         return offsetof(struct flow_tnl, metadata.opts)
  66             + src->metadata.present.len;
  67     }
  68     if (!src->metadata.present.map) {
  69         /* No TLVs, opts is irrelevant. */
  70         return offsetof(struct flow_tnl, metadata.opts);
  71     }
  72     /* Have decoded TLVs, opts is relevant. */
  73     return sizeof *src;
  74 }
  75
  76 /* Copy flow_tnl, but avoid copying unused portions of tun_metadata.  Unused
  77  * data in 'dst' is NOT cleared, so this must not be used in cases where the
  78  * uninitialized portion may be hashed over. */
  79 static inline void
  80 flow_tnl_copy__(struct flow_tnl *dst, const struct flow_tnl *src)
  81 {
  82     memcpy(dst, src, flow_tnl_size(src));
  83 }
  84
  85 static inline bool
  86 flow_tnl_equal(const struct flow_tnl *a, const struct flow_tnl *b)
  87 {
  88     size_t a_size = flow_tnl_size(a);
  89
  90     return a_size == flow_tnl_size(b) && !memcmp(a, b, a_size);
  91 }
  92
  93 /* Datapath packet metadata */
  94 struct pkt_metadata {
  95     uint32_t recirc_id;         /* Recirculation id carried with the
  96                                    recirculating packets. 0 for packets
  97                                    received from the wire. */
  98     uint32_t dp_hash;           /* hash value computed by the recirculation
  99                                    action. */
 100     uint32_t skb_priority;      /* Packet priority for QoS. */
 101     uint32_t pkt_mark;          /* Packet mark. */
 102     uint8_t  ct_state;          /* Connection state. */
 103     bool ct_orig_tuple_ipv6;
 104     uint16_t ct_zone;           /* Connection zone. */
 105     uint32_t ct_mark;           /* Connection mark. */
 106     ovs_u128 ct_label;          /* Connection label. */
 107     union {                     /* Populated only for non-zero 'ct_state'. */
 108         struct ovs_key_ct_tuple_ipv4 ipv4;
 109         struct ovs_key_ct_tuple_ipv6 ipv6;   /* Used only if                */
 110     } ct_orig_tuple;                         /* 'ct_orig_tuple_ipv6' is set */
 111     union flow_in_port in_port; /* Input port. */
 112     struct flow_tnl tunnel;     /* Encapsulating tunnel parameters. Note that
 113                                  * if 'ip_dst' == 0, the rest of the fields may
 114                                  * be uninitialized. */
 115 };
 116
 117 static inline void
 118 pkt_metadata_init_tnl(struct pkt_metadata *md)
 119 {
 120     /* Zero up through the tunnel metadata options. The length and table
 121      * are before this and as long as they are empty, the options won't
 122      * be looked at. */
 123     memset(md, 0, offsetof(struct pkt_metadata, tunnel.metadata.opts));
 124 }
 125
 126 static inline void
 127 pkt_metadata_init(struct pkt_metadata *md, odp_port_t port)
 128 {
 129     /* It can be expensive to zero out all of the tunnel metadata. However,
 130      * we can just zero out ip_dst and the rest of the data will never be
 131      * looked at. */
 132     memset(md, 0, offsetof(struct pkt_metadata, in_port));
 133     md->tunnel.ip_dst = 0;
 134     md->tunnel.ipv6_dst = in6addr_any;
 135
 136     md->in_port.odp_port = port;
 137 }
 138
 139 /* This function prefetches the cachelines touched by pkt_metadata_init()
 140  * For performance reasons the two functions should be kept in sync. */
 141 static inline void
 142 pkt_metadata_prefetch_init(struct pkt_metadata *md)
 143 {
 144     ovs_prefetch_range(md, offsetof(struct pkt_metadata, tunnel.ip_src));
 145 }
 146
 147 bool dpid_from_string(const char *s, uint64_t *dpidp);
 148
 149 #define ETH_ADDR_LEN           6
 150
 151 static const struct eth_addr eth_addr_broadcast OVS_UNUSED
 152     = { { { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff } } };
 153
 154 static const struct eth_addr eth_addr_exact OVS_UNUSED
 155     = { { { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff } } };
 156
 157 static const struct eth_addr eth_addr_zero OVS_UNUSED
 158     = { { { 0x00, 0x00, 0x00, 0x00, 0x00, 0x00 } } };
 159 static const struct eth_addr64 eth_addr64_zero OVS_UNUSED
 160     = { { { 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00 } } };
 161
 162 static const struct eth_addr eth_addr_stp OVS_UNUSED
 163     = { { { 0x01, 0x80, 0xC2, 0x00, 0x00, 0x00 } } };
 164
 165 static const struct eth_addr eth_addr_lacp OVS_UNUSED
 166     = { { { 0x01, 0x80, 0xC2, 0x00, 0x00, 0x02 } } };
 167
 168 static const struct eth_addr eth_addr_bfd OVS_UNUSED
 169     = { { { 0x00, 0x23, 0x20, 0x00, 0x00, 0x01 } } };
 170
 171 static inline bool eth_addr_is_broadcast(const struct eth_addr a)
 172 {
 173     return (a.be16[0] & a.be16[1] & a.be16[2]) == htons(0xffff);
 174 }
 175
 176 static inline bool eth_addr_is_multicast(const struct eth_addr a)
 177 {
 178     return a.ea[0] & 1;
 179 }
 180
 181 static inline bool eth_addr_is_local(const struct eth_addr a)
 182 {
 183     /* Local if it is either a locally administered address or a Nicira random
 184      * address. */
 185     return a.ea[0] & 2
 186         || (a.be16[0] == htons(0x0023)
 187             && (a.be16[1] & htons(0xff80)) == htons(0x2080));
 188 }
 189 static inline bool eth_addr_is_zero(const struct eth_addr a)
 190 {
 191     return !(a.be16[0] | a.be16[1] | a.be16[2]);
 192 }
 193 static inline bool eth_addr64_is_zero(const struct eth_addr64 a)
 194 {
 195     return !(a.be16[0] | a.be16[1] | a.be16[2] | a.be16[3]);
 196 }
 197
 198 static inline int eth_mask_is_exact(const struct eth_addr a)
 199 {
 200     return (a.be16[0] & a.be16[1] & a.be16[2]) == htons(0xffff);
 201 }
 202
 203 static inline int eth_addr_compare_3way(const struct eth_addr a,
 204                                         const struct eth_addr b)
 205 {
 206     return memcmp(&a, &b, sizeof a);
 207 }
 208 static inline int eth_addr64_compare_3way(const struct eth_addr64 a,
 209                                           const struct eth_addr64 b)
 210 {
 211     return memcmp(&a, &b, sizeof a);
 212 }
 213
 214 static inline bool eth_addr_equals(const struct eth_addr a,
 215                                    const struct eth_addr b)
 216 {
 217     return !eth_addr_compare_3way(a, b);
 218 }
 219 static inline bool eth_addr64_equals(const struct eth_addr64 a,
 220                                      const struct eth_addr64 b)
 221 {
 222     return !eth_addr64_compare_3way(a, b);
 223 }
 224
 225 static inline bool eth_addr_equal_except(const struct eth_addr a,
 226                                          const struct eth_addr b,
 227                                          const struct eth_addr mask)
 228 {
 229     return !(((a.be16[0] ^ b.be16[0]) & mask.be16[0])
 230              || ((a.be16[1] ^ b.be16[1]) & mask.be16[1])
 231              || ((a.be16[2] ^ b.be16[2]) & mask.be16[2]));
 232 }
 233
 234 static inline uint64_t eth_addr_to_uint64(const struct eth_addr ea)
 235 {
 236     return (((uint64_t) ntohs(ea.be16[0]) << 32)
 237             | ((uint64_t) ntohs(ea.be16[1]) << 16)
 238             | ntohs(ea.be16[2]));
 239 }
 240
 241 static inline uint64_t eth_addr_vlan_to_uint64(const struct eth_addr ea,
 242                                                uint16_t vlan)
 243 {
 244     return (((uint64_t)vlan << 48) | eth_addr_to_uint64(ea));
 245 }
 246
 247 static inline void eth_addr_from_uint64(uint64_t x, struct eth_addr *ea)
 248 {
 249     ea->be16[0] = htons(x >> 32);
 250     ea->be16[1] = htons((x & 0xFFFF0000) >> 16);
 251     ea->be16[2] = htons(x & 0xFFFF);
 252 }
 253
 254 static inline struct eth_addr eth_addr_invert(const struct eth_addr src)
 255 {
 256     struct eth_addr dst;
 257
 258     for (int i = 0; i < ARRAY_SIZE(src.be16); i++) {
 259         dst.be16[i] = ~src.be16[i];
 260     }
 261
 262     return dst;
 263 }
 264
 265 static inline void eth_addr_mark_random(struct eth_addr *ea)
 266 {
 267     ea->ea[0] &= ~1;                /* Unicast. */
 268     ea->ea[0] |= 2;                 /* Private. */
 269 }
 270
 271 static inline void eth_addr_random(struct eth_addr *ea)
 272 {
 273     random_bytes((uint8_t *)ea, sizeof *ea);
 274     eth_addr_mark_random(ea);
 275 }
 276
 277 static inline void eth_addr_nicira_random(struct eth_addr *ea)
 278 {
 279     eth_addr_random(ea);
 280
 281     /* Set the OUI to the Nicira one. */
 282     ea->ea[0] = 0x00;
 283     ea->ea[1] = 0x23;
 284     ea->ea[2] = 0x20;
 285
 286     /* Set the top bit to indicate random Nicira address. */
 287     ea->ea[3] |= 0x80;
 288 }
 289 static inline uint32_t hash_mac(const struct eth_addr ea,
 290                                 const uint16_t vlan, const uint32_t basis)
 291 {
 292     return hash_uint64_basis(eth_addr_vlan_to_uint64(ea, vlan), basis);
 293 }
 294
 295 bool eth_addr_is_reserved(const struct eth_addr);
 296 bool eth_addr_from_string(const char *, struct eth_addr *);
 297
 298 void compose_rarp(struct dp_packet *, const struct eth_addr);
 299
 300 void eth_push_vlan(struct dp_packet *, ovs_be16 tpid, ovs_be16 tci);
 301 void eth_pop_vlan(struct dp_packet *);
 302
 303 const char *eth_from_hex(const char *hex, struct dp_packet **packetp);
 304 void eth_format_masked(const struct eth_addr ea,
 305                        const struct eth_addr *mask, struct ds *s);
 306
 307 void set_mpls_lse(struct dp_packet *, ovs_be32 label);
 308 void push_mpls(struct dp_packet *packet, ovs_be16 ethtype, ovs_be32 lse);
 309 void pop_mpls(struct dp_packet *, ovs_be16 ethtype);
 310
 311 void set_mpls_lse_ttl(ovs_be32 *lse, uint8_t ttl);
 312 void set_mpls_lse_tc(ovs_be32 *lse, uint8_t tc);
 313 void set_mpls_lse_label(ovs_be32 *lse, ovs_be32 label);
 314 void set_mpls_lse_bos(ovs_be32 *lse, uint8_t bos);
 315 ovs_be32 set_mpls_lse_values(uint8_t ttl, uint8_t tc, uint8_t bos,
 316                              ovs_be32 label);
 317
 318 /* Example:
 319  *
 320  * struct eth_addr mac;
 321  *    [...]
 322  * printf("The Ethernet address is "ETH_ADDR_FMT"\n", ETH_ADDR_ARGS(mac));
 323  *
 324  */
 325 #define ETH_ADDR_FMT                                                    \
 326     "%02"PRIx8":%02"PRIx8":%02"PRIx8":%02"PRIx8":%02"PRIx8":%02"PRIx8
 327 #define ETH_ADDR_ARGS(EA) ETH_ADDR_BYTES_ARGS((EA).ea)
 328 #define ETH_ADDR_BYTES_ARGS(EAB) \
 329          (EAB)[0], (EAB)[1], (EAB)[2], (EAB)[3], (EAB)[4], (EAB)[5]
 330 #define ETH_ADDR_STRLEN 17
 331
 332 /* Example:
 333  *
 334  * struct eth_addr64 eui64;
 335  *    [...]
 336  * printf("The EUI-64 address is "ETH_ADDR64_FMT"\n", ETH_ADDR64_ARGS(mac));
 337  *
 338  */
 339 #define ETH_ADDR64_FMT \
 340     "%02"PRIx8":%02"PRIx8":%02"PRIx8":%02"PRIx8":" \
 341     "%02"PRIx8":%02"PRIx8":%02"PRIx8":%02"PRIx8
 342 #define ETH_ADDR64_ARGS(EA) ETH_ADDR64_BYTES_ARGS((EA).ea64)
 343 #define ETH_ADDR64_BYTES_ARGS(EAB) \
 344          (EAB)[0], (EAB)[1], (EAB)[2], (EAB)[3], \
 345          (EAB)[4], (EAB)[5], (EAB)[6], (EAB)[7]
 346 #define ETH_ADDR64_STRLEN 23
 347
 348 /* Example:
 349  *
 350  * char *string = "1 00:11:22:33:44:55 2";
 351  * struct eth_addr mac;
 352  * int a, b;
 353  *
 354  * if (ovs_scan(string, "%d"ETH_ADDR_SCAN_FMT"%d",
 355  *              &a, ETH_ADDR_SCAN_ARGS(mac), &b)) {
 356  *     ...
 357  * }
 358  */
 359 #define ETH_ADDR_SCAN_FMT "%"SCNx8":%"SCNx8":%"SCNx8":%"SCNx8":%"SCNx8":%"SCNx8
 360 #define ETH_ADDR_SCAN_ARGS(EA) \
 361     &(EA).ea[0], &(EA).ea[1], &(EA).ea[2], &(EA).ea[3], &(EA).ea[4], &(EA).ea[5]
 362
 363 #define ETH_TYPE_IP            0x0800
 364 #define ETH_TYPE_ARP           0x0806
 365 #define ETH_TYPE_TEB           0x6558
 366 #define ETH_TYPE_VLAN_8021Q    0x8100
 367 #define ETH_TYPE_VLAN          ETH_TYPE_VLAN_8021Q
 368 #define ETH_TYPE_VLAN_8021AD   0x88a8
 369 #define ETH_TYPE_IPV6          0x86dd
 370 #define ETH_TYPE_LACP          0x8809
 371 #define ETH_TYPE_RARP          0x8035
 372 #define ETH_TYPE_MPLS          0x8847
 373 #define ETH_TYPE_MPLS_MCAST    0x8848
 374
 375 static inline bool eth_type_mpls(ovs_be16 eth_type)
 376 {
 377     return eth_type == htons(ETH_TYPE_MPLS) ||
 378         eth_type == htons(ETH_TYPE_MPLS_MCAST);
 379 }
 380
 381 static inline bool eth_type_vlan(ovs_be16 eth_type)
 382 {
 383     return eth_type == htons(ETH_TYPE_VLAN_8021Q) ||
 384         eth_type == htons(ETH_TYPE_VLAN_8021AD);
 385 }
 386
 387
 388 /* Minimum value for an Ethernet type.  Values below this are IEEE 802.2 frame
 389  * lengths. */
 390 #define ETH_TYPE_MIN           0x600
 391
 392 #define ETH_HEADER_LEN 14
 393 #define ETH_PAYLOAD_MIN 46
 394 #define ETH_PAYLOAD_MAX 1500
 395 #define ETH_TOTAL_MIN (ETH_HEADER_LEN + ETH_PAYLOAD_MIN)
 396 #define ETH_TOTAL_MAX (ETH_HEADER_LEN + ETH_PAYLOAD_MAX)
 397 #define ETH_VLAN_TOTAL_MAX (ETH_HEADER_LEN + VLAN_HEADER_LEN + ETH_PAYLOAD_MAX)
 398 OVS_PACKED(
 399 struct eth_header {
 400     struct eth_addr eth_dst;
 401     struct eth_addr eth_src;
 402     ovs_be16 eth_type;
 403 });
 404 BUILD_ASSERT_DECL(ETH_HEADER_LEN == sizeof(struct eth_header));
 405
 406 void push_eth(struct dp_packet *packet, const struct eth_addr *dst,
 407               const struct eth_addr *src);
 408 void pop_eth(struct dp_packet *packet);
 409
 410 #define LLC_DSAP_SNAP 0xaa
 411 #define LLC_SSAP_SNAP 0xaa
 412 #define LLC_CNTL_SNAP 3
 413
 414 #define LLC_HEADER_LEN 3
 415 OVS_PACKED(
 416 struct llc_header {
 417     uint8_t llc_dsap;
 418     uint8_t llc_ssap;
 419     uint8_t llc_cntl;
 420 });
 421 BUILD_ASSERT_DECL(LLC_HEADER_LEN == sizeof(struct llc_header));
 422
 423 /* LLC field values used for STP frames. */
 424 #define STP_LLC_SSAP 0x42
 425 #define STP_LLC_DSAP 0x42
 426 #define STP_LLC_CNTL 0x03
 427
 428 #define SNAP_ORG_ETHERNET "\0\0" /* The compiler adds a null byte, so
 429                                     sizeof(SNAP_ORG_ETHERNET) == 3. */
 430 #define SNAP_HEADER_LEN 5
 431 OVS_PACKED(
 432 struct snap_header {
 433     uint8_t snap_org[3];
 434     ovs_be16 snap_type;
 435 });
 436 BUILD_ASSERT_DECL(SNAP_HEADER_LEN == sizeof(struct snap_header));
 437
 438 #define LLC_SNAP_HEADER_LEN (LLC_HEADER_LEN + SNAP_HEADER_LEN)
 439 OVS_PACKED(
 440 struct llc_snap_header {
 441     struct llc_header llc;
 442     struct snap_header snap;
 443 });
 444 BUILD_ASSERT_DECL(LLC_SNAP_HEADER_LEN == sizeof(struct llc_snap_header));
 445
 446 #define VLAN_VID_MASK 0x0fff
 447 #define VLAN_VID_SHIFT 0
 448
 449 #define VLAN_PCP_MASK 0xe000
 450 #define VLAN_PCP_SHIFT 13
 451
 452 #define VLAN_CFI 0x1000
 453 #define VLAN_CFI_SHIFT 12
 454
 455 /* Given the vlan_tci field from an 802.1Q header, in network byte order,
 456  * returns the VLAN ID in host byte order. */
 457 static inline uint16_t
 458 vlan_tci_to_vid(ovs_be16 vlan_tci)
 459 {
 460     return (ntohs(vlan_tci) & VLAN_VID_MASK) >> VLAN_VID_SHIFT;
 461 }
 462
 463 /* Given the vlan_tci field from an 802.1Q header, in network byte order,
 464  * returns the priority code point (PCP) in host byte order. */
 465 static inline int
 466 vlan_tci_to_pcp(ovs_be16 vlan_tci)
 467 {
 468     return (ntohs(vlan_tci) & VLAN_PCP_MASK) >> VLAN_PCP_SHIFT;
 469 }
 470
 471 /* Given the vlan_tci field from an 802.1Q header, in network byte order,
 472  * returns the Canonical Format Indicator (CFI). */
 473 static inline int
 474 vlan_tci_to_cfi(ovs_be16 vlan_tci)
 475 {
 476     return (vlan_tci & htons(VLAN_CFI)) != 0;
 477 }
 478
 479 #define VLAN_HEADER_LEN 4
 480 struct vlan_header {
 481     ovs_be16 vlan_tci;          /* Lowest 12 bits are VLAN ID. */
 482     ovs_be16 vlan_next_type;
 483 };
 484 BUILD_ASSERT_DECL(VLAN_HEADER_LEN == sizeof(struct vlan_header));
 485
 486 #define VLAN_ETH_HEADER_LEN (ETH_HEADER_LEN + VLAN_HEADER_LEN)
 487 OVS_PACKED(
 488 struct vlan_eth_header {
 489     struct eth_addr veth_dst;
 490     struct eth_addr veth_src;
 491     ovs_be16 veth_type;         /* Always htons(ETH_TYPE_VLAN). */
 492     ovs_be16 veth_tci;          /* Lowest 12 bits are VLAN ID. */
 493     ovs_be16 veth_next_type;
 494 });
 495 BUILD_ASSERT_DECL(VLAN_ETH_HEADER_LEN == sizeof(struct vlan_eth_header));
 496
 497 /* MPLS related definitions */
 498 #define MPLS_TTL_MASK       0x000000ff
 499 #define MPLS_TTL_SHIFT      0
 500
 501 #define MPLS_BOS_MASK       0x00000100
 502 #define MPLS_BOS_SHIFT      8
 503
 504 #define MPLS_TC_MASK        0x00000e00
 505 #define MPLS_TC_SHIFT       9
 506
 507 #define MPLS_LABEL_MASK     0xfffff000
 508 #define MPLS_LABEL_SHIFT    12
 509
 510 #define MPLS_HLEN           4
 511
 512 struct mpls_hdr {
 513     ovs_16aligned_be32 mpls_lse;
 514 };
 515 BUILD_ASSERT_DECL(MPLS_HLEN == sizeof(struct mpls_hdr));
 516
 517 /* Given a mpls label stack entry in network byte order
 518  * return mpls label in host byte order */
 519 static inline uint32_t
 520 mpls_lse_to_label(ovs_be32 mpls_lse)
 521 {
 522     return (ntohl(mpls_lse) & MPLS_LABEL_MASK) >> MPLS_LABEL_SHIFT;
 523 }
 524
 525 /* Given a mpls label stack entry in network byte order
 526  * return mpls tc */
 527 static inline uint8_t
 528 mpls_lse_to_tc(ovs_be32 mpls_lse)
 529 {
 530     return (ntohl(mpls_lse) & MPLS_TC_MASK) >> MPLS_TC_SHIFT;
 531 }
 532
 533 /* Given a mpls label stack entry in network byte order
 534  * return mpls ttl */
 535 static inline uint8_t
 536 mpls_lse_to_ttl(ovs_be32 mpls_lse)
 537 {
 538     return (ntohl(mpls_lse) & MPLS_TTL_MASK) >> MPLS_TTL_SHIFT;
 539 }
 540
 541 /* Set TTL in mpls lse. */
 542 static inline void
 543 flow_set_mpls_lse_ttl(ovs_be32 *mpls_lse, uint8_t ttl)
 544 {
 545     *mpls_lse &= ~htonl(MPLS_TTL_MASK);
 546     *mpls_lse |= htonl(ttl << MPLS_TTL_SHIFT);
 547 }
 548
 549 /* Given a mpls label stack entry in network byte order
 550  * return mpls BoS bit  */
 551 static inline uint8_t
 552 mpls_lse_to_bos(ovs_be32 mpls_lse)
 553 {
 554     return (mpls_lse & htonl(MPLS_BOS_MASK)) != 0;
 555 }
 556
 557 #define IP_FMT "%"PRIu32".%"PRIu32".%"PRIu32".%"PRIu32
 558 #define IP_ARGS(ip)                             \
 559     ntohl(ip) >> 24,                            \
 560     (ntohl(ip) >> 16) & 0xff,                   \
 561     (ntohl(ip) >> 8) & 0xff,                    \
 562     ntohl(ip) & 0xff
 563
 564 /* Example:
 565  *
 566  * char *string = "1 33.44.55.66 2";
 567  * ovs_be32 ip;
 568  * int a, b;
 569  *
 570  * if (ovs_scan(string, "%d"IP_SCAN_FMT"%d", &a, IP_SCAN_ARGS(&ip), &b)) {
 571  *     ...
 572  * }
 573  */
 574 #define IP_SCAN_FMT "%"SCNu8".%"SCNu8".%"SCNu8".%"SCNu8
 575 #define IP_SCAN_ARGS(ip)                                    \
 576         ((void) (ovs_be32) *(ip), &((uint8_t *) ip)[0]),    \
 577         &((uint8_t *) ip)[1],                               \
 578         &((uint8_t *) ip)[2],                               \
 579         &((uint8_t *) ip)[3]
 580
 581 #define IP_PORT_SCAN_FMT "%"SCNu8".%"SCNu8".%"SCNu8".%"SCNu8":%"SCNu16
 582 #define IP_PORT_SCAN_ARGS(ip, port)                                    \
 583         ((void) (ovs_be32) *(ip), &((uint8_t *) ip)[0]),    \
 584         &((uint8_t *) ip)[1],                               \
 585         &((uint8_t *) ip)[2],                               \
 586         &((uint8_t *) ip)[3],                               \
 587         ((void) (ovs_be16) *(port), (uint16_t *) port)
 588
 589 /* Returns true if 'netmask' is a CIDR netmask, that is, if it consists of N
 590  * high-order 1-bits and 32-N low-order 0-bits. */
 591 static inline bool
 592 ip_is_cidr(ovs_be32 netmask)
 593 {
 594     uint32_t x = ~ntohl(netmask);
 595     return !(x & (x + 1));
 596 }
 597 static inline bool
 598 ip_is_multicast(ovs_be32 ip)
 599 {
 600     return (ip & htonl(0xf0000000)) == htonl(0xe0000000);
 601 }
 602 static inline bool
 603 ip_is_local_multicast(ovs_be32 ip)
 604 {
 605     return (ip & htonl(0xffffff00)) == htonl(0xe0000000);
 606 }
 607 int ip_count_cidr_bits(ovs_be32 netmask);
 608 void ip_format_masked(ovs_be32 ip, ovs_be32 mask, struct ds *);
 609 bool ip_parse(const char *s, ovs_be32 *ip);
 610 char *ip_parse_port(const char *s, ovs_be32 *ip, ovs_be16 *port)
 611     OVS_WARN_UNUSED_RESULT;
 612 char *ip_parse_masked(const char *s, ovs_be32 *ip, ovs_be32 *mask)
 613     OVS_WARN_UNUSED_RESULT;
 614 char *ip_parse_cidr(const char *s, ovs_be32 *ip, unsigned int *plen)
 615     OVS_WARN_UNUSED_RESULT;
 616 char *ip_parse_masked_len(const char *s, int *n, ovs_be32 *ip, ovs_be32 *mask)
 617     OVS_WARN_UNUSED_RESULT;
 618 char *ip_parse_cidr_len(const char *s, int *n, ovs_be32 *ip,
 619                         unsigned int *plen)
 620     OVS_WARN_UNUSED_RESULT;
 621
 622 #define IP_VER(ip_ihl_ver) ((ip_ihl_ver) >> 4)
 623 #define IP_IHL(ip_ihl_ver) ((ip_ihl_ver) & 15)
 624 #define IP_IHL_VER(ihl, ver) (((ver) << 4) | (ihl))
 625
 626 #ifndef IPPROTO_SCTP
 627 #define IPPROTO_SCTP 132
 628 #endif
 629
 630 #ifndef IPPROTO_DCCP
 631 #define IPPROTO_DCCP 33
 632 #endif
 633
 634 #ifndef IPPROTO_IGMP
 635 #define IPPROTO_IGMP 2
 636 #endif
 637
 638 #ifndef IPPROTO_UDPLITE
 639 #define IPPROTO_UDPLITE 136
 640 #endif
 641
 642 /* TOS fields. */
 643 #define IP_ECN_NOT_ECT 0x0
 644 #define IP_ECN_ECT_1 0x01
 645 #define IP_ECN_ECT_0 0x02
 646 #define IP_ECN_CE 0x03
 647 #define IP_ECN_MASK 0x03
 648 #define IP_DSCP_MASK 0xfc
 649
 650 static inline int
 651 IP_ECN_is_ce(uint8_t dsfield)
 652 {
 653     return (dsfield & IP_ECN_MASK) == IP_ECN_CE;
 654 }
 655
 656 #define IP_VERSION 4
 657
 658 #define IP_DONT_FRAGMENT  0x4000 /* Don't fragment. */
 659 #define IP_MORE_FRAGMENTS 0x2000 /* More fragments. */
 660 #define IP_FRAG_OFF_MASK  0x1fff /* Fragment offset. */
 661 #define IP_IS_FRAGMENT(ip_frag_off) \
 662         ((ip_frag_off) & htons(IP_MORE_FRAGMENTS | IP_FRAG_OFF_MASK))
 663
 664 #define IP_HEADER_LEN 20
 665 struct ip_header {
 666     uint8_t ip_ihl_ver;
 667     uint8_t ip_tos;
 668     ovs_be16 ip_tot_len;
 669     ovs_be16 ip_id;
 670     ovs_be16 ip_frag_off;
 671     uint8_t ip_ttl;
 672     uint8_t ip_proto;
 673     ovs_be16 ip_csum;
 674     ovs_16aligned_be32 ip_src;
 675     ovs_16aligned_be32 ip_dst;
 676 };
 677 BUILD_ASSERT_DECL(IP_HEADER_LEN == sizeof(struct ip_header));
 678
 679 /* ICMPv4 types. */
 680 #define ICMP4_ECHO_REPLY 0
 681 #define ICMP4_DST_UNREACH 3
 682 #define ICMP4_SOURCEQUENCH 4
 683 #define ICMP4_REDIRECT 5
 684 #define ICMP4_ECHO_REQUEST 8
 685 #define ICMP4_TIME_EXCEEDED 11
 686 #define ICMP4_PARAM_PROB 12
 687 #define ICMP4_TIMESTAMP 13
 688 #define ICMP4_TIMESTAMPREPLY 14
 689 #define ICMP4_INFOREQUEST 15
 690 #define ICMP4_INFOREPLY 16
 691
 692 #define ICMP_HEADER_LEN 8
 693 struct icmp_header {
 694     uint8_t icmp_type;
 695     uint8_t icmp_code;
 696     ovs_be16 icmp_csum;
 697     union {
 698         struct {
 699             ovs_be16 id;
 700             ovs_be16 seq;
 701         } echo;
 702         struct {
 703             ovs_be16 empty;
 704             ovs_be16 mtu;
 705         } frag;
 706         ovs_16aligned_be32 gateway;
 707     } icmp_fields;
 708 };
 709 BUILD_ASSERT_DECL(ICMP_HEADER_LEN == sizeof(struct icmp_header));
 710
 711 #define IGMP_HEADER_LEN 8
 712 struct igmp_header {
 713     uint8_t igmp_type;
 714     uint8_t igmp_code;
 715     ovs_be16 igmp_csum;
 716     ovs_16aligned_be32 group;
 717 };
 718 BUILD_ASSERT_DECL(IGMP_HEADER_LEN == sizeof(struct igmp_header));
 719
 720 #define IGMPV3_HEADER_LEN 8
 721 struct igmpv3_header {
 722     uint8_t type;
 723     uint8_t rsvr1;
 724     ovs_be16 csum;
 725     ovs_be16 rsvr2;
 726     ovs_be16 ngrp;
 727 };
 728 BUILD_ASSERT_DECL(IGMPV3_HEADER_LEN == sizeof(struct igmpv3_header));
 729
 730 #define IGMPV3_RECORD_LEN 8
 731 struct igmpv3_record {
 732     uint8_t type;
 733     uint8_t aux_len;
 734     ovs_be16 nsrcs;
 735     ovs_16aligned_be32 maddr;
 736 };
 737 BUILD_ASSERT_DECL(IGMPV3_RECORD_LEN == sizeof(struct igmpv3_record));
 738
 739 #define IGMP_HOST_MEMBERSHIP_QUERY    0x11 /* From RFC1112 */
 740 #define IGMP_HOST_MEMBERSHIP_REPORT   0x12 /* Ditto */
 741 #define IGMPV2_HOST_MEMBERSHIP_REPORT 0x16 /* V2 version of 0x12 */
 742 #define IGMP_HOST_LEAVE_MESSAGE       0x17
 743 #define IGMPV3_HOST_MEMBERSHIP_REPORT 0x22 /* V3 version of 0x12 */
 744
 745 /*
 746  * IGMPv3 and MLDv2 use the same codes.
 747  */
 748 #define IGMPV3_MODE_IS_INCLUDE 1
 749 #define IGMPV3_MODE_IS_EXCLUDE 2
 750 #define IGMPV3_CHANGE_TO_INCLUDE_MODE 3
 751 #define IGMPV3_CHANGE_TO_EXCLUDE_MODE 4
 752 #define IGMPV3_ALLOW_NEW_SOURCES 5
 753 #define IGMPV3_BLOCK_OLD_SOURCES 6
 754
 755 #define SCTP_HEADER_LEN 12
 756 struct sctp_header {
 757     ovs_be16 sctp_src;
 758     ovs_be16 sctp_dst;
 759     ovs_16aligned_be32 sctp_vtag;
 760     ovs_16aligned_be32 sctp_csum;
 761 };
 762 BUILD_ASSERT_DECL(SCTP_HEADER_LEN == sizeof(struct sctp_header));
 763
 764 #define UDP_HEADER_LEN 8
 765 struct udp_header {
 766     ovs_be16 udp_src;
 767     ovs_be16 udp_dst;
 768     ovs_be16 udp_len;
 769     ovs_be16 udp_csum;
 770 };
 771 BUILD_ASSERT_DECL(UDP_HEADER_LEN == sizeof(struct udp_header));
 772
 773 #define TCP_FIN 0x001
 774 #define TCP_SYN 0x002
 775 #define TCP_RST 0x004
 776 #define TCP_PSH 0x008
 777 #define TCP_ACK 0x010
 778 #define TCP_URG 0x020
 779 #define TCP_ECE 0x040
 780 #define TCP_CWR 0x080
 781 #define TCP_NS  0x100
 782
 783 #define TCP_CTL(flags, offset) (htons((flags) | ((offset) << 12)))
 784 #define TCP_FLAGS(tcp_ctl) (ntohs(tcp_ctl) & 0x0fff)
 785 #define TCP_FLAGS_BE16(tcp_ctl) ((tcp_ctl) & htons(0x0fff))
 786 #define TCP_OFFSET(tcp_ctl) (ntohs(tcp_ctl) >> 12)
 787
 788 #define TCP_HEADER_LEN 20
 789 struct tcp_header {
 790     ovs_be16 tcp_src;
 791     ovs_be16 tcp_dst;
 792     ovs_16aligned_be32 tcp_seq;
 793     ovs_16aligned_be32 tcp_ack;
 794     ovs_be16 tcp_ctl;
 795     ovs_be16 tcp_winsz;
 796     ovs_be16 tcp_csum;
 797     ovs_be16 tcp_urg;
 798 };
 799 BUILD_ASSERT_DECL(TCP_HEADER_LEN == sizeof(struct tcp_header));
 800
 801 /* Connection states.
 802  *
 803  * Names like CS_RELATED are bit values, e.g. 1 << 2.
 804  * Names like CS_RELATED_BIT are bit indexes, e.g. 2. */
 805 #define CS_STATES                               \
 806     CS_STATE(NEW,         0, "new")             \
 807     CS_STATE(ESTABLISHED, 1, "est")             \
 808     CS_STATE(RELATED,     2, "rel")             \
 809     CS_STATE(REPLY_DIR,   3, "rpl")             \
 810     CS_STATE(INVALID,     4, "inv")             \
 811     CS_STATE(TRACKED,     5, "trk")             \
 812     CS_STATE(SRC_NAT,     6, "snat")            \
 813     CS_STATE(DST_NAT,     7, "dnat")
 814
 815 enum {
 816 #define CS_STATE(ENUM, INDEX, NAME) \
 817     CS_##ENUM = 1 << INDEX, \
 818     CS_##ENUM##_BIT = INDEX,
 819     CS_STATES
 820 #undef CS_STATE
 821 };
 822
 823 /* Undefined connection state bits. */
 824 enum {
 825 #define CS_STATE(ENUM, INDEX, NAME) +CS_##ENUM
 826     CS_SUPPORTED_MASK = CS_STATES
 827 #undef CS_STATE
 828 };
 829 #define CS_UNSUPPORTED_MASK  (~(uint32_t)CS_SUPPORTED_MASK)
 830
 831 #define ARP_HRD_ETHERNET 1
 832 #define ARP_PRO_IP 0x0800
 833 #define ARP_OP_REQUEST 1
 834 #define ARP_OP_REPLY 2
 835 #define ARP_OP_RARP 3
 836
 837 #define ARP_ETH_HEADER_LEN 28
 838 struct arp_eth_header {
 839     /* Generic members. */
 840     ovs_be16 ar_hrd;           /* Hardware type. */
 841     ovs_be16 ar_pro;           /* Protocol type. */
 842     uint8_t ar_hln;            /* Hardware address length. */
 843     uint8_t ar_pln;            /* Protocol address length. */
 844     ovs_be16 ar_op;            /* Opcode. */
 845
 846     /* Ethernet+IPv4 specific members. */
 847     struct eth_addr ar_sha;     /* Sender hardware address. */
 848     ovs_16aligned_be32 ar_spa;  /* Sender protocol address. */
 849     struct eth_addr ar_tha;     /* Target hardware address. */
 850     ovs_16aligned_be32 ar_tpa;  /* Target protocol address. */
 851 };
 852 BUILD_ASSERT_DECL(ARP_ETH_HEADER_LEN == sizeof(struct arp_eth_header));
 853
 854 #define IPV6_HEADER_LEN 40
 855
 856 /* Like struct in6_addr, but whereas that struct requires 32-bit alignment on
 857  * most implementations, this one only requires 16-bit alignment. */
 858 union ovs_16aligned_in6_addr {
 859     ovs_be16 be16[8];
 860     ovs_16aligned_be32 be32[4];
 861 };
 862
 863 /* Like struct in6_hdr, but whereas that struct requires 32-bit alignment, this
 864  * one only requires 16-bit alignment. */
 865 struct ovs_16aligned_ip6_hdr {
 866     union {
 867         struct ovs_16aligned_ip6_hdrctl {
 868             ovs_16aligned_be32 ip6_un1_flow;
 869             ovs_be16 ip6_un1_plen;
 870             uint8_t ip6_un1_nxt;
 871             uint8_t ip6_un1_hlim;
 872         } ip6_un1;
 873         uint8_t ip6_un2_vfc;
 874     } ip6_ctlun;
 875     union ovs_16aligned_in6_addr ip6_src;
 876     union ovs_16aligned_in6_addr ip6_dst;
 877 };
 878
 879 /* Like struct in6_frag, but whereas that struct requires 32-bit alignment,
 880  * this one only requires 16-bit alignment. */
 881 struct ovs_16aligned_ip6_frag {
 882     uint8_t ip6f_nxt;
 883     uint8_t ip6f_reserved;
 884     ovs_be16 ip6f_offlg;
 885     ovs_16aligned_be32 ip6f_ident;
 886 };
 887
 888 #define ICMP6_HEADER_LEN 4
 889 struct icmp6_header {
 890     uint8_t icmp6_type;
 891     uint8_t icmp6_code;
 892     ovs_be16 icmp6_cksum;
 893 };
 894 BUILD_ASSERT_DECL(ICMP6_HEADER_LEN == sizeof(struct icmp6_header));
 895
 896 uint32_t packet_csum_pseudoheader6(const struct ovs_16aligned_ip6_hdr *);
 897 uint16_t packet_csum_upperlayer6(const struct ovs_16aligned_ip6_hdr *,
 898                                  const void *, uint8_t, uint16_t);
 899
 900 /* Neighbor Discovery option field.
 901  * ND options are always a multiple of 8 bytes in size. */
 902 #define ND_LLA_OPT_LEN 8
 903 struct ovs_nd_lla_opt {
 904     uint8_t type;               /* One of ND_OPT_*_LINKADDR. */
 905     uint8_t len;
 906     struct eth_addr mac;
 907 };
 908 BUILD_ASSERT_DECL(ND_LLA_OPT_LEN == sizeof(struct ovs_nd_lla_opt));
 909
 910 /* Neighbor Discovery option: Prefix Information. */
 911 #define ND_PREFIX_OPT_LEN 32
 912 struct ovs_nd_prefix_opt {
 913     uint8_t type;               /* ND_OPT_PREFIX_INFORMATION. */
 914     uint8_t len;                /* Always 4. */
 915     uint8_t prefix_len;
 916     uint8_t la_flags;           /* ND_PREFIX_* flags. */
 917     ovs_16aligned_be32 valid_lifetime;
 918     ovs_16aligned_be32 preferred_lifetime;
 919     ovs_16aligned_be32 reserved;          /* Always 0. */
 920     union ovs_16aligned_in6_addr prefix;
 921 };
 922 BUILD_ASSERT_DECL(ND_PREFIX_OPT_LEN == sizeof(struct ovs_nd_prefix_opt));
 923
 924 #define ND_PREFIX_ON_LINK            0x80
 925 #define ND_PREFIX_AUTONOMOUS_ADDRESS 0x40
 926
 927 /* Neighbor Discovery option: MTU. */
 928 #define ND_MTU_OPT_LEN 8
 929 struct ovs_nd_mtu_opt {
 930     uint8_t  type;      /* ND_OPT_MTU */
 931     uint8_t  len;       /* Always 1. */
 932     ovs_be16 reserved;  /* Always 0. */
 933     ovs_16aligned_be32 mtu;
 934 };
 935 BUILD_ASSERT_DECL(ND_MTU_OPT_LEN == sizeof(struct ovs_nd_mtu_opt));
 936
 937 /* Like struct nd_msg (from ndisc.h), but whereas that struct requires 32-bit
 938  * alignment, this one only requires 16-bit alignment. */
 939 #define ND_MSG_LEN 24
 940 struct ovs_nd_msg {
 941     struct icmp6_header icmph;
 942     ovs_16aligned_be32 rso_flags;
 943     union ovs_16aligned_in6_addr target;
 944     struct ovs_nd_lla_opt options[0];
 945 };
 946 BUILD_ASSERT_DECL(ND_MSG_LEN == sizeof(struct ovs_nd_msg));
 947
 948 /* Neighbor Discovery packet flags. */
 949 #define ND_RSO_ROUTER    0x80000000
 950 #define ND_RSO_SOLICITED 0x40000000
 951 #define ND_RSO_OVERRIDE  0x20000000
 952
 953 #define RA_MSG_LEN 16
 954 struct ovs_ra_msg {
 955     struct icmp6_header icmph;
 956     uint8_t cur_hop_limit;
 957     uint8_t mo_flags;  /* ND_RA_MANAGED_ADDRESS and ND_RA_OTHER_CONFIG flags. */
 958     ovs_be16 router_lifetime;
 959     ovs_be32 reachable_time;
 960     ovs_be32 retrans_timer;
 961     struct ovs_nd_lla_opt options[0];
 962 };
 963 BUILD_ASSERT_DECL(RA_MSG_LEN == sizeof(struct ovs_ra_msg));
 964
 965 #define ND_RA_MANAGED_ADDRESS 0x80
 966 #define ND_RA_OTHER_CONFIG    0x40
 967
 968 /*
 969  * Use the same struct for MLD and MLD2, naming members as the defined fields in
 970  * in the corresponding version of the protocol, though they are reserved in the
 971  * other one.
 972  */
 973 #define MLD_HEADER_LEN 8
 974 struct mld_header {
 975     uint8_t type;
 976     uint8_t code;
 977     ovs_be16 csum;
 978     ovs_be16 mrd;
 979     ovs_be16 ngrp;
 980 };
 981 BUILD_ASSERT_DECL(MLD_HEADER_LEN == sizeof(struct mld_header));
 982
 983 #define MLD2_RECORD_LEN 20
 984 struct mld2_record {
 985     uint8_t type;
 986     uint8_t aux_len;
 987     ovs_be16 nsrcs;
 988     union ovs_16aligned_in6_addr maddr;
 989 };
 990 BUILD_ASSERT_DECL(MLD2_RECORD_LEN == sizeof(struct mld2_record));
 991
 992 #define MLD_QUERY 130
 993 #define MLD_REPORT 131
 994 #define MLD_DONE 132
 995 #define MLD2_REPORT 143
 996
 997 /* The IPv6 flow label is in the lower 20 bits of the first 32-bit word. */
 998 #define IPV6_LABEL_MASK 0x000fffff
 999
1000 /* Example:
1001  *
1002  * char *string = "1 ::1 2";
1003  * char ipv6_s[IPV6_SCAN_LEN + 1];
1004  * struct in6_addr ipv6;
1005  *
1006  * if (ovs_scan(string, "%d"IPV6_SCAN_FMT"%d", &a, ipv6_s, &b)
1007  *     && inet_pton(AF_INET6, ipv6_s, &ipv6) == 1) {
1008  *     ...
1009  * }
1010  */
1011 #define IPV6_SCAN_FMT "%46[0123456789abcdefABCDEF:.]"
1012 #define IPV6_SCAN_LEN 46
1013
1014 extern const struct in6_addr in6addr_exact;
1015 #define IN6ADDR_EXACT_INIT { { { 0xff,0xff,0xff,0xff,0xff,0xff,0xff,0xff, \
1016                                  0xff,0xff,0xff,0xff,0xff,0xff,0xff,0xff } } }
1017
1018 extern const struct in6_addr in6addr_all_hosts;
1019 #define IN6ADDR_ALL_HOSTS_INIT { { { 0xff,0x02,0x00,0x00,0x00,0x00,0x00,0x00, \
1020                                      0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x01 } } }
1021
1022 extern const struct in6_addr in6addr_all_routers;
1023 #define IN6ADDR_ALL_ROUTERS_INIT { { { 0xff,0x02,0x00,0x00,0x00,0x00,0x00,0x00, \
1024                                        0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x02 } } }
1025
1026 static inline bool ipv6_addr_equals(const struct in6_addr *a,
1027                                     const struct in6_addr *b)
1028 {
1029 #ifdef IN6_ARE_ADDR_EQUAL
1030     return IN6_ARE_ADDR_EQUAL(a, b);
1031 #else
1032     return !memcmp(a, b, sizeof(*a));
1033 #endif
1034 }
1035
1036 /* Checks the IPv6 address in 'mask' for all zeroes. */
1037 static inline bool ipv6_mask_is_any(const struct in6_addr *mask) {
1038     return ipv6_addr_equals(mask, &in6addr_any);
1039 }
1040
1041 static inline bool ipv6_mask_is_exact(const struct in6_addr *mask) {
1042     return ipv6_addr_equals(mask, &in6addr_exact);
1043 }
1044
1045 static inline bool ipv6_is_all_hosts(const struct in6_addr *addr) {
1046     return ipv6_addr_equals(addr, &in6addr_all_hosts);
1047 }
1048
1049 static inline bool ipv6_addr_is_set(const struct in6_addr *addr) {
1050     return !ipv6_addr_equals(addr, &in6addr_any);
1051 }
1052
1053 static inline bool ipv6_addr_is_multicast(const struct in6_addr *ip) {
1054     return ip->s6_addr[0] == 0xff;
1055 }
1056
1057 static inline struct in6_addr
1058 in6_addr_mapped_ipv4(ovs_be32 ip4)
1059 {
1060     struct in6_addr ip6 = { .s6_addr = { [10] = 0xff, [11] = 0xff } };
1061     memcpy(&ip6.s6_addr[12], &ip4, 4);
1062     return ip6;
1063 }
1064
1065 static inline void
1066 in6_addr_set_mapped_ipv4(struct in6_addr *ip6, ovs_be32 ip4)
1067 {
1068     *ip6 = in6_addr_mapped_ipv4(ip4);
1069 }
1070
1071 static inline ovs_be32
1072 in6_addr_get_mapped_ipv4(const struct in6_addr *addr)
1073 {
1074     union ovs_16aligned_in6_addr *taddr = (void *) addr;
1075     if (IN6_IS_ADDR_V4MAPPED(addr)) {
1076         return get_16aligned_be32(&taddr->be32[3]);
1077     } else {
1078         return INADDR_ANY;
1079     }
1080 }
1081
1082 static inline void
1083 in6_addr_solicited_node(struct in6_addr *addr, const struct in6_addr *ip6)
1084 {
1085     union ovs_16aligned_in6_addr *taddr = (void *) addr;
1086     memset(taddr->be16, 0, sizeof(taddr->be16));
1087     taddr->be16[0] = htons(0xff02);
1088     taddr->be16[5] = htons(0x1);
1089     taddr->be16[6] = htons(0xff00);
1090     memcpy(&addr->s6_addr[13], &ip6->s6_addr[13], 3);
1091 }
1092
1093 /*
1094  * Generates ipv6 EUI64 address from the given eth addr
1095  * and prefix and stores it in 'lla'
1096  */
1097 static inline void
1098 in6_generate_eui64(struct eth_addr ea, struct in6_addr *prefix,
1099                    struct in6_addr *lla)
1100 {
1101     union ovs_16aligned_in6_addr *taddr = (void *) lla;
1102     union ovs_16aligned_in6_addr *prefix_taddr = (void *) prefix;
1103     taddr->be16[0] = prefix_taddr->be16[0];
1104     taddr->be16[1] = prefix_taddr->be16[1];
1105     taddr->be16[2] = prefix_taddr->be16[2];
1106     taddr->be16[3] = prefix_taddr->be16[3];
1107     taddr->be16[4] = htons(((ea.ea[0] ^ 0x02) << 8) | ea.ea[1]);
1108     taddr->be16[5] = htons(ea.ea[2] << 8 | 0x00ff);
1109     taddr->be16[6] = htons(0xfe << 8 | ea.ea[3]);
1110     taddr->be16[7] = ea.be16[2];
1111 }
1112
1113 /*
1114  * Generates ipv6 link local address from the given eth addr
1115  * with prefix 'fe80::/64' and stores it in 'lla'
1116  */
1117 static inline void
1118 in6_generate_lla(struct eth_addr ea, struct in6_addr *lla)
1119 {
1120     union ovs_16aligned_in6_addr *taddr = (void *) lla;
1121     memset(taddr->be16, 0, sizeof(taddr->be16));
1122     taddr->be16[0] = htons(0xfe80);
1123     taddr->be16[4] = htons(((ea.ea[0] ^ 0x02) << 8) | ea.ea[1]);
1124     taddr->be16[5] = htons(ea.ea[2] << 8 | 0x00ff);
1125     taddr->be16[6] = htons(0xfe << 8 | ea.ea[3]);
1126     taddr->be16[7] = ea.be16[2];
1127 }
1128
1129 /* Returns true if 'addr' is a link local address.  Otherwise, false. */
1130 static inline bool
1131 in6_is_lla(struct in6_addr *addr)
1132 {
1133 #ifdef s6_addr32
1134     return addr->s6_addr32[0] == htonl(0xfe800000) && !(addr->s6_addr32[1]);
1135 #else
1136     return addr->s6_addr[0] == 0xfe && addr->s6_addr[1] == 0x80 &&
1137          !(addr->s6_addr[2] | addr->s6_addr[3] | addr->s6_addr[4] |
1138            addr->s6_addr[5] | addr->s6_addr[6] | addr->s6_addr[7]);
1139 #endif
1140 }
1141
1142 static inline void
1143 ipv6_multicast_to_ethernet(struct eth_addr *eth, const struct in6_addr *ip6)
1144 {
1145     eth->ea[0] = 0x33;
1146     eth->ea[1] = 0x33;
1147     eth->ea[2] = ip6->s6_addr[12];
1148     eth->ea[3] = ip6->s6_addr[13];
1149     eth->ea[4] = ip6->s6_addr[14];
1150     eth->ea[5] = ip6->s6_addr[15];
1151 }
1152
1153 static inline bool dl_type_is_ip_any(ovs_be16 dl_type)
1154 {
1155     return dl_type == htons(ETH_TYPE_IP)
1156         || dl_type == htons(ETH_TYPE_IPV6);
1157 }
1158
1159 /* Tunnel header */
1160
1161 /* GRE protocol header */
1162 struct gre_base_hdr {
1163     ovs_be16 flags;
1164     ovs_be16 protocol;
1165 };
1166
1167 #define GRE_CSUM        0x8000
1168 #define GRE_ROUTING     0x4000
1169 #define GRE_KEY         0x2000
1170 #define GRE_SEQ         0x1000
1171 #define GRE_STRICT      0x0800
1172 #define GRE_REC         0x0700
1173 #define GRE_FLAGS       0x00F8
1174 #define GRE_VERSION     0x0007
1175
1176 /* VXLAN protocol header */
1177 struct vxlanhdr {
1178     ovs_16aligned_be32 vx_flags;
1179     ovs_16aligned_be32 vx_vni;
1180 };
1181
1182 #define VXLAN_FLAGS 0x08000000  /* struct vxlanhdr.vx_flags required value. */
1183
1184 /* Input values for PACKET_TYPE macros have to be in host byte order.
1185  * The _BE postfix indicates result is in network byte order. Otherwise result
1186  * is in host byte order. */
1187 #define PACKET_TYPE(NS, NS_TYPE) ((uint32_t) ((NS) << 16 | (NS_TYPE)))
1188 #define PACKET_TYPE_BE(NS, NS_TYPE) (htonl((NS) << 16 | (NS_TYPE)))
1189
1190 /* Returns the host byte ordered namespace of 'packet type'. */
1191 static inline uint16_t
1192 pt_ns(ovs_be32 packet_type)
1193 {
1194     return ntohl(packet_type) >> 16;
1195 }
1196
1197 /* Returns the network byte ordered namespace type of 'packet type'. */
1198 static inline ovs_be16
1199 pt_ns_type_be(ovs_be32 packet_type)
1200 {
1201     return be32_to_be16(packet_type);
1202 }
1203
1204 /* Returns the host byte ordered namespace type of 'packet type'. */
1205 static inline uint16_t
1206 pt_ns_type(ovs_be32 packet_type)
1207 {
1208     return ntohs(pt_ns_type_be(packet_type));
1209 }
1210
1211 /* Well-known packet_type field values. */
1212 enum packet_type {
1213     PT_ETH  = PACKET_TYPE(OFPHTN_ONF, 0x0000),  /* Default: Ethernet */
1214     PT_IPV4 = PACKET_TYPE(OFPHTN_ETHERTYPE, ETH_TYPE_IP),
1215     PT_IPV6 = PACKET_TYPE(OFPHTN_ETHERTYPE, ETH_TYPE_IPV6),
1216     PT_MPLS = PACKET_TYPE(OFPHTN_ETHERTYPE, ETH_TYPE_MPLS),
1217     PT_MPLS_MC = PACKET_TYPE(OFPHTN_ETHERTYPE, ETH_TYPE_MPLS_MCAST),
1218     PT_UNKNOWN = PACKET_TYPE(0xffff, 0xffff),   /* Unknown packet type. */
1219 };
1220
1221
1222 void ipv6_format_addr(const struct in6_addr *addr, struct ds *);
1223 void ipv6_format_addr_bracket(const struct in6_addr *addr, struct ds *,
1224                               bool bracket);
1225 void ipv6_format_mapped(const struct in6_addr *addr, struct ds *);
1226 void ipv6_format_masked(const struct in6_addr *addr,
1227                         const struct in6_addr *mask, struct ds *);
1228 const char * ipv6_string_mapped(char *addr_str, const struct in6_addr *addr);
1229 struct in6_addr ipv6_addr_bitand(const struct in6_addr *src,
1230                                  const struct in6_addr *mask);
1231 struct in6_addr ipv6_addr_bitxor(const struct in6_addr *a,
1232                                  const struct in6_addr *b);
1233 bool ipv6_is_zero(const struct in6_addr *a);
1234 struct in6_addr ipv6_create_mask(int mask);
1235 int ipv6_count_cidr_bits(const struct in6_addr *netmask);
1236 bool ipv6_is_cidr(const struct in6_addr *netmask);
1237
1238 bool ipv6_parse(const char *s, struct in6_addr *ip);
1239 char *ipv6_parse_masked(const char *s, struct in6_addr *ipv6,
1240                         struct in6_addr *mask);
1241 char *ipv6_parse_cidr(const char *s, struct in6_addr *ip, unsigned int *plen)
1242     OVS_WARN_UNUSED_RESULT;
1243 char *ipv6_parse_masked_len(const char *s, int *n, struct in6_addr *ipv6,
1244                             struct in6_addr *mask);
1245 char *ipv6_parse_cidr_len(const char *s, int *n, struct in6_addr *ip,
1246                           unsigned int *plen)
1247     OVS_WARN_UNUSED_RESULT;
1248
1249 void *eth_compose(struct dp_packet *, const struct eth_addr eth_dst,
1250                   const struct eth_addr eth_src, uint16_t eth_type,
1251                   size_t size);
1252 void *snap_compose(struct dp_packet *, const struct eth_addr eth_dst,
1253                    const struct eth_addr eth_src,
1254                    unsigned int oui, uint16_t snap_type, size_t size);
1255 void packet_set_ipv4(struct dp_packet *, ovs_be32 src, ovs_be32 dst, uint8_t tos,
1256                      uint8_t ttl);
1257 void packet_set_ipv4_addr(struct dp_packet *packet, ovs_16aligned_be32 *addr,
1258                           ovs_be32 new_addr);
1259 void packet_set_ipv6(struct dp_packet *, const struct in6_addr *src,
1260                      const struct in6_addr *dst, uint8_t tc,
1261                      ovs_be32 fl, uint8_t hlmit);
1262 void packet_set_ipv6_addr(struct dp_packet *packet, uint8_t proto,
1263                           ovs_16aligned_be32 addr[4],
1264                           const struct in6_addr *new_addr,
1265                           bool recalculate_csum);
1266 void packet_set_tcp_port(struct dp_packet *, ovs_be16 src, ovs_be16 dst);
1267 void packet_set_udp_port(struct dp_packet *, ovs_be16 src, ovs_be16 dst);
1268 void packet_set_sctp_port(struct dp_packet *, ovs_be16 src, ovs_be16 dst);
1269 void packet_set_icmp(struct dp_packet *, uint8_t type, uint8_t code);
1270 void packet_set_nd(struct dp_packet *, const struct in6_addr *target,
1271                    const struct eth_addr sll, const struct eth_addr tll);
1272
1273 void packet_format_tcp_flags(struct ds *, uint16_t);
1274 const char *packet_tcp_flag_to_string(uint32_t flag);
1275 void compose_arp__(struct dp_packet *);
1276 void compose_arp(struct dp_packet *, uint16_t arp_op,
1277                  const struct eth_addr arp_sha,
1278                  const struct eth_addr arp_tha, bool broadcast,
1279                  ovs_be32 arp_spa, ovs_be32 arp_tpa);
1280 void compose_nd_ns(struct dp_packet *, const struct eth_addr eth_src,
1281                    const struct in6_addr *ipv6_src,
1282                    const struct in6_addr *ipv6_dst);
1283 void compose_nd_na(struct dp_packet *, const struct eth_addr eth_src,
1284                    const struct eth_addr eth_dst,
1285                    const struct in6_addr *ipv6_src,
1286                    const struct in6_addr *ipv6_dst,
1287                    ovs_be32 rso_flags);
1288 void compose_nd_ra(struct dp_packet *,
1289                    const struct eth_addr eth_src,
1290                    const struct eth_addr eth_dst,
1291                    const struct in6_addr *ipv6_src,
1292                    const struct in6_addr *ipv6_dst,
1293                    uint8_t cur_hop_limit, uint8_t mo_flags,
1294                    ovs_be16 router_lt, ovs_be32 reachable_time,
1295                    ovs_be32 retrans_timer, ovs_be32 mtu);
1296 void packet_put_ra_prefix_opt(struct dp_packet *,
1297                               uint8_t plen, uint8_t la_flags,
1298                               ovs_be32 valid_lifetime,
1299                               ovs_be32 preferred_lifetime,
1300                               const ovs_be128 router_prefix);
1301 uint32_t packet_csum_pseudoheader(const struct ip_header *);
1302 void IP_ECN_set_ce(struct dp_packet *pkt, bool is_ipv6);
1303
1304 #define DNS_HEADER_LEN 12
1305 struct dns_header {
1306     ovs_be16 id;
1307     uint8_t lo_flag; /* QR (1), OPCODE (4), AA (1), TC (1) and RD (1) */
1308     uint8_t hi_flag; /* RA (1), Z (3) and RCODE (4) */
1309     ovs_be16 qdcount; /* Num of entries in the question section. */
1310     ovs_be16 ancount; /* Num of resource records in the answer section. */
1311
1312     /* Num of name server records in the authority record section. */
1313     ovs_be16 nscount;
1314
1315     /* Num of resource records in the additional records section. */
1316     ovs_be16 arcount;
1317 };
1318
1319 BUILD_ASSERT_DECL(DNS_HEADER_LEN == sizeof(struct dns_header));
1320
1321 #define DNS_QUERY_TYPE_A        0x01
1322 #define DNS_QUERY_TYPE_AAAA     0x1c
1323 #define DNS_QUERY_TYPE_ANY      0xff
1324
1325 #define DNS_CLASS_IN            0x01
1326 #define DNS_DEFAULT_RR_TTL      3600
1327
1328 #endif /* packets.h */