1 /* 2 * CDDL HEADER START 3 * 4 * The contents of this file are subject to the terms of the 5 * Common Development and Distribution License (the "License"). 6 * You may not use this file except in compliance with the License. 7 * 8 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE 9 * or http://www.opensolaris.org/os/licensing. 10 * See the License for the specific language governing permissions 11 * and limitations under the License. 12 * 13 * When distributing Covered Code, include this CDDL HEADER in each 14 * file and include the License file at usr/src/OPENSOLARIS.LICENSE. 15 * If applicable, add the following below this CDDL HEADER, with the 16 * fields enclosed by brackets "[]" replaced with your own identifying 17 * information: Portions Copyright [yyyy] [name of copyright owner] 18 * 19 * CDDL HEADER END 20 */ 21 22 /* 23 * Copyright (c) 1991, 2010, Oracle and/or its affiliates. All rights reserved. 24 * Copyright (c) 1990 Mentat Inc. 25 */ 26 27 #ifndef _INET_IP_H 28 #define _INET_IP_H 29 30 #ifdef __cplusplus 31 extern "C" { 32 #endif 33 34 #include <sys/isa_defs.h> 35 #include <sys/types.h> 36 #include <inet/mib2.h> 37 #include <inet/nd.h> 38 #include <sys/atomic.h> 39 #include <net/if_dl.h> 40 #include <net/if.h> 41 #include <netinet/ip.h> 42 #include <netinet/igmp.h> 43 #include <sys/neti.h> 44 #include <sys/hook.h> 45 #include <sys/hook_event.h> 46 #include <sys/hook_impl.h> 47 #include <inet/ip_stack.h> 48 49 #ifdef _KERNEL 50 #include <netinet/ip6.h> 51 #include <sys/avl.h> 52 #include <sys/list.h> 53 #include <sys/vmem.h> 54 #include <sys/squeue.h> 55 #include <net/route.h> 56 #include <sys/systm.h> 57 #include <net/radix.h> 58 #include <sys/modhash.h> 59 60 #ifdef DEBUG 61 #define CONN_DEBUG 62 #endif 63 64 #define IP_DEBUG 65 /* 66 * The mt-streams(9F) flags for the IP module; put here so that other 67 * "drivers" that are actually IP (e.g., ICMP, UDP) can use the same set 68 * of flags. 69 */ 70 #define IP_DEVMTFLAGS D_MP 71 #endif /* _KERNEL */ 72 73 #define IP_MOD_NAME "ip" 74 #define IP_DEV_NAME "/dev/ip" 75 #define IP6_DEV_NAME "/dev/ip6" 76 77 #define UDP_MOD_NAME "udp" 78 #define UDP_DEV_NAME "/dev/udp" 79 #define UDP6_DEV_NAME "/dev/udp6" 80 81 #define TCP_MOD_NAME "tcp" 82 #define TCP_DEV_NAME "/dev/tcp" 83 #define TCP6_DEV_NAME "/dev/tcp6" 84 85 #define SCTP_MOD_NAME "sctp" 86 87 #ifndef _IPADDR_T 88 #define _IPADDR_T 89 typedef uint32_t ipaddr_t; 90 #endif 91 92 /* Number of bits in an address */ 93 #define IP_ABITS 32 94 #define IPV4_ABITS IP_ABITS 95 #define IPV6_ABITS 128 96 97 #define IP_HOST_MASK (ipaddr_t)0xffffffffU 98 99 #define IP_CSUM(mp, off, sum) (~ip_cksum(mp, off, sum) & 0xFFFF) 100 #define IP_CSUM_PARTIAL(mp, off, sum) ip_cksum(mp, off, sum) 101 #define IP_BCSUM_PARTIAL(bp, len, sum) bcksum(bp, len, sum) 102 103 #define ILL_FRAG_HASH_TBL_COUNT ((unsigned int)64) 104 #define ILL_FRAG_HASH_TBL_SIZE (ILL_FRAG_HASH_TBL_COUNT * sizeof (ipfb_t)) 105 106 #define IPV4_ADDR_LEN 4 107 #define IP_ADDR_LEN IPV4_ADDR_LEN 108 #define IP_ARP_PROTO_TYPE 0x0800 109 110 #define IPV4_VERSION 4 111 #define IP_VERSION IPV4_VERSION 112 #define IP_SIMPLE_HDR_LENGTH_IN_WORDS 5 113 #define IP_SIMPLE_HDR_LENGTH 20 114 #define IP_MAX_HDR_LENGTH 60 115 116 #define IP_MAX_OPT_LENGTH (IP_MAX_HDR_LENGTH-IP_SIMPLE_HDR_LENGTH) 117 118 #define IP_MIN_MTU (IP_MAX_HDR_LENGTH + 8) /* 68 bytes */ 119 120 /* 121 * XXX IP_MAXPACKET is defined in <netinet/ip.h> as well. At some point the 122 * 2 files should be cleaned up to remove all redundant definitions. 123 */ 124 #define IP_MAXPACKET 65535 125 #define IP_SIMPLE_HDR_VERSION \ 126 ((IP_VERSION << 4) | IP_SIMPLE_HDR_LENGTH_IN_WORDS) 127 128 #define UDPH_SIZE 8 129 130 /* 131 * Constants and type definitions to support IP IOCTL commands 132 */ 133 #define IP_IOCTL (('i'<<8)|'p') 134 #define IP_IOC_IRE_DELETE 4 135 #define IP_IOC_IRE_DELETE_NO_REPLY 5 136 #define IP_IOC_RTS_REQUEST 7 137 138 /* Common definitions used by IP IOCTL data structures */ 139 typedef struct ipllcmd_s { 140 uint_t ipllc_cmd; 141 uint_t ipllc_name_offset; 142 uint_t ipllc_name_length; 143 } ipllc_t; 144 145 /* IP IRE Delete Command Structure. */ 146 typedef struct ipid_s { 147 ipllc_t ipid_ipllc; 148 uint_t ipid_ire_type; 149 uint_t ipid_addr_offset; 150 uint_t ipid_addr_length; 151 uint_t ipid_mask_offset; 152 uint_t ipid_mask_length; 153 } ipid_t; 154 155 #define ipid_cmd ipid_ipllc.ipllc_cmd 156 157 #ifdef _KERNEL 158 /* 159 * Temporary state for ip options parser. 160 */ 161 typedef struct ipoptp_s 162 { 163 uint8_t *ipoptp_next; /* next option to look at */ 164 uint8_t *ipoptp_end; /* end of options */ 165 uint8_t *ipoptp_cur; /* start of current option */ 166 uint8_t ipoptp_len; /* length of current option */ 167 uint32_t ipoptp_flags; 168 } ipoptp_t; 169 170 /* 171 * Flag(s) for ipoptp_flags 172 */ 173 #define IPOPTP_ERROR 0x00000001 174 #endif /* _KERNEL */ 175 176 /* Controls forwarding of IP packets, set via ipadm(1M)/ndd(1M) */ 177 #define IP_FORWARD_NEVER 0 178 #define IP_FORWARD_ALWAYS 1 179 180 #define WE_ARE_FORWARDING(ipst) ((ipst)->ips_ip_forwarding == IP_FORWARD_ALWAYS) 181 182 #define IPH_HDR_LENGTH(ipha) \ 183 ((int)(((ipha_t *)ipha)->ipha_version_and_hdr_length & 0xF) << 2) 184 185 #define IPH_HDR_VERSION(ipha) \ 186 ((int)(((ipha_t *)ipha)->ipha_version_and_hdr_length) >> 4) 187 188 #ifdef _KERNEL 189 /* 190 * IP reassembly macros. We hide starting and ending offsets in b_next and 191 * b_prev of messages on the reassembly queue. The messages are chained using 192 * b_cont. These macros are used in ip_reassemble() so we don't have to see 193 * the ugly casts and assignments. 194 * Note that the offsets are <= 64k i.e. a uint_t is sufficient to represent 195 * them. 196 */ 197 #define IP_REASS_START(mp) ((uint_t)(uintptr_t)((mp)->b_next)) 198 #define IP_REASS_SET_START(mp, u) \ 199 ((mp)->b_next = (mblk_t *)(uintptr_t)(u)) 200 #define IP_REASS_END(mp) ((uint_t)(uintptr_t)((mp)->b_prev)) 201 #define IP_REASS_SET_END(mp, u) \ 202 ((mp)->b_prev = (mblk_t *)(uintptr_t)(u)) 203 204 #define IP_REASS_COMPLETE 0x1 205 #define IP_REASS_PARTIAL 0x2 206 #define IP_REASS_FAILED 0x4 207 208 /* 209 * Test to determine whether this is a module instance of IP or a 210 * driver instance of IP. 211 */ 212 #define CONN_Q(q) (WR(q)->q_next == NULL) 213 214 #define Q_TO_CONN(q) ((conn_t *)(q)->q_ptr) 215 #define Q_TO_TCP(q) (Q_TO_CONN((q))->conn_tcp) 216 #define Q_TO_UDP(q) (Q_TO_CONN((q))->conn_udp) 217 #define Q_TO_ICMP(q) (Q_TO_CONN((q))->conn_icmp) 218 #define Q_TO_RTS(q) (Q_TO_CONN((q))->conn_rts) 219 220 #define CONNP_TO_WQ(connp) ((connp)->conn_wq) 221 #define CONNP_TO_RQ(connp) ((connp)->conn_rq) 222 223 #define GRAB_CONN_LOCK(q) { \ 224 if (q != NULL && CONN_Q(q)) \ 225 mutex_enter(&(Q_TO_CONN(q))->conn_lock); \ 226 } 227 228 #define RELEASE_CONN_LOCK(q) { \ 229 if (q != NULL && CONN_Q(q)) \ 230 mutex_exit(&(Q_TO_CONN(q))->conn_lock); \ 231 } 232 233 /* 234 * Ref counter macros for ioctls. This provides a guard for TCP to stop 235 * tcp_close from removing the rq/wq whilst an ioctl is still in flight on the 236 * stream. The ioctl could have been queued on e.g. an ipsq. tcp_close will wait 237 * until the ioctlref count is zero before proceeding. 238 * Ideally conn_oper_pending_ill would be used for this purpose. However, in the 239 * case where an ioctl is aborted or interrupted, it can be cleared prematurely. 240 * There are also some race possibilities between ip and the stream head which 241 * can also end up with conn_oper_pending_ill being cleared prematurely. So, to 242 * avoid these situations, we use a dedicated ref counter for ioctls which is 243 * used in addition to and in parallel with the normal conn_ref count. 244 */ 245 #define CONN_INC_IOCTLREF_LOCKED(connp) { \ 246 ASSERT(MUTEX_HELD(&(connp)->conn_lock)); \ 247 DTRACE_PROBE1(conn__inc__ioctlref, conn_t *, (connp)); \ 248 (connp)->conn_ioctlref++; \ 249 mutex_exit(&(connp)->conn_lock); \ 250 } 251 252 #define CONN_INC_IOCTLREF(connp) { \ 253 mutex_enter(&(connp)->conn_lock); \ 254 CONN_INC_IOCTLREF_LOCKED(connp); \ 255 } 256 257 #define CONN_DEC_IOCTLREF(connp) { \ 258 mutex_enter(&(connp)->conn_lock); \ 259 DTRACE_PROBE1(conn__dec__ioctlref, conn_t *, (connp)); \ 260 /* Make sure conn_ioctlref will not underflow. */ \ 261 ASSERT((connp)->conn_ioctlref != 0); \ 262 if ((--(connp)->conn_ioctlref == 0) && \ 263 ((connp)->conn_state_flags & CONN_CLOSING)) { \ 264 cv_broadcast(&(connp)->conn_cv); \ 265 } \ 266 mutex_exit(&(connp)->conn_lock); \ 267 } 268 269 270 /* 271 * Complete the pending operation. Usually an ioctl. Can also 272 * be a bind or option management request that got enqueued 273 * in an ipsq_t. Called on completion of the operation. 274 */ 275 #define CONN_OPER_PENDING_DONE(connp) { \ 276 mutex_enter(&(connp)->conn_lock); \ 277 (connp)->conn_oper_pending_ill = NULL; \ 278 cv_broadcast(&(connp)->conn_refcv); \ 279 mutex_exit(&(connp)->conn_lock); \ 280 CONN_DEC_REF(connp); \ 281 } 282 283 /* 284 * Values for squeue switch: 285 */ 286 #define IP_SQUEUE_ENTER_NODRAIN 1 287 #define IP_SQUEUE_ENTER 2 288 #define IP_SQUEUE_FILL 3 289 290 extern int ip_squeue_flag; 291 292 /* IP Fragmentation Reassembly Header */ 293 typedef struct ipf_s { 294 struct ipf_s *ipf_hash_next; 295 struct ipf_s **ipf_ptphn; /* Pointer to previous hash next. */ 296 uint32_t ipf_ident; /* Ident to match. */ 297 uint8_t ipf_protocol; /* Protocol to match. */ 298 uchar_t ipf_last_frag_seen : 1; /* Last fragment seen ? */ 299 time_t ipf_timestamp; /* Reassembly start time. */ 300 mblk_t *ipf_mp; /* mblk we live in. */ 301 mblk_t *ipf_tail_mp; /* Frag queue tail pointer. */ 302 int ipf_hole_cnt; /* Number of holes (hard-case). */ 303 int ipf_end; /* Tail end offset (0 -> hard-case). */ 304 uint_t ipf_gen; /* Frag queue generation */ 305 size_t ipf_count; /* Count of bytes used by frag */ 306 uint_t ipf_nf_hdr_len; /* Length of nonfragmented header */ 307 in6_addr_t ipf_v6src; /* IPv6 source address */ 308 in6_addr_t ipf_v6dst; /* IPv6 dest address */ 309 uint_t ipf_prev_nexthdr_offset; /* Offset for nexthdr value */ 310 uint8_t ipf_ecn; /* ECN info for the fragments */ 311 uint8_t ipf_num_dups; /* Number of times dup frags recvd */ 312 uint16_t ipf_checksum_flags; /* Hardware checksum flags */ 313 uint32_t ipf_checksum; /* Partial checksum of fragment data */ 314 } ipf_t; 315 316 /* 317 * IPv4 Fragments 318 */ 319 #define IS_V4_FRAGMENT(ipha_fragment_offset_and_flags) \ 320 (((ntohs(ipha_fragment_offset_and_flags) & IPH_OFFSET) != 0) || \ 321 ((ntohs(ipha_fragment_offset_and_flags) & IPH_MF) != 0)) 322 323 #define ipf_src V4_PART_OF_V6(ipf_v6src) 324 #define ipf_dst V4_PART_OF_V6(ipf_v6dst) 325 326 #endif /* _KERNEL */ 327 328 /* ICMP types */ 329 #define ICMP_ECHO_REPLY 0 330 #define ICMP_DEST_UNREACHABLE 3 331 #define ICMP_SOURCE_QUENCH 4 332 #define ICMP_REDIRECT 5 333 #define ICMP_ECHO_REQUEST 8 334 #define ICMP_ROUTER_ADVERTISEMENT 9 335 #define ICMP_ROUTER_SOLICITATION 10 336 #define ICMP_TIME_EXCEEDED 11 337 #define ICMP_PARAM_PROBLEM 12 338 #define ICMP_TIME_STAMP_REQUEST 13 339 #define ICMP_TIME_STAMP_REPLY 14 340 #define ICMP_INFO_REQUEST 15 341 #define ICMP_INFO_REPLY 16 342 #define ICMP_ADDRESS_MASK_REQUEST 17 343 #define ICMP_ADDRESS_MASK_REPLY 18 344 345 /* Evaluates to true if the ICMP type is an ICMP error */ 346 #define ICMP_IS_ERROR(type) ( \ 347 (type) == ICMP_DEST_UNREACHABLE || \ 348 (type) == ICMP_SOURCE_QUENCH || \ 349 (type) == ICMP_TIME_EXCEEDED || \ 350 (type) == ICMP_PARAM_PROBLEM) 351 352 /* ICMP_TIME_EXCEEDED codes */ 353 #define ICMP_TTL_EXCEEDED 0 354 #define ICMP_REASSEMBLY_TIME_EXCEEDED 1 355 356 /* ICMP_DEST_UNREACHABLE codes */ 357 #define ICMP_NET_UNREACHABLE 0 358 #define ICMP_HOST_UNREACHABLE 1 359 #define ICMP_PROTOCOL_UNREACHABLE 2 360 #define ICMP_PORT_UNREACHABLE 3 361 #define ICMP_FRAGMENTATION_NEEDED 4 362 #define ICMP_SOURCE_ROUTE_FAILED 5 363 #define ICMP_DEST_NET_UNKNOWN 6 364 #define ICMP_DEST_HOST_UNKNOWN 7 365 #define ICMP_SRC_HOST_ISOLATED 8 366 #define ICMP_DEST_NET_UNREACH_ADMIN 9 367 #define ICMP_DEST_HOST_UNREACH_ADMIN 10 368 #define ICMP_DEST_NET_UNREACH_TOS 11 369 #define ICMP_DEST_HOST_UNREACH_TOS 12 370 371 /* ICMP Header Structure */ 372 typedef struct icmph_s { 373 uint8_t icmph_type; 374 uint8_t icmph_code; 375 uint16_t icmph_checksum; 376 union { 377 struct { /* ECHO request/response structure */ 378 uint16_t u_echo_ident; 379 uint16_t u_echo_seqnum; 380 } u_echo; 381 struct { /* Destination unreachable structure */ 382 uint16_t u_du_zero; 383 uint16_t u_du_mtu; 384 } u_du; 385 struct { /* Parameter problem structure */ 386 uint8_t u_pp_ptr; 387 uint8_t u_pp_rsvd[3]; 388 } u_pp; 389 struct { /* Redirect structure */ 390 ipaddr_t u_rd_gateway; 391 } u_rd; 392 } icmph_u; 393 } icmph_t; 394 395 #define icmph_echo_ident icmph_u.u_echo.u_echo_ident 396 #define icmph_echo_seqnum icmph_u.u_echo.u_echo_seqnum 397 #define icmph_du_zero icmph_u.u_du.u_du_zero 398 #define icmph_du_mtu icmph_u.u_du.u_du_mtu 399 #define icmph_pp_ptr icmph_u.u_pp.u_pp_ptr 400 #define icmph_rd_gateway icmph_u.u_rd.u_rd_gateway 401 402 #define ICMPH_SIZE 8 403 404 /* 405 * Minimum length of transport layer header included in an ICMP error 406 * message for it to be considered valid. 407 */ 408 #define ICMP_MIN_TP_HDR_LEN 8 409 410 /* Aligned IP header */ 411 typedef struct ipha_s { 412 uint8_t ipha_version_and_hdr_length; 413 uint8_t ipha_type_of_service; 414 uint16_t ipha_length; 415 uint16_t ipha_ident; 416 uint16_t ipha_fragment_offset_and_flags; 417 uint8_t ipha_ttl; 418 uint8_t ipha_protocol; 419 uint16_t ipha_hdr_checksum; 420 ipaddr_t ipha_src; 421 ipaddr_t ipha_dst; 422 } ipha_t; 423 424 /* 425 * IP Flags 426 * 427 * Some of these constant names are copied for the DTrace IP provider in 428 * usr/src/lib/libdtrace/common/{ip.d.in, ip.sed.in}, which should be kept 429 * in sync. 430 */ 431 #define IPH_DF 0x4000 /* Don't fragment */ 432 #define IPH_MF 0x2000 /* More fragments to come */ 433 #define IPH_OFFSET 0x1FFF /* Where the offset lives */ 434 435 /* Byte-order specific values */ 436 #ifdef _BIG_ENDIAN 437 #define IPH_DF_HTONS 0x4000 /* Don't fragment */ 438 #define IPH_MF_HTONS 0x2000 /* More fragments to come */ 439 #define IPH_OFFSET_HTONS 0x1FFF /* Where the offset lives */ 440 #else 441 #define IPH_DF_HTONS 0x0040 /* Don't fragment */ 442 #define IPH_MF_HTONS 0x0020 /* More fragments to come */ 443 #define IPH_OFFSET_HTONS 0xFF1F /* Where the offset lives */ 444 #endif 445 446 /* ECN code points for IPv4 TOS byte and IPv6 traffic class octet. */ 447 #define IPH_ECN_NECT 0x0 /* Not ECN-Capable Transport */ 448 #define IPH_ECN_ECT1 0x1 /* ECN-Capable Transport, ECT(1) */ 449 #define IPH_ECN_ECT0 0x2 /* ECN-Capable Transport, ECT(0) */ 450 #define IPH_ECN_CE 0x3 /* ECN-Congestion Experienced (CE) */ 451 452 struct ill_s; 453 454 typedef void ip_v6intfid_func_t(struct ill_s *, in6_addr_t *); 455 typedef void ip_v6mapinfo_func_t(struct ill_s *, uchar_t *, uchar_t *); 456 typedef void ip_v4mapinfo_func_t(struct ill_s *, uchar_t *, uchar_t *); 457 458 /* IP Mac info structure */ 459 typedef struct ip_m_s { 460 t_uscalar_t ip_m_mac_type; /* From <sys/dlpi.h> */ 461 int ip_m_type; /* From <net/if_types.h> */ 462 t_uscalar_t ip_m_ipv4sap; 463 t_uscalar_t ip_m_ipv6sap; 464 ip_v4mapinfo_func_t *ip_m_v4mapping; 465 ip_v6mapinfo_func_t *ip_m_v6mapping; 466 ip_v6intfid_func_t *ip_m_v6intfid; 467 ip_v6intfid_func_t *ip_m_v6destintfid; 468 } ip_m_t; 469 470 /* 471 * The following functions attempt to reduce the link layer dependency 472 * of the IP stack. The current set of link specific operations are: 473 * a. map from IPv4 class D (224.0/4) multicast address range or the 474 * IPv6 multicast address range (ff00::/8) to the link layer multicast 475 * address. 476 * b. derive the default IPv6 interface identifier from the interface. 477 * c. derive the default IPv6 destination interface identifier from 478 * the interface (point-to-point only). 479 */ 480 extern void ip_mcast_mapping(struct ill_s *, uchar_t *, uchar_t *); 481 /* ip_m_v6*intfid return void and are never NULL */ 482 #define MEDIA_V6INTFID(ip_m, ill, v6ptr) (ip_m)->ip_m_v6intfid(ill, v6ptr) 483 #define MEDIA_V6DESTINTFID(ip_m, ill, v6ptr) \ 484 (ip_m)->ip_m_v6destintfid(ill, v6ptr) 485 486 /* Router entry types */ 487 #define IRE_BROADCAST 0x0001 /* Route entry for broadcast address */ 488 #define IRE_DEFAULT 0x0002 /* Route entry for default gateway */ 489 #define IRE_LOCAL 0x0004 /* Route entry for local address */ 490 #define IRE_LOOPBACK 0x0008 /* Route entry for loopback address */ 491 #define IRE_PREFIX 0x0010 /* Route entry for prefix routes */ 492 #ifndef _KERNEL 493 /* Keep so user-level still compiles */ 494 #define IRE_CACHE 0x0020 /* Cached Route entry */ 495 #endif 496 #define IRE_IF_NORESOLVER 0x0040 /* Route entry for local interface */ 497 /* net without any address mapping. */ 498 #define IRE_IF_RESOLVER 0x0080 /* Route entry for local interface */ 499 /* net with resolver. */ 500 #define IRE_HOST 0x0100 /* Host route entry */ 501 /* Keep so user-level still compiles */ 502 #define IRE_HOST_REDIRECT 0x0200 /* only used for T_SVR4_OPTMGMT_REQ */ 503 #define IRE_IF_CLONE 0x0400 /* Per host clone of IRE_IF */ 504 #define IRE_MULTICAST 0x0800 /* Special - not in table */ 505 #define IRE_NOROUTE 0x1000 /* Special - not in table */ 506 507 #define IRE_INTERFACE (IRE_IF_NORESOLVER | IRE_IF_RESOLVER) 508 509 #define IRE_IF_ALL (IRE_IF_NORESOLVER | IRE_IF_RESOLVER | \ 510 IRE_IF_CLONE) 511 #define IRE_OFFSUBNET (IRE_DEFAULT | IRE_PREFIX | IRE_HOST) 512 #define IRE_OFFLINK IRE_OFFSUBNET 513 /* 514 * Note that we view IRE_NOROUTE as ONLINK since we can "send" to them without 515 * going through a router; the result of sending will be an error/icmp error. 516 */ 517 #define IRE_ONLINK (IRE_IF_ALL|IRE_LOCAL|IRE_LOOPBACK| \ 518 IRE_BROADCAST|IRE_MULTICAST|IRE_NOROUTE) 519 520 /* Arguments to ire_flush_cache() */ 521 #define IRE_FLUSH_DELETE 0 522 #define IRE_FLUSH_ADD 1 523 #define IRE_FLUSH_GWCHANGE 2 524 525 /* 526 * Flags to ire_route_recursive 527 */ 528 #define IRR_NONE 0 529 #define IRR_ALLOCATE 1 /* OK to allocate IRE_IF_CLONE */ 530 #define IRR_INCOMPLETE 2 /* OK to return incomplete chain */ 531 532 /* 533 * Open/close synchronization flags. 534 * These are kept in a separate field in the conn and the synchronization 535 * depends on the atomic 32 bit access to that field. 536 */ 537 #define CONN_CLOSING 0x01 /* ip_close waiting for ip_wsrv */ 538 #define CONN_CONDEMNED 0x02 /* conn is closing, no more refs */ 539 #define CONN_INCIPIENT 0x04 /* conn not yet visible, no refs */ 540 #define CONN_QUIESCED 0x08 /* conn is now quiescent */ 541 #define CONN_UPDATE_ILL 0x10 /* conn_update_ill in progress */ 542 543 /* 544 * Flags for dce_flags field. Specifies which information has been set. 545 * dce_ident is always present, but the other ones are identified by the flags. 546 */ 547 #define DCEF_DEFAULT 0x0001 /* Default DCE - no pmtu or uinfo */ 548 #define DCEF_PMTU 0x0002 /* Different than interface MTU */ 549 #define DCEF_UINFO 0x0004 /* dce_uinfo set */ 550 #define DCEF_TOO_SMALL_PMTU 0x0008 /* Smaller than IPv4/IPv6 MIN */ 551 552 #ifdef _KERNEL 553 /* 554 * Extra structures need for per-src-addr filtering (IGMPv3/MLDv2) 555 */ 556 #define MAX_FILTER_SIZE 64 557 558 typedef struct slist_s { 559 int sl_numsrc; 560 in6_addr_t sl_addr[MAX_FILTER_SIZE]; 561 } slist_t; 562 563 /* 564 * Following struct is used to maintain retransmission state for 565 * a multicast group. One rtx_state_t struct is an in-line field 566 * of the ilm_t struct; the slist_ts in the rtx_state_t struct are 567 * alloc'd as needed. 568 */ 569 typedef struct rtx_state_s { 570 uint_t rtx_timer; /* retrans timer */ 571 int rtx_cnt; /* retrans count */ 572 int rtx_fmode_cnt; /* retrans count for fmode change */ 573 slist_t *rtx_allow; 574 slist_t *rtx_block; 575 } rtx_state_t; 576 577 /* 578 * Used to construct list of multicast address records that will be 579 * sent in a single listener report. 580 */ 581 typedef struct mrec_s { 582 struct mrec_s *mrec_next; 583 uint8_t mrec_type; 584 uint8_t mrec_auxlen; /* currently unused */ 585 in6_addr_t mrec_group; 586 slist_t mrec_srcs; 587 } mrec_t; 588 589 /* Group membership list per upper conn */ 590 591 /* 592 * We record the multicast information from the socket option in 593 * ilg_ifaddr/ilg_ifindex. This allows rejoining the group in the case when 594 * the ifaddr (or ifindex) disappears and later reappears, potentially on 595 * a different ill. The IPv6 multicast socket options and ioctls all specify 596 * the interface using an ifindex. For IPv4 some socket options/ioctls use 597 * the interface address and others use the index. We record here the method 598 * that was actually used (and leave the other of ilg_ifaddr or ilg_ifindex) 599 * at zero so that we can rejoin the way the application intended. 600 * 601 * We track the ill on which we will or already have joined an ilm using 602 * ilg_ill. When we have succeeded joining the ilm and have a refhold on it 603 * then we set ilg_ilm. Thus intentionally there is a window where ilg_ill is 604 * set and ilg_ilm is not set. This allows clearing ilg_ill as a signal that 605 * the ill is being unplumbed and the ilm should be discarded. 606 * 607 * ilg records the state of multicast memberships of a socket end point. 608 * ilm records the state of multicast memberships with the driver and is 609 * maintained per interface. 610 * 611 * The ilg state is protected by conn_ilg_lock. 612 * The ilg will not be freed until ilg_refcnt drops to zero. 613 */ 614 typedef struct ilg_s { 615 struct ilg_s *ilg_next; 616 struct ilg_s **ilg_ptpn; 617 struct conn_s *ilg_connp; /* Back pointer to get lock */ 618 in6_addr_t ilg_v6group; 619 ipaddr_t ilg_ifaddr; /* For some IPv4 cases */ 620 uint_t ilg_ifindex; /* IPv6 and some other IPv4 cases */ 621 struct ill_s *ilg_ill; /* Where ilm is joined. No refhold */ 622 struct ilm_s *ilg_ilm; /* With ilm_refhold */ 623 uint_t ilg_refcnt; 624 mcast_record_t ilg_fmode; /* MODE_IS_INCLUDE/MODE_IS_EXCLUDE */ 625 slist_t *ilg_filter; 626 boolean_t ilg_condemned; /* Conceptually deleted */ 627 } ilg_t; 628 629 /* 630 * Multicast address list entry for ill. 631 * ilm_ill is used by IPv4 and IPv6 632 * 633 * The ilm state (and other multicast state on the ill) is protected by 634 * ill_mcast_lock. Operations that change state on both an ilg and ilm 635 * in addition use ill_mcast_serializer to ensure that we can't have 636 * interleaving between e.g., add and delete operations for the same conn_t, 637 * group, and ill. The ill_mcast_serializer is also used to ensure that 638 * multicast group joins do not occur on an interface that is in the process 639 * of joining an IPMP group. 640 * 641 * The comment below (and for other netstack_t references) refers 642 * to the fact that we only do netstack_hold in particular cases, 643 * such as the references from open endpoints (ill_t and conn_t's 644 * pointers). Internally within IP we rely on IP's ability to cleanup e.g. 645 * ire_t's when an ill goes away. 646 */ 647 typedef struct ilm_s { 648 in6_addr_t ilm_v6addr; 649 int ilm_refcnt; 650 uint_t ilm_timer; /* IGMP/MLD query resp timer, in msec */ 651 struct ilm_s *ilm_next; /* Linked list for each ill */ 652 uint_t ilm_state; /* state of the membership */ 653 struct ill_s *ilm_ill; /* Back pointer to ill - ill_ilm_cnt */ 654 zoneid_t ilm_zoneid; 655 int ilm_no_ilg_cnt; /* number of joins w/ no ilg */ 656 mcast_record_t ilm_fmode; /* MODE_IS_INCLUDE/MODE_IS_EXCLUDE */ 657 slist_t *ilm_filter; /* source filter list */ 658 slist_t *ilm_pendsrcs; /* relevant src addrs for pending req */ 659 rtx_state_t ilm_rtx; /* SCR retransmission state */ 660 ipaddr_t ilm_ifaddr; /* For IPv4 netstat */ 661 ip_stack_t *ilm_ipst; /* Does not have a netstack_hold */ 662 } ilm_t; 663 664 #define ilm_addr V4_PART_OF_V6(ilm_v6addr) 665 666 /* 667 * Soft reference to an IPsec SA. 668 * 669 * On relative terms, conn's can be persistent (living as long as the 670 * processes which create them), while SA's are ephemeral (dying when 671 * they hit their time-based or byte-based lifetimes). 672 * 673 * We could hold a hard reference to an SA from an ipsec_latch_t, 674 * but this would cause expired SA's to linger for a potentially 675 * unbounded time. 676 * 677 * Instead, we remember the hash bucket number and bucket generation 678 * in addition to the pointer. The bucket generation is incremented on 679 * each deletion. 680 */ 681 typedef struct ipsa_ref_s 682 { 683 struct ipsa_s *ipsr_sa; 684 struct isaf_s *ipsr_bucket; 685 uint64_t ipsr_gen; 686 } ipsa_ref_t; 687 688 /* 689 * IPsec "latching" state. 690 * 691 * In the presence of IPsec policy, fully-bound conn's bind a connection 692 * to more than just the 5-tuple, but also a specific IPsec action and 693 * identity-pair. 694 * The identity pair is accessed from both the receive and transmit side 695 * hence it is maintained in the ipsec_latch_t structure. conn_latch and 696 * ixa_ipsec_latch points to it. 697 * The policy and actions are stored in conn_latch_in_policy and 698 * conn_latch_in_action for the inbound side, and in ixa_ipsec_policy and 699 * ixa_ipsec_action for the transmit side. 700 * 701 * As an optimization, we also cache soft references to IPsec SA's in 702 * ip_xmit_attr_t so that we can fast-path around most of the work needed for 703 * outbound IPsec SA selection. 704 */ 705 typedef struct ipsec_latch_s 706 { 707 kmutex_t ipl_lock; 708 uint32_t ipl_refcnt; 709 710 struct ipsid_s *ipl_local_cid; 711 struct ipsid_s *ipl_remote_cid; 712 unsigned int 713 ipl_ids_latched : 1, 714 715 ipl_pad_to_bit_31 : 31; 716 } ipsec_latch_t; 717 718 #define IPLATCH_REFHOLD(ipl) { \ 719 atomic_add_32(&(ipl)->ipl_refcnt, 1); \ 720 ASSERT((ipl)->ipl_refcnt != 0); \ 721 } 722 723 #define IPLATCH_REFRELE(ipl) { \ 724 ASSERT((ipl)->ipl_refcnt != 0); \ 725 membar_exit(); \ 726 if (atomic_add_32_nv(&(ipl)->ipl_refcnt, -1) == 0) \ 727 iplatch_free(ipl); \ 728 } 729 730 /* 731 * peer identity structure. 732 */ 733 typedef struct conn_s conn_t; 734 735 /* 736 * This is used to match an inbound/outbound datagram with policy. 737 */ 738 typedef struct ipsec_selector { 739 in6_addr_t ips_local_addr_v6; 740 in6_addr_t ips_remote_addr_v6; 741 uint16_t ips_local_port; 742 uint16_t ips_remote_port; 743 uint8_t ips_icmp_type; 744 uint8_t ips_icmp_code; 745 uint8_t ips_protocol; 746 uint8_t ips_isv4 : 1, 747 ips_is_icmp_inv_acq: 1; 748 } ipsec_selector_t; 749 750 /* 751 * Note that we put v4 addresses in the *first* 32-bit word of the 752 * selector rather than the last to simplify the prefix match/mask code 753 * in spd.c 754 */ 755 #define ips_local_addr_v4 ips_local_addr_v6.s6_addr32[0] 756 #define ips_remote_addr_v4 ips_remote_addr_v6.s6_addr32[0] 757 758 /* Values used in IP by IPSEC Code */ 759 #define IPSEC_OUTBOUND B_TRUE 760 #define IPSEC_INBOUND B_FALSE 761 762 /* 763 * There are two variants in policy failures. The packet may come in 764 * secure when not needed (IPSEC_POLICY_???_NOT_NEEDED) or it may not 765 * have the desired level of protection (IPSEC_POLICY_MISMATCH). 766 */ 767 #define IPSEC_POLICY_NOT_NEEDED 0 768 #define IPSEC_POLICY_MISMATCH 1 769 #define IPSEC_POLICY_AUTH_NOT_NEEDED 2 770 #define IPSEC_POLICY_ENCR_NOT_NEEDED 3 771 #define IPSEC_POLICY_SE_NOT_NEEDED 4 772 #define IPSEC_POLICY_MAX 5 /* Always max + 1. */ 773 774 /* 775 * Check with IPSEC inbound policy if 776 * 777 * 1) per-socket policy is present - indicated by conn_in_enforce_policy. 778 * 2) Or if we have not cached policy on the conn and the global policy is 779 * non-empty. 780 */ 781 #define CONN_INBOUND_POLICY_PRESENT(connp, ipss) \ 782 ((connp)->conn_in_enforce_policy || \ 783 (!((connp)->conn_policy_cached) && \ 784 (ipss)->ipsec_inbound_v4_policy_present)) 785 786 #define CONN_INBOUND_POLICY_PRESENT_V6(connp, ipss) \ 787 ((connp)->conn_in_enforce_policy || \ 788 (!(connp)->conn_policy_cached && \ 789 (ipss)->ipsec_inbound_v6_policy_present)) 790 791 #define CONN_OUTBOUND_POLICY_PRESENT(connp, ipss) \ 792 ((connp)->conn_out_enforce_policy || \ 793 (!((connp)->conn_policy_cached) && \ 794 (ipss)->ipsec_outbound_v4_policy_present)) 795 796 #define CONN_OUTBOUND_POLICY_PRESENT_V6(connp, ipss) \ 797 ((connp)->conn_out_enforce_policy || \ 798 (!(connp)->conn_policy_cached && \ 799 (ipss)->ipsec_outbound_v6_policy_present)) 800 801 /* 802 * Information cached in IRE for upper layer protocol (ULP). 803 */ 804 typedef struct iulp_s { 805 boolean_t iulp_set; /* Is any metric set? */ 806 uint32_t iulp_ssthresh; /* Slow start threshold (TCP). */ 807 clock_t iulp_rtt; /* Guestimate in millisecs. */ 808 clock_t iulp_rtt_sd; /* Cached value of RTT variance. */ 809 uint32_t iulp_spipe; /* Send pipe size. */ 810 uint32_t iulp_rpipe; /* Receive pipe size. */ 811 uint32_t iulp_rtomax; /* Max round trip timeout. */ 812 uint32_t iulp_sack; /* Use SACK option (TCP)? */ 813 uint32_t iulp_mtu; /* Setable with routing sockets */ 814 815 uint32_t 816 iulp_tstamp_ok : 1, /* Use timestamp option (TCP)? */ 817 iulp_wscale_ok : 1, /* Use window scale option (TCP)? */ 818 iulp_ecn_ok : 1, /* Enable ECN (for TCP)? */ 819 iulp_pmtud_ok : 1, /* Enable PMTUd? */ 820 821 /* These three are passed out by ip_set_destination */ 822 iulp_localnet: 1, /* IRE_ONLINK */ 823 iulp_loopback: 1, /* IRE_LOOPBACK */ 824 iulp_local: 1, /* IRE_LOCAL */ 825 826 iulp_not_used : 25; 827 } iulp_t; 828 829 /* 830 * The conn drain list structure (idl_t), protected by idl_lock. Each conn_t 831 * inserted in the list points back at this idl_t using conn_idl, and is 832 * chained by conn_drain_next and conn_drain_prev, which are also protected by 833 * idl_lock. When flow control is relieved, either ip_wsrv() (STREAMS) or 834 * ill_flow_enable() (non-STREAMS) will call conn_drain(). 835 * 836 * The conn drain list, idl_t, itself is part of tx cookie list structure. 837 * A tx cookie list points to a blocked Tx ring and contains the list of 838 * all conn's that are blocked due to the flow-controlled Tx ring (via 839 * the idl drain list). Note that a link can have multiple Tx rings. The 840 * drain list will store the conn's blocked due to Tx ring being flow 841 * controlled. 842 */ 843 844 typedef uintptr_t ip_mac_tx_cookie_t; 845 typedef struct idl_s idl_t; 846 typedef struct idl_tx_list_s idl_tx_list_t; 847 848 struct idl_tx_list_s { 849 ip_mac_tx_cookie_t txl_cookie; 850 kmutex_t txl_lock; /* Lock for this list */ 851 idl_t *txl_drain_list; 852 int txl_drain_index; 853 }; 854 855 struct idl_s { 856 conn_t *idl_conn; /* Head of drain list */ 857 kmutex_t idl_lock; /* Lock for this list */ 858 idl_tx_list_t *idl_itl; 859 }; 860 861 /* 862 * Interface route structure which holds the necessary information to recreate 863 * routes that are tied to an interface i.e. have ire_ill set. 864 * 865 * These routes which were initially created via a routing socket or via the 866 * SIOCADDRT ioctl may be gateway routes (RTF_GATEWAY being set) or may be 867 * traditional interface routes. When an ill comes back up after being 868 * down, this information will be used to recreate the routes. These 869 * are part of an mblk_t chain that hangs off of the ILL (ill_saved_ire_mp). 870 */ 871 typedef struct ifrt_s { 872 ushort_t ifrt_type; /* Type of IRE */ 873 in6_addr_t ifrt_v6addr; /* Address IRE represents. */ 874 in6_addr_t ifrt_v6gateway_addr; /* Gateway if IRE_OFFLINK */ 875 in6_addr_t ifrt_v6setsrc_addr; /* Src addr if RTF_SETSRC */ 876 in6_addr_t ifrt_v6mask; /* Mask for matching IRE. */ 877 uint32_t ifrt_flags; /* flags related to route */ 878 iulp_t ifrt_metrics; /* Routing socket metrics */ 879 zoneid_t ifrt_zoneid; /* zoneid for route */ 880 } ifrt_t; 881 882 #define ifrt_addr V4_PART_OF_V6(ifrt_v6addr) 883 #define ifrt_gateway_addr V4_PART_OF_V6(ifrt_v6gateway_addr) 884 #define ifrt_mask V4_PART_OF_V6(ifrt_v6mask) 885 #define ifrt_setsrc_addr V4_PART_OF_V6(ifrt_v6setsrc_addr) 886 887 /* Number of IP addresses that can be hosted on a physical interface */ 888 #define MAX_ADDRS_PER_IF 8192 889 /* 890 * Number of Source addresses to be considered for source address 891 * selection. Used by ipif_select_source_v4/v6. 892 */ 893 #define MAX_IPIF_SELECT_SOURCE 50 894 895 #ifdef IP_DEBUG 896 /* 897 * Trace refholds and refreles for debugging. 898 */ 899 #define TR_STACK_DEPTH 14 900 typedef struct tr_buf_s { 901 int tr_depth; 902 clock_t tr_time; 903 pc_t tr_stack[TR_STACK_DEPTH]; 904 } tr_buf_t; 905 906 typedef struct th_trace_s { 907 int th_refcnt; 908 uint_t th_trace_lastref; 909 kthread_t *th_id; 910 #define TR_BUF_MAX 38 911 tr_buf_t th_trbuf[TR_BUF_MAX]; 912 } th_trace_t; 913 914 typedef struct th_hash_s { 915 list_node_t thh_link; 916 mod_hash_t *thh_hash; 917 ip_stack_t *thh_ipst; 918 } th_hash_t; 919 #endif 920 921 /* The following are ipif_state_flags */ 922 #define IPIF_CONDEMNED 0x1 /* The ipif is being removed */ 923 #define IPIF_CHANGING 0x2 /* A critcal ipif field is changing */ 924 #define IPIF_SET_LINKLOCAL 0x10 /* transient flag during bringup */ 925 926 /* IP interface structure, one per local address */ 927 typedef struct ipif_s { 928 struct ipif_s *ipif_next; 929 struct ill_s *ipif_ill; /* Back pointer to our ill */ 930 int ipif_id; /* Logical unit number */ 931 in6_addr_t ipif_v6lcl_addr; /* Local IP address for this if. */ 932 in6_addr_t ipif_v6subnet; /* Subnet prefix for this if. */ 933 in6_addr_t ipif_v6net_mask; /* Net mask for this interface. */ 934 in6_addr_t ipif_v6brd_addr; /* Broadcast addr for this interface. */ 935 in6_addr_t ipif_v6pp_dst_addr; /* Point-to-point dest address. */ 936 uint64_t ipif_flags; /* Interface flags. */ 937 uint_t ipif_ire_type; /* IRE_LOCAL or IRE_LOOPBACK */ 938 939 /* 940 * The packet count in the ipif contain the sum of the 941 * packet counts in dead IRE_LOCAL/LOOPBACK for this ipif. 942 */ 943 uint_t ipif_ib_pkt_count; /* Inbound packets for our dead IREs */ 944 945 /* Exclusive bit fields, protected by ipsq_t */ 946 unsigned int 947 ipif_was_up : 1, /* ipif was up before */ 948 ipif_addr_ready : 1, /* DAD is done */ 949 ipif_was_dup : 1, /* DAD had failed */ 950 ipif_added_nce : 1, /* nce added for local address */ 951 952 ipif_pad_to_31 : 28; 953 954 ilm_t *ipif_allhosts_ilm; /* For all-nodes join */ 955 ilm_t *ipif_solmulti_ilm; /* For IPv6 solicited multicast join */ 956 957 uint_t ipif_seqid; /* unique index across all ills */ 958 uint_t ipif_state_flags; /* See IPIF_* flag defs above */ 959 uint_t ipif_refcnt; /* active consistent reader cnt */ 960 961 zoneid_t ipif_zoneid; /* zone ID number */ 962 timeout_id_t ipif_recovery_id; /* Timer for DAD recovery */ 963 boolean_t ipif_trace_disable; /* True when alloc fails */ 964 /* 965 * For an IPMP interface, ipif_bound_ill tracks the ill whose hardware 966 * information this ipif is associated with via ARP/NDP. We can use 967 * an ill pointer (rather than an index) because only ills that are 968 * part of a group will be pointed to, and an ill cannot disappear 969 * while it's in a group. 970 */ 971 struct ill_s *ipif_bound_ill; 972 struct ipif_s *ipif_bound_next; /* bound ipif chain */ 973 boolean_t ipif_bound; /* B_TRUE if we successfully bound */ 974 975 struct ire_s *ipif_ire_local; /* Our IRE_LOCAL or LOOPBACK */ 976 struct ire_s *ipif_ire_if; /* Our IRE_INTERFACE */ 977 } ipif_t; 978 979 /* 980 * The following table lists the protection levels of the various members 981 * of the ipif_t. The following notation is used. 982 * 983 * Write once - Written to only once at the time of bringing up 984 * the interface and can be safely read after the bringup without any lock. 985 * 986 * ipsq - Need to execute in the ipsq to perform the indicated access. 987 * 988 * ill_lock - Need to hold this mutex to perform the indicated access. 989 * 990 * ill_g_lock - Need to hold this rw lock as reader/writer for read access or 991 * write access respectively. 992 * 993 * down ill - Written to only when the ill is down (i.e all ipifs are down) 994 * up ill - Read only when the ill is up (i.e. at least 1 ipif is up) 995 * 996 * Table of ipif_t members and their protection 997 * 998 * ipif_next ipsq + ill_lock + ipsq OR ill_lock OR 999 * ill_g_lock ill_g_lock 1000 * ipif_ill ipsq + down ipif write once 1001 * ipif_id ipsq + down ipif write once 1002 * ipif_v6lcl_addr ipsq + down ipif up ipif 1003 * ipif_v6subnet ipsq + down ipif up ipif 1004 * ipif_v6net_mask ipsq + down ipif up ipif 1005 * 1006 * ipif_v6brd_addr 1007 * ipif_v6pp_dst_addr 1008 * ipif_flags ill_lock ill_lock 1009 * ipif_ire_type ipsq + down ill up ill 1010 * 1011 * ipif_ib_pkt_count Approx 1012 * 1013 * bit fields ill_lock ill_lock 1014 * 1015 * ipif_allhosts_ilm ipsq ipsq 1016 * ipif_solmulti_ilm ipsq ipsq 1017 * 1018 * ipif_seqid ipsq Write once 1019 * 1020 * ipif_state_flags ill_lock ill_lock 1021 * ipif_refcnt ill_lock ill_lock 1022 * ipif_bound_ill ipsq + ipmp_lock ipsq OR ipmp_lock 1023 * ipif_bound_next ipsq ipsq 1024 * ipif_bound ipsq ipsq 1025 * 1026 * ipif_ire_local ipsq + ips_ill_g_lock ipsq OR ips_ill_g_lock 1027 * ipif_ire_if ipsq + ips_ill_g_lock ipsq OR ips_ill_g_lock 1028 */ 1029 1030 /* 1031 * Return values from ip_laddr_verify_{v4,v6} 1032 */ 1033 typedef enum { IPVL_UNICAST_UP, IPVL_UNICAST_DOWN, IPVL_MCAST, IPVL_BCAST, 1034 IPVL_BAD} ip_laddr_t; 1035 1036 1037 #define IP_TR_HASH(tid) ((((uintptr_t)tid) >> 6) & (IP_TR_HASH_MAX - 1)) 1038 1039 #ifdef DEBUG 1040 #define IPIF_TRACE_REF(ipif) ipif_trace_ref(ipif) 1041 #define ILL_TRACE_REF(ill) ill_trace_ref(ill) 1042 #define IPIF_UNTRACE_REF(ipif) ipif_untrace_ref(ipif) 1043 #define ILL_UNTRACE_REF(ill) ill_untrace_ref(ill) 1044 #else 1045 #define IPIF_TRACE_REF(ipif) 1046 #define ILL_TRACE_REF(ill) 1047 #define IPIF_UNTRACE_REF(ipif) 1048 #define ILL_UNTRACE_REF(ill) 1049 #endif 1050 1051 /* IPv4 compatibility macros */ 1052 #define ipif_lcl_addr V4_PART_OF_V6(ipif_v6lcl_addr) 1053 #define ipif_subnet V4_PART_OF_V6(ipif_v6subnet) 1054 #define ipif_net_mask V4_PART_OF_V6(ipif_v6net_mask) 1055 #define ipif_brd_addr V4_PART_OF_V6(ipif_v6brd_addr) 1056 #define ipif_pp_dst_addr V4_PART_OF_V6(ipif_v6pp_dst_addr) 1057 1058 /* Macros for easy backreferences to the ill. */ 1059 #define ipif_isv6 ipif_ill->ill_isv6 1060 1061 #define SIOCLIFADDR_NDX 112 /* ndx of SIOCLIFADDR in the ndx ioctl table */ 1062 1063 /* 1064 * mode value for ip_ioctl_finish for finishing an ioctl 1065 */ 1066 #define CONN_CLOSE 1 /* No mi_copy */ 1067 #define COPYOUT 2 /* do an mi_copyout if needed */ 1068 #define NO_COPYOUT 3 /* do an mi_copy_done */ 1069 #define IPI2MODE(ipi) ((ipi)->ipi_flags & IPI_GET_CMD ? COPYOUT : NO_COPYOUT) 1070 1071 /* 1072 * The IP-MT design revolves around the serialization objects ipsq_t (IPSQ) 1073 * and ipxop_t (exclusive operation or "xop"). Becoming "writer" on an IPSQ 1074 * ensures that no other threads can become "writer" on any IPSQs sharing that 1075 * IPSQ's xop until the writer thread is done. 1076 * 1077 * Each phyint points to one IPSQ that remains fixed over the phyint's life. 1078 * Each IPSQ points to one xop that can change over the IPSQ's life. If a 1079 * phyint is *not* in an IPMP group, then its IPSQ will refer to the IPSQ's 1080 * "own" xop (ipsq_ownxop). If a phyint *is* part of an IPMP group, then its 1081 * IPSQ will refer to the "group" xop, which is shorthand for the xop of the 1082 * IPSQ of the IPMP meta-interface's phyint. Thus, all phyints that are part 1083 * of the same IPMP group will have their IPSQ's point to the group xop, and 1084 * thus becoming "writer" on any phyint in the group will prevent any other 1085 * writer on any other phyint in the group. All IPSQs sharing the same xop 1086 * are chained together through ipsq_next (in the degenerate common case, 1087 * ipsq_next simply refers to itself). Note that the group xop is guaranteed 1088 * to exist at least as long as there are members in the group, since the IPMP 1089 * meta-interface can only be destroyed if the group is empty. 1090 * 1091 * Incoming exclusive operation requests are enqueued on the IPSQ they arrived 1092 * on rather than the xop. This makes switching xop's (as would happen when a 1093 * phyint leaves an IPMP group) simple, because after the phyint leaves the 1094 * group, any operations enqueued on its IPSQ can be safely processed with 1095 * respect to its new xop, and any operations enqueued on the IPSQs of its 1096 * former group can be processed with respect to their existing group xop. 1097 * Even so, switching xops is a subtle dance; see ipsq_dq() for details. 1098 * 1099 * An IPSQ's "own" xop is embedded within the IPSQ itself since they have have 1100 * identical lifetimes, and because doing so simplifies pointer management. 1101 * While each phyint and IPSQ point to each other, it is not possible to free 1102 * the IPSQ when the phyint is freed, since we may still *inside* the IPSQ 1103 * when the phyint is being freed. Thus, ipsq_phyint is set to NULL when the 1104 * phyint is freed, and the IPSQ free is later done in ipsq_exit(). 1105 * 1106 * ipsq_t synchronization: read write 1107 * 1108 * ipsq_xopq_mphead ipx_lock ipx_lock 1109 * ipsq_xopq_mptail ipx_lock ipx_lock 1110 * ipsq_xop_switch_mp ipsq_lock ipsq_lock 1111 * ipsq_phyint write once write once 1112 * ipsq_next RW_READER ill_g_lock RW_WRITER ill_g_lock 1113 * ipsq_xop ipsq_lock or ipsq ipsq_lock + ipsq 1114 * ipsq_swxop ipsq ipsq 1115 * ipsq_ownxop see ipxop_t see ipxop_t 1116 * ipsq_ipst write once write once 1117 * 1118 * ipxop_t synchronization: read write 1119 * 1120 * ipx_writer ipx_lock ipx_lock 1121 * ipx_xop_queued ipx_lock ipx_lock 1122 * ipx_mphead ipx_lock ipx_lock 1123 * ipx_mptail ipx_lock ipx_lock 1124 * ipx_ipsq write once write once 1125 * ips_ipsq_queued ipx_lock ipx_lock 1126 * ipx_waitfor ipsq or ipx_lock ipsq + ipx_lock 1127 * ipx_reentry_cnt ipsq or ipx_lock ipsq + ipx_lock 1128 * ipx_current_done ipsq ipsq 1129 * ipx_current_ioctl ipsq ipsq 1130 * ipx_current_ipif ipsq or ipx_lock ipsq + ipx_lock 1131 * ipx_pending_ipif ipsq or ipx_lock ipsq + ipx_lock 1132 * ipx_pending_mp ipsq or ipx_lock ipsq + ipx_lock 1133 * ipx_forced ipsq ipsq 1134 * ipx_depth ipsq ipsq 1135 * ipx_stack ipsq ipsq 1136 */ 1137 typedef struct ipxop_s { 1138 kmutex_t ipx_lock; /* see above */ 1139 kthread_t *ipx_writer; /* current owner */ 1140 mblk_t *ipx_mphead; /* messages tied to this op */ 1141 mblk_t *ipx_mptail; 1142 struct ipsq_s *ipx_ipsq; /* associated ipsq */ 1143 boolean_t ipx_ipsq_queued; /* ipsq using xop has queued op */ 1144 int ipx_waitfor; /* waiting; values encoded below */ 1145 int ipx_reentry_cnt; 1146 boolean_t ipx_current_done; /* is the current operation done? */ 1147 int ipx_current_ioctl; /* current ioctl, or 0 if no ioctl */ 1148 ipif_t *ipx_current_ipif; /* ipif for current op */ 1149 ipif_t *ipx_pending_ipif; /* ipif for ipx_pending_mp */ 1150 mblk_t *ipx_pending_mp; /* current ioctl mp while waiting */ 1151 boolean_t ipx_forced; /* debugging aid */ 1152 #ifdef DEBUG 1153 int ipx_depth; /* debugging aid */ 1154 #define IPX_STACK_DEPTH 15 1155 pc_t ipx_stack[IPX_STACK_DEPTH]; /* debugging aid */ 1156 #endif 1157 } ipxop_t; 1158 1159 typedef struct ipsq_s { 1160 kmutex_t ipsq_lock; /* see above */ 1161 mblk_t *ipsq_switch_mp; /* op to handle right after switch */ 1162 mblk_t *ipsq_xopq_mphead; /* list of excl ops (mostly ioctls) */ 1163 mblk_t *ipsq_xopq_mptail; 1164 struct phyint *ipsq_phyint; /* associated phyint */ 1165 struct ipsq_s *ipsq_next; /* next ipsq sharing ipsq_xop */ 1166 struct ipxop_s *ipsq_xop; /* current xop synchronization info */ 1167 struct ipxop_s *ipsq_swxop; /* switch xop to on ipsq_exit() */ 1168 struct ipxop_s ipsq_ownxop; /* our own xop (may not be in-use) */ 1169 ip_stack_t *ipsq_ipst; /* does not have a netstack_hold */ 1170 } ipsq_t; 1171 1172 /* 1173 * ipx_waitfor values: 1174 */ 1175 enum { 1176 IPIF_DOWN = 1, /* ipif_down() waiting for refcnts to drop */ 1177 ILL_DOWN, /* ill_down() waiting for refcnts to drop */ 1178 IPIF_FREE, /* ipif_free() waiting for refcnts to drop */ 1179 ILL_FREE /* ill unplumb waiting for refcnts to drop */ 1180 }; 1181 1182 /* Operation types for ipsq_try_enter() */ 1183 #define CUR_OP 0 /* request writer within current operation */ 1184 #define NEW_OP 1 /* request writer for a new operation */ 1185 #define SWITCH_OP 2 /* request writer once IPSQ XOP switches */ 1186 1187 /* 1188 * Kstats tracked on each IPMP meta-interface. Order here must match 1189 * ipmp_kstats[] in ip/ipmp.c. 1190 */ 1191 enum { 1192 IPMP_KSTAT_OBYTES, IPMP_KSTAT_OBYTES64, IPMP_KSTAT_RBYTES, 1193 IPMP_KSTAT_RBYTES64, IPMP_KSTAT_OPACKETS, IPMP_KSTAT_OPACKETS64, 1194 IPMP_KSTAT_OERRORS, IPMP_KSTAT_IPACKETS, IPMP_KSTAT_IPACKETS64, 1195 IPMP_KSTAT_IERRORS, IPMP_KSTAT_MULTIRCV, IPMP_KSTAT_MULTIXMT, 1196 IPMP_KSTAT_BRDCSTRCV, IPMP_KSTAT_BRDCSTXMT, IPMP_KSTAT_LINK_UP, 1197 IPMP_KSTAT_MAX /* keep last */ 1198 }; 1199 1200 /* 1201 * phyint represents state that is common to both IPv4 and IPv6 interfaces. 1202 * There is a separate ill_t representing IPv4 and IPv6 which has a 1203 * backpointer to the phyint structure for accessing common state. 1204 */ 1205 typedef struct phyint { 1206 struct ill_s *phyint_illv4; 1207 struct ill_s *phyint_illv6; 1208 uint_t phyint_ifindex; /* SIOCSLIFINDEX */ 1209 uint64_t phyint_flags; 1210 avl_node_t phyint_avl_by_index; /* avl tree by index */ 1211 avl_node_t phyint_avl_by_name; /* avl tree by name */ 1212 kmutex_t phyint_lock; 1213 struct ipsq_s *phyint_ipsq; /* back pointer to ipsq */ 1214 struct ipmp_grp_s *phyint_grp; /* associated IPMP group */ 1215 char phyint_name[LIFNAMSIZ]; /* physical interface name */ 1216 uint64_t phyint_kstats0[IPMP_KSTAT_MAX]; /* baseline kstats */ 1217 } phyint_t; 1218 1219 #define CACHE_ALIGN_SIZE 64 1220 #define CACHE_ALIGN(align_struct) P2ROUNDUP(sizeof (struct align_struct),\ 1221 CACHE_ALIGN_SIZE) 1222 struct _phyint_list_s_ { 1223 avl_tree_t phyint_list_avl_by_index; /* avl tree by index */ 1224 avl_tree_t phyint_list_avl_by_name; /* avl tree by name */ 1225 }; 1226 1227 typedef union phyint_list_u { 1228 struct _phyint_list_s_ phyint_list_s; 1229 char phyint_list_filler[CACHE_ALIGN(_phyint_list_s_)]; 1230 } phyint_list_t; 1231 1232 #define phyint_list_avl_by_index phyint_list_s.phyint_list_avl_by_index 1233 #define phyint_list_avl_by_name phyint_list_s.phyint_list_avl_by_name 1234 1235 /* 1236 * Fragmentation hash bucket 1237 */ 1238 typedef struct ipfb_s { 1239 struct ipf_s *ipfb_ipf; /* List of ... */ 1240 size_t ipfb_count; /* Count of bytes used by frag(s) */ 1241 kmutex_t ipfb_lock; /* Protect all ipf in list */ 1242 uint_t ipfb_frag_pkts; /* num of distinct fragmented pkts */ 1243 } ipfb_t; 1244 1245 /* 1246 * IRE bucket structure. Usually there is an array of such structures, 1247 * each pointing to a linked list of ires. irb_refcnt counts the number 1248 * of walkers of a given hash bucket. Usually the reference count is 1249 * bumped up if the walker wants no IRES to be DELETED while walking the 1250 * list. Bumping up does not PREVENT ADDITION. This allows walking a given 1251 * hash bucket without stumbling up on a free pointer. 1252 * 1253 * irb_t structures in ip_ftable are dynamically allocated and freed. 1254 * In order to identify the irb_t structures that can be safely kmem_free'd 1255 * we need to ensure that 1256 * - the irb_refcnt is quiescent, indicating no other walkers, 1257 * - no other threads or ire's are holding references to the irb, 1258 * i.e., irb_nire == 0, 1259 * - there are no active ire's in the bucket, i.e., irb_ire_cnt == 0 1260 */ 1261 typedef struct irb { 1262 struct ire_s *irb_ire; /* First ire in this bucket */ 1263 /* Should be first in this struct */ 1264 krwlock_t irb_lock; /* Protect this bucket */ 1265 uint_t irb_refcnt; /* Protected by irb_lock */ 1266 uchar_t irb_marks; /* CONDEMNED ires in this bucket ? */ 1267 #define IRB_MARK_CONDEMNED 0x0001 /* Contains some IRE_IS_CONDEMNED */ 1268 #define IRB_MARK_DYNAMIC 0x0002 /* Dynamically allocated */ 1269 /* Once IPv6 uses radix then IRB_MARK_DYNAMIC will be always be set */ 1270 uint_t irb_ire_cnt; /* Num of active IRE in this bucket */ 1271 int irb_nire; /* Num of ftable ire's that ref irb */ 1272 ip_stack_t *irb_ipst; /* Does not have a netstack_hold */ 1273 } irb_t; 1274 1275 #define IRB2RT(irb) (rt_t *)((caddr_t)(irb) - offsetof(rt_t, rt_irb)) 1276 1277 /* Forward declarations */ 1278 struct dce_s; 1279 typedef struct dce_s dce_t; 1280 struct ire_s; 1281 typedef struct ire_s ire_t; 1282 struct ncec_s; 1283 typedef struct ncec_s ncec_t; 1284 struct nce_s; 1285 typedef struct nce_s nce_t; 1286 struct ip_recv_attr_s; 1287 typedef struct ip_recv_attr_s ip_recv_attr_t; 1288 struct ip_xmit_attr_s; 1289 typedef struct ip_xmit_attr_s ip_xmit_attr_t; 1290 1291 struct tsol_ire_gw_secattr_s; 1292 typedef struct tsol_ire_gw_secattr_s tsol_ire_gw_secattr_t; 1293 1294 /* 1295 * This is a structure for a one-element route cache that is passed 1296 * by reference between ip_input and ill_inputfn. 1297 */ 1298 typedef struct { 1299 ire_t *rtc_ire; 1300 ipaddr_t rtc_ipaddr; 1301 in6_addr_t rtc_ip6addr; 1302 } rtc_t; 1303 1304 /* 1305 * Note: Temporarily use 64 bits, and will probably go back to 32 bits after 1306 * more cleanup work is done. 1307 */ 1308 typedef uint64_t iaflags_t; 1309 1310 /* The ill input function pointer type */ 1311 typedef void (*pfillinput_t)(mblk_t *, void *, void *, ip_recv_attr_t *, 1312 rtc_t *); 1313 1314 /* The ire receive function pointer type */ 1315 typedef void (*pfirerecv_t)(ire_t *, mblk_t *, void *, ip_recv_attr_t *); 1316 1317 /* The ire send and postfrag function pointer types */ 1318 typedef int (*pfiresend_t)(ire_t *, mblk_t *, void *, 1319 ip_xmit_attr_t *, uint32_t *); 1320 typedef int (*pfirepostfrag_t)(mblk_t *, nce_t *, iaflags_t, uint_t, uint32_t, 1321 zoneid_t, zoneid_t, uintptr_t *); 1322 1323 1324 #define IP_V4_G_HEAD 0 1325 #define IP_V6_G_HEAD 1 1326 1327 #define MAX_G_HEADS 2 1328 1329 /* 1330 * unpadded ill_if structure 1331 */ 1332 struct _ill_if_s_ { 1333 union ill_if_u *illif_next; 1334 union ill_if_u *illif_prev; 1335 avl_tree_t illif_avl_by_ppa; /* AVL tree sorted on ppa */ 1336 vmem_t *illif_ppa_arena; /* ppa index space */ 1337 uint16_t illif_mcast_v1; /* hints for */ 1338 uint16_t illif_mcast_v2; /* [igmp|mld]_slowtimo */ 1339 int illif_name_len; /* name length */ 1340 char illif_name[LIFNAMSIZ]; /* name of interface type */ 1341 }; 1342 1343 /* cache aligned ill_if structure */ 1344 typedef union ill_if_u { 1345 struct _ill_if_s_ ill_if_s; 1346 char illif_filler[CACHE_ALIGN(_ill_if_s_)]; 1347 } ill_if_t; 1348 1349 #define illif_next ill_if_s.illif_next 1350 #define illif_prev ill_if_s.illif_prev 1351 #define illif_avl_by_ppa ill_if_s.illif_avl_by_ppa 1352 #define illif_ppa_arena ill_if_s.illif_ppa_arena 1353 #define illif_mcast_v1 ill_if_s.illif_mcast_v1 1354 #define illif_mcast_v2 ill_if_s.illif_mcast_v2 1355 #define illif_name ill_if_s.illif_name 1356 #define illif_name_len ill_if_s.illif_name_len 1357 1358 typedef struct ill_walk_context_s { 1359 int ctx_current_list; /* current list being searched */ 1360 int ctx_last_list; /* last list to search */ 1361 } ill_walk_context_t; 1362 1363 /* 1364 * ill_g_heads structure, one for IPV4 and one for IPV6 1365 */ 1366 struct _ill_g_head_s_ { 1367 ill_if_t *ill_g_list_head; 1368 ill_if_t *ill_g_list_tail; 1369 }; 1370 1371 typedef union ill_g_head_u { 1372 struct _ill_g_head_s_ ill_g_head_s; 1373 char ill_g_head_filler[CACHE_ALIGN(_ill_g_head_s_)]; 1374 } ill_g_head_t; 1375 1376 #define ill_g_list_head ill_g_head_s.ill_g_list_head 1377 #define ill_g_list_tail ill_g_head_s.ill_g_list_tail 1378 1379 #define IP_V4_ILL_G_LIST(ipst) \ 1380 (ipst)->ips_ill_g_heads[IP_V4_G_HEAD].ill_g_list_head 1381 #define IP_V6_ILL_G_LIST(ipst) \ 1382 (ipst)->ips_ill_g_heads[IP_V6_G_HEAD].ill_g_list_head 1383 #define IP_VX_ILL_G_LIST(i, ipst) \ 1384 (ipst)->ips_ill_g_heads[i].ill_g_list_head 1385 1386 #define ILL_START_WALK_V4(ctx_ptr, ipst) \ 1387 ill_first(IP_V4_G_HEAD, IP_V4_G_HEAD, ctx_ptr, ipst) 1388 #define ILL_START_WALK_V6(ctx_ptr, ipst) \ 1389 ill_first(IP_V6_G_HEAD, IP_V6_G_HEAD, ctx_ptr, ipst) 1390 #define ILL_START_WALK_ALL(ctx_ptr, ipst) \ 1391 ill_first(MAX_G_HEADS, MAX_G_HEADS, ctx_ptr, ipst) 1392 1393 /* 1394 * Capabilities, possible flags for ill_capabilities. 1395 */ 1396 #define ILL_CAPAB_LSO 0x04 /* Large Send Offload */ 1397 #define ILL_CAPAB_HCKSUM 0x08 /* Hardware checksumming */ 1398 #define ILL_CAPAB_ZEROCOPY 0x10 /* Zero-copy */ 1399 #define ILL_CAPAB_DLD 0x20 /* DLD capabilities */ 1400 #define ILL_CAPAB_DLD_POLL 0x40 /* Polling */ 1401 #define ILL_CAPAB_DLD_DIRECT 0x80 /* Direct function call */ 1402 1403 /* 1404 * Per-ill Hardware Checksumming capbilities. 1405 */ 1406 typedef struct ill_hcksum_capab_s ill_hcksum_capab_t; 1407 1408 /* 1409 * Per-ill Zero-copy capabilities. 1410 */ 1411 typedef struct ill_zerocopy_capab_s ill_zerocopy_capab_t; 1412 1413 /* 1414 * DLD capbilities. 1415 */ 1416 typedef struct ill_dld_capab_s ill_dld_capab_t; 1417 1418 /* 1419 * Per-ill polling resource map. 1420 */ 1421 typedef struct ill_rx_ring ill_rx_ring_t; 1422 1423 /* 1424 * Per-ill Large Send Offload capabilities. 1425 */ 1426 typedef struct ill_lso_capab_s ill_lso_capab_t; 1427 1428 /* The following are ill_state_flags */ 1429 #define ILL_LL_SUBNET_PENDING 0x01 /* Waiting for DL_INFO_ACK from drv */ 1430 #define ILL_CONDEMNED 0x02 /* No more new ref's to the ILL */ 1431 #define ILL_DL_UNBIND_IN_PROGRESS 0x04 /* UNBIND_REQ is sent */ 1432 #define ILL_DOWN_IN_PROGRESS 0x08 /* ILL is going down - no new nce's */ 1433 #define ILL_LL_BIND_PENDING 0x0020 /* XXX Reuse ILL_LL_SUBNET_PENDING ? */ 1434 #define ILL_LL_UP 0x0040 1435 #define ILL_LL_DOWN 0x0080 1436 1437 /* Is this an ILL whose source address is used by other ILL's ? */ 1438 #define IS_USESRC_ILL(ill) \ 1439 (((ill)->ill_usesrc_ifindex == 0) && \ 1440 ((ill)->ill_usesrc_grp_next != NULL)) 1441 1442 /* Is this a client/consumer of the usesrc ILL ? */ 1443 #define IS_USESRC_CLI_ILL(ill) \ 1444 (((ill)->ill_usesrc_ifindex != 0) && \ 1445 ((ill)->ill_usesrc_grp_next != NULL)) 1446 1447 /* Is this an virtual network interface (vni) ILL ? */ 1448 #define IS_VNI(ill) \ 1449 (((ill)->ill_phyint->phyint_flags & (PHYI_LOOPBACK|PHYI_VIRTUAL)) == \ 1450 PHYI_VIRTUAL) 1451 1452 /* Is this a loopback ILL? */ 1453 #define IS_LOOPBACK(ill) \ 1454 ((ill)->ill_phyint->phyint_flags & PHYI_LOOPBACK) 1455 1456 /* Is this an IPMP meta-interface ILL? */ 1457 #define IS_IPMP(ill) \ 1458 ((ill)->ill_phyint->phyint_flags & PHYI_IPMP) 1459 1460 /* Is this ILL under an IPMP meta-interface? (aka "in a group?") */ 1461 #define IS_UNDER_IPMP(ill) \ 1462 ((ill)->ill_grp != NULL && !IS_IPMP(ill)) 1463 1464 /* Is ill1 in the same illgrp as ill2? */ 1465 #define IS_IN_SAME_ILLGRP(ill1, ill2) \ 1466 ((ill1)->ill_grp != NULL && ((ill1)->ill_grp == (ill2)->ill_grp)) 1467 1468 /* Is ill1 on the same LAN as ill2? */ 1469 #define IS_ON_SAME_LAN(ill1, ill2) \ 1470 ((ill1) == (ill2) || IS_IN_SAME_ILLGRP(ill1, ill2)) 1471 1472 #define ILL_OTHER(ill) \ 1473 ((ill)->ill_isv6 ? (ill)->ill_phyint->phyint_illv4 : \ 1474 (ill)->ill_phyint->phyint_illv6) 1475 1476 /* 1477 * IPMP group ILL state structure -- up to two per IPMP group (V4 and V6). 1478 * Created when the V4 and/or V6 IPMP meta-interface is I_PLINK'd. It is 1479 * guaranteed to persist while there are interfaces of that type in the group. 1480 * In general, most fields are accessed outside of the IPSQ (e.g., in the 1481 * datapath), and thus use locks in addition to the IPSQ for protection. 1482 * 1483 * synchronization: read write 1484 * 1485 * ig_if ipsq or ill_g_lock ipsq and ill_g_lock 1486 * ig_actif ipsq or ipmp_lock ipsq and ipmp_lock 1487 * ig_nactif ipsq or ipmp_lock ipsq and ipmp_lock 1488 * ig_next_ill ipsq or ipmp_lock ipsq and ipmp_lock 1489 * ig_ipmp_ill write once write once 1490 * ig_cast_ill ipsq or ipmp_lock ipsq and ipmp_lock 1491 * ig_arpent ipsq ipsq 1492 * ig_mtu ipsq ipsq 1493 */ 1494 typedef struct ipmp_illgrp_s { 1495 list_t ig_if; /* list of all interfaces */ 1496 list_t ig_actif; /* list of active interfaces */ 1497 uint_t ig_nactif; /* number of active interfaces */ 1498 struct ill_s *ig_next_ill; /* next active interface to use */ 1499 struct ill_s *ig_ipmp_ill; /* backpointer to IPMP meta-interface */ 1500 struct ill_s *ig_cast_ill; /* nominated ill for multi/broadcast */ 1501 list_t ig_arpent; /* list of ARP entries */ 1502 uint_t ig_mtu; /* ig_ipmp_ill->ill_max_mtu */ 1503 } ipmp_illgrp_t; 1504 1505 /* 1506 * IPMP group state structure -- one per IPMP group. Created when the 1507 * IPMP meta-interface is plumbed; it is guaranteed to persist while there 1508 * are interfaces in it. 1509 * 1510 * ipmp_grp_t synchronization: read write 1511 * 1512 * gr_name ipmp_lock ipmp_lock 1513 * gr_ifname write once write once 1514 * gr_mactype ipmp_lock ipmp_lock 1515 * gr_phyint write once write once 1516 * gr_nif ipmp_lock ipmp_lock 1517 * gr_nactif ipsq ipsq 1518 * gr_v4 ipmp_lock ipmp_lock 1519 * gr_v6 ipmp_lock ipmp_lock 1520 * gr_nv4 ipmp_lock ipmp_lock 1521 * gr_nv6 ipmp_lock ipmp_lock 1522 * gr_pendv4 ipmp_lock ipmp_lock 1523 * gr_pendv6 ipmp_lock ipmp_lock 1524 * gr_linkdownmp ipsq ipsq 1525 * gr_ksp ipmp_lock ipmp_lock 1526 * gr_kstats0 atomic atomic 1527 */ 1528 typedef struct ipmp_grp_s { 1529 char gr_name[LIFGRNAMSIZ]; /* group name */ 1530 char gr_ifname[LIFNAMSIZ]; /* interface name */ 1531 t_uscalar_t gr_mactype; /* DLPI mactype of group */ 1532 phyint_t *gr_phyint; /* IPMP group phyint */ 1533 uint_t gr_nif; /* number of interfaces in group */ 1534 uint_t gr_nactif; /* number of active interfaces */ 1535 ipmp_illgrp_t *gr_v4; /* V4 group information */ 1536 ipmp_illgrp_t *gr_v6; /* V6 group information */ 1537 uint_t gr_nv4; /* number of ills in V4 group */ 1538 uint_t gr_nv6; /* number of ills in V6 group */ 1539 uint_t gr_pendv4; /* number of pending ills in V4 group */ 1540 uint_t gr_pendv6; /* number of pending ills in V6 group */ 1541 mblk_t *gr_linkdownmp; /* message used to bring link down */ 1542 kstat_t *gr_ksp; /* group kstat pointer */ 1543 uint64_t gr_kstats0[IPMP_KSTAT_MAX]; /* baseline group kstats */ 1544 } ipmp_grp_t; 1545 1546 /* 1547 * IPMP ARP entry -- one per SIOCS*ARP entry tied to the group. Used to keep 1548 * ARP up-to-date as the active set of interfaces in the group changes. 1549 */ 1550 typedef struct ipmp_arpent_s { 1551 ipaddr_t ia_ipaddr; /* IP address for this entry */ 1552 boolean_t ia_proxyarp; /* proxy ARP entry? */ 1553 boolean_t ia_notified; /* ARP notified about this entry? */ 1554 list_node_t ia_node; /* next ARP entry in list */ 1555 uint16_t ia_flags; /* nce_flags for the address */ 1556 size_t ia_lladdr_len; 1557 uchar_t *ia_lladdr; 1558 } ipmp_arpent_t; 1559 1560 struct arl_s; 1561 1562 /* 1563 * Per-ill capabilities. 1564 */ 1565 struct ill_hcksum_capab_s { 1566 uint_t ill_hcksum_version; /* interface version */ 1567 uint_t ill_hcksum_txflags; /* capabilities on transmit */ 1568 }; 1569 1570 struct ill_zerocopy_capab_s { 1571 uint_t ill_zerocopy_version; /* interface version */ 1572 uint_t ill_zerocopy_flags; /* capabilities */ 1573 }; 1574 1575 struct ill_lso_capab_s { 1576 uint_t ill_lso_flags; /* capabilities */ 1577 uint_t ill_lso_max; /* maximum size of payload */ 1578 }; 1579 1580 /* 1581 * IP Lower level Structure. 1582 * Instance data structure in ip_open when there is a device below us. 1583 */ 1584 typedef struct ill_s { 1585 pfillinput_t ill_inputfn; /* Fast input function selector */ 1586 ill_if_t *ill_ifptr; /* pointer to interface type */ 1587 queue_t *ill_rq; /* Read queue. */ 1588 queue_t *ill_wq; /* Write queue. */ 1589 1590 int ill_error; /* Error value sent up by device. */ 1591 1592 ipif_t *ill_ipif; /* Interface chain for this ILL. */ 1593 1594 uint_t ill_ipif_up_count; /* Number of IPIFs currently up. */ 1595 uint_t ill_max_frag; /* Max IDU from DLPI. */ 1596 uint_t ill_current_frag; /* Current IDU from DLPI. */ 1597 uint_t ill_mtu; /* User-specified MTU; SIOCSLIFMTU */ 1598 uint_t ill_metric; /* BSD if metric, for compatibility. */ 1599 char *ill_name; /* Our name. */ 1600 uint_t ill_ipif_dup_count; /* Number of duplicate addresses. */ 1601 uint_t ill_name_length; /* Name length, incl. terminator. */ 1602 uint_t ill_net_type; /* IRE_IF_RESOLVER/IRE_IF_NORESOLVER. */ 1603 /* 1604 * Physical Point of Attachment num. If DLPI style 1 provider 1605 * then this is derived from the devname. 1606 */ 1607 uint_t ill_ppa; 1608 t_uscalar_t ill_sap; 1609 t_scalar_t ill_sap_length; /* Including sign (for position) */ 1610 uint_t ill_phys_addr_length; /* Excluding the sap. */ 1611 uint_t ill_bcast_addr_length; /* Only set when the DL provider */ 1612 /* supports broadcast. */ 1613 t_uscalar_t ill_mactype; 1614 uint8_t *ill_frag_ptr; /* Reassembly state. */ 1615 timeout_id_t ill_frag_timer_id; /* timeout id for the frag timer */ 1616 ipfb_t *ill_frag_hash_tbl; /* Fragment hash list head. */ 1617 1618 krwlock_t ill_mcast_lock; /* Protects multicast state */ 1619 kmutex_t ill_mcast_serializer; /* Serialize across ilg and ilm state */ 1620 ilm_t *ill_ilm; /* Multicast membership for ill */ 1621 uint_t ill_global_timer; /* for IGMPv3/MLDv2 general queries */ 1622 int ill_mcast_type; /* type of router which is querier */ 1623 /* on this interface */ 1624 uint16_t ill_mcast_v1_time; /* # slow timeouts since last v1 qry */ 1625 uint16_t ill_mcast_v2_time; /* # slow timeouts since last v2 qry */ 1626 uint8_t ill_mcast_v1_tset; /* 1 => timer is set; 0 => not set */ 1627 uint8_t ill_mcast_v2_tset; /* 1 => timer is set; 0 => not set */ 1628 1629 uint8_t ill_mcast_rv; /* IGMPv3/MLDv2 robustness variable */ 1630 int ill_mcast_qi; /* IGMPv3/MLDv2 query interval var */ 1631 1632 /* 1633 * All non-NULL cells between 'ill_first_mp_to_free' and 1634 * 'ill_last_mp_to_free' are freed in ill_delete. 1635 */ 1636 #define ill_first_mp_to_free ill_bcast_mp 1637 mblk_t *ill_bcast_mp; /* DLPI header for broadcasts. */ 1638 mblk_t *ill_unbind_mp; /* unbind mp from ill_dl_up() */ 1639 mblk_t *ill_promiscoff_mp; /* for ill_leave_allmulti() */ 1640 mblk_t *ill_dlpi_deferred; /* b_next chain of control messages */ 1641 mblk_t *ill_dest_addr_mp; /* mblk which holds ill_dest_addr */ 1642 mblk_t *ill_replumb_mp; /* replumb mp from ill_replumb() */ 1643 mblk_t *ill_phys_addr_mp; /* mblk which holds ill_phys_addr */ 1644 mblk_t *ill_mcast_deferred; /* b_next chain of IGMP/MLD packets */ 1645 #define ill_last_mp_to_free ill_mcast_deferred 1646 1647 cred_t *ill_credp; /* opener's credentials */ 1648 uint8_t *ill_phys_addr; /* ill_phys_addr_mp->b_rptr + off */ 1649 uint8_t *ill_dest_addr; /* ill_dest_addr_mp->b_rptr + off */ 1650 1651 uint_t ill_state_flags; /* see ILL_* flags above */ 1652 1653 /* Following bit fields protected by ipsq_t */ 1654 uint_t 1655 ill_needs_attach : 1, 1656 ill_reserved : 1, 1657 ill_isv6 : 1, 1658 ill_dlpi_style_set : 1, 1659 1660 ill_ifname_pending : 1, 1661 ill_logical_down : 1, 1662 ill_dl_up : 1, 1663 ill_up_ipifs : 1, 1664 1665 ill_note_link : 1, /* supports link-up notification */ 1666 ill_capab_reneg : 1, /* capability renegotiation to be done */ 1667 ill_dld_capab_inprog : 1, /* direct dld capab call in prog */ 1668 ill_need_recover_multicast : 1, 1669 1670 ill_replumbing : 1, 1671 ill_arl_dlpi_pending : 1, 1672 ill_grp_pending : 1, 1673 1674 ill_pad_to_bit_31 : 17; 1675 1676 /* Following bit fields protected by ill_lock */ 1677 uint_t 1678 ill_fragtimer_executing : 1, 1679 ill_fragtimer_needrestart : 1, 1680 ill_manual_token : 1, /* system won't override ill_token */ 1681 /* 1682 * ill_manual_linklocal : system will not change the 1683 * linklocal whenever ill_token changes. 1684 */ 1685 ill_manual_linklocal : 1, 1686 1687 ill_manual_dst_linklocal : 1, /* same for pt-pt dst linklocal */ 1688 1689 ill_pad_bit_31 : 27; 1690 1691 /* 1692 * Used in SIOCSIFMUXID and SIOCGIFMUXID for 'ifconfig unplumb'. 1693 */ 1694 int ill_muxid; /* muxid returned from plink */ 1695 1696 /* Used for IP frag reassembly throttling on a per ILL basis. */ 1697 uint_t ill_ipf_gen; /* Generation of next fragment queue */ 1698 uint_t ill_frag_count; /* Count of all reassembly mblk bytes */ 1699 uint_t ill_frag_free_num_pkts; /* num of fragmented packets to free */ 1700 clock_t ill_last_frag_clean_time; /* time when frag's were pruned */ 1701 int ill_type; /* From <net/if_types.h> */ 1702 uint_t ill_dlpi_multicast_state; /* See below IDS_* */ 1703 uint_t ill_dlpi_fastpath_state; /* See below IDS_* */ 1704 1705 /* 1706 * Capabilities related fields. 1707 */ 1708 uint_t ill_dlpi_capab_state; /* State of capability query, IDCS_* */ 1709 uint_t ill_capab_pending_cnt; 1710 uint64_t ill_capabilities; /* Enabled capabilities, ILL_CAPAB_* */ 1711 ill_hcksum_capab_t *ill_hcksum_capab; /* H/W cksumming capabilities */ 1712 ill_zerocopy_capab_t *ill_zerocopy_capab; /* Zero-copy capabilities */ 1713 ill_dld_capab_t *ill_dld_capab; /* DLD capabilities */ 1714 ill_lso_capab_t *ill_lso_capab; /* Large Segment Offload capabilities */ 1715 mblk_t *ill_capab_reset_mp; /* Preallocated mblk for capab reset */ 1716 1717 uint8_t ill_max_hops; /* Maximum hops for any logical interface */ 1718 uint_t ill_user_mtu; /* User-specified MTU via SIOCSLIFLNKINFO */ 1719 uint32_t ill_reachable_time; /* Value for ND algorithm in msec */ 1720 uint32_t ill_reachable_retrans_time; /* Value for ND algorithm msec */ 1721 uint_t ill_max_buf; /* Max # of req to buffer for ND */ 1722 in6_addr_t ill_token; /* IPv6 interface id */ 1723 in6_addr_t ill_dest_token; /* Destination IPv6 interface id */ 1724 uint_t ill_token_length; 1725 uint32_t ill_xmit_count; /* ndp max multicast xmits */ 1726 mib2_ipIfStatsEntry_t *ill_ip_mib; /* ver indep. interface mib */ 1727 mib2_ipv6IfIcmpEntry_t *ill_icmp6_mib; /* Per interface mib */ 1728 1729 phyint_t *ill_phyint; 1730 uint64_t ill_flags; 1731 1732 kmutex_t ill_lock; /* Please see table below */ 1733 /* 1734 * The ill_nd_lla* fields handle the link layer address option 1735 * from neighbor discovery. This is used for external IPv6 1736 * address resolution. 1737 */ 1738 mblk_t *ill_nd_lla_mp; /* mblk which holds ill_nd_lla */ 1739 uint8_t *ill_nd_lla; /* Link Layer Address */ 1740 uint_t ill_nd_lla_len; /* Link Layer Address length */ 1741 /* 1742 * We have 4 phys_addr_req's sent down. This field keeps track 1743 * of which one is pending. 1744 */ 1745 t_uscalar_t ill_phys_addr_pend; /* which dl_phys_addr_req pending */ 1746 /* 1747 * Used to save errors that occur during plumbing 1748 */ 1749 uint_t ill_ifname_pending_err; 1750 avl_node_t ill_avl_byppa; /* avl node based on ppa */ 1751 list_t ill_nce; /* pointer to nce_s list */ 1752 uint_t ill_refcnt; /* active refcnt by threads */ 1753 uint_t ill_ire_cnt; /* ires associated with this ill */ 1754 kcondvar_t ill_cv; 1755 uint_t ill_ncec_cnt; /* ncecs associated with this ill */ 1756 uint_t ill_nce_cnt; /* nces associated with this ill */ 1757 uint_t ill_waiters; /* threads waiting in ipsq_enter */ 1758 /* 1759 * Contains the upper read queue pointer of the module immediately 1760 * beneath IP. This field allows IP to validate sub-capability 1761 * acknowledgments coming up from downstream. 1762 */ 1763 queue_t *ill_lmod_rq; /* read queue pointer of module below */ 1764 uint_t ill_lmod_cnt; /* number of modules beneath IP */ 1765 ip_m_t *ill_media; /* media specific params/functions */ 1766 t_uscalar_t ill_dlpi_pending; /* Last DLPI primitive issued */ 1767 uint_t ill_usesrc_ifindex; /* use src addr from this ILL */ 1768 struct ill_s *ill_usesrc_grp_next; /* Next ILL in the usesrc group */ 1769 boolean_t ill_trace_disable; /* True when alloc fails */ 1770 zoneid_t ill_zoneid; 1771 ip_stack_t *ill_ipst; /* Corresponds to a netstack_hold */ 1772 uint32_t ill_dhcpinit; /* IP_DHCPINIT_IFs for ill */ 1773 void *ill_flownotify_mh; /* Tx flow ctl, mac cb handle */ 1774 uint_t ill_ilm_cnt; /* ilms referencing this ill */ 1775 uint_t ill_ipallmulti_cnt; /* ip_join_allmulti() calls */ 1776 ilm_t *ill_ipallmulti_ilm; 1777 1778 mblk_t *ill_saved_ire_mp; /* Allocated for each extra IRE */ 1779 /* with ire_ill set so they can */ 1780 /* survive the ill going down and up. */ 1781 kmutex_t ill_saved_ire_lock; /* Protects ill_saved_ire_mp, cnt */ 1782 uint_t ill_saved_ire_cnt; /* # entries */ 1783 struct arl_ill_common_s *ill_common; 1784 ire_t *ill_ire_multicast; /* IRE_MULTICAST for ill */ 1785 clock_t ill_defend_start; /* start of 1 hour period */ 1786 uint_t ill_defend_count; /* # of announce/defends per ill */ 1787 /* 1788 * IPMP fields. 1789 */ 1790 ipmp_illgrp_t *ill_grp; /* IPMP group information */ 1791 list_node_t ill_actnode; /* next active ill in group */ 1792 list_node_t ill_grpnode; /* next ill in group */ 1793 ipif_t *ill_src_ipif; /* source address selection rotor */ 1794 ipif_t *ill_move_ipif; /* ipif awaiting move to new ill */ 1795 boolean_t ill_nom_cast; /* nominated for mcast/bcast */ 1796 uint_t ill_bound_cnt; /* # of data addresses bound to ill */ 1797 ipif_t *ill_bound_ipif; /* ipif chain bound to ill */ 1798 timeout_id_t ill_refresh_tid; /* ill refresh retry timeout id */ 1799 1800 uint32_t ill_mrouter_cnt; /* mrouter allmulti joins */ 1801 } ill_t; 1802 1803 /* 1804 * ILL_FREE_OK() means that there are no incoming pointer references 1805 * to the ill. 1806 */ 1807 #define ILL_FREE_OK(ill) \ 1808 ((ill)->ill_ire_cnt == 0 && (ill)->ill_ilm_cnt == 0 && \ 1809 (ill)->ill_ncec_cnt == 0 && (ill)->ill_nce_cnt == 0) 1810 1811 /* 1812 * An ipif/ill can be marked down only when the ire and ncec references 1813 * to that ipif/ill goes to zero. ILL_DOWN_OK() is a necessary condition 1814 * quiescence checks. See comments above IPIF_DOWN_OK for details 1815 * on why ires and nces are selectively considered for this macro. 1816 */ 1817 #define ILL_DOWN_OK(ill) \ 1818 (ill->ill_ire_cnt == 0 && ill->ill_ncec_cnt == 0 && \ 1819 ill->ill_nce_cnt == 0) 1820 1821 /* 1822 * The following table lists the protection levels of the various members 1823 * of the ill_t. Same notation as that used for ipif_t above is used. 1824 * 1825 * Write Read 1826 * 1827 * ill_ifptr ill_g_lock + s Write once 1828 * ill_rq ipsq Write once 1829 * ill_wq ipsq Write once 1830 * 1831 * ill_error ipsq None 1832 * ill_ipif ill_g_lock + ipsq ill_g_lock OR ipsq 1833 * ill_ipif_up_count ill_lock + ipsq ill_lock OR ipsq 1834 * ill_max_frag ill_lock ill_lock 1835 * ill_current_frag ill_lock ill_lock 1836 * 1837 * ill_name ill_g_lock + ipsq Write once 1838 * ill_name_length ill_g_lock + ipsq Write once 1839 * ill_ndd_name ipsq Write once 1840 * ill_net_type ipsq Write once 1841 * ill_ppa ill_g_lock + ipsq Write once 1842 * ill_sap ipsq + down ill Write once 1843 * ill_sap_length ipsq + down ill Write once 1844 * ill_phys_addr_length ipsq + down ill Write once 1845 * 1846 * ill_bcast_addr_length ipsq ipsq 1847 * ill_mactype ipsq ipsq 1848 * ill_frag_ptr ipsq ipsq 1849 * 1850 * ill_frag_timer_id ill_lock ill_lock 1851 * ill_frag_hash_tbl ipsq up ill 1852 * ill_ilm ill_mcast_lock(WRITER) ill_mcast_lock(READER) 1853 * ill_global_timer ill_mcast_lock(WRITER) ill_mcast_lock(READER) 1854 * ill_mcast_type ill_mcast_lock(WRITER) ill_mcast_lock(READER) 1855 * ill_mcast_v1_time ill_mcast_lock(WRITER) ill_mcast_lock(READER) 1856 * ill_mcast_v2_time ill_mcast_lock(WRITER) ill_mcast_lock(READER) 1857 * ill_mcast_v1_tset ill_mcast_lock(WRITER) ill_mcast_lock(READER) 1858 * ill_mcast_v2_tset ill_mcast_lock(WRITER) ill_mcast_lock(READER) 1859 * ill_mcast_rv ill_mcast_lock(WRITER) ill_mcast_lock(READER) 1860 * ill_mcast_qi ill_mcast_lock(WRITER) ill_mcast_lock(READER) 1861 * 1862 * ill_down_mp ipsq ipsq 1863 * ill_dlpi_deferred ill_lock ill_lock 1864 * ill_dlpi_pending ipsq + ill_lock ipsq or ill_lock or 1865 * absence of ipsq writer. 1866 * ill_phys_addr_mp ipsq + down ill only when ill is up 1867 * ill_mcast_deferred ill_lock ill_lock 1868 * ill_phys_addr ipsq + down ill only when ill is up 1869 * ill_dest_addr_mp ipsq + down ill only when ill is up 1870 * ill_dest_addr ipsq + down ill only when ill is up 1871 * 1872 * ill_state_flags ill_lock ill_lock 1873 * exclusive bit flags ipsq_t ipsq_t 1874 * shared bit flags ill_lock ill_lock 1875 * 1876 * ill_muxid ipsq Not atomic 1877 * 1878 * ill_ipf_gen Not atomic 1879 * ill_frag_count atomics atomics 1880 * ill_type ipsq + down ill only when ill is up 1881 * ill_dlpi_multicast_state ill_lock ill_lock 1882 * ill_dlpi_fastpath_state ill_lock ill_lock 1883 * ill_dlpi_capab_state ipsq ipsq 1884 * ill_max_hops ipsq Not atomic 1885 * 1886 * ill_mtu ill_lock None 1887 * 1888 * ill_user_mtu ipsq + ill_lock ill_lock 1889 * ill_reachable_time ipsq + ill_lock ill_lock 1890 * ill_reachable_retrans_time ipsq + ill_lock ill_lock 1891 * ill_max_buf ipsq + ill_lock ill_lock 1892 * 1893 * Next 2 fields need ill_lock because of the get ioctls. They should not 1894 * report partially updated results without executing in the ipsq. 1895 * ill_token ipsq + ill_lock ill_lock 1896 * ill_token_length ipsq + ill_lock ill_lock 1897 * ill_dest_token ipsq + down ill only when ill is up 1898 * ill_xmit_count ipsq + down ill write once 1899 * ill_ip6_mib ipsq + down ill only when ill is up 1900 * ill_icmp6_mib ipsq + down ill only when ill is up 1901 * 1902 * ill_phyint ipsq, ill_g_lock, ill_lock Any of them 1903 * ill_flags ill_lock ill_lock 1904 * ill_nd_lla_mp ipsq + down ill only when ill is up 1905 * ill_nd_lla ipsq + down ill only when ill is up 1906 * ill_nd_lla_len ipsq + down ill only when ill is up 1907 * ill_phys_addr_pend ipsq + down ill only when ill is up 1908 * ill_ifname_pending_err ipsq ipsq 1909 * ill_avl_byppa ipsq, ill_g_lock write once 1910 * 1911 * ill_fastpath_list ill_lock ill_lock 1912 * ill_refcnt ill_lock ill_lock 1913 * ill_ire_cnt ill_lock ill_lock 1914 * ill_cv ill_lock ill_lock 1915 * ill_ncec_cnt ill_lock ill_lock 1916 * ill_nce_cnt ill_lock ill_lock 1917 * ill_ilm_cnt ill_lock ill_lock 1918 * ill_src_ipif ill_g_lock ill_g_lock 1919 * ill_trace ill_lock ill_lock 1920 * ill_usesrc_grp_next ill_g_usesrc_lock ill_g_usesrc_lock 1921 * ill_dhcpinit atomics atomics 1922 * ill_flownotify_mh write once write once 1923 * ill_capab_pending_cnt ipsq ipsq 1924 * ill_ipallmulti_cnt ill_lock ill_lock 1925 * ill_ipallmulti_ilm ill_lock ill_lock 1926 * ill_saved_ire_mp ill_saved_ire_lock ill_saved_ire_lock 1927 * ill_saved_ire_cnt ill_saved_ire_lock ill_saved_ire_lock 1928 * ill_arl ??? ??? 1929 * ill_ire_multicast ipsq + quiescent none 1930 * ill_bound_ipif ipsq ipsq 1931 * ill_actnode ipsq + ipmp_lock ipsq OR ipmp_lock 1932 * ill_grpnode ipsq + ill_g_lock ipsq OR ill_g_lock 1933 * ill_src_ipif ill_g_lock ill_g_lock 1934 * ill_move_ipif ipsq ipsq 1935 * ill_nom_cast ipsq ipsq OR advisory 1936 * ill_refresh_tid ill_lock ill_lock 1937 * ill_grp (for IPMP ill) write once write once 1938 * ill_grp (for underlying ill) ipsq + ill_g_lock ipsq OR ill_g_lock 1939 * ill_grp_pending ill_mcast_serializer ill_mcast_serializer 1940 * ill_mrouter_cnt atomics atomics 1941 * 1942 * NOTE: It's OK to make heuristic decisions on an underlying interface 1943 * by using IS_UNDER_IPMP() or comparing ill_grp's raw pointer value. 1944 */ 1945 1946 /* 1947 * For ioctl restart mechanism see ip_reprocess_ioctl() 1948 */ 1949 struct ip_ioctl_cmd_s; 1950 1951 typedef int (*ifunc_t)(ipif_t *, struct sockaddr_in *, queue_t *, mblk_t *, 1952 struct ip_ioctl_cmd_s *, void *); 1953 1954 typedef struct ip_ioctl_cmd_s { 1955 int ipi_cmd; 1956 size_t ipi_copyin_size; 1957 uint_t ipi_flags; 1958 uint_t ipi_cmd_type; 1959 ifunc_t ipi_func; 1960 ifunc_t ipi_func_restart; 1961 } ip_ioctl_cmd_t; 1962 1963 /* 1964 * ipi_cmd_type: 1965 * 1966 * IF_CMD 1 old style ifreq cmd 1967 * LIF_CMD 2 new style lifreq cmd 1968 * ARP_CMD 3 arpreq cmd 1969 * XARP_CMD 4 xarpreq cmd 1970 * MSFILT_CMD 5 multicast source filter cmd 1971 * MISC_CMD 6 misc cmd (not a more specific one above) 1972 */ 1973 1974 enum { IF_CMD = 1, LIF_CMD, ARP_CMD, XARP_CMD, MSFILT_CMD, MISC_CMD }; 1975 1976 #define IPI_DONTCARE 0 /* For ioctl encoded values that don't matter */ 1977 1978 /* Flag values in ipi_flags */ 1979 #define IPI_PRIV 0x1 /* Root only command */ 1980 #define IPI_MODOK 0x2 /* Permitted on mod instance of IP */ 1981 #define IPI_WR 0x4 /* Need to grab writer access */ 1982 #define IPI_GET_CMD 0x8 /* branch to mi_copyout on success */ 1983 /* unused 0x10 */ 1984 #define IPI_NULL_BCONT 0x20 /* ioctl has not data and hence no b_cont */ 1985 1986 extern ip_ioctl_cmd_t ip_ndx_ioctl_table[]; 1987 extern ip_ioctl_cmd_t ip_misc_ioctl_table[]; 1988 extern int ip_ndx_ioctl_count; 1989 extern int ip_misc_ioctl_count; 1990 1991 /* Passed down by ARP to IP during I_PLINK/I_PUNLINK */ 1992 typedef struct ipmx_s { 1993 char ipmx_name[LIFNAMSIZ]; /* if name */ 1994 uint_t 1995 ipmx_arpdev_stream : 1, /* This is the arp stream */ 1996 ipmx_notused : 31; 1997 } ipmx_t; 1998 1999 /* 2000 * State for detecting if a driver supports certain features. 2001 * Support for DL_ENABMULTI_REQ uses ill_dlpi_multicast_state. 2002 * Support for DLPI M_DATA fastpath uses ill_dlpi_fastpath_state. 2003 */ 2004 #define IDS_UNKNOWN 0 /* No DLPI request sent */ 2005 #define IDS_INPROGRESS 1 /* DLPI request sent */ 2006 #define IDS_OK 2 /* DLPI request completed successfully */ 2007 #define IDS_FAILED 3 /* DLPI request failed */ 2008 2009 /* Support for DL_CAPABILITY_REQ uses ill_dlpi_capab_state. */ 2010 enum { 2011 IDCS_UNKNOWN, 2012 IDCS_PROBE_SENT, 2013 IDCS_OK, 2014 IDCS_RESET_SENT, 2015 IDCS_RENEG, 2016 IDCS_FAILED 2017 }; 2018 2019 /* Extended NDP Management Structure */ 2020 typedef struct ipndp_s { 2021 ndgetf_t ip_ndp_getf; 2022 ndsetf_t ip_ndp_setf; 2023 caddr_t ip_ndp_data; 2024 char *ip_ndp_name; 2025 } ipndp_t; 2026 2027 /* IXA Notification types */ 2028 typedef enum { 2029 IXAN_LSO, /* LSO capability change */ 2030 IXAN_PMTU, /* PMTU change */ 2031 IXAN_ZCOPY /* ZEROCOPY capability change */ 2032 } ixa_notify_type_t; 2033 2034 typedef uint_t ixa_notify_arg_t; 2035 2036 typedef void (*ixa_notify_t)(void *, ip_xmit_attr_t *ixa, ixa_notify_type_t, 2037 ixa_notify_arg_t); 2038 2039 /* 2040 * Attribute flags that are common to the transmit and receive attributes 2041 */ 2042 #define IAF_IS_IPV4 0x80000000 /* ipsec_*_v4 */ 2043 #define IAF_TRUSTED_ICMP 0x40000000 /* ipsec_*_icmp_loopback */ 2044 #define IAF_NO_LOOP_ZONEID_SET 0x20000000 /* Zone that shouldn't have */ 2045 /* a copy */ 2046 #define IAF_LOOPBACK_COPY 0x10000000 /* For multi and broadcast */ 2047 2048 #define IAF_MASK 0xf0000000 /* Flags that are common */ 2049 2050 /* 2051 * Transmit side attributes used between the transport protocols and IP as 2052 * well as inside IP. It is also used to cache information in the conn_t i.e. 2053 * replaces conn_ire and the IPsec caching in the conn_t. 2054 */ 2055 struct ip_xmit_attr_s { 2056 iaflags_t ixa_flags; /* IXAF_*. See below */ 2057 2058 uint32_t ixa_free_flags; /* IXA_FREE_*. See below */ 2059 uint32_t ixa_refcnt; /* Using atomics */ 2060 2061 /* 2062 * Always initialized independently of ixa_flags settings. 2063 * Used by ip_xmit so we keep them up front for cache locality. 2064 */ 2065 uint32_t ixa_xmit_hint; /* For ECMP and GLD TX ring fanout */ 2066 uint_t ixa_pktlen; /* Always set. For frag and stats */ 2067 zoneid_t ixa_zoneid; /* Assumed always set */ 2068 2069 /* Always set for conn_ip_output(); might be stale */ 2070 /* 2071 * Since TCP keeps the conn_t around past the process going away 2072 * we need to use the "notr" (e.g, ire_refhold_notr) for ixa_ire, 2073 * ixa_nce, and ixa_dce. 2074 */ 2075 ire_t *ixa_ire; /* Forwarding table entry */ 2076 uint_t ixa_ire_generation; 2077 nce_t *ixa_nce; /* Neighbor cache entry */ 2078 dce_t *ixa_dce; /* Destination cache entry */ 2079 uint_t ixa_dce_generation; 2080 uint_t ixa_src_generation; /* If IXAF_VERIFY_SOURCE */ 2081 2082 uint32_t ixa_src_preferences; /* prefs for src addr select */ 2083 uint32_t ixa_pmtu; /* IXAF_VERIFY_PMTU */ 2084 2085 /* Set by ULP if IXAF_VERIFY_PMTU; otherwise set by IP */ 2086 uint32_t ixa_fragsize; 2087 2088 int8_t ixa_use_min_mtu; /* IXAF_USE_MIN_MTU values */ 2089 2090 pfirepostfrag_t ixa_postfragfn; /* Set internally in IP */ 2091 2092 in6_addr_t ixa_nexthop_v6; /* IXAF_NEXTHOP_SET */ 2093 #define ixa_nexthop_v4 V4_PART_OF_V6(ixa_nexthop_v6) 2094 2095 zoneid_t ixa_no_loop_zoneid; /* IXAF_NO_LOOP_ZONEID_SET */ 2096 2097 uint_t ixa_scopeid; /* For IPv6 link-locals */ 2098 2099 uint_t ixa_broadcast_ttl; /* IXAF_BROACAST_TTL_SET */ 2100 2101 uint_t ixa_multicast_ttl; /* Assumed set for multicast */ 2102 uint_t ixa_multicast_ifindex; /* Assumed set for multicast */ 2103 ipaddr_t ixa_multicast_ifaddr; /* Assumed set for multicast */ 2104 2105 int ixa_raw_cksum_offset; /* If IXAF_SET_RAW_CKSUM */ 2106 2107 uint32_t ixa_ident; /* For IPv6 fragment header */ 2108 2109 /* 2110 * Cached LSO information. 2111 */ 2112 ill_lso_capab_t ixa_lso_capab; /* Valid when IXAF_LSO_CAPAB */ 2113 2114 uint64_t ixa_ipsec_policy_gen; /* Generation from iph_gen */ 2115 /* 2116 * The following IPsec fields are only initialized when 2117 * IXAF_IPSEC_SECURE is set. Otherwise they contain garbage. 2118 */ 2119 ipsec_latch_t *ixa_ipsec_latch; /* Just the ids */ 2120 struct ipsa_s *ixa_ipsec_ah_sa; /* Hard reference SA for AH */ 2121 struct ipsa_s *ixa_ipsec_esp_sa; /* Hard reference SA for ESP */ 2122 struct ipsec_policy_s *ixa_ipsec_policy; /* why are we here? */ 2123 struct ipsec_action_s *ixa_ipsec_action; /* For reflected packets */ 2124 ipsa_ref_t ixa_ipsec_ref[2]; /* Soft reference to SA */ 2125 /* 0: ESP, 1: AH */ 2126 2127 /* 2128 * The selectors here are potentially different than the SPD rule's 2129 * selectors, and we need to have both available for IKEv2. 2130 * 2131 * NOTE: "Source" and "Dest" are w.r.t. outbound datagrams. Ports can 2132 * be zero, and the protocol number is needed to make the ports 2133 * significant. 2134 */ 2135 uint16_t ixa_ipsec_src_port; /* Source port number of d-gram. */ 2136 uint16_t ixa_ipsec_dst_port; /* Destination port number of d-gram. */ 2137 uint8_t ixa_ipsec_icmp_type; /* ICMP type of d-gram */ 2138 uint8_t ixa_ipsec_icmp_code; /* ICMP code of d-gram */ 2139 2140 sa_family_t ixa_ipsec_inaf; /* Inner address family */ 2141 #define IXA_MAX_ADDRLEN 4 /* Max addr len. (in 32-bit words) */ 2142 uint32_t ixa_ipsec_insrc[IXA_MAX_ADDRLEN]; /* Inner src address */ 2143 uint32_t ixa_ipsec_indst[IXA_MAX_ADDRLEN]; /* Inner dest address */ 2144 uint8_t ixa_ipsec_insrcpfx; /* Inner source prefix */ 2145 uint8_t ixa_ipsec_indstpfx; /* Inner destination prefix */ 2146 2147 uint8_t ixa_ipsec_proto; /* IP protocol number for d-gram. */ 2148 2149 /* Always initialized independently of ixa_flags settings */ 2150 uint_t ixa_ifindex; /* Assumed always set */ 2151 uint16_t ixa_ip_hdr_length; /* Points to ULP header */ 2152 uint8_t ixa_protocol; /* Protocol number for ULP cksum */ 2153 ts_label_t *ixa_tsl; /* Always set. NULL if not TX */ 2154 ip_stack_t *ixa_ipst; /* Always set */ 2155 uint32_t ixa_extra_ident; /* Set if LSO */ 2156 cred_t *ixa_cred; /* For getpeerucred */ 2157 pid_t ixa_cpid; /* For getpeerucred */ 2158 2159 #ifdef DEBUG 2160 kthread_t *ixa_curthread; /* For serialization assert */ 2161 #endif 2162 squeue_t *ixa_sqp; /* Set from conn_sqp as a hint */ 2163 uintptr_t ixa_cookie; /* cookie to use for tx flow control */ 2164 2165 /* 2166 * Must be set by ULP if any of IXAF_VERIFY_LSO, IXAF_VERIFY_PMTU, 2167 * or IXAF_VERIFY_ZCOPY is set. 2168 */ 2169 ixa_notify_t ixa_notify; /* Registered upcall notify function */ 2170 void *ixa_notify_cookie; /* ULP cookie for ixa_notify */ 2171 }; 2172 2173 /* 2174 * Flags to indicate which transmit attributes are set. 2175 * Split into "xxx_SET" ones which indicate that the "xxx" field it set, and 2176 * single flags. 2177 */ 2178 #define IXAF_REACH_CONF 0x00000001 /* Reachability confirmation */ 2179 #define IXAF_BROADCAST_TTL_SET 0x00000002 /* ixa_broadcast_ttl valid */ 2180 #define IXAF_SET_SOURCE 0x00000004 /* Replace if broadcast */ 2181 #define IXAF_USE_MIN_MTU 0x00000008 /* IPV6_USE_MIN_MTU */ 2182 2183 #define IXAF_DONTFRAG 0x00000010 /* IP*_DONTFRAG */ 2184 #define IXAF_VERIFY_PMTU 0x00000020 /* ixa_pmtu/ixa_fragsize set */ 2185 #define IXAF_PMTU_DISCOVERY 0x00000040 /* Create/use PMTU state */ 2186 #define IXAF_MULTICAST_LOOP 0x00000080 /* IP_MULTICAST_LOOP */ 2187 2188 #define IXAF_IPSEC_SECURE 0x00000100 /* Need IPsec processing */ 2189 #define IXAF_UCRED_TSL 0x00000200 /* ixa_tsl from SCM_UCRED */ 2190 #define IXAF_DONTROUTE 0x00000400 /* SO_DONTROUTE */ 2191 #define IXAF_NO_IPSEC 0x00000800 /* Ignore policy */ 2192 2193 #define IXAF_PMTU_TOO_SMALL 0x00001000 /* PMTU too small */ 2194 #define IXAF_SET_ULP_CKSUM 0x00002000 /* Calculate ULP checksum */ 2195 #define IXAF_VERIFY_SOURCE 0x00004000 /* Check that source is ok */ 2196 #define IXAF_NEXTHOP_SET 0x00008000 /* ixa_nexthop set */ 2197 2198 #define IXAF_PMTU_IPV4_DF 0x00010000 /* Set IPv4 DF */ 2199 #define IXAF_NO_DEV_FLOW_CTL 0x00020000 /* Protocol needs no flow ctl */ 2200 #define IXAF_NO_TTL_CHANGE 0x00040000 /* Internal to IP */ 2201 #define IXAF_IPV6_ADD_FRAGHDR 0x00080000 /* Add fragment header */ 2202 2203 #define IXAF_IPSEC_TUNNEL 0x00100000 /* Tunnel mode */ 2204 #define IXAF_NO_PFHOOK 0x00200000 /* Skip xmit pfhook */ 2205 #define IXAF_NO_TRACE 0x00400000 /* When back from ARP/ND */ 2206 #define IXAF_SCOPEID_SET 0x00800000 /* ixa_scopeid set */ 2207 2208 #define IXAF_MULTIRT_MULTICAST 0x01000000 /* MULTIRT for multicast */ 2209 #define IXAF_NO_HW_CKSUM 0x02000000 /* Force software cksum */ 2210 #define IXAF_SET_RAW_CKSUM 0x04000000 /* Use ixa_raw_cksum_offset */ 2211 #define IXAF_IPSEC_GLOBAL_POLICY 0x08000000 /* Policy came from global */ 2212 2213 /* Note the following uses bits 0x10000000 through 0x80000000 */ 2214 #define IXAF_IS_IPV4 IAF_IS_IPV4 2215 #define IXAF_TRUSTED_ICMP IAF_TRUSTED_ICMP 2216 #define IXAF_NO_LOOP_ZONEID_SET IAF_NO_LOOP_ZONEID_SET 2217 #define IXAF_LOOPBACK_COPY IAF_LOOPBACK_COPY 2218 2219 /* Note: use the upper 32 bits */ 2220 #define IXAF_VERIFY_LSO 0x100000000 /* Check LSO capability */ 2221 #define IXAF_LSO_CAPAB 0x200000000 /* Capable of LSO */ 2222 #define IXAF_VERIFY_ZCOPY 0x400000000 /* Check Zero Copy capability */ 2223 #define IXAF_ZCOPY_CAPAB 0x800000000 /* Capable of ZEROCOPY */ 2224 2225 /* 2226 * The normal flags for sending packets e.g., icmp errors 2227 */ 2228 #define IXAF_BASIC_SIMPLE_V4 \ 2229 (IXAF_SET_ULP_CKSUM | IXAF_IS_IPV4 | IXAF_VERIFY_SOURCE) 2230 #define IXAF_BASIC_SIMPLE_V6 (IXAF_SET_ULP_CKSUM | IXAF_VERIFY_SOURCE) 2231 2232 /* 2233 * Normally these fields do not have a hold. But in some cases they do, for 2234 * instance when we've gone through ip_*_attr_to/from_mblk. 2235 * We use ixa_free_flags to indicate that they have a hold and need to be 2236 * released on cleanup. 2237 */ 2238 #define IXA_FREE_CRED 0x00000001 /* ixa_cred needs to be rele */ 2239 #define IXA_FREE_TSL 0x00000002 /* ixa_tsl needs to be rele */ 2240 2241 /* 2242 * Simplistic way to set the ixa_xmit_hint for locally generated traffic 2243 * and forwarded traffic. The shift amount are based on the size of the 2244 * structs to discard the low order bits which don't have much if any variation 2245 * (coloring in kmem_cache_alloc might provide some variation). 2246 * 2247 * Basing the locally generated hint on the address of the conn_t means that 2248 * the packets from the same socket/connection do not get reordered. 2249 * Basing the hint for forwarded traffic on the ill_ring_t means that 2250 * packets from the same NIC+ring are likely to use the same outbound ring 2251 * hence we get low contention on the ring in the transmitting driver. 2252 */ 2253 #define CONN_TO_XMIT_HINT(connp) ((uint32_t)(((uintptr_t)connp) >> 11)) 2254 #define ILL_RING_TO_XMIT_HINT(ring) ((uint32_t)(((uintptr_t)ring) >> 7)) 2255 2256 /* 2257 * IP set Destination Flags used by function ip_set_destination, 2258 * ip_attr_connect, and conn_connect. 2259 */ 2260 #define IPDF_ALLOW_MCBC 0x1 /* Allow multi/broadcast */ 2261 #define IPDF_VERIFY_DST 0x2 /* Verify destination addr */ 2262 #define IPDF_SELECT_SRC 0x4 /* Select source address */ 2263 #define IPDF_LSO 0x8 /* Try LSO */ 2264 #define IPDF_IPSEC 0x10 /* Set IPsec policy */ 2265 #define IPDF_ZONE_IS_GLOBAL 0x20 /* From conn_zone_is_global */ 2266 #define IPDF_ZCOPY 0x40 /* Try ZEROCOPY */ 2267 #define IPDF_UNIQUE_DCE 0x80 /* Get a per-destination DCE */ 2268 2269 /* 2270 * Receive side attributes used between the transport protocols and IP as 2271 * well as inside IP. 2272 */ 2273 struct ip_recv_attr_s { 2274 iaflags_t ira_flags; /* See below */ 2275 2276 uint32_t ira_free_flags; /* IRA_FREE_*. See below */ 2277 2278 /* 2279 * This is a hint for TCP SYN packets. 2280 * Always initialized independently of ira_flags settings 2281 */ 2282 squeue_t *ira_sqp; 2283 ill_rx_ring_t *ira_ring; /* Internal to IP */ 2284 2285 /* For ip_accept_tcp when IRAF_TARGET_SQP is set */ 2286 squeue_t *ira_target_sqp; 2287 mblk_t *ira_target_sqp_mp; 2288 2289 /* Always initialized independently of ira_flags settings */ 2290 uint32_t ira_xmit_hint; /* For ECMP and GLD TX ring fanout */ 2291 zoneid_t ira_zoneid; /* ALL_ZONES unless local delivery */ 2292 uint_t ira_pktlen; /* Always set. For frag and stats */ 2293 uint16_t ira_ip_hdr_length; /* Points to ULP header */ 2294 uint8_t ira_protocol; /* Protocol number for ULP cksum */ 2295 uint_t ira_rifindex; /* Received ifindex */ 2296 uint_t ira_ruifindex; /* Received upper ifindex */ 2297 ts_label_t *ira_tsl; /* Always set. NULL if not TX */ 2298 /* 2299 * ira_rill and ira_ill is set inside IP, but not when conn_recv is 2300 * called; ULPs should use ira_ruifindex instead. 2301 */ 2302 ill_t *ira_rill; /* ill where packet came */ 2303 ill_t *ira_ill; /* ill where IP address hosted */ 2304 cred_t *ira_cred; /* For getpeerucred */ 2305 pid_t ira_cpid; /* For getpeerucred */ 2306 2307 /* Used when IRAF_VERIFIED_SRC is set; this source was ok */ 2308 ipaddr_t ira_verified_src; 2309 2310 /* 2311 * The following IPsec fields are only initialized when 2312 * IRAF_IPSEC_SECURE is set. Otherwise they contain garbage. 2313 */ 2314 struct ipsec_action_s *ira_ipsec_action; /* how we made it in.. */ 2315 struct ipsa_s *ira_ipsec_ah_sa; /* SA for AH */ 2316 struct ipsa_s *ira_ipsec_esp_sa; /* SA for ESP */ 2317 2318 ipaddr_t ira_mroute_tunnel; /* IRAF_MROUTE_TUNNEL_SET */ 2319 2320 zoneid_t ira_no_loop_zoneid; /* IRAF_NO_LOOP_ZONEID_SET */ 2321 2322 uint32_t ira_esp_udp_ports; /* IRAF_ESP_UDP_PORTS */ 2323 2324 /* 2325 * For IP_RECVSLLA and ip_ndp_conflict/find_solicitation. 2326 * Same size as max for sockaddr_dl 2327 */ 2328 #define IRA_L2SRC_SIZE 244 2329 uint8_t ira_l2src[IRA_L2SRC_SIZE]; /* If IRAF_L2SRC_SET */ 2330 2331 /* 2332 * Local handle that we use to do lazy setting of ira_l2src. 2333 * We defer setting l2src until needed but we do before any 2334 * ip_input pullupmsg or copymsg. 2335 */ 2336 struct mac_header_info_s *ira_mhip; /* Could be NULL */ 2337 }; 2338 2339 /* 2340 * Flags to indicate which receive attributes are set. 2341 */ 2342 #define IRAF_SYSTEM_LABELED 0x00000001 /* is_system_labeled() */ 2343 #define IRAF_IPV4_OPTIONS 0x00000002 /* Performance */ 2344 #define IRAF_MULTICAST 0x00000004 /* Was multicast at L3 */ 2345 #define IRAF_BROADCAST 0x00000008 /* Was broadcast at L3 */ 2346 #define IRAF_MULTIBROADCAST (IRAF_MULTICAST|IRAF_BROADCAST) 2347 2348 #define IRAF_LOOPBACK 0x00000010 /* Looped back by IP */ 2349 #define IRAF_VERIFY_IP_CKSUM 0x00000020 /* Need to verify IP */ 2350 #define IRAF_VERIFY_ULP_CKSUM 0x00000040 /* Need to verify TCP,UDP,etc */ 2351 #define IRAF_SCTP_CSUM_ERR 0x00000080 /* sctp pkt has failed chksum */ 2352 2353 #define IRAF_IPSEC_SECURE 0x00000100 /* Passed AH and/or ESP */ 2354 #define IRAF_DHCP_UNICAST 0x00000200 2355 #define IRAF_IPSEC_DECAPS 0x00000400 /* Was packet decapsulated */ 2356 /* from a matching inner packet? */ 2357 #define IRAF_TARGET_SQP 0x00000800 /* ira_target_sqp is set */ 2358 #define IRAF_VERIFIED_SRC 0x00001000 /* ira_verified_src set */ 2359 #define IRAF_RSVP 0x00002000 /* RSVP packet for rsvpd */ 2360 #define IRAF_MROUTE_TUNNEL_SET 0x00004000 /* From ip_mroute_decap */ 2361 #define IRAF_PIM_REGISTER 0x00008000 /* From register_mforward */ 2362 2363 #define IRAF_TX_MAC_EXEMPTABLE 0x00010000 /* Allow MAC_EXEMPT readdown */ 2364 #define IRAF_TX_SHARED_ADDR 0x00020000 /* Arrived on ALL_ZONES addr */ 2365 #define IRAF_ESP_UDP_PORTS 0x00040000 /* NAT-traversal packet */ 2366 #define IRAF_NO_HW_CKSUM 0x00080000 /* Force software cksum */ 2367 2368 #define IRAF_ICMP_ERROR 0x00100000 /* Send to conn_recvicmp */ 2369 #define IRAF_ROUTER_ALERT 0x00200000 /* IPv6 router alert */ 2370 #define IRAF_L2SRC_SET 0x00400000 /* ira_l2src has been set */ 2371 #define IRAF_L2SRC_LOOPBACK 0x00800000 /* Came from us */ 2372 2373 #define IRAF_L2DST_MULTICAST 0x01000000 /* Multicast at L2 */ 2374 #define IRAF_L2DST_BROADCAST 0x02000000 /* Broadcast at L2 */ 2375 /* Unused 0x04000000 */ 2376 /* Unused 0x08000000 */ 2377 2378 /* Below starts with 0x10000000 */ 2379 #define IRAF_IS_IPV4 IAF_IS_IPV4 2380 #define IRAF_TRUSTED_ICMP IAF_TRUSTED_ICMP 2381 #define IRAF_NO_LOOP_ZONEID_SET IAF_NO_LOOP_ZONEID_SET 2382 #define IRAF_LOOPBACK_COPY IAF_LOOPBACK_COPY 2383 2384 /* 2385 * Normally these fields do not have a hold. But in some cases they do, for 2386 * instance when we've gone through ip_*_attr_to/from_mblk. 2387 * We use ira_free_flags to indicate that they have a hold and need to be 2388 * released on cleanup. 2389 */ 2390 #define IRA_FREE_CRED 0x00000001 /* ira_cred needs to be rele */ 2391 #define IRA_FREE_TSL 0x00000002 /* ira_tsl needs to be rele */ 2392 2393 /* 2394 * Optional destination cache entry for path MTU information, 2395 * and ULP metrics. 2396 */ 2397 struct dce_s { 2398 uint_t dce_generation; /* Changed since cached? */ 2399 uint_t dce_flags; /* See below */ 2400 uint_t dce_ipversion; /* IPv4/IPv6 version */ 2401 uint32_t dce_pmtu; /* Path MTU if DCEF_PMTU */ 2402 uint32_t dce_ident; /* Per destination IP ident. */ 2403 iulp_t dce_uinfo; /* Metrics if DCEF_UINFO */ 2404 2405 struct dce_s *dce_next; 2406 struct dce_s **dce_ptpn; 2407 struct dcb_s *dce_bucket; 2408 2409 union { 2410 in6_addr_t dceu_v6addr; 2411 ipaddr_t dceu_v4addr; 2412 } dce_u; 2413 #define dce_v4addr dce_u.dceu_v4addr 2414 #define dce_v6addr dce_u.dceu_v6addr 2415 /* Note that for IPv6+IPMP we use the ifindex for the upper interface */ 2416 uint_t dce_ifindex; /* For IPv6 link-locals */ 2417 2418 kmutex_t dce_lock; 2419 uint_t dce_refcnt; 2420 uint64_t dce_last_change_time; /* Path MTU. In seconds */ 2421 2422 ip_stack_t *dce_ipst; /* Does not have a netstack_hold */ 2423 }; 2424 2425 /* 2426 * Values for dce_generation. 2427 * 2428 * If a DCE has DCE_GENERATION_CONDEMNED, the last dce_refrele should delete 2429 * it. 2430 * 2431 * DCE_GENERATION_VERIFY is never stored in dce_generation but it is 2432 * stored in places that cache DCE (such as ixa_dce_generation). 2433 * It is used as a signal that the cache is stale and needs to be reverified. 2434 */ 2435 #define DCE_GENERATION_CONDEMNED 0 2436 #define DCE_GENERATION_VERIFY 1 2437 #define DCE_GENERATION_INITIAL 2 2438 #define DCE_IS_CONDEMNED(dce) \ 2439 ((dce)->dce_generation == DCE_GENERATION_CONDEMNED) 2440 2441 2442 /* 2443 * Values for ips_src_generation. 2444 * 2445 * SRC_GENERATION_VERIFY is never stored in ips_src_generation but it is 2446 * stored in places that cache IREs (ixa_src_generation). It is used as a 2447 * signal that the cache is stale and needs to be reverified. 2448 */ 2449 #define SRC_GENERATION_VERIFY 0 2450 #define SRC_GENERATION_INITIAL 1 2451 2452 /* 2453 * The kernel stores security attributes of all gateways in a database made 2454 * up of one or more tsol_gcdb_t elements. Each tsol_gcdb_t contains the 2455 * security-related credentials of the gateway. More than one gateways may 2456 * share entries in the database. 2457 * 2458 * The tsol_gc_t structure represents the gateway to credential association, 2459 * and refers to an entry in the database. One or more tsol_gc_t entities are 2460 * grouped together to form one or more tsol_gcgrp_t, each representing the 2461 * list of security attributes specific to the gateway. A gateway may be 2462 * associated with at most one credentials group. 2463 */ 2464 struct tsol_gcgrp_s; 2465 2466 extern uchar_t ip6opt_ls; /* TX IPv6 enabler */ 2467 2468 /* 2469 * Gateway security credential record. 2470 */ 2471 typedef struct tsol_gcdb_s { 2472 uint_t gcdb_refcnt; /* reference count */ 2473 struct rtsa_s gcdb_attr; /* security attributes */ 2474 #define gcdb_mask gcdb_attr.rtsa_mask 2475 #define gcdb_doi gcdb_attr.rtsa_doi 2476 #define gcdb_slrange gcdb_attr.rtsa_slrange 2477 } tsol_gcdb_t; 2478 2479 /* 2480 * Gateway to credential association. 2481 */ 2482 typedef struct tsol_gc_s { 2483 uint_t gc_refcnt; /* reference count */ 2484 struct tsol_gcgrp_s *gc_grp; /* pointer to group */ 2485 struct tsol_gc_s *gc_prev; /* previous in list */ 2486 struct tsol_gc_s *gc_next; /* next in list */ 2487 tsol_gcdb_t *gc_db; /* pointer to actual credentials */ 2488 } tsol_gc_t; 2489 2490 /* 2491 * Gateway credentials group address. 2492 */ 2493 typedef struct tsol_gcgrp_addr_s { 2494 int ga_af; /* address family */ 2495 in6_addr_t ga_addr; /* IPv4 mapped or IPv6 address */ 2496 } tsol_gcgrp_addr_t; 2497 2498 /* 2499 * Gateway credentials group. 2500 */ 2501 typedef struct tsol_gcgrp_s { 2502 uint_t gcgrp_refcnt; /* reference count */ 2503 krwlock_t gcgrp_rwlock; /* lock to protect following */ 2504 uint_t gcgrp_count; /* number of credentials */ 2505 tsol_gc_t *gcgrp_head; /* first credential in list */ 2506 tsol_gc_t *gcgrp_tail; /* last credential in list */ 2507 tsol_gcgrp_addr_t gcgrp_addr; /* next-hop gateway address */ 2508 } tsol_gcgrp_t; 2509 2510 extern kmutex_t gcgrp_lock; 2511 2512 #define GC_REFRELE(p) { \ 2513 ASSERT((p)->gc_grp != NULL); \ 2514 rw_enter(&(p)->gc_grp->gcgrp_rwlock, RW_WRITER); \ 2515 ASSERT((p)->gc_refcnt > 0); \ 2516 if (--((p)->gc_refcnt) == 0) \ 2517 gc_inactive(p); \ 2518 else \ 2519 rw_exit(&(p)->gc_grp->gcgrp_rwlock); \ 2520 } 2521 2522 #define GCGRP_REFHOLD(p) { \ 2523 mutex_enter(&gcgrp_lock); \ 2524 ++((p)->gcgrp_refcnt); \ 2525 ASSERT((p)->gcgrp_refcnt != 0); \ 2526 mutex_exit(&gcgrp_lock); \ 2527 } 2528 2529 #define GCGRP_REFRELE(p) { \ 2530 mutex_enter(&gcgrp_lock); \ 2531 ASSERT((p)->gcgrp_refcnt > 0); \ 2532 if (--((p)->gcgrp_refcnt) == 0) \ 2533 gcgrp_inactive(p); \ 2534 ASSERT(MUTEX_HELD(&gcgrp_lock)); \ 2535 mutex_exit(&gcgrp_lock); \ 2536 } 2537 2538 /* 2539 * IRE gateway security attributes structure, pointed to by tsol_ire_gw_secattr 2540 */ 2541 struct tsol_tnrhc; 2542 2543 struct tsol_ire_gw_secattr_s { 2544 kmutex_t igsa_lock; /* lock to protect following */ 2545 struct tsol_tnrhc *igsa_rhc; /* host entry for gateway */ 2546 tsol_gc_t *igsa_gc; /* for prefix IREs */ 2547 }; 2548 2549 void irb_refrele_ftable(irb_t *); 2550 2551 extern struct kmem_cache *rt_entry_cache; 2552 2553 typedef struct ire4 { 2554 ipaddr_t ire4_mask; /* Mask for matching this IRE. */ 2555 ipaddr_t ire4_addr; /* Address this IRE represents. */ 2556 ipaddr_t ire4_gateway_addr; /* Gateway including for IRE_ONLINK */ 2557 ipaddr_t ire4_setsrc_addr; /* RTF_SETSRC */ 2558 } ire4_t; 2559 2560 typedef struct ire6 { 2561 in6_addr_t ire6_mask; /* Mask for matching this IRE. */ 2562 in6_addr_t ire6_addr; /* Address this IRE represents. */ 2563 in6_addr_t ire6_gateway_addr; /* Gateway including for IRE_ONLINK */ 2564 in6_addr_t ire6_setsrc_addr; /* RTF_SETSRC */ 2565 } ire6_t; 2566 2567 typedef union ire_addr { 2568 ire6_t ire6_u; 2569 ire4_t ire4_u; 2570 } ire_addr_u_t; 2571 2572 /* 2573 * Internet Routing Entry 2574 * When we have multiple identical IREs we logically add them by manipulating 2575 * ire_identical_ref and ire_delete first decrements 2576 * that and when it reaches 1 we know it is the last IRE. 2577 * "identical" is defined as being the same for: 2578 * ire_addr, ire_netmask, ire_gateway, ire_ill, ire_zoneid, and ire_type 2579 * For instance, multiple IRE_BROADCASTs for the same subnet number are 2580 * viewed as identical, and so are the IRE_INTERFACEs when there are 2581 * multiple logical interfaces (on the same ill) with the same subnet prefix. 2582 */ 2583 struct ire_s { 2584 struct ire_s *ire_next; /* The hash chain must be first. */ 2585 struct ire_s **ire_ptpn; /* Pointer to previous next. */ 2586 uint32_t ire_refcnt; /* Number of references */ 2587 ill_t *ire_ill; 2588 uint32_t ire_identical_ref; /* IRE_INTERFACE, IRE_BROADCAST */ 2589 uchar_t ire_ipversion; /* IPv4/IPv6 version */ 2590 ushort_t ire_type; /* Type of IRE */ 2591 uint_t ire_generation; /* Generation including CONDEMNED */ 2592 uint_t ire_ib_pkt_count; /* Inbound packets for ire_addr */ 2593 uint_t ire_ob_pkt_count; /* Outbound packets to ire_addr */ 2594 time_t ire_create_time; /* Time (in secs) IRE was created. */ 2595 uint32_t ire_flags; /* flags related to route (RTF_*) */ 2596 /* 2597 * ire_testhidden is TRUE for INTERFACE IREs of IS_UNDER_IPMP(ill) 2598 * interfaces 2599 */ 2600 boolean_t ire_testhidden; 2601 pfirerecv_t ire_recvfn; /* Receive side handling */ 2602 pfiresend_t ire_sendfn; /* Send side handling */ 2603 pfirepostfrag_t ire_postfragfn; /* Bottom end of send handling */ 2604 2605 uint_t ire_masklen; /* # bits in ire_mask{,_v6} */ 2606 ire_addr_u_t ire_u; /* IPv4/IPv6 address info. */ 2607 2608 irb_t *ire_bucket; /* Hash bucket when ire_ptphn is set */ 2609 kmutex_t ire_lock; 2610 clock_t ire_last_used_time; /* For IRE_LOCAL reception */ 2611 tsol_ire_gw_secattr_t *ire_gw_secattr; /* gateway security attributes */ 2612 zoneid_t ire_zoneid; 2613 2614 /* 2615 * Cached information of where to send packets that match this route. 2616 * The ire_dep_* information is used to determine when ire_nce_cache 2617 * needs to be updated. 2618 * ire_nce_cache is the fastpath for the Neighbor Cache Entry 2619 * for IPv6; arp info for IPv4 2620 * Since this is a cache setup and torn down independently of 2621 * applications we need to use nce_ref{rele,hold}_notr for it. 2622 */ 2623 nce_t *ire_nce_cache; 2624 2625 /* 2626 * Quick check whether the ire_type and ire_masklen indicates 2627 * that the IRE can have ire_nce_cache set i.e., whether it is 2628 * IRE_ONLINK and for a single destination. 2629 */ 2630 boolean_t ire_nce_capable; 2631 2632 /* 2633 * Dependency tracking so we can safely cache IRE and NCE pointers 2634 * in offlink and onlink IREs. 2635 * These are locked under the ips_ire_dep_lock rwlock. Write held 2636 * when modifying the linkage. 2637 * ire_dep_parent (Also chain towards IRE for nexthop) 2638 * ire_dep_parent_generation: ire_generation of ire_dep_parent 2639 * ire_dep_children (From parent to first child) 2640 * ire_dep_sib_next (linked list of siblings) 2641 * ire_dep_sib_ptpn (linked list of siblings) 2642 * 2643 * The parent has a ire_refhold on each child, and each child has 2644 * an ire_refhold on its parent. 2645 * Since ire_dep_parent is a cache setup and torn down independently of 2646 * applications we need to use ire_ref{rele,hold}_notr for it. 2647 */ 2648 ire_t *ire_dep_parent; 2649 ire_t *ire_dep_children; 2650 ire_t *ire_dep_sib_next; 2651 ire_t **ire_dep_sib_ptpn; /* Pointer to previous next */ 2652 uint_t ire_dep_parent_generation; 2653 2654 uint_t ire_badcnt; /* Number of times ND_UNREACHABLE */ 2655 uint64_t ire_last_badcnt; /* In seconds */ 2656 2657 /* ire_defense* and ire_last_used_time are only used on IRE_LOCALs */ 2658 uint_t ire_defense_count; /* number of ARP conflicts */ 2659 uint_t ire_defense_time; /* last time defended (secs) */ 2660 2661 boolean_t ire_trace_disable; /* True when alloc fails */ 2662 ip_stack_t *ire_ipst; /* Does not have a netstack_hold */ 2663 iulp_t ire_metrics; 2664 /* 2665 * default and prefix routes that are added without explicitly 2666 * specifying the interface are termed "unbound" routes, and will 2667 * have ire_unbound set to true. 2668 */ 2669 boolean_t ire_unbound; 2670 }; 2671 2672 /* IPv4 compatibility macros */ 2673 #define ire_mask ire_u.ire4_u.ire4_mask 2674 #define ire_addr ire_u.ire4_u.ire4_addr 2675 #define ire_gateway_addr ire_u.ire4_u.ire4_gateway_addr 2676 #define ire_setsrc_addr ire_u.ire4_u.ire4_setsrc_addr 2677 2678 #define ire_mask_v6 ire_u.ire6_u.ire6_mask 2679 #define ire_addr_v6 ire_u.ire6_u.ire6_addr 2680 #define ire_gateway_addr_v6 ire_u.ire6_u.ire6_gateway_addr 2681 #define ire_setsrc_addr_v6 ire_u.ire6_u.ire6_setsrc_addr 2682 2683 /* 2684 * Values for ire_generation. 2685 * 2686 * If an IRE is marked with IRE_IS_CONDEMNED, the last walker of 2687 * the bucket should delete this IRE from this bucket. 2688 * 2689 * IRE_GENERATION_VERIFY is never stored in ire_generation but it is 2690 * stored in places that cache IREs (such as ixa_ire_generation and 2691 * ire_dep_parent_generation). It is used as a signal that the cache is 2692 * stale and needs to be reverified. 2693 */ 2694 #define IRE_GENERATION_CONDEMNED 0 2695 #define IRE_GENERATION_VERIFY 1 2696 #define IRE_GENERATION_INITIAL 2 2697 #define IRE_IS_CONDEMNED(ire) \ 2698 ((ire)->ire_generation == IRE_GENERATION_CONDEMNED) 2699 2700 /* Convenient typedefs for sockaddrs */ 2701 typedef struct sockaddr_in sin_t; 2702 typedef struct sockaddr_in6 sin6_t; 2703 2704 /* Name/Value Descriptor. */ 2705 typedef struct nv_s { 2706 uint64_t nv_value; 2707 char *nv_name; 2708 } nv_t; 2709 2710 #define ILL_FRAG_HASH(s, i) \ 2711 ((ntohl(s) ^ ((i) ^ ((i) >> 8))) % ILL_FRAG_HASH_TBL_COUNT) 2712 2713 /* 2714 * The MAX number of allowed fragmented packets per hash bucket 2715 * calculation is based on the most common mtu size of 1500. This limit 2716 * will work well for other mtu sizes as well. 2717 */ 2718 #define COMMON_IP_MTU 1500 2719 #define MAX_FRAG_MIN 10 2720 #define MAX_FRAG_PKTS(ipst) \ 2721 MAX(MAX_FRAG_MIN, (2 * (ipst->ips_ip_reass_queue_bytes / \ 2722 (COMMON_IP_MTU * ILL_FRAG_HASH_TBL_COUNT)))) 2723 2724 /* 2725 * Maximum dups allowed per packet. 2726 */ 2727 extern uint_t ip_max_frag_dups; 2728 2729 /* 2730 * Per-packet information for received packets and transmitted. 2731 * Used by the transport protocols when converting between the packet 2732 * and ancillary data and socket options. 2733 * 2734 * Note: This private data structure and related IPPF_* constant 2735 * definitions are exposed to enable compilation of some debugging tools 2736 * like lsof which use struct tcp_t in <inet/tcp.h>. This is intended to be 2737 * a temporary hack and long term alternate interfaces should be defined 2738 * to support the needs of such tools and private definitions moved to 2739 * private headers. 2740 */ 2741 struct ip_pkt_s { 2742 uint_t ipp_fields; /* Which fields are valid */ 2743 in6_addr_t ipp_addr; /* pktinfo src/dst addr */ 2744 #define ipp_addr_v4 V4_PART_OF_V6(ipp_addr) 2745 uint_t ipp_unicast_hops; /* IPV6_UNICAST_HOPS, IP_TTL */ 2746 uint_t ipp_hoplimit; /* IPV6_HOPLIMIT */ 2747 uint_t ipp_hopoptslen; 2748 uint_t ipp_rthdrdstoptslen; 2749 uint_t ipp_rthdrlen; 2750 uint_t ipp_dstoptslen; 2751 uint_t ipp_fraghdrlen; 2752 ip6_hbh_t *ipp_hopopts; 2753 ip6_dest_t *ipp_rthdrdstopts; 2754 ip6_rthdr_t *ipp_rthdr; 2755 ip6_dest_t *ipp_dstopts; 2756 ip6_frag_t *ipp_fraghdr; 2757 uint8_t ipp_tclass; /* IPV6_TCLASS */ 2758 uint8_t ipp_type_of_service; /* IP_TOS */ 2759 uint_t ipp_ipv4_options_len; /* Len of IPv4 options */ 2760 uint8_t *ipp_ipv4_options; /* Ptr to IPv4 options */ 2761 uint_t ipp_label_len_v4; /* Len of TX label for IPv4 */ 2762 uint8_t *ipp_label_v4; /* TX label for IPv4 */ 2763 uint_t ipp_label_len_v6; /* Len of TX label for IPv6 */ 2764 uint8_t *ipp_label_v6; /* TX label for IPv6 */ 2765 }; 2766 typedef struct ip_pkt_s ip_pkt_t; 2767 2768 extern void ip_pkt_free(ip_pkt_t *); /* free storage inside ip_pkt_t */ 2769 extern ipaddr_t ip_pkt_source_route_v4(const ip_pkt_t *); 2770 extern in6_addr_t *ip_pkt_source_route_v6(const ip_pkt_t *); 2771 extern int ip_pkt_copy(ip_pkt_t *, ip_pkt_t *, int); 2772 extern void ip_pkt_source_route_reverse_v4(ip_pkt_t *); 2773 2774 /* ipp_fields values */ 2775 #define IPPF_ADDR 0x0001 /* Part of in6_pktinfo: src/dst addr */ 2776 #define IPPF_HOPLIMIT 0x0002 /* Overrides unicast and multicast */ 2777 #define IPPF_TCLASS 0x0004 /* Overrides class in sin6_flowinfo */ 2778 2779 #define IPPF_HOPOPTS 0x0010 /* ipp_hopopts set */ 2780 #define IPPF_RTHDR 0x0020 /* ipp_rthdr set */ 2781 #define IPPF_RTHDRDSTOPTS 0x0040 /* ipp_rthdrdstopts set */ 2782 #define IPPF_DSTOPTS 0x0080 /* ipp_dstopts set */ 2783 2784 #define IPPF_IPV4_OPTIONS 0x0100 /* ipp_ipv4_options set */ 2785 #define IPPF_LABEL_V4 0x0200 /* ipp_label_v4 set */ 2786 #define IPPF_LABEL_V6 0x0400 /* ipp_label_v6 set */ 2787 2788 #define IPPF_FRAGHDR 0x0800 /* Used for IPsec receive side */ 2789 2790 /* 2791 * Data structure which is passed to conn_opt_get/set. 2792 * The conn_t is included even though it can be inferred from queue_t. 2793 * setsockopt and getsockopt use conn_ixa and conn_xmit_ipp. However, 2794 * when handling ancillary data we use separate ixa and ipps. 2795 */ 2796 typedef struct conn_opt_arg_s { 2797 conn_t *coa_connp; 2798 ip_xmit_attr_t *coa_ixa; 2799 ip_pkt_t *coa_ipp; 2800 boolean_t coa_ancillary; /* Ancillary data and not setsockopt */ 2801 uint_t coa_changed; /* See below */ 2802 } conn_opt_arg_t; 2803 2804 /* 2805 * Flags for what changed. 2806 * If we want to be more efficient in the future we can have more fine 2807 * grained flags e.g., a flag for just IP_TOS changing. 2808 * For now we either call ip_set_destination (for "route changed") 2809 * and/or conn_build_hdr_template/conn_prepend_hdr (for "header changed"). 2810 */ 2811 #define COA_HEADER_CHANGED 0x0001 2812 #define COA_ROUTE_CHANGED 0x0002 2813 #define COA_RCVBUF_CHANGED 0x0004 /* SO_RCVBUF */ 2814 #define COA_SNDBUF_CHANGED 0x0008 /* SO_SNDBUF */ 2815 #define COA_WROFF_CHANGED 0x0010 /* Header size changed */ 2816 #define COA_ICMP_BIND_NEEDED 0x0020 2817 #define COA_OOBINLINE_CHANGED 0x0040 2818 2819 #define TCP_PORTS_OFFSET 0 2820 #define UDP_PORTS_OFFSET 0 2821 2822 /* 2823 * lookups return the ill/ipif only if the flags are clear OR Iam writer. 2824 * ill / ipif lookup functions increment the refcnt on the ill / ipif only 2825 * after calling these macros. This ensures that the refcnt on the ipif or 2826 * ill will eventually drop down to zero. 2827 */ 2828 #define ILL_LOOKUP_FAILED 1 /* Used as error code */ 2829 #define IPIF_LOOKUP_FAILED 2 /* Used as error code */ 2830 2831 #define ILL_CAN_LOOKUP(ill) \ 2832 (!((ill)->ill_state_flags & ILL_CONDEMNED) || \ 2833 IAM_WRITER_ILL(ill)) 2834 2835 #define ILL_IS_CONDEMNED(ill) \ 2836 ((ill)->ill_state_flags & ILL_CONDEMNED) 2837 2838 #define IPIF_CAN_LOOKUP(ipif) \ 2839 (!((ipif)->ipif_state_flags & IPIF_CONDEMNED) || \ 2840 IAM_WRITER_IPIF(ipif)) 2841 2842 #define IPIF_IS_CONDEMNED(ipif) \ 2843 ((ipif)->ipif_state_flags & IPIF_CONDEMNED) 2844 2845 #define IPIF_IS_CHANGING(ipif) \ 2846 ((ipif)->ipif_state_flags & IPIF_CHANGING) 2847 2848 /* Macros used to assert that this thread is a writer */ 2849 #define IAM_WRITER_IPSQ(ipsq) ((ipsq)->ipsq_xop->ipx_writer == curthread) 2850 #define IAM_WRITER_ILL(ill) IAM_WRITER_IPSQ((ill)->ill_phyint->phyint_ipsq) 2851 #define IAM_WRITER_IPIF(ipif) IAM_WRITER_ILL((ipif)->ipif_ill) 2852 2853 /* 2854 * Grab ill locks in the proper order. The order is highest addressed 2855 * ill is locked first. 2856 */ 2857 #define GRAB_ILL_LOCKS(ill_1, ill_2) \ 2858 { \ 2859 if ((ill_1) > (ill_2)) { \ 2860 if (ill_1 != NULL) \ 2861 mutex_enter(&(ill_1)->ill_lock); \ 2862 if (ill_2 != NULL) \ 2863 mutex_enter(&(ill_2)->ill_lock); \ 2864 } else { \ 2865 if (ill_2 != NULL) \ 2866 mutex_enter(&(ill_2)->ill_lock); \ 2867 if (ill_1 != NULL && ill_1 != ill_2) \ 2868 mutex_enter(&(ill_1)->ill_lock); \ 2869 } \ 2870 } 2871 2872 #define RELEASE_ILL_LOCKS(ill_1, ill_2) \ 2873 { \ 2874 if (ill_1 != NULL) \ 2875 mutex_exit(&(ill_1)->ill_lock); \ 2876 if (ill_2 != NULL && ill_2 != ill_1) \ 2877 mutex_exit(&(ill_2)->ill_lock); \ 2878 } 2879 2880 /* Get the other protocol instance ill */ 2881 #define ILL_OTHER(ill) \ 2882 ((ill)->ill_isv6 ? (ill)->ill_phyint->phyint_illv4 : \ 2883 (ill)->ill_phyint->phyint_illv6) 2884 2885 /* ioctl command info: Ioctl properties extracted and stored in here */ 2886 typedef struct cmd_info_s 2887 { 2888 ipif_t *ci_ipif; /* ipif associated with [l]ifreq ioctl's */ 2889 sin_t *ci_sin; /* the sin struct passed down */ 2890 sin6_t *ci_sin6; /* the sin6_t struct passed down */ 2891 struct lifreq *ci_lifr; /* the lifreq struct passed down */ 2892 } cmd_info_t; 2893 2894 extern struct kmem_cache *ire_cache; 2895 2896 extern ipaddr_t ip_g_all_ones; 2897 2898 extern uint_t ip_loopback_mtu; /* /etc/system */ 2899 extern uint_t ip_loopback_mtuplus; 2900 extern uint_t ip_loopback_mtu_v6plus; 2901 2902 extern vmem_t *ip_minor_arena_sa; 2903 extern vmem_t *ip_minor_arena_la; 2904 2905 /* 2906 * ip_g_forward controls IP forwarding. It takes two values: 2907 * 0: IP_FORWARD_NEVER Don't forward packets ever. 2908 * 1: IP_FORWARD_ALWAYS Forward packets for elsewhere. 2909 * 2910 * RFC1122 says there must be a configuration switch to control forwarding, 2911 * but that the default MUST be to not forward packets ever. Implicit 2912 * control based on configuration of multiple interfaces MUST NOT be 2913 * implemented (Section 3.1). SunOS 4.1 did provide the "automatic" capability 2914 * and, in fact, it was the default. That capability is now provided in the 2915 * /etc/rc2.d/S69inet script. 2916 */ 2917 2918 #define ips_ip_respond_to_address_mask_broadcast \ 2919 ips_propinfo_tbl[0].prop_cur_bval 2920 #define ips_ip_g_resp_to_echo_bcast ips_propinfo_tbl[1].prop_cur_bval 2921 #define ips_ip_g_resp_to_echo_mcast ips_propinfo_tbl[2].prop_cur_bval 2922 #define ips_ip_g_resp_to_timestamp ips_propinfo_tbl[3].prop_cur_bval 2923 #define ips_ip_g_resp_to_timestamp_bcast ips_propinfo_tbl[4].prop_cur_bval 2924 #define ips_ip_g_send_redirects ips_propinfo_tbl[5].prop_cur_bval 2925 #define ips_ip_g_forward_directed_bcast ips_propinfo_tbl[6].prop_cur_bval 2926 #define ips_ip_mrtdebug ips_propinfo_tbl[7].prop_cur_uval 2927 #define ips_ip_ire_reclaim_fraction ips_propinfo_tbl[8].prop_cur_uval 2928 #define ips_ip_nce_reclaim_fraction ips_propinfo_tbl[9].prop_cur_uval 2929 #define ips_ip_dce_reclaim_fraction ips_propinfo_tbl[10].prop_cur_uval 2930 #define ips_ip_def_ttl ips_propinfo_tbl[11].prop_cur_uval 2931 #define ips_ip_forward_src_routed ips_propinfo_tbl[12].prop_cur_bval 2932 #define ips_ip_wroff_extra ips_propinfo_tbl[13].prop_cur_uval 2933 #define ips_ip_pathmtu_interval ips_propinfo_tbl[14].prop_cur_uval 2934 #define ips_ip_icmp_return ips_propinfo_tbl[15].prop_cur_uval 2935 #define ips_ip_path_mtu_discovery ips_propinfo_tbl[16].prop_cur_bval 2936 #define ips_ip_pmtu_min ips_propinfo_tbl[17].prop_cur_uval 2937 #define ips_ip_ignore_redirect ips_propinfo_tbl[18].prop_cur_bval 2938 #define ips_ip_arp_icmp_error ips_propinfo_tbl[19].prop_cur_bval 2939 #define ips_ip_broadcast_ttl ips_propinfo_tbl[20].prop_cur_uval 2940 #define ips_ip_icmp_err_interval ips_propinfo_tbl[21].prop_cur_uval 2941 #define ips_ip_icmp_err_burst ips_propinfo_tbl[22].prop_cur_uval 2942 #define ips_ip_reass_queue_bytes ips_propinfo_tbl[23].prop_cur_uval 2943 #define ips_ip_strict_dst_multihoming ips_propinfo_tbl[24].prop_cur_uval 2944 #define ips_ip_addrs_per_if ips_propinfo_tbl[25].prop_cur_uval 2945 #define ips_ipsec_override_persocket_policy ips_propinfo_tbl[26].prop_cur_bval 2946 #define ips_icmp_accept_clear_messages ips_propinfo_tbl[27].prop_cur_bval 2947 #define ips_igmp_accept_clear_messages ips_propinfo_tbl[28].prop_cur_bval 2948 2949 /* IPv6 configuration knobs */ 2950 #define ips_delay_first_probe_time ips_propinfo_tbl[29].prop_cur_uval 2951 #define ips_max_unicast_solicit ips_propinfo_tbl[30].prop_cur_uval 2952 #define ips_ipv6_def_hops ips_propinfo_tbl[31].prop_cur_uval 2953 #define ips_ipv6_icmp_return ips_propinfo_tbl[32].prop_cur_uval 2954 #define ips_ipv6_forward_src_routed ips_propinfo_tbl[33].prop_cur_bval 2955 #define ips_ipv6_resp_echo_mcast ips_propinfo_tbl[34].prop_cur_bval 2956 #define ips_ipv6_send_redirects ips_propinfo_tbl[35].prop_cur_bval 2957 #define ips_ipv6_ignore_redirect ips_propinfo_tbl[36].prop_cur_bval 2958 #define ips_ipv6_strict_dst_multihoming ips_propinfo_tbl[37].prop_cur_uval 2959 #define ips_src_check ips_propinfo_tbl[38].prop_cur_uval 2960 #define ips_ipsec_policy_log_interval ips_propinfo_tbl[39].prop_cur_uval 2961 #define ips_pim_accept_clear_messages ips_propinfo_tbl[40].prop_cur_bval 2962 #define ips_ip_ndp_unsolicit_interval ips_propinfo_tbl[41].prop_cur_uval 2963 #define ips_ip_ndp_unsolicit_count ips_propinfo_tbl[42].prop_cur_uval 2964 #define ips_ipv6_ignore_home_address_opt ips_propinfo_tbl[43].prop_cur_bval 2965 2966 /* Misc IP configuration knobs */ 2967 #define ips_ip_policy_mask ips_propinfo_tbl[44].prop_cur_uval 2968 #define ips_ip_ecmp_behavior ips_propinfo_tbl[45].prop_cur_uval 2969 #define ips_ip_multirt_ttl ips_propinfo_tbl[46].prop_cur_uval 2970 #define ips_ip_ire_badcnt_lifetime ips_propinfo_tbl[47].prop_cur_uval 2971 #define ips_ip_max_temp_idle ips_propinfo_tbl[48].prop_cur_uval 2972 #define ips_ip_max_temp_defend ips_propinfo_tbl[49].prop_cur_uval 2973 #define ips_ip_max_defend ips_propinfo_tbl[50].prop_cur_uval 2974 #define ips_ip_defend_interval ips_propinfo_tbl[51].prop_cur_uval 2975 #define ips_ip_dup_recovery ips_propinfo_tbl[52].prop_cur_uval 2976 #define ips_ip_restrict_interzone_loopback ips_propinfo_tbl[53].prop_cur_bval 2977 #define ips_ip_lso_outbound ips_propinfo_tbl[54].prop_cur_bval 2978 #define ips_igmp_max_version ips_propinfo_tbl[55].prop_cur_uval 2979 #define ips_mld_max_version ips_propinfo_tbl[56].prop_cur_uval 2980 #define ips_ip_forwarding ips_propinfo_tbl[57].prop_cur_bval 2981 #define ips_ipv6_forwarding ips_propinfo_tbl[58].prop_cur_bval 2982 #define ips_ip_reassembly_timeout ips_propinfo_tbl[59].prop_cur_uval 2983 #define ips_ipv6_reassembly_timeout ips_propinfo_tbl[60].prop_cur_uval 2984 #define ips_ip_cgtp_filter ips_propinfo_tbl[61].prop_cur_bval 2985 #define ips_arp_probe_delay ips_propinfo_tbl[62].prop_cur_uval 2986 #define ips_arp_fastprobe_delay ips_propinfo_tbl[63].prop_cur_uval 2987 #define ips_arp_probe_interval ips_propinfo_tbl[64].prop_cur_uval 2988 #define ips_arp_fastprobe_interval ips_propinfo_tbl[65].prop_cur_uval 2989 #define ips_arp_probe_count ips_propinfo_tbl[66].prop_cur_uval 2990 #define ips_arp_fastprobe_count ips_propinfo_tbl[67].prop_cur_uval 2991 #define ips_ipv4_dad_announce_interval ips_propinfo_tbl[68].prop_cur_uval 2992 #define ips_ipv6_dad_announce_interval ips_propinfo_tbl[69].prop_cur_uval 2993 #define ips_arp_defend_interval ips_propinfo_tbl[70].prop_cur_uval 2994 #define ips_arp_defend_rate ips_propinfo_tbl[71].prop_cur_uval 2995 #define ips_ndp_defend_interval ips_propinfo_tbl[72].prop_cur_uval 2996 #define ips_ndp_defend_rate ips_propinfo_tbl[73].prop_cur_uval 2997 #define ips_arp_defend_period ips_propinfo_tbl[74].prop_cur_uval 2998 #define ips_ndp_defend_period ips_propinfo_tbl[75].prop_cur_uval 2999 #define ips_ipv4_icmp_return_pmtu ips_propinfo_tbl[76].prop_cur_bval 3000 #define ips_ipv6_icmp_return_pmtu ips_propinfo_tbl[77].prop_cur_bval 3001 #define ips_ip_arp_publish_count ips_propinfo_tbl[78].prop_cur_uval 3002 #define ips_ip_arp_publish_interval ips_propinfo_tbl[79].prop_cur_uval 3003 #define ips_ip_strict_src_multihoming ips_propinfo_tbl[80].prop_cur_uval 3004 #define ips_ipv6_strict_src_multihoming ips_propinfo_tbl[81].prop_cur_uval 3005 #define ips_ipv6_drop_inbound_icmpv6 ips_propinfo_tbl[82].prop_cur_bval 3006 3007 extern int dohwcksum; /* use h/w cksum if supported by the h/w */ 3008 #ifdef ZC_TEST 3009 extern int noswcksum; 3010 #endif 3011 3012 extern char ipif_loopback_name[]; 3013 3014 extern nv_t *ire_nv_tbl; 3015 3016 extern struct module_info ip_mod_info; 3017 3018 #define HOOKS4_INTERESTED_PHYSICAL_IN(ipst) \ 3019 ((ipst)->ips_ip4_physical_in_event.he_interested) 3020 #define HOOKS6_INTERESTED_PHYSICAL_IN(ipst) \ 3021 ((ipst)->ips_ip6_physical_in_event.he_interested) 3022 #define HOOKS4_INTERESTED_PHYSICAL_OUT(ipst) \ 3023 ((ipst)->ips_ip4_physical_out_event.he_interested) 3024 #define HOOKS6_INTERESTED_PHYSICAL_OUT(ipst) \ 3025 ((ipst)->ips_ip6_physical_out_event.he_interested) 3026 #define HOOKS4_INTERESTED_FORWARDING(ipst) \ 3027 ((ipst)->ips_ip4_forwarding_event.he_interested) 3028 #define HOOKS6_INTERESTED_FORWARDING(ipst) \ 3029 ((ipst)->ips_ip6_forwarding_event.he_interested) 3030 #define HOOKS4_INTERESTED_LOOPBACK_IN(ipst) \ 3031 ((ipst)->ips_ip4_loopback_in_event.he_interested) 3032 #define HOOKS6_INTERESTED_LOOPBACK_IN(ipst) \ 3033 ((ipst)->ips_ip6_loopback_in_event.he_interested) 3034 #define HOOKS4_INTERESTED_LOOPBACK_OUT(ipst) \ 3035 ((ipst)->ips_ip4_loopback_out_event.he_interested) 3036 #define HOOKS6_INTERESTED_LOOPBACK_OUT(ipst) \ 3037 ((ipst)->ips_ip6_loopback_out_event.he_interested) 3038 /* 3039 * Hooks marcos used inside of ip 3040 * The callers use the above INTERESTED macros first, hence 3041 * the he_interested check is superflous. 3042 */ 3043 #define FW_HOOKS(_hook, _event, _ilp, _olp, _iph, _fm, _m, _llm, ipst, _err) \ 3044 if ((_hook).he_interested) { \ 3045 hook_pkt_event_t info; \ 3046 \ 3047 _NOTE(CONSTCOND) \ 3048 ASSERT((_ilp != NULL) || (_olp != NULL)); \ 3049 \ 3050 FW_SET_ILL_INDEX(info.hpe_ifp, (ill_t *)_ilp); \ 3051 FW_SET_ILL_INDEX(info.hpe_ofp, (ill_t *)_olp); \ 3052 info.hpe_protocol = ipst->ips_ipv4_net_data; \ 3053 info.hpe_hdr = _iph; \ 3054 info.hpe_mp = &(_fm); \ 3055 info.hpe_mb = _m; \ 3056 info.hpe_flags = _llm; \ 3057 _err = hook_run(ipst->ips_ipv4_net_data->netd_hooks, \ 3058 _event, (hook_data_t)&info); \ 3059 if (_err != 0) { \ 3060 ip2dbg(("%s hook dropped mblk chain %p hdr %p\n",\ 3061 (_hook).he_name, (void *)_fm, (void *)_m)); \ 3062 if (_fm != NULL) { \ 3063 freemsg(_fm); \ 3064 _fm = NULL; \ 3065 } \ 3066 _iph = NULL; \ 3067 _m = NULL; \ 3068 } else { \ 3069 _iph = info.hpe_hdr; \ 3070 _m = info.hpe_mb; \ 3071 } \ 3072 } 3073 3074 #define FW_HOOKS6(_hook, _event, _ilp, _olp, _iph, _fm, _m, _llm, ipst, _err) \ 3075 if ((_hook).he_interested) { \ 3076 hook_pkt_event_t info; \ 3077 \ 3078 _NOTE(CONSTCOND) \ 3079 ASSERT((_ilp != NULL) || (_olp != NULL)); \ 3080 \ 3081 FW_SET_ILL_INDEX(info.hpe_ifp, (ill_t *)_ilp); \ 3082 FW_SET_ILL_INDEX(info.hpe_ofp, (ill_t *)_olp); \ 3083 info.hpe_protocol = ipst->ips_ipv6_net_data; \ 3084 info.hpe_hdr = _iph; \ 3085 info.hpe_mp = &(_fm); \ 3086 info.hpe_mb = _m; \ 3087 info.hpe_flags = _llm; \ 3088 _err = hook_run(ipst->ips_ipv6_net_data->netd_hooks, \ 3089 _event, (hook_data_t)&info); \ 3090 if (_err != 0) { \ 3091 ip2dbg(("%s hook dropped mblk chain %p hdr %p\n",\ 3092 (_hook).he_name, (void *)_fm, (void *)_m)); \ 3093 if (_fm != NULL) { \ 3094 freemsg(_fm); \ 3095 _fm = NULL; \ 3096 } \ 3097 _iph = NULL; \ 3098 _m = NULL; \ 3099 } else { \ 3100 _iph = info.hpe_hdr; \ 3101 _m = info.hpe_mb; \ 3102 } \ 3103 } 3104 3105 #define FW_SET_ILL_INDEX(fp, ill) \ 3106 _NOTE(CONSTCOND) \ 3107 if ((ill) == NULL || (ill)->ill_phyint == NULL) { \ 3108 (fp) = 0; \ 3109 _NOTE(CONSTCOND) \ 3110 } else if (IS_UNDER_IPMP(ill)) { \ 3111 (fp) = ipmp_ill_get_ipmp_ifindex(ill); \ 3112 } else { \ 3113 (fp) = (ill)->ill_phyint->phyint_ifindex; \ 3114 } 3115 3116 /* 3117 * Network byte order macros 3118 */ 3119 #ifdef _BIG_ENDIAN 3120 #define N_IN_CLASSA_NET IN_CLASSA_NET 3121 #define N_IN_CLASSD_NET IN_CLASSD_NET 3122 #define N_INADDR_UNSPEC_GROUP INADDR_UNSPEC_GROUP 3123 #define N_IN_LOOPBACK_NET (ipaddr_t)0x7f000000U 3124 #else /* _BIG_ENDIAN */ 3125 #define N_IN_CLASSA_NET (ipaddr_t)0x000000ffU 3126 #define N_IN_CLASSD_NET (ipaddr_t)0x000000f0U 3127 #define N_INADDR_UNSPEC_GROUP (ipaddr_t)0x000000e0U 3128 #define N_IN_LOOPBACK_NET (ipaddr_t)0x0000007fU 3129 #endif /* _BIG_ENDIAN */ 3130 #define CLASSD(addr) (((addr) & N_IN_CLASSD_NET) == N_INADDR_UNSPEC_GROUP) 3131 #define CLASSE(addr) (((addr) & N_IN_CLASSD_NET) == N_IN_CLASSD_NET) 3132 #define IP_LOOPBACK_ADDR(addr) \ 3133 (((addr) & N_IN_CLASSA_NET == N_IN_LOOPBACK_NET)) 3134 3135 extern int ip_debug; 3136 extern uint_t ip_thread_data; 3137 extern krwlock_t ip_thread_rwlock; 3138 extern list_t ip_thread_list; 3139 3140 #ifdef IP_DEBUG 3141 #include <sys/debug.h> 3142 #include <sys/promif.h> 3143 3144 #define ip0dbg(a) printf a 3145 #define ip1dbg(a) if (ip_debug > 2) printf a 3146 #define ip2dbg(a) if (ip_debug > 3) printf a 3147 #define ip3dbg(a) if (ip_debug > 4) printf a 3148 #else 3149 #define ip0dbg(a) /* */ 3150 #define ip1dbg(a) /* */ 3151 #define ip2dbg(a) /* */ 3152 #define ip3dbg(a) /* */ 3153 #endif /* IP_DEBUG */ 3154 3155 /* Default MAC-layer address string length for mac_colon_addr */ 3156 #define MAC_STR_LEN 128 3157 3158 struct mac_header_info_s; 3159 3160 extern void ill_frag_timer(void *); 3161 extern ill_t *ill_first(int, int, ill_walk_context_t *, ip_stack_t *); 3162 extern ill_t *ill_next(ill_walk_context_t *, ill_t *); 3163 extern void ill_frag_timer_start(ill_t *); 3164 extern void ill_nic_event_dispatch(ill_t *, lif_if_t, nic_event_t, 3165 nic_event_data_t, size_t); 3166 extern mblk_t *ip_carve_mp(mblk_t **, ssize_t); 3167 extern mblk_t *ip_dlpi_alloc(size_t, t_uscalar_t); 3168 extern mblk_t *ip_dlnotify_alloc(uint_t, uint_t); 3169 extern char *ip_dot_addr(ipaddr_t, char *); 3170 extern const char *mac_colon_addr(const uint8_t *, size_t, char *, size_t); 3171 extern void ip_lwput(queue_t *, mblk_t *); 3172 extern boolean_t icmp_err_rate_limit(ip_stack_t *); 3173 extern void icmp_frag_needed(mblk_t *, int, ip_recv_attr_t *); 3174 extern mblk_t *icmp_inbound_v4(mblk_t *, ip_recv_attr_t *); 3175 extern void icmp_time_exceeded(mblk_t *, uint8_t, ip_recv_attr_t *); 3176 extern void icmp_unreachable(mblk_t *, uint8_t, ip_recv_attr_t *); 3177 extern boolean_t ip_ipsec_policy_inherit(conn_t *, conn_t *, ip_recv_attr_t *); 3178 extern void *ip_pullup(mblk_t *, ssize_t, ip_recv_attr_t *); 3179 extern void ip_setl2src(mblk_t *, ip_recv_attr_t *, ill_t *); 3180 extern mblk_t *ip_check_and_align_header(mblk_t *, uint_t, ip_recv_attr_t *); 3181 extern mblk_t *ip_check_length(mblk_t *, uchar_t *, ssize_t, uint_t, uint_t, 3182 ip_recv_attr_t *); 3183 extern mblk_t *ip_check_optlen(mblk_t *, ipha_t *, uint_t, uint_t, 3184 ip_recv_attr_t *); 3185 extern mblk_t *ip_fix_dbref(mblk_t *, ip_recv_attr_t *); 3186 extern uint_t ip_cksum(mblk_t *, int, uint32_t); 3187 extern int ip_close(queue_t *, int); 3188 extern uint16_t ip_csum_hdr(ipha_t *); 3189 extern void ip_forward_xmit_v4(nce_t *, ill_t *, mblk_t *, ipha_t *, 3190 ip_recv_attr_t *, uint32_t, uint32_t); 3191 extern boolean_t ip_forward_options(mblk_t *, ipha_t *, ill_t *, 3192 ip_recv_attr_t *); 3193 extern int ip_fragment_v4(mblk_t *, nce_t *, iaflags_t, uint_t, uint32_t, 3194 uint32_t, zoneid_t, zoneid_t, pfirepostfrag_t postfragfn, 3195 uintptr_t *cookie); 3196 extern void ip_proto_not_sup(mblk_t *, ip_recv_attr_t *); 3197 extern void ip_ire_g_fini(void); 3198 extern void ip_ire_g_init(void); 3199 extern void ip_ire_fini(ip_stack_t *); 3200 extern void ip_ire_init(ip_stack_t *); 3201 extern void ip_mdata_to_mhi(ill_t *, mblk_t *, struct mac_header_info_s *); 3202 extern int ip_openv4(queue_t *q, dev_t *devp, int flag, int sflag, 3203 cred_t *credp); 3204 extern int ip_openv6(queue_t *q, dev_t *devp, int flag, int sflag, 3205 cred_t *credp); 3206 extern int ip_reassemble(mblk_t *, ipf_t *, uint_t, boolean_t, ill_t *, 3207 size_t); 3208 extern void ip_rput(queue_t *, mblk_t *); 3209 extern void ip_input(ill_t *, ill_rx_ring_t *, mblk_t *, 3210 struct mac_header_info_s *); 3211 extern void ip_input_v6(ill_t *, ill_rx_ring_t *, mblk_t *, 3212 struct mac_header_info_s *); 3213 extern mblk_t *ip_input_common_v4(ill_t *, ill_rx_ring_t *, mblk_t *, 3214 struct mac_header_info_s *, squeue_t *, mblk_t **, uint_t *); 3215 extern mblk_t *ip_input_common_v6(ill_t *, ill_rx_ring_t *, mblk_t *, 3216 struct mac_header_info_s *, squeue_t *, mblk_t **, uint_t *); 3217 extern void ill_input_full_v4(mblk_t *, void *, void *, 3218 ip_recv_attr_t *, rtc_t *); 3219 extern void ill_input_short_v4(mblk_t *, void *, void *, 3220 ip_recv_attr_t *, rtc_t *); 3221 extern void ill_input_full_v6(mblk_t *, void *, void *, 3222 ip_recv_attr_t *, rtc_t *); 3223 extern void ill_input_short_v6(mblk_t *, void *, void *, 3224 ip_recv_attr_t *, rtc_t *); 3225 extern ipaddr_t ip_input_options(ipha_t *, ipaddr_t, mblk_t *, 3226 ip_recv_attr_t *, int *); 3227 extern boolean_t ip_input_local_options(mblk_t *, ipha_t *, ip_recv_attr_t *); 3228 extern mblk_t *ip_input_fragment(mblk_t *, ipha_t *, ip_recv_attr_t *); 3229 extern mblk_t *ip_input_fragment_v6(mblk_t *, ip6_t *, ip6_frag_t *, uint_t, 3230 ip_recv_attr_t *); 3231 extern void ip_input_post_ipsec(mblk_t *, ip_recv_attr_t *); 3232 extern void ip_fanout_v4(mblk_t *, ipha_t *, ip_recv_attr_t *); 3233 extern void ip_fanout_v6(mblk_t *, ip6_t *, ip_recv_attr_t *); 3234 extern void ip_fanout_proto_conn(conn_t *, mblk_t *, ipha_t *, ip6_t *, 3235 ip_recv_attr_t *); 3236 extern void ip_fanout_proto_v4(mblk_t *, ipha_t *, ip_recv_attr_t *); 3237 extern void ip_fanout_send_icmp_v4(mblk_t *, uint_t, uint_t, 3238 ip_recv_attr_t *); 3239 extern void ip_fanout_udp_conn(conn_t *, mblk_t *, ipha_t *, ip6_t *, 3240 ip_recv_attr_t *); 3241 extern void ip_fanout_udp_multi_v4(mblk_t *, ipha_t *, uint16_t, uint16_t, 3242 ip_recv_attr_t *); 3243 extern mblk_t *zero_spi_check(mblk_t *, ip_recv_attr_t *); 3244 extern void ip_build_hdrs_v4(uchar_t *, uint_t, const ip_pkt_t *, uint8_t); 3245 extern int ip_find_hdr_v4(ipha_t *, ip_pkt_t *, boolean_t); 3246 extern int ip_total_hdrs_len_v4(const ip_pkt_t *); 3247 3248 extern mblk_t *ip_accept_tcp(ill_t *, ill_rx_ring_t *, squeue_t *, 3249 mblk_t *, mblk_t **, uint_t *cnt); 3250 extern void ip_rput_dlpi(ill_t *, mblk_t *); 3251 extern void ip_rput_notdata(ill_t *, mblk_t *); 3252 3253 extern void ip_mib2_add_ip_stats(mib2_ipIfStatsEntry_t *, 3254 mib2_ipIfStatsEntry_t *); 3255 extern void ip_mib2_add_icmp6_stats(mib2_ipv6IfIcmpEntry_t *, 3256 mib2_ipv6IfIcmpEntry_t *); 3257 extern void ip_rput_other(ipsq_t *, queue_t *, mblk_t *, void *); 3258 extern ire_t *ip_check_multihome(void *, ire_t *, ill_t *); 3259 extern void ip_send_potential_redirect_v4(mblk_t *, ipha_t *, ire_t *, 3260 ip_recv_attr_t *); 3261 extern int ip_set_destination_v4(ipaddr_t *, ipaddr_t, ipaddr_t, 3262 ip_xmit_attr_t *, iulp_t *, uint32_t, uint_t); 3263 extern int ip_set_destination_v6(in6_addr_t *, const in6_addr_t *, 3264 const in6_addr_t *, ip_xmit_attr_t *, iulp_t *, uint32_t, uint_t); 3265 3266 extern int ip_output_simple(mblk_t *, ip_xmit_attr_t *); 3267 extern int ip_output_simple_v4(mblk_t *, ip_xmit_attr_t *); 3268 extern int ip_output_simple_v6(mblk_t *, ip_xmit_attr_t *); 3269 extern int ip_output_options(mblk_t *, ipha_t *, ip_xmit_attr_t *, 3270 ill_t *); 3271 extern void ip_output_local_options(ipha_t *, ip_stack_t *); 3272 3273 extern ip_xmit_attr_t *conn_get_ixa(conn_t *, boolean_t); 3274 extern ip_xmit_attr_t *conn_get_ixa_tryhard(conn_t *, boolean_t); 3275 extern ip_xmit_attr_t *conn_replace_ixa(conn_t *, ip_xmit_attr_t *); 3276 extern ip_xmit_attr_t *conn_get_ixa_exclusive(conn_t *); 3277 extern ip_xmit_attr_t *ip_xmit_attr_duplicate(ip_xmit_attr_t *); 3278 extern void ip_xmit_attr_replace_tsl(ip_xmit_attr_t *, ts_label_t *); 3279 extern void ip_xmit_attr_restore_tsl(ip_xmit_attr_t *, cred_t *); 3280 boolean_t ip_recv_attr_replace_label(ip_recv_attr_t *, ts_label_t *); 3281 extern void ixa_inactive(ip_xmit_attr_t *); 3282 extern void ixa_refrele(ip_xmit_attr_t *); 3283 extern boolean_t ixa_check_drain_insert(conn_t *, ip_xmit_attr_t *); 3284 extern void ixa_cleanup(ip_xmit_attr_t *); 3285 extern void ira_cleanup(ip_recv_attr_t *, boolean_t); 3286 extern void ixa_safe_copy(ip_xmit_attr_t *, ip_xmit_attr_t *); 3287 3288 extern int conn_ip_output(mblk_t *, ip_xmit_attr_t *); 3289 extern boolean_t ip_output_verify_local(ip_xmit_attr_t *); 3290 extern mblk_t *ip_output_process_local(mblk_t *, ip_xmit_attr_t *, boolean_t, 3291 boolean_t, conn_t *); 3292 3293 extern int conn_opt_get(conn_opt_arg_t *, t_scalar_t, t_scalar_t, 3294 uchar_t *); 3295 extern int conn_opt_set(conn_opt_arg_t *, t_scalar_t, t_scalar_t, uint_t, 3296 uchar_t *, boolean_t, cred_t *); 3297 extern boolean_t conn_same_as_last_v4(conn_t *, sin_t *); 3298 extern boolean_t conn_same_as_last_v6(conn_t *, sin6_t *); 3299 extern int conn_update_label(const conn_t *, const ip_xmit_attr_t *, 3300 const in6_addr_t *, ip_pkt_t *); 3301 3302 extern int ip_opt_set_multicast_group(conn_t *, t_scalar_t, 3303 uchar_t *, boolean_t, boolean_t); 3304 extern int ip_opt_set_multicast_sources(conn_t *, t_scalar_t, 3305 uchar_t *, boolean_t, boolean_t); 3306 extern int conn_getsockname(conn_t *, struct sockaddr *, uint_t *); 3307 extern int conn_getpeername(conn_t *, struct sockaddr *, uint_t *); 3308 3309 extern int conn_build_hdr_template(conn_t *, uint_t, uint_t, 3310 const in6_addr_t *, const in6_addr_t *, uint32_t); 3311 extern mblk_t *conn_prepend_hdr(ip_xmit_attr_t *, const ip_pkt_t *, 3312 const in6_addr_t *, const in6_addr_t *, uint8_t, uint32_t, uint_t, 3313 mblk_t *, uint_t, uint_t, uint32_t *, int *); 3314 extern void ip_attr_newdst(ip_xmit_attr_t *); 3315 extern void ip_attr_nexthop(const ip_pkt_t *, const ip_xmit_attr_t *, 3316 const in6_addr_t *, in6_addr_t *); 3317 extern int conn_connect(conn_t *, iulp_t *, uint32_t); 3318 extern int ip_attr_connect(const conn_t *, ip_xmit_attr_t *, 3319 const in6_addr_t *, const in6_addr_t *, const in6_addr_t *, in_port_t, 3320 in6_addr_t *, iulp_t *, uint32_t); 3321 extern int conn_inherit_parent(conn_t *, conn_t *); 3322 3323 extern void conn_ixa_cleanup(conn_t *connp, void *arg); 3324 3325 extern boolean_t conn_wantpacket(conn_t *, ip_recv_attr_t *, ipha_t *); 3326 extern uint_t ip_type_v4(ipaddr_t, ip_stack_t *); 3327 extern uint_t ip_type_v6(const in6_addr_t *, ip_stack_t *); 3328 3329 extern void ip_wput_nondata(queue_t *, mblk_t *); 3330 extern void ip_wsrv(queue_t *); 3331 extern char *ip_nv_lookup(nv_t *, int); 3332 extern boolean_t ip_local_addr_ok_v6(const in6_addr_t *, const in6_addr_t *); 3333 extern boolean_t ip_remote_addr_ok_v6(const in6_addr_t *, const in6_addr_t *); 3334 extern ipaddr_t ip_massage_options(ipha_t *, netstack_t *); 3335 extern ipaddr_t ip_net_mask(ipaddr_t); 3336 extern void arp_bringup_done(ill_t *, int); 3337 extern void arp_replumb_done(ill_t *, int); 3338 3339 extern struct qinit iprinitv6; 3340 3341 extern void ipmp_init(ip_stack_t *); 3342 extern void ipmp_destroy(ip_stack_t *); 3343 extern ipmp_grp_t *ipmp_grp_create(const char *, phyint_t *); 3344 extern void ipmp_grp_destroy(ipmp_grp_t *); 3345 extern void ipmp_grp_info(const ipmp_grp_t *, lifgroupinfo_t *); 3346 extern int ipmp_grp_rename(ipmp_grp_t *, const char *); 3347 extern ipmp_grp_t *ipmp_grp_lookup(const char *, ip_stack_t *); 3348 extern int ipmp_grp_vet_phyint(ipmp_grp_t *, phyint_t *); 3349 extern ipmp_illgrp_t *ipmp_illgrp_create(ill_t *); 3350 extern void ipmp_illgrp_destroy(ipmp_illgrp_t *); 3351 extern ill_t *ipmp_illgrp_add_ipif(ipmp_illgrp_t *, ipif_t *); 3352 extern void ipmp_illgrp_del_ipif(ipmp_illgrp_t *, ipif_t *); 3353 extern ill_t *ipmp_illgrp_next_ill(ipmp_illgrp_t *); 3354 extern ill_t *ipmp_illgrp_hold_next_ill(ipmp_illgrp_t *); 3355 extern ill_t *ipmp_illgrp_hold_cast_ill(ipmp_illgrp_t *); 3356 extern ill_t *ipmp_illgrp_ipmp_ill(ipmp_illgrp_t *); 3357 extern void ipmp_illgrp_refresh_mtu(ipmp_illgrp_t *); 3358 extern ipmp_arpent_t *ipmp_illgrp_create_arpent(ipmp_illgrp_t *, 3359 boolean_t, ipaddr_t, uchar_t *, size_t, uint16_t); 3360 extern void ipmp_illgrp_destroy_arpent(ipmp_illgrp_t *, ipmp_arpent_t *); 3361 extern ipmp_arpent_t *ipmp_illgrp_lookup_arpent(ipmp_illgrp_t *, ipaddr_t *); 3362 extern void ipmp_illgrp_refresh_arpent(ipmp_illgrp_t *); 3363 extern void ipmp_illgrp_mark_arpent(ipmp_illgrp_t *, ipmp_arpent_t *); 3364 extern ill_t *ipmp_illgrp_find_ill(ipmp_illgrp_t *, uchar_t *, uint_t); 3365 extern void ipmp_illgrp_link_grp(ipmp_illgrp_t *, ipmp_grp_t *); 3366 extern int ipmp_illgrp_unlink_grp(ipmp_illgrp_t *); 3367 extern uint_t ipmp_ill_get_ipmp_ifindex(const ill_t *); 3368 extern void ipmp_ill_join_illgrp(ill_t *, ipmp_illgrp_t *); 3369 extern void ipmp_ill_leave_illgrp(ill_t *); 3370 extern ill_t *ipmp_ill_hold_ipmp_ill(ill_t *); 3371 extern ill_t *ipmp_ill_hold_xmit_ill(ill_t *, boolean_t); 3372 extern boolean_t ipmp_ill_is_active(ill_t *); 3373 extern void ipmp_ill_refresh_active(ill_t *); 3374 extern void ipmp_phyint_join_grp(phyint_t *, ipmp_grp_t *); 3375 extern void ipmp_phyint_leave_grp(phyint_t *); 3376 extern void ipmp_phyint_refresh_active(phyint_t *); 3377 extern ill_t *ipmp_ipif_bound_ill(const ipif_t *); 3378 extern ill_t *ipmp_ipif_hold_bound_ill(const ipif_t *); 3379 extern boolean_t ipmp_ipif_is_dataaddr(const ipif_t *); 3380 extern boolean_t ipmp_ipif_is_stubaddr(const ipif_t *); 3381 extern boolean_t ipmp_packet_is_probe(mblk_t *, ill_t *); 3382 extern void ipmp_ncec_delete_nce(ncec_t *); 3383 extern void ipmp_ncec_refresh_nce(ncec_t *); 3384 3385 extern void conn_drain_insert(conn_t *, idl_tx_list_t *); 3386 extern void conn_setqfull(conn_t *, boolean_t *); 3387 extern void conn_clrqfull(conn_t *, boolean_t *); 3388 extern int conn_ipsec_length(conn_t *); 3389 extern ipaddr_t ip_get_dst(ipha_t *); 3390 extern uint_t ip_get_pmtu(ip_xmit_attr_t *); 3391 extern uint_t ip_get_base_mtu(ill_t *, ire_t *); 3392 extern mblk_t *ip_output_attach_policy(mblk_t *, ipha_t *, ip6_t *, 3393 const conn_t *, ip_xmit_attr_t *); 3394 extern int ipsec_out_extra_length(ip_xmit_attr_t *); 3395 extern int ipsec_out_process(mblk_t *, ip_xmit_attr_t *); 3396 extern int ip_output_post_ipsec(mblk_t *, ip_xmit_attr_t *); 3397 extern void ipsec_out_to_in(ip_xmit_attr_t *, ill_t *ill, 3398 ip_recv_attr_t *); 3399 3400 extern void ire_cleanup(ire_t *); 3401 extern void ire_inactive(ire_t *); 3402 extern boolean_t irb_inactive(irb_t *); 3403 extern ire_t *ire_unlink(irb_t *); 3404 3405 #ifdef DEBUG 3406 extern boolean_t th_trace_ref(const void *, ip_stack_t *); 3407 extern void th_trace_unref(const void *); 3408 extern void th_trace_cleanup(const void *, boolean_t); 3409 extern void ire_trace_ref(ire_t *); 3410 extern void ire_untrace_ref(ire_t *); 3411 #endif 3412 3413 extern int ip_srcid_insert(const in6_addr_t *, zoneid_t, ip_stack_t *); 3414 extern int ip_srcid_remove(const in6_addr_t *, zoneid_t, ip_stack_t *); 3415 extern void ip_srcid_find_id(uint_t, in6_addr_t *, zoneid_t, netstack_t *); 3416 extern uint_t ip_srcid_find_addr(const in6_addr_t *, zoneid_t, netstack_t *); 3417 3418 extern uint8_t ipoptp_next(ipoptp_t *); 3419 extern uint8_t ipoptp_first(ipoptp_t *, ipha_t *); 3420 extern int ip_opt_get_user(conn_t *, uchar_t *); 3421 extern int ipsec_req_from_conn(conn_t *, ipsec_req_t *, int); 3422 extern int ip_snmp_get(queue_t *q, mblk_t *mctl, int level); 3423 extern int ip_snmp_set(queue_t *q, int, int, uchar_t *, int); 3424 extern void ip_process_ioctl(ipsq_t *, queue_t *, mblk_t *, void *); 3425 extern void ip_quiesce_conn(conn_t *); 3426 extern void ip_reprocess_ioctl(ipsq_t *, queue_t *, mblk_t *, void *); 3427 extern void ip_ioctl_finish(queue_t *, mblk_t *, int, int, ipsq_t *); 3428 3429 extern boolean_t ip_cmpbuf(const void *, uint_t, boolean_t, const void *, 3430 uint_t); 3431 extern boolean_t ip_allocbuf(void **, uint_t *, boolean_t, const void *, 3432 uint_t); 3433 extern void ip_savebuf(void **, uint_t *, boolean_t, const void *, uint_t); 3434 3435 extern boolean_t ipsq_pending_mp_cleanup(ill_t *, conn_t *); 3436 extern void conn_ioctl_cleanup(conn_t *); 3437 3438 extern void ip_unbind(conn_t *); 3439 3440 extern void tnet_init(void); 3441 extern void tnet_fini(void); 3442 3443 /* 3444 * Hook functions to enable cluster networking 3445 * On non-clustered systems these vectors must always be NULL. 3446 */ 3447 extern int (*cl_inet_isclusterwide)(netstackid_t stack_id, uint8_t protocol, 3448 sa_family_t addr_family, uint8_t *laddrp, void *args); 3449 extern uint32_t (*cl_inet_ipident)(netstackid_t stack_id, uint8_t protocol, 3450 sa_family_t addr_family, uint8_t *laddrp, uint8_t *faddrp, 3451 void *args); 3452 extern int (*cl_inet_connect2)(netstackid_t stack_id, uint8_t protocol, 3453 boolean_t is_outgoing, sa_family_t addr_family, uint8_t *laddrp, 3454 in_port_t lport, uint8_t *faddrp, in_port_t fport, void *args); 3455 extern void (*cl_inet_getspi)(netstackid_t, uint8_t, uint8_t *, size_t, 3456 void *); 3457 extern void (*cl_inet_getspi)(netstackid_t stack_id, uint8_t protocol, 3458 uint8_t *ptr, size_t len, void *args); 3459 extern int (*cl_inet_checkspi)(netstackid_t stack_id, uint8_t protocol, 3460 uint32_t spi, void *args); 3461 extern void (*cl_inet_deletespi)(netstackid_t stack_id, uint8_t protocol, 3462 uint32_t spi, void *args); 3463 extern void (*cl_inet_idlesa)(netstackid_t, uint8_t, uint32_t, 3464 sa_family_t, in6_addr_t, in6_addr_t, void *); 3465 3466 3467 /* Hooks for CGTP (multirt routes) filtering module */ 3468 #define CGTP_FILTER_REV_1 1 3469 #define CGTP_FILTER_REV_2 2 3470 #define CGTP_FILTER_REV_3 3 3471 #define CGTP_FILTER_REV CGTP_FILTER_REV_3 3472 3473 /* cfo_filter and cfo_filter_v6 hooks return values */ 3474 #define CGTP_IP_PKT_NOT_CGTP 0 3475 #define CGTP_IP_PKT_PREMIUM 1 3476 #define CGTP_IP_PKT_DUPLICATE 2 3477 3478 /* Version 3 of the filter interface */ 3479 typedef struct cgtp_filter_ops { 3480 int cfo_filter_rev; /* CGTP_FILTER_REV_3 */ 3481 int (*cfo_change_state)(netstackid_t, int); 3482 int (*cfo_add_dest_v4)(netstackid_t, ipaddr_t, ipaddr_t, 3483 ipaddr_t, ipaddr_t); 3484 int (*cfo_del_dest_v4)(netstackid_t, ipaddr_t, ipaddr_t); 3485 int (*cfo_add_dest_v6)(netstackid_t, in6_addr_t *, in6_addr_t *, 3486 in6_addr_t *, in6_addr_t *); 3487 int (*cfo_del_dest_v6)(netstackid_t, in6_addr_t *, in6_addr_t *); 3488 int (*cfo_filter)(netstackid_t, uint_t, mblk_t *); 3489 int (*cfo_filter_v6)(netstackid_t, uint_t, ip6_t *, 3490 ip6_frag_t *); 3491 } cgtp_filter_ops_t; 3492 3493 #define CGTP_MCAST_SUCCESS 1 3494 3495 /* 3496 * The separate CGTP module needs this global symbol so that it 3497 * can check the version and determine whether to use the old or the new 3498 * version of the filtering interface. 3499 */ 3500 extern int ip_cgtp_filter_rev; 3501 3502 extern int ip_cgtp_filter_supported(void); 3503 extern int ip_cgtp_filter_register(netstackid_t, cgtp_filter_ops_t *); 3504 extern int ip_cgtp_filter_unregister(netstackid_t); 3505 extern int ip_cgtp_filter_is_registered(netstackid_t); 3506 3507 /* 3508 * rr_ring_state cycles in the order shown below from RR_FREE through 3509 * RR_FREE_IN_PROG and back to RR_FREE. 3510 */ 3511 typedef enum { 3512 RR_FREE, /* Free slot */ 3513 RR_SQUEUE_UNBOUND, /* Ring's squeue is unbound */ 3514 RR_SQUEUE_BIND_INPROG, /* Ring's squeue bind in progress */ 3515 RR_SQUEUE_BOUND, /* Ring's squeue bound to cpu */ 3516 RR_FREE_INPROG /* Ring is being freed */ 3517 } ip_ring_state_t; 3518 3519 #define ILL_MAX_RINGS 256 /* Max num of rx rings we can manage */ 3520 #define ILL_POLLING 0x01 /* Polling in use */ 3521 3522 /* 3523 * These functions pointer types are exported by the mac/dls layer. 3524 * we need to duplicate the definitions here because we cannot 3525 * include mac/dls header files here. 3526 */ 3527 typedef boolean_t (*ip_mac_intr_disable_t)(void *); 3528 typedef void (*ip_mac_intr_enable_t)(void *); 3529 typedef ip_mac_tx_cookie_t (*ip_dld_tx_t)(void *, mblk_t *, 3530 uint64_t, uint16_t); 3531 typedef void (*ip_flow_enable_t)(void *, ip_mac_tx_cookie_t); 3532 typedef void *(*ip_dld_callb_t)(void *, 3533 ip_flow_enable_t, void *); 3534 typedef boolean_t (*ip_dld_fctl_t)(void *, ip_mac_tx_cookie_t); 3535 typedef int (*ip_capab_func_t)(void *, uint_t, 3536 void *, uint_t); 3537 3538 /* 3539 * POLLING README 3540 * sq_get_pkts() is called to pick packets from softring in poll mode. It 3541 * calls rr_rx to get the chain and process it with rr_ip_accept. 3542 * rr_rx = mac_soft_ring_poll() to pick packets 3543 * rr_ip_accept = ip_accept_tcp() to process packets 3544 */ 3545 3546 /* 3547 * XXX: With protocol, service specific squeues, they will have 3548 * specific acceptor functions. 3549 */ 3550 typedef mblk_t *(*ip_mac_rx_t)(void *, size_t); 3551 typedef mblk_t *(*ip_accept_t)(ill_t *, ill_rx_ring_t *, 3552 squeue_t *, mblk_t *, mblk_t **, uint_t *); 3553 3554 /* 3555 * rr_intr_enable, rr_intr_disable, rr_rx_handle, rr_rx: 3556 * May be accessed while in the squeue AND after checking that SQS_POLL_CAPAB 3557 * is set. 3558 * 3559 * rr_ring_state: Protected by ill_lock. 3560 */ 3561 struct ill_rx_ring { 3562 ip_mac_intr_disable_t rr_intr_disable; /* Interrupt disabling func */ 3563 ip_mac_intr_enable_t rr_intr_enable; /* Interrupt enabling func */ 3564 void *rr_intr_handle; /* Handle interrupt funcs */ 3565 ip_mac_rx_t rr_rx; /* Driver receive function */ 3566 ip_accept_t rr_ip_accept; /* IP accept function */ 3567 void *rr_rx_handle; /* Handle for Rx ring */ 3568 squeue_t *rr_sqp; /* Squeue the ring is bound to */ 3569 ill_t *rr_ill; /* back pointer to ill */ 3570 ip_ring_state_t rr_ring_state; /* State of this ring */ 3571 }; 3572 3573 /* 3574 * IP - DLD direct function call capability 3575 * Suffixes, df - dld function, dh - dld handle, 3576 * cf - client (IP) function, ch - client handle 3577 */ 3578 typedef struct ill_dld_direct_s { /* DLD provided driver Tx */ 3579 ip_dld_tx_t idd_tx_df; /* str_mdata_fastpath_put */ 3580 void *idd_tx_dh; /* dld_str_t *dsp */ 3581 ip_dld_callb_t idd_tx_cb_df; /* mac_tx_srs_notify */ 3582 void *idd_tx_cb_dh; /* mac_client_handle_t *mch */ 3583 ip_dld_fctl_t idd_tx_fctl_df; /* mac_tx_is_flow_blocked */ 3584 void *idd_tx_fctl_dh; /* mac_client_handle */ 3585 } ill_dld_direct_t; 3586 3587 /* IP - DLD polling capability */ 3588 typedef struct ill_dld_poll_s { 3589 ill_rx_ring_t idp_ring_tbl[ILL_MAX_RINGS]; 3590 } ill_dld_poll_t; 3591 3592 /* Describes ill->ill_dld_capab */ 3593 struct ill_dld_capab_s { 3594 ip_capab_func_t idc_capab_df; /* dld_capab_func */ 3595 void *idc_capab_dh; /* dld_str_t *dsp */ 3596 ill_dld_direct_t idc_direct; 3597 ill_dld_poll_t idc_poll; 3598 }; 3599 3600 /* 3601 * IP squeues exports 3602 */ 3603 extern boolean_t ip_squeue_fanout; 3604 3605 #define IP_SQUEUE_GET(hint) ip_squeue_random(hint) 3606 3607 extern void ip_squeue_init(void (*)(squeue_t *)); 3608 extern squeue_t *ip_squeue_random(uint_t); 3609 extern squeue_t *ip_squeue_get(ill_rx_ring_t *); 3610 extern squeue_t *ip_squeue_getfree(pri_t); 3611 extern int ip_squeue_cpu_move(squeue_t *, processorid_t); 3612 extern void *ip_squeue_add_ring(ill_t *, void *); 3613 extern void ip_squeue_bind_ring(ill_t *, ill_rx_ring_t *, processorid_t); 3614 extern void ip_squeue_clean_ring(ill_t *, ill_rx_ring_t *); 3615 extern void ip_squeue_quiesce_ring(ill_t *, ill_rx_ring_t *); 3616 extern void ip_squeue_restart_ring(ill_t *, ill_rx_ring_t *); 3617 extern void ip_squeue_clean_all(ill_t *); 3618 extern boolean_t ip_source_routed(ipha_t *, ip_stack_t *); 3619 3620 extern void tcp_wput(queue_t *, mblk_t *); 3621 3622 extern int ip_fill_mtuinfo(conn_t *, ip_xmit_attr_t *, 3623 struct ip6_mtuinfo *); 3624 extern hook_t *ipobs_register_hook(netstack_t *, pfv_t); 3625 extern void ipobs_unregister_hook(netstack_t *, hook_t *); 3626 extern void ipobs_hook(mblk_t *, int, zoneid_t, zoneid_t, const ill_t *, 3627 ip_stack_t *); 3628 typedef void (*ipsq_func_t)(ipsq_t *, queue_t *, mblk_t *, void *); 3629 3630 extern void dce_g_init(void); 3631 extern void dce_g_destroy(void); 3632 extern void dce_stack_init(ip_stack_t *); 3633 extern void dce_stack_destroy(ip_stack_t *); 3634 extern void dce_cleanup(uint_t, ip_stack_t *); 3635 extern dce_t *dce_get_default(ip_stack_t *); 3636 extern dce_t *dce_lookup_pkt(mblk_t *, ip_xmit_attr_t *, uint_t *); 3637 extern dce_t *dce_lookup_v4(ipaddr_t, ip_stack_t *, uint_t *); 3638 extern dce_t *dce_lookup_v6(const in6_addr_t *, uint_t, ip_stack_t *, 3639 uint_t *); 3640 extern dce_t *dce_lookup_and_add_v4(ipaddr_t, ip_stack_t *); 3641 extern dce_t *dce_lookup_and_add_v6(const in6_addr_t *, uint_t, 3642 ip_stack_t *); 3643 extern int dce_update_uinfo_v4(ipaddr_t, iulp_t *, ip_stack_t *); 3644 extern int dce_update_uinfo_v6(const in6_addr_t *, uint_t, iulp_t *, 3645 ip_stack_t *); 3646 extern int dce_update_uinfo(const in6_addr_t *, uint_t, iulp_t *, 3647 ip_stack_t *); 3648 extern void dce_increment_generation(dce_t *); 3649 extern void dce_increment_all_generations(boolean_t, ip_stack_t *); 3650 extern void dce_refrele(dce_t *); 3651 extern void dce_refhold(dce_t *); 3652 extern void dce_refrele_notr(dce_t *); 3653 extern void dce_refhold_notr(dce_t *); 3654 mblk_t *ip_snmp_get_mib2_ip_dce(queue_t *, mblk_t *, ip_stack_t *ipst); 3655 3656 extern ip_laddr_t ip_laddr_verify_v4(ipaddr_t, zoneid_t, 3657 ip_stack_t *, boolean_t); 3658 extern ip_laddr_t ip_laddr_verify_v6(const in6_addr_t *, zoneid_t, 3659 ip_stack_t *, boolean_t, uint_t); 3660 extern int ip_laddr_fanout_insert(conn_t *); 3661 3662 extern boolean_t ip_verify_src(mblk_t *, ip_xmit_attr_t *, uint_t *); 3663 extern int ip_verify_ire(mblk_t *, ip_xmit_attr_t *); 3664 3665 extern mblk_t *ip_xmit_attr_to_mblk(ip_xmit_attr_t *); 3666 extern boolean_t ip_xmit_attr_from_mblk(mblk_t *, ip_xmit_attr_t *); 3667 extern mblk_t *ip_xmit_attr_free_mblk(mblk_t *); 3668 extern mblk_t *ip_recv_attr_to_mblk(ip_recv_attr_t *); 3669 extern boolean_t ip_recv_attr_from_mblk(mblk_t *, ip_recv_attr_t *); 3670 extern mblk_t *ip_recv_attr_free_mblk(mblk_t *); 3671 extern boolean_t ip_recv_attr_is_mblk(mblk_t *); 3672 3673 /* 3674 * Squeue tags. Tags only need to be unique when the callback function is the 3675 * same to distinguish between different calls, but we use unique tags for 3676 * convenience anyway. 3677 */ 3678 #define SQTAG_IP_INPUT 1 3679 #define SQTAG_TCP_INPUT_ICMP_ERR 2 3680 #define SQTAG_TCP6_INPUT_ICMP_ERR 3 3681 #define SQTAG_IP_TCP_INPUT 4 3682 #define SQTAG_IP6_TCP_INPUT 5 3683 #define SQTAG_IP_TCP_CLOSE 6 3684 #define SQTAG_TCP_OUTPUT 7 3685 #define SQTAG_TCP_TIMER 8 3686 #define SQTAG_TCP_TIMEWAIT 9 3687 #define SQTAG_TCP_ACCEPT_FINISH 10 3688 #define SQTAG_TCP_ACCEPT_FINISH_Q0 11 3689 #define SQTAG_TCP_ACCEPT_PENDING 12 3690 #define SQTAG_TCP_LISTEN_DISCON 13 3691 #define SQTAG_TCP_CONN_REQ_1 14 3692 #define SQTAG_TCP_EAGER_BLOWOFF 15 3693 #define SQTAG_TCP_EAGER_CLEANUP 16 3694 #define SQTAG_TCP_EAGER_CLEANUP_Q0 17 3695 #define SQTAG_TCP_CONN_IND 18 3696 #define SQTAG_TCP_RSRV 19 3697 #define SQTAG_TCP_ABORT_BUCKET 20 3698 #define SQTAG_TCP_REINPUT 21 3699 #define SQTAG_TCP_REINPUT_EAGER 22 3700 #define SQTAG_TCP_INPUT_MCTL 23 3701 #define SQTAG_TCP_RPUTOTHER 24 3702 #define SQTAG_IP_PROTO_AGAIN 25 3703 #define SQTAG_IP_FANOUT_TCP 26 3704 #define SQTAG_IPSQ_CLEAN_RING 27 3705 #define SQTAG_TCP_WPUT_OTHER 28 3706 #define SQTAG_TCP_CONN_REQ_UNBOUND 29 3707 #define SQTAG_TCP_SEND_PENDING 30 3708 #define SQTAG_BIND_RETRY 31 3709 #define SQTAG_UDP_FANOUT 32 3710 #define SQTAG_UDP_INPUT 33 3711 #define SQTAG_UDP_WPUT 34 3712 #define SQTAG_UDP_OUTPUT 35 3713 #define SQTAG_TCP_KSSL_INPUT 36 3714 #define SQTAG_TCP_DROP_Q0 37 3715 #define SQTAG_TCP_CONN_REQ_2 38 3716 #define SQTAG_IP_INPUT_RX_RING 39 3717 #define SQTAG_SQUEUE_CHANGE 40 3718 #define SQTAG_CONNECT_FINISH 41 3719 #define SQTAG_SYNCHRONOUS_OP 42 3720 #define SQTAG_TCP_SHUTDOWN_OUTPUT 43 3721 #define SQTAG_TCP_IXA_CLEANUP 44 3722 #define SQTAG_TCP_SEND_SYNACK 45 3723 3724 extern sin_t sin_null; /* Zero address for quick clears */ 3725 extern sin6_t sin6_null; /* Zero address for quick clears */ 3726 3727 #endif /* _KERNEL */ 3728 3729 #ifdef __cplusplus 3730 } 3731 #endif 3732 3733 #endif /* _INET_IP_H */ 3734