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 * Copyright 2010 Sun Microsystems, Inc. All rights reserved. 23 * Use is subject to license terms. 24 */ 25 26 /* 27 * Copyright 2020 OmniOS Community Edition (OmniOSce) Association. 28 */ 29 30 #ifndef _INET_IPCLASSIFIER_H 31 #define _INET_IPCLASSIFIER_H 32 33 #ifdef __cplusplus 34 extern "C" { 35 #endif 36 37 #include <inet/common.h> 38 #include <inet/ip.h> 39 #include <inet/mi.h> 40 #include <inet/tcp.h> 41 #include <inet/ip6.h> 42 #include <netinet/in.h> /* for IPPROTO_* constants */ 43 #include <sys/sdt.h> 44 #include <sys/socket_proto.h> 45 #include <sys/sunddi.h> 46 #include <sys/sunldi.h> 47 48 typedef void (*edesc_rpf)(void *, mblk_t *, void *, ip_recv_attr_t *); 49 struct icmph_s; 50 struct icmp6_hdr; 51 typedef boolean_t (*edesc_vpf)(conn_t *, void *, struct icmph_s *, 52 struct icmp6_hdr *, ip_recv_attr_t *); 53 54 /* 55 * ============================== 56 * = The CONNECTION = 57 * ============================== 58 */ 59 60 /* 61 * The connection structure contains the common information/flags/ref needed. 62 * Implementation will keep the connection struct, the layers (with their 63 * respective data for event i.e. tcp_t if event was tcp_input_data) all in one 64 * contiguous memory location. 65 */ 66 67 /* Conn Flags */ 68 /* Unused 0x00020000 */ 69 /* Unused 0x00040000 */ 70 #define IPCL_FULLY_BOUND 0x00080000 /* Bound to correct squeue */ 71 /* Unused 0x00100000 */ 72 /* Unused 0x00200000 */ 73 /* Unused 0x00400000 */ 74 #define IPCL_CL_LISTENER 0x00800000 /* Cluster listener */ 75 /* Unused 0x01000000 */ 76 /* Unused 0x02000000 */ 77 /* Unused 0x04000000 */ 78 /* Unused 0x08000000 */ 79 /* Unused 0x10000000 */ 80 /* Unused 0x20000000 */ 81 #define IPCL_CONNECTED 0x40000000 /* Conn in connected table */ 82 #define IPCL_BOUND 0x80000000 /* Conn in bind table */ 83 84 /* Flags identifying the type of conn */ 85 #define IPCL_TCPCONN 0x00000001 /* From tcp_conn_cache */ 86 #define IPCL_SCTPCONN 0x00000002 /* From sctp_conn_cache */ 87 #define IPCL_IPCCONN 0x00000004 /* From ip_conn_cache */ 88 #define IPCL_UDPCONN 0x00000008 /* From udp_conn_cache */ 89 #define IPCL_RAWIPCONN 0x00000010 /* From rawip_conn_cache */ 90 #define IPCL_RTSCONN 0x00000020 /* From rts_conn_cache */ 91 /* Unused 0x00000040 */ 92 #define IPCL_IPTUN 0x00000080 /* iptun module above us */ 93 94 #define IPCL_NONSTR 0x00001000 /* A non-STREAMS socket */ 95 /* Unused 0x10000000 */ 96 97 #define IPCL_REMOVED 0x00000100 98 #define IPCL_REUSED 0x00000200 99 100 #define IPCL_IS_CONNECTED(connp) \ 101 ((connp)->conn_flags & IPCL_CONNECTED) 102 103 #define IPCL_IS_BOUND(connp) \ 104 ((connp)->conn_flags & IPCL_BOUND) 105 106 /* 107 * Can't use conn_proto since we need to tell difference 108 * between a real TCP socket and a SOCK_RAW, IPPROTO_TCP. 109 */ 110 #define IPCL_IS_TCP(connp) \ 111 ((connp)->conn_flags & IPCL_TCPCONN) 112 113 #define IPCL_IS_SCTP(connp) \ 114 ((connp)->conn_flags & IPCL_SCTPCONN) 115 116 #define IPCL_IS_UDP(connp) \ 117 ((connp)->conn_flags & IPCL_UDPCONN) 118 119 #define IPCL_IS_RAWIP(connp) \ 120 ((connp)->conn_flags & IPCL_RAWIPCONN) 121 122 #define IPCL_IS_RTS(connp) \ 123 ((connp)->conn_flags & IPCL_RTSCONN) 124 125 #define IPCL_IS_IPTUN(connp) \ 126 ((connp)->conn_flags & IPCL_IPTUN) 127 128 #define IPCL_IS_NONSTR(connp) ((connp)->conn_flags & IPCL_NONSTR) 129 130 typedef struct connf_s connf_t; 131 132 typedef struct 133 { 134 int ctb_depth; 135 #define CONN_STACK_DEPTH 15 136 pc_t ctb_stack[CONN_STACK_DEPTH]; 137 } conn_trace_t; 138 139 typedef struct ip_helper_minor_info_s { 140 dev_t ip_minfo_dev; /* Device */ 141 vmem_t *ip_minfo_arena; /* Arena */ 142 } ip_helper_minfo_t; 143 144 /* 145 * ip helper stream info 146 */ 147 typedef struct ip_helper_stream_info_s { 148 ldi_handle_t iphs_handle; 149 queue_t *iphs_rq; 150 queue_t *iphs_wq; 151 ip_helper_minfo_t *iphs_minfo; 152 } ip_helper_stream_info_t; 153 154 /* 155 * Mandatory Access Control mode, in conn_t's conn_mac_mode field. 156 * CONN_MAC_DEFAULT: strict enforcement of MAC. 157 * CONN_MAC_AWARE: allows communications between unlabeled systems 158 * and privileged daemons 159 * CONN_MAC_IMPLICIT: allows communications without explicit labels 160 * on the wire with privileged daemons. 161 * 162 * CONN_MAC_IMPLICIT is intended specifically for labeled IPsec key management 163 * in networks which don't pass CIPSO-labeled packets. 164 */ 165 #define CONN_MAC_DEFAULT 0 166 #define CONN_MAC_AWARE 1 167 #define CONN_MAC_IMPLICIT 2 168 169 /* 170 * conn receive ancillary definition. 171 * 172 * These are the set of socket options that make the receive side 173 * potentially pass up ancillary data items. 174 * We have a union with an integer so that we can quickly check whether 175 * any ancillary data items need to be added. 176 */ 177 typedef struct crb_s { 178 union { 179 uint32_t crbu_all; 180 struct { 181 uint32_t 182 crbb_recvdstaddr : 1, /* IP_RECVDSTADDR option */ 183 crbb_recvopts : 1, /* IP_RECVOPTS option */ 184 crbb_recvif : 1, /* IP_RECVIF option */ 185 crbb_recvslla : 1, /* IP_RECVSLLA option */ 186 187 crbb_recvttl : 1, /* IP_RECVTTL option */ 188 crbb_recvtos : 1, /* IP_RECVTOS option */ 189 crbb_ip_recvpktinfo : 1, /* IP*_RECVPKTINFO option */ 190 crbb_ipv6_recvhoplimit : 1, /* IPV6_RECVHOPLIMIT option */ 191 crbb_ipv6_recvhopopts : 1, /* IPV6_RECVHOPOPTS option */ 192 193 crbb_ipv6_recvdstopts : 1, /* IPV6_RECVDSTOPTS option */ 194 crbb_ipv6_recvrthdr : 1, /* IPV6_RECVRTHDR option */ 195 crbb_old_ipv6_recvdstopts : 1, /* old form of IPV6_DSTOPTS */ 196 crbb_ipv6_recvrthdrdstopts : 1, /* IPV6_RECVRTHDRDSTOPTS */ 197 198 crbb_ipv6_recvtclass : 1, /* IPV6_RECVTCLASS */ 199 crbb_recvucred : 1, /* IP_RECVUCRED option */ 200 crbb_timestamp : 1; /* SO_TIMESTAMP "socket" option */ 201 202 } crbb; 203 } crbu; 204 } crb_t; 205 206 #define crb_all crbu.crbu_all 207 #define crb_recvdstaddr crbu.crbb.crbb_recvdstaddr 208 #define crb_recvopts crbu.crbb.crbb_recvopts 209 #define crb_recvif crbu.crbb.crbb_recvif 210 #define crb_recvslla crbu.crbb.crbb_recvslla 211 #define crb_recvttl crbu.crbb.crbb_recvttl 212 #define crb_recvtos crbu.crbb.crbb_recvtos 213 #define crb_ip_recvpktinfo crbu.crbb.crbb_ip_recvpktinfo 214 #define crb_ipv6_recvhoplimit crbu.crbb.crbb_ipv6_recvhoplimit 215 #define crb_ipv6_recvhopopts crbu.crbb.crbb_ipv6_recvhopopts 216 #define crb_ipv6_recvdstopts crbu.crbb.crbb_ipv6_recvdstopts 217 #define crb_ipv6_recvrthdr crbu.crbb.crbb_ipv6_recvrthdr 218 #define crb_old_ipv6_recvdstopts crbu.crbb.crbb_old_ipv6_recvdstopts 219 #define crb_ipv6_recvrthdrdstopts crbu.crbb.crbb_ipv6_recvrthdrdstopts 220 #define crb_ipv6_recvtclass crbu.crbb.crbb_ipv6_recvtclass 221 #define crb_recvucred crbu.crbb.crbb_recvucred 222 #define crb_timestamp crbu.crbb.crbb_timestamp 223 224 /* 225 * The initial fields in the conn_t are setup by the kmem_cache constructor, 226 * and are preserved when it is freed. Fields after that are bzero'ed when 227 * the conn_t is freed. 228 * 229 * Much of the conn_t is protected by conn_lock. 230 * 231 * conn_lock is also used by some ULPs (like UDP and RAWIP) to protect 232 * their state. 233 */ 234 struct conn_s { 235 kmutex_t conn_lock; 236 uint32_t conn_ref; /* Reference counter */ 237 uint32_t conn_flags; /* Conn Flags */ 238 239 union { 240 tcp_t *cp_tcp; /* Pointer to the tcp struct */ 241 struct udp_s *cp_udp; /* Pointer to the udp struct */ 242 struct icmp_s *cp_icmp; /* Pointer to rawip struct */ 243 struct rts_s *cp_rts; /* Pointer to rts struct */ 244 struct iptun_s *cp_iptun; /* Pointer to iptun_t */ 245 struct sctp_s *cp_sctp; /* For IPCL_SCTPCONN */ 246 void *cp_priv; 247 } conn_proto_priv; 248 #define conn_tcp conn_proto_priv.cp_tcp 249 #define conn_udp conn_proto_priv.cp_udp 250 #define conn_icmp conn_proto_priv.cp_icmp 251 #define conn_rts conn_proto_priv.cp_rts 252 #define conn_iptun conn_proto_priv.cp_iptun 253 #define conn_sctp conn_proto_priv.cp_sctp 254 #define conn_priv conn_proto_priv.cp_priv 255 256 kcondvar_t conn_cv; 257 uint8_t conn_proto; /* protocol type */ 258 259 edesc_rpf conn_recv; /* Pointer to recv routine */ 260 edesc_rpf conn_recvicmp; /* For ICMP error */ 261 edesc_vpf conn_verifyicmp; /* Verify ICMP error */ 262 263 ip_xmit_attr_t *conn_ixa; /* Options if no ancil data */ 264 265 /* Fields after this are bzero'ed when the conn_t is freed. */ 266 #define conn_start_clr conn_recv_ancillary 267 268 /* Options for receive-side ancillary data */ 269 crb_t conn_recv_ancillary; 270 271 squeue_t *conn_sqp; /* Squeue for processing */ 272 uint_t conn_state_flags; /* IP state flags */ 273 274 int conn_lingertime; /* linger time (in seconds) */ 275 276 unsigned int 277 conn_on_sqp : 1, /* Conn is being processed */ 278 conn_linger : 1, /* SO_LINGER state */ 279 conn_useloopback : 1, /* SO_USELOOPBACK state */ 280 conn_broadcast : 1, /* SO_BROADCAST state */ 281 282 conn_reuseaddr : 1, /* SO_REUSEADDR state */ 283 conn_keepalive : 1, /* SO_KEEPALIVE state */ 284 conn_multi_router : 1, /* Wants all multicast pkts */ 285 conn_unspec_src : 1, /* IP_UNSPEC_SRC */ 286 287 conn_policy_cached : 1, /* Is policy cached/latched ? */ 288 conn_in_enforce_policy : 1, /* Enforce Policy on inbound */ 289 conn_out_enforce_policy : 1, /* Enforce Policy on outbound */ 290 conn_debug : 1, /* SO_DEBUG */ 291 292 conn_ipv6_v6only : 1, /* IPV6_V6ONLY */ 293 conn_oobinline : 1, /* SO_OOBINLINE state */ 294 conn_dgram_errind : 1, /* SO_DGRAM_ERRIND state */ 295 conn_exclbind : 1, /* SO_EXCLBIND state */ 296 297 conn_mdt_ok : 1, /* MDT is permitted */ 298 conn_allzones : 1, /* SO_ALLZONES */ 299 conn_ipv6_recvpathmtu : 1, /* IPV6_RECVPATHMTU */ 300 conn_mcbc_bind : 1, /* Bound to multi/broadcast */ 301 302 conn_pad_to_bit_31 : 12; 303 304 boolean_t conn_blocked; /* conn is flow-controlled */ 305 306 squeue_t *conn_initial_sqp; /* Squeue at open time */ 307 squeue_t *conn_final_sqp; /* Squeue after connect */ 308 ill_t *conn_dhcpinit_ill; /* IP_DHCPINIT_IF */ 309 ipsec_latch_t *conn_latch; /* latched IDS */ 310 struct ipsec_policy_s *conn_latch_in_policy; /* latched policy (in) */ 311 struct ipsec_action_s *conn_latch_in_action; /* latched action (in) */ 312 uint_t conn_bound_if; /* IP*_BOUND_IF */ 313 queue_t *conn_rq; /* Read queue */ 314 queue_t *conn_wq; /* Write queue */ 315 dev_t conn_dev; /* Minor number */ 316 vmem_t *conn_minor_arena; /* Minor arena */ 317 ip_helper_stream_info_t *conn_helper_info; 318 319 cred_t *conn_cred; /* Credentials */ 320 pid_t conn_cpid; /* pid from open/connect */ 321 uint64_t conn_open_time; /* time when this was opened */ 322 323 connf_t *conn_g_fanout; /* Global Hash bucket head */ 324 struct conn_s *conn_g_next; /* Global Hash chain next */ 325 struct conn_s *conn_g_prev; /* Global Hash chain prev */ 326 struct ipsec_policy_head_s *conn_policy; /* Configured policy */ 327 in6_addr_t conn_bound_addr_v6; /* Address in bind() */ 328 #define conn_bound_addr_v4 V4_PART_OF_V6(conn_bound_addr_v6) 329 connf_t *conn_fanout; /* Hash bucket we're part of */ 330 struct conn_s *conn_next; /* Hash chain next */ 331 struct conn_s *conn_prev; /* Hash chain prev */ 332 333 struct { 334 in6_addr_t connua_laddr; /* Local address - match */ 335 in6_addr_t connua_faddr; /* Remote address */ 336 } connua_v6addr; 337 #define conn_laddr_v4 V4_PART_OF_V6(connua_v6addr.connua_laddr) 338 #define conn_faddr_v4 V4_PART_OF_V6(connua_v6addr.connua_faddr) 339 #define conn_laddr_v6 connua_v6addr.connua_laddr 340 #define conn_faddr_v6 connua_v6addr.connua_faddr 341 in6_addr_t conn_saddr_v6; /* Local address - source */ 342 #define conn_saddr_v4 V4_PART_OF_V6(conn_saddr_v6) 343 344 union { 345 /* Used for classifier match performance */ 346 uint32_t connu_ports2; 347 struct { 348 in_port_t connu_fport; /* Remote port */ 349 in_port_t connu_lport; /* Local port */ 350 } connu_ports; 351 } u_port; 352 #define conn_fport u_port.connu_ports.connu_fport 353 #define conn_lport u_port.connu_ports.connu_lport 354 #define conn_ports u_port.connu_ports2 355 356 uint_t conn_incoming_ifindex; /* IP{,V6}_BOUND_IF, scopeid */ 357 ill_t *conn_oper_pending_ill; /* pending shared ioctl */ 358 359 krwlock_t conn_ilg_lock; /* Protects conn_ilg_* */ 360 ilg_t *conn_ilg; /* Group memberships */ 361 362 kcondvar_t conn_refcv; /* For conn_oper_pending_ill */ 363 364 struct conn_s *conn_drain_next; /* Next conn in drain list */ 365 struct conn_s *conn_drain_prev; /* Prev conn in drain list */ 366 idl_t *conn_idl; /* Ptr to the drain list head */ 367 mblk_t *conn_ipsec_opt_mp; /* ipsec option mblk */ 368 zoneid_t conn_zoneid; /* zone connection is in */ 369 int conn_rtaware; /* RT_AWARE sockopt value */ 370 kcondvar_t conn_sq_cv; /* For non-STREAMS socket IO */ 371 sock_upcalls_t *conn_upcalls; /* Upcalls to sockfs */ 372 sock_upper_handle_t conn_upper_handle; /* Upper handle: sonode * */ 373 374 unsigned int 375 conn_mlp_type : 2, /* mlp_type_t; tsol/tndb.h */ 376 conn_anon_mlp : 1, /* user wants anon MLP */ 377 conn_anon_port : 1, /* user bound anonymously */ 378 379 conn_mac_mode : 2, /* normal/loose/implicit MAC */ 380 conn_anon_priv_bind : 1, /* *_ANON_PRIV_BIND state */ 381 conn_zone_is_global : 1, /* GLOBAL_ZONEID */ 382 conn_isvrrp : 1, /* VRRP control socket */ 383 conn_spare : 23; 384 385 boolean_t conn_flow_cntrld; 386 netstack_t *conn_netstack; /* Corresponds to a netstack_hold */ 387 388 /* 389 * IP format that packets received for this struct should use. 390 * Value can be IP4_VERSION or IPV6_VERSION. 391 * The sending version is encoded using IXAF_IS_IPV4. 392 */ 393 ushort_t conn_ipversion; 394 395 /* Written to only once at the time of opening the endpoint */ 396 sa_family_t conn_family; /* Family from socket() call */ 397 uint_t conn_so_type; /* Type from socket() call */ 398 399 uint_t conn_sndbuf; /* SO_SNDBUF state */ 400 uint_t conn_rcvbuf; /* SO_RCVBUF state */ 401 uint_t conn_wroff; /* Current write offset */ 402 403 uint_t conn_sndlowat; /* Send buffer low water mark */ 404 uint_t conn_rcvlowat; /* Recv buffer low water mark */ 405 406 uint8_t conn_default_ttl; /* Default TTL/hoplimit */ 407 uint8_t conn_min_ttl; /* IP_MINTTL+IPV6_MINHOPLIMIT */ 408 409 uint32_t conn_flowinfo; /* Connected flow id and tclass */ 410 411 /* 412 * The most recent address for sendto. Initially set to zero 413 * which is always different than then the destination address 414 * since the send interprets zero as the loopback address. 415 */ 416 in6_addr_t conn_v6lastdst; 417 #define conn_v4lastdst V4_PART_OF_V6(conn_v6lastdst) 418 ushort_t conn_lastipversion; 419 in_port_t conn_lastdstport; 420 uint32_t conn_lastflowinfo; /* IPv6-only */ 421 uint_t conn_lastscopeid; /* IPv6-only */ 422 uint_t conn_lastsrcid; /* Only for AF_INET6 */ 423 /* 424 * When we are not connected conn_saddr might be unspecified. 425 * We track the source that was used with conn_v6lastdst here. 426 */ 427 in6_addr_t conn_v6lastsrc; 428 #define conn_v4lastsrc V4_PART_OF_V6(conn_v6lastsrc) 429 430 /* Templates for transmitting packets */ 431 ip_pkt_t conn_xmit_ipp; /* Options if no ancil data */ 432 433 /* 434 * Header template - conn_ht_ulp is a pointer into conn_ht_iphc. 435 * Note that ixa_ip_hdr_length indicates the offset of ht_ulp in 436 * ht_iphc 437 * 438 * The header template is maintained for connected endpoints (and 439 * updated when sticky options are changed) and also for the lastdst. 440 * There is no conflict between those usages since SOCK_DGRAM and 441 * SOCK_RAW can not be used to specify a destination address (with 442 * sendto/sendmsg) if the socket has been connected. 443 */ 444 uint8_t *conn_ht_iphc; /* Start of IP header */ 445 uint_t conn_ht_iphc_allocated; /* Allocated buffer size */ 446 uint_t conn_ht_iphc_len; /* IP+ULP size */ 447 uint8_t *conn_ht_ulp; /* Upper-layer header */ 448 uint_t conn_ht_ulp_len; /* ULP header len */ 449 450 /* Checksum to compensate for source routed packets. Host byte order */ 451 uint32_t conn_sum; 452 453 uint32_t conn_ioctlref; /* ioctl ref count */ 454 #ifdef CONN_DEBUG 455 #define CONN_TRACE_MAX 10 456 int conn_trace_last; /* ndx of last used tracebuf */ 457 conn_trace_t conn_trace_buf[CONN_TRACE_MAX]; 458 #endif 459 }; 460 461 /* 462 * connf_t - connection fanout data. 463 * 464 * The hash tables and their linkage (conn_t.{hashnextp, hashprevp} are 465 * protected by the per-bucket lock. Each conn_t inserted in the list 466 * points back at the connf_t that heads the bucket. 467 */ 468 struct connf_s { 469 struct conn_s *connf_head; 470 kmutex_t connf_lock; 471 }; 472 473 #define CONN_INC_REF(connp) { \ 474 mutex_enter(&(connp)->conn_lock); \ 475 DTRACE_PROBE1(conn__inc__ref, conn_t *, connp); \ 476 ASSERT(conn_trace_ref(connp)); \ 477 (connp)->conn_ref++; \ 478 ASSERT((connp)->conn_ref != 0); \ 479 mutex_exit(&(connp)->conn_lock); \ 480 } 481 482 #define CONN_INC_REF_LOCKED(connp) { \ 483 DTRACE_PROBE1(conn__inc__ref, conn_t *, connp); \ 484 ASSERT(MUTEX_HELD(&(connp)->conn_lock)); \ 485 ASSERT(conn_trace_ref(connp)); \ 486 (connp)->conn_ref++; \ 487 ASSERT((connp)->conn_ref != 0); \ 488 } 489 490 #define CONN_DEC_REF(connp) { \ 491 mutex_enter(&(connp)->conn_lock); \ 492 DTRACE_PROBE1(conn__dec__ref, conn_t *, connp); \ 493 /* \ 494 * The squeue framework always does a CONN_DEC_REF after return \ 495 * from TCP. Hence the refcnt must be at least 2 if conn_on_sqp \ 496 * is B_TRUE and conn_ref is being decremented. This is to \ 497 * account for the mblk being currently processed. \ 498 */ \ 499 if ((connp)->conn_ref == 0 || \ 500 ((connp)->conn_ref == 1 && (connp)->conn_on_sqp)) \ 501 cmn_err(CE_PANIC, "CONN_DEC_REF: connp(%p) has ref " \ 502 "= %d\n", (void *)(connp), (connp)->conn_ref); \ 503 ASSERT(conn_untrace_ref(connp)); \ 504 (connp)->conn_ref--; \ 505 if ((connp)->conn_ref == 0) { \ 506 /* Refcnt can't increase again, safe to drop lock */ \ 507 mutex_exit(&(connp)->conn_lock); \ 508 ipcl_conn_destroy(connp); \ 509 } else { \ 510 cv_broadcast(&(connp)->conn_cv); \ 511 mutex_exit(&(connp)->conn_lock); \ 512 } \ 513 } 514 515 /* 516 * For use with subsystems within ip which use ALL_ZONES as a wildcard 517 */ 518 #define IPCL_ZONEID(connp) \ 519 ((connp)->conn_allzones ? ALL_ZONES : (connp)->conn_zoneid) 520 521 /* 522 * For matching between a conn_t and a zoneid. 523 */ 524 #define IPCL_ZONE_MATCH(connp, zoneid) \ 525 (((connp)->conn_allzones) || \ 526 ((zoneid) == ALL_ZONES) || \ 527 (connp)->conn_zoneid == (zoneid)) 528 529 /* 530 * On a labeled system, we must treat bindings to ports 531 * on shared IP addresses by sockets with MAC exemption 532 * privilege as being in all zones, as there's 533 * otherwise no way to identify the right receiver. 534 */ 535 536 #define IPCL_CONNS_MAC(conn1, conn2) \ 537 (((conn1)->conn_mac_mode != CONN_MAC_DEFAULT) || \ 538 ((conn2)->conn_mac_mode != CONN_MAC_DEFAULT)) 539 540 #define IPCL_BIND_ZONE_MATCH(conn1, conn2) \ 541 (IPCL_CONNS_MAC(conn1, conn2) || \ 542 IPCL_ZONE_MATCH(conn1, conn2->conn_zoneid) || \ 543 IPCL_ZONE_MATCH(conn2, conn1->conn_zoneid)) 544 545 546 #define _IPCL_V4_MATCH(v6addr, v4addr) \ 547 (V4_PART_OF_V6((v6addr)) == (v4addr) && IN6_IS_ADDR_V4MAPPED(&(v6addr))) 548 549 #define _IPCL_V4_MATCH_ANY(addr) \ 550 (IN6_IS_ADDR_V4MAPPED_ANY(&(addr)) || IN6_IS_ADDR_UNSPECIFIED(&(addr))) 551 552 553 /* 554 * IPCL_PROTO_MATCH() and IPCL_PROTO_MATCH_V6() only matches conns with 555 * the specified ira_zoneid or conn_allzones by calling conn_wantpacket. 556 */ 557 #define IPCL_PROTO_MATCH(connp, ira, ipha) \ 558 ((((connp)->conn_laddr_v4 == INADDR_ANY) || \ 559 (((connp)->conn_laddr_v4 == ((ipha)->ipha_dst)) && \ 560 (((connp)->conn_faddr_v4 == INADDR_ANY) || \ 561 ((connp)->conn_faddr_v4 == ((ipha)->ipha_src))))) && \ 562 conn_wantpacket((connp), (ira), (ipha))) 563 564 #define IPCL_PROTO_MATCH_V6(connp, ira, ip6h) \ 565 ((IN6_IS_ADDR_UNSPECIFIED(&(connp)->conn_laddr_v6) || \ 566 (IN6_ARE_ADDR_EQUAL(&(connp)->conn_laddr_v6, &((ip6h)->ip6_dst)) && \ 567 (IN6_IS_ADDR_UNSPECIFIED(&(connp)->conn_faddr_v6) || \ 568 IN6_ARE_ADDR_EQUAL(&(connp)->conn_faddr_v6, &((ip6h)->ip6_src))))) && \ 569 (conn_wantpacket_v6((connp), (ira), (ip6h)))) 570 571 #define IPCL_CONN_HASH(src, ports, ipst) \ 572 ((unsigned)(ntohl((src)) ^ ((ports) >> 24) ^ ((ports) >> 16) ^ \ 573 ((ports) >> 8) ^ (ports)) % (ipst)->ips_ipcl_conn_fanout_size) 574 575 #define IPCL_CONN_HASH_V6(src, ports, ipst) \ 576 IPCL_CONN_HASH(V4_PART_OF_V6((src)), (ports), (ipst)) 577 578 #define IPCL_CONN_MATCH(connp, proto, src, dst, ports) \ 579 ((connp)->conn_proto == (proto) && \ 580 (connp)->conn_ports == (ports) && \ 581 _IPCL_V4_MATCH((connp)->conn_faddr_v6, (src)) && \ 582 _IPCL_V4_MATCH((connp)->conn_laddr_v6, (dst)) && \ 583 !(connp)->conn_ipv6_v6only) 584 585 #define IPCL_CONN_MATCH_V6(connp, proto, src, dst, ports) \ 586 ((connp)->conn_proto == (proto) && \ 587 (connp)->conn_ports == (ports) && \ 588 IN6_ARE_ADDR_EQUAL(&(connp)->conn_faddr_v6, &(src)) && \ 589 IN6_ARE_ADDR_EQUAL(&(connp)->conn_laddr_v6, &(dst))) 590 591 #define IPCL_PORT_HASH(port, size) \ 592 ((((port) >> 8) ^ (port)) & ((size) - 1)) 593 594 #define IPCL_BIND_HASH(lport, ipst) \ 595 ((unsigned)(((lport) >> 8) ^ (lport)) % \ 596 (ipst)->ips_ipcl_bind_fanout_size) 597 598 #define IPCL_BIND_MATCH(connp, proto, laddr, lport) \ 599 ((connp)->conn_proto == (proto) && \ 600 (connp)->conn_lport == (lport) && \ 601 (_IPCL_V4_MATCH_ANY((connp)->conn_laddr_v6) || \ 602 _IPCL_V4_MATCH((connp)->conn_laddr_v6, (laddr))) && \ 603 !(connp)->conn_ipv6_v6only) 604 605 #define IPCL_BIND_MATCH_V6(connp, proto, laddr, lport) \ 606 ((connp)->conn_proto == (proto) && \ 607 (connp)->conn_lport == (lport) && \ 608 (IN6_ARE_ADDR_EQUAL(&(connp)->conn_laddr_v6, &(laddr)) || \ 609 IN6_IS_ADDR_UNSPECIFIED(&(connp)->conn_laddr_v6))) 610 611 /* 612 * We compare conn_laddr since it captures both connected and a bind to 613 * a multicast or broadcast address. 614 * The caller needs to match the zoneid and also call conn_wantpacket 615 * for multicast, broadcast, or when conn_incoming_ifindex is set. 616 */ 617 #define IPCL_UDP_MATCH(connp, lport, laddr, fport, faddr) \ 618 (((connp)->conn_lport == (lport)) && \ 619 ((_IPCL_V4_MATCH_ANY((connp)->conn_laddr_v6) || \ 620 (_IPCL_V4_MATCH((connp)->conn_laddr_v6, (laddr)) && \ 621 (_IPCL_V4_MATCH_ANY((connp)->conn_faddr_v6) || \ 622 (_IPCL_V4_MATCH((connp)->conn_faddr_v6, (faddr)) && \ 623 (connp)->conn_fport == (fport)))))) && \ 624 !(connp)->conn_ipv6_v6only) 625 626 /* 627 * We compare conn_laddr since it captures both connected and a bind to 628 * a multicast or broadcast address. 629 * The caller needs to match the zoneid and also call conn_wantpacket_v6 630 * for multicast or when conn_incoming_ifindex is set. 631 */ 632 #define IPCL_UDP_MATCH_V6(connp, lport, laddr, fport, faddr) \ 633 (((connp)->conn_lport == (lport)) && \ 634 (IN6_IS_ADDR_UNSPECIFIED(&(connp)->conn_laddr_v6) || \ 635 (IN6_ARE_ADDR_EQUAL(&(connp)->conn_laddr_v6, &(laddr)) && \ 636 (IN6_IS_ADDR_UNSPECIFIED(&(connp)->conn_faddr_v6) || \ 637 (IN6_ARE_ADDR_EQUAL(&(connp)->conn_faddr_v6, &(faddr)) && \ 638 (connp)->conn_fport == (fport)))))) 639 640 #define IPCL_IPTUN_HASH(laddr, faddr) \ 641 ((ntohl(laddr) ^ ((ntohl(faddr) << 24) | (ntohl(faddr) >> 8))) % \ 642 ipcl_iptun_fanout_size) 643 644 #define IPCL_IPTUN_HASH_V6(laddr, faddr) \ 645 IPCL_IPTUN_HASH((laddr)->s6_addr32[0] ^ (laddr)->s6_addr32[1] ^ \ 646 (faddr)->s6_addr32[2] ^ (faddr)->s6_addr32[3], \ 647 (faddr)->s6_addr32[0] ^ (faddr)->s6_addr32[1] ^ \ 648 (laddr)->s6_addr32[2] ^ (laddr)->s6_addr32[3]) 649 650 #define IPCL_IPTUN_MATCH(connp, laddr, faddr) \ 651 (_IPCL_V4_MATCH((connp)->conn_laddr_v6, (laddr)) && \ 652 _IPCL_V4_MATCH((connp)->conn_faddr_v6, (faddr))) 653 654 #define IPCL_IPTUN_MATCH_V6(connp, laddr, faddr) \ 655 (IN6_ARE_ADDR_EQUAL(&(connp)->conn_laddr_v6, (laddr)) && \ 656 IN6_ARE_ADDR_EQUAL(&(connp)->conn_faddr_v6, (faddr))) 657 658 #define IPCL_UDP_HASH(lport, ipst) \ 659 IPCL_PORT_HASH(lport, (ipst)->ips_ipcl_udp_fanout_size) 660 661 #define CONN_G_HASH_SIZE 1024 662 663 /* Raw socket hash function. */ 664 #define IPCL_RAW_HASH(lport, ipst) \ 665 IPCL_PORT_HASH(lport, (ipst)->ips_ipcl_raw_fanout_size) 666 667 /* 668 * This is similar to IPCL_BIND_MATCH except that the local port check 669 * is changed to a wildcard port check. 670 * We compare conn_laddr since it captures both connected and a bind to 671 * a multicast or broadcast address. 672 */ 673 #define IPCL_RAW_MATCH(connp, proto, laddr) \ 674 ((connp)->conn_proto == (proto) && \ 675 (connp)->conn_lport == 0 && \ 676 (_IPCL_V4_MATCH_ANY((connp)->conn_laddr_v6) || \ 677 _IPCL_V4_MATCH((connp)->conn_laddr_v6, (laddr)))) 678 679 #define IPCL_RAW_MATCH_V6(connp, proto, laddr) \ 680 ((connp)->conn_proto == (proto) && \ 681 (connp)->conn_lport == 0 && \ 682 (IN6_IS_ADDR_UNSPECIFIED(&(connp)->conn_laddr_v6) || \ 683 IN6_ARE_ADDR_EQUAL(&(connp)->conn_laddr_v6, &(laddr)))) 684 685 /* Function prototypes */ 686 extern void ipcl_g_init(void); 687 extern void ipcl_init(ip_stack_t *); 688 extern void ipcl_g_destroy(void); 689 extern void ipcl_destroy(ip_stack_t *); 690 extern conn_t *ipcl_conn_create(uint32_t, int, netstack_t *); 691 extern void ipcl_conn_destroy(conn_t *); 692 693 void ipcl_hash_insert_wildcard(connf_t *, conn_t *); 694 void ipcl_hash_remove(conn_t *); 695 void ipcl_hash_remove_locked(conn_t *connp, connf_t *connfp); 696 697 extern int ipcl_bind_insert(conn_t *); 698 extern int ipcl_bind_insert_v4(conn_t *); 699 extern int ipcl_bind_insert_v6(conn_t *); 700 extern int ipcl_conn_insert(conn_t *); 701 extern int ipcl_conn_insert_v4(conn_t *); 702 extern int ipcl_conn_insert_v6(conn_t *); 703 extern conn_t *ipcl_get_next_conn(connf_t *, conn_t *, uint32_t); 704 705 conn_t *ipcl_classify_v4(mblk_t *, uint8_t, uint_t, ip_recv_attr_t *, 706 ip_stack_t *); 707 conn_t *ipcl_classify_v6(mblk_t *, uint8_t, uint_t, ip_recv_attr_t *, 708 ip_stack_t *); 709 conn_t *ipcl_classify(mblk_t *, ip_recv_attr_t *, ip_stack_t *); 710 conn_t *ipcl_classify_raw(mblk_t *, uint8_t, uint32_t, ipha_t *, 711 ip6_t *, ip_recv_attr_t *, ip_stack_t *); 712 conn_t *ipcl_iptun_classify_v4(ipaddr_t *, ipaddr_t *, ip_stack_t *); 713 conn_t *ipcl_iptun_classify_v6(in6_addr_t *, in6_addr_t *, ip_stack_t *); 714 void ipcl_globalhash_insert(conn_t *); 715 void ipcl_globalhash_remove(conn_t *); 716 void ipcl_walk(pfv_t, void *, ip_stack_t *); 717 conn_t *ipcl_tcp_lookup_reversed_ipv4(ipha_t *, tcpha_t *, int, ip_stack_t *); 718 conn_t *ipcl_tcp_lookup_reversed_ipv6(ip6_t *, tcpha_t *, int, uint_t, 719 ip_stack_t *); 720 conn_t *ipcl_lookup_listener_v4(uint16_t, ipaddr_t, zoneid_t, ip_stack_t *); 721 conn_t *ipcl_lookup_listener_v6(uint16_t, in6_addr_t *, uint_t, zoneid_t, 722 ip_stack_t *); 723 int conn_trace_ref(conn_t *); 724 int conn_untrace_ref(conn_t *); 725 void ipcl_conn_cleanup(conn_t *); 726 extern uint_t conn_recvancillary_size(conn_t *, crb_t, ip_recv_attr_t *, 727 mblk_t *, ip_pkt_t *); 728 extern void conn_recvancillary_add(conn_t *, crb_t, ip_recv_attr_t *, 729 ip_pkt_t *, uchar_t *, uint_t); 730 conn_t *ipcl_conn_tcp_lookup_reversed_ipv4(conn_t *, ipha_t *, tcpha_t *, 731 ip_stack_t *); 732 conn_t *ipcl_conn_tcp_lookup_reversed_ipv6(conn_t *, ip6_t *, tcpha_t *, 733 ip_stack_t *); 734 735 extern int ip_create_helper_stream(conn_t *, ldi_ident_t); 736 extern void ip_free_helper_stream(conn_t *); 737 extern int ip_helper_stream_setup(queue_t *, dev_t *, int, int, 738 cred_t *, boolean_t); 739 extern mib2_socketInfoEntry_t *conn_get_socket_info(conn_t *, 740 mib2_socketInfoEntry_t *); 741 742 #ifdef __cplusplus 743 } 744 #endif 745 746 #endif /* _INET_IPCLASSIFIER_H */ 747