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 2007 Sun Microsystems, Inc. All rights reserved. 23 * Use is subject to license terms. 24 */ 25 /* Copyright (c) 1990 Mentat Inc. */ 26 27 #ifndef _INET_IP_IRE_H 28 #define _INET_IP_IRE_H 29 30 #pragma ident "%Z%%M% %I% %E% SMI" 31 32 #ifdef __cplusplus 33 extern "C" { 34 #endif 35 36 #define IPV6_LL_PREFIXLEN 10 /* Number of bits in link-local pref */ 37 38 #define IP_CACHE_TABLE_SIZE 256 39 #define IP_MASK_TABLE_SIZE (IP_ABITS + 1) /* 33 ptrs */ 40 41 #define IP6_FTABLE_HASH_SIZE 32 /* size of each hash table in ptrs */ 42 #define IP6_CACHE_TABLE_SIZE 256 43 #define IP6_MASK_TABLE_SIZE (IPV6_ABITS + 1) /* 129 ptrs */ 44 45 /* 46 * We use the common modulo hash function. In ip_ire_init(), we make 47 * sure that the cache table size is always a power of 2. That's why 48 * we can use & instead of %. Also note that we try hard to make sure 49 * the lower bits of an address capture most info from the whole address. 50 * The reason being that since our hash table is probably a lot smaller 51 * than 2^32 buckets so the lower bits are the most important. 52 */ 53 #define IRE_ADDR_HASH(addr, table_size) \ 54 (((addr) ^ ((addr) >> 8) ^ ((addr) >> 16) ^ ((addr) >> 24)) & \ 55 ((table_size) - 1)) 56 57 /* 58 * To make a byte-order neutral hash for IPv6, just take all the 59 * bytes in the bottom 32 bits into account. 60 */ 61 #define IRE_ADDR_HASH_V6(addr, table_size) \ 62 IRE_ADDR_HASH((addr).s6_addr32[3], table_size) 63 64 /* This assumes that the ftable size is a power of 2. */ 65 #define IRE_ADDR_MASK_HASH_V6(addr, mask, table_size) \ 66 ((((addr).s6_addr8[8] & (mask).s6_addr8[8]) ^ \ 67 ((addr).s6_addr8[9] & (mask).s6_addr8[9]) ^ \ 68 ((addr).s6_addr8[10] & (mask).s6_addr8[10]) ^ \ 69 ((addr).s6_addr8[13] & (mask).s6_addr8[13]) ^ \ 70 ((addr).s6_addr8[14] & (mask).s6_addr8[14]) ^ \ 71 ((addr).s6_addr8[15] & (mask).s6_addr8[15])) & ((table_size) - 1)) 72 73 /* 74 * match parameter definitions for 75 * IRE lookup routines. 76 */ 77 #define MATCH_IRE_DSTONLY 0x0000 /* Match just the address */ 78 #define MATCH_IRE_TYPE 0x0001 /* Match IRE type */ 79 #define MATCH_IRE_SRC 0x0002 /* Match IRE source address */ 80 #define MATCH_IRE_MASK 0x0004 /* Match IRE mask */ 81 #define MATCH_IRE_WQ 0x0008 /* Match IRE Write Q */ 82 #define MATCH_IRE_GW 0x0010 /* Match IRE gateway */ 83 #define MATCH_IRE_IPIF 0x0020 /* Match IRE ipif */ 84 #define MATCH_IRE_RECURSIVE 0x0040 /* Do recursive lookup if necessary */ 85 #define MATCH_IRE_DEFAULT 0x0080 /* Return default route if no route */ 86 /* found. */ 87 #define MATCH_IRE_RJ_BHOLE 0x0100 /* During lookup if we hit an ire */ 88 /* with RTF_REJECT or RTF_BLACKHOLE, */ 89 /* return the ire. No recursive */ 90 /* lookup should be done. */ 91 #define MATCH_IRE_IHANDLE 0x0200 /* Match IRE on ihandle */ 92 #define MATCH_IRE_MARK_HIDDEN 0x0400 /* Match IRE ire_marks with */ 93 /* IRE_MARK_HIDDEN. */ 94 /* 95 * MATCH_IRE_ILL is used whenever we want to specifically match an IRE 96 * whose ire_ipif->ipif_ill or (ill_t *)ire_stq->q_ptr matches a given 97 * ill. When MATCH_IRE_ILL is used to locate an IRE_CACHE, it implies 98 * that the packet will not be load balanced. This is normally used 99 * by in.mpathd to send out failure detection probes. 100 * 101 * MATCH_IRE_ILL_GROUP is used whenever we are not specific about which 102 * interface (ill) the packet should be sent out. This implies that the 103 * packets will be subjected to load balancing and it might go out on 104 * any interface in the group. When there is only interface in the group, 105 * MATCH_IRE_ILL_GROUP becomes MATCH_IRE_ILL. Most of the code uses 106 * MATCH_IRE_ILL_GROUP and MATCH_IRE_ILL is used in very few cases where 107 * we want to disable load balancing. 108 * 109 * MATCH_IRE_PARENT is used whenever we unconditionally want to get the 110 * parent IRE (sire) while recursively searching IREs for an offsubnet 111 * destination. With this flag, even if no IRE_CACHETABLE or IRE_INTERFACE 112 * is found to help resolving IRE_OFFSUBNET in lookup routines, the 113 * IRE_OFFSUBNET sire, if any, is returned to the caller. 114 */ 115 #define MATCH_IRE_ILL_GROUP 0x0800 /* Match IRE on ill or the ill_group. */ 116 #define MATCH_IRE_ILL 0x1000 /* Match IRE on the ill only */ 117 118 #define MATCH_IRE_PARENT 0x2000 /* Match parent ire, if any, */ 119 /* even if ire is not matched. */ 120 #define MATCH_IRE_ZONEONLY 0x4000 /* Match IREs in specified zone, ie */ 121 /* don't match IRE_LOCALs from other */ 122 /* zones or shared IREs */ 123 #define MATCH_IRE_MARK_PRIVATE_ADDR 0x8000 /* Match IRE ire_marks with */ 124 /* IRE_MARK_PRIVATE_ADDR. */ 125 #define MATCH_IRE_SECATTR 0x10000 /* Match gateway security attributes */ 126 #define MATCH_IRE_COMPLETE 0x20000 /* ire_ftable_lookup() can return */ 127 /* IRE_CACHE entry only if it is */ 128 /* ND_REACHABLE */ 129 130 /* 131 * Any ire to nce association is long term, and 132 * the refhold and refrele may be done by different 133 * threads. So all cases of making or breaking ire to 134 * nce association should all effectively use the NOTR variants. 135 * To understand the *effectively* part read on. 136 * 137 * ndp_lookup() and ndp_add_v4()/ndp_add_v6() implicitly do 138 * NCE_REFHOLD. So wherever we make ire to nce association after 139 * calling these functions, we effectively want to end up with 140 * NCE_REFHOLD_NOTR. We call this macro to achieve this effect. This 141 * macro changes a NCE_REFHOLD to a NCE_REFHOLD_NOTR. The macro's 142 * NCE_REFRELE cancels off ndp_lookup[ndp_add]'s implicit NCE_REFHOLD, 143 * and what you are left with is a NCE_REFHOLD_NOTR 144 */ 145 #define NCE_REFHOLD_TO_REFHOLD_NOTR(nce) { \ 146 NCE_REFHOLD_NOTR(nce); \ 147 NCE_REFRELE(nce); \ 148 } 149 150 /* 151 * find the next ire_t entry in the ire_next chain starting at ire 152 * that is not CONDEMNED. ire is set to NULL if we reach the end of the list. 153 * Caller must hold the ire_bucket lock. 154 */ 155 156 #define IRE_FIND_NEXT_ORIGIN(ire) { \ 157 while ((ire) != NULL && ((ire)->ire_marks & IRE_MARK_CONDEMNED))\ 158 (ire) = (ire)->ire_next; \ 159 } 160 161 162 /* Structure for ire_cache_count() */ 163 typedef struct { 164 int icc_total; /* Total number of IRE_CACHE */ 165 int icc_unused; /* # off/no PMTU unused since last reclaim */ 166 int icc_offlink; /* # offlink without PMTU information */ 167 int icc_pmtu; /* # offlink with PMTU information */ 168 int icc_onlink; /* # onlink */ 169 } ire_cache_count_t; 170 171 /* 172 * Structure for ire_cache_reclaim(). Each field is a fraction i.e. 1 meaning 173 * reclaim all, N meaning reclaim 1/Nth of all entries, 0 meaning reclaim none. 174 * 175 * The comment below (and for other netstack_t references) refers 176 * to the fact that we only do netstack_hold in particular cases, 177 * such as the references from open streams (ill_t and conn_t's 178 * pointers). Internally within IP we rely on IP's ability to cleanup e.g. 179 * ire_t's when an ill goes away. 180 */ 181 typedef struct { 182 int icr_unused; /* Fraction for unused since last reclaim */ 183 int icr_offlink; /* Fraction for offlink without PMTU info */ 184 int icr_pmtu; /* Fraction for offlink with PMTU info */ 185 int icr_onlink; /* Fraction for onlink */ 186 ip_stack_t *icr_ipst; /* Does not have a netstack_hold */ 187 } ire_cache_reclaim_t; 188 189 /* 190 * We use atomics so that we get an accurate accounting on the ires. 191 * Otherwise we can't determine leaks correctly. 192 */ 193 #define BUMP_IRE_STATS(ire_stats, x) atomic_add_64(&(ire_stats).x, 1) 194 195 #ifdef _KERNEL 196 struct ts_label_s; 197 struct nce_s; 198 199 extern ipaddr_t ip_plen_to_mask(uint_t); 200 extern in6_addr_t *ip_plen_to_mask_v6(uint_t, in6_addr_t *); 201 202 extern int ip_ire_advise(queue_t *, mblk_t *, cred_t *); 203 extern int ip_ire_delete(queue_t *, mblk_t *, cred_t *); 204 extern boolean_t ip_ire_clookup_and_delete(ipaddr_t, ipif_t *, ip_stack_t *); 205 extern void ip_ire_clookup_and_delete_v6(const in6_addr_t *, 206 ip_stack_t *); 207 208 extern int ip_ire_report(queue_t *, mblk_t *, caddr_t, cred_t *); 209 extern int ip_ire_report_v6(queue_t *, mblk_t *, caddr_t, cred_t *); 210 extern void ire_report_ftable(ire_t *, char *); 211 212 extern void ip_ire_req(queue_t *, mblk_t *); 213 214 extern int ip_mask_to_plen(ipaddr_t); 215 extern int ip_mask_to_plen_v6(const in6_addr_t *); 216 217 extern ire_t *ipif_to_ire(const ipif_t *); 218 extern ire_t *ipif_to_ire_v6(const ipif_t *); 219 220 extern int ire_add(ire_t **, queue_t *, mblk_t *, ipsq_func_t, boolean_t); 221 extern void ire_add_then_send(queue_t *, ire_t *, mblk_t *); 222 extern int ire_add_v6(ire_t **, queue_t *, mblk_t *, ipsq_func_t); 223 extern int ire_atomic_start(irb_t *irb_ptr, ire_t *ire, queue_t *q, 224 mblk_t *mp, ipsq_func_t func); 225 extern void ire_atomic_end(irb_t *irb_ptr, ire_t *ire); 226 227 extern void ire_cache_count(ire_t *, char *); 228 extern ire_t *ire_cache_lookup(ipaddr_t, zoneid_t, 229 const struct ts_label_s *, ip_stack_t *); 230 extern ire_t *ire_cache_lookup_v6(const in6_addr_t *, zoneid_t, 231 const struct ts_label_s *, ip_stack_t *); 232 extern void ire_cache_reclaim(ire_t *, char *); 233 234 extern void ire_check_bcast_present(ipif_t *, ipaddr_t, int, boolean_t *, 235 boolean_t *); 236 237 extern ire_t *ire_create_mp(uchar_t *, uchar_t *, uchar_t *, uchar_t *, 238 uint_t, struct nce_s *, queue_t *, queue_t *, ushort_t, ipif_t *, ipaddr_t, 239 uint32_t, uint32_t, uint32_t, const iulp_t *, tsol_gc_t *, tsol_gcgrp_t *, 240 ip_stack_t *); 241 extern ire_t *ire_create(uchar_t *, uchar_t *, uchar_t *, uchar_t *, 242 uint_t *, struct nce_s *, queue_t *, queue_t *, ushort_t, ipif_t *, 243 ipaddr_t, uint32_t, uint32_t, uint32_t, const iulp_t *, tsol_gc_t *, 244 tsol_gcgrp_t *, ip_stack_t *); 245 246 extern ire_t **ire_check_and_create_bcast(ipif_t *, ipaddr_t, 247 ire_t **, int); 248 extern ire_t **ire_create_bcast(ipif_t *, ipaddr_t, ire_t **); 249 extern ire_t *ire_init(ire_t *, uchar_t *, uchar_t *, uchar_t *, uchar_t *, 250 uint_t *, struct nce_s *, queue_t *, queue_t *, ushort_t, ipif_t *, 251 ipaddr_t, uint32_t, uint32_t, uint32_t, const iulp_t *, tsol_gc_t *, 252 tsol_gcgrp_t *, ip_stack_t *); 253 254 extern boolean_t ire_init_common(ire_t *, uint_t *, struct nce_s *, queue_t *, 255 queue_t *, ushort_t, ipif_t *, uint32_t, uint32_t, uint32_t, uchar_t, 256 const iulp_t *, tsol_gc_t *, tsol_gcgrp_t *, ip_stack_t *); 257 258 extern ire_t *ire_create_v6(const in6_addr_t *, const in6_addr_t *, 259 const in6_addr_t *, const in6_addr_t *, uint_t *, struct nce_s *, queue_t *, 260 queue_t *, ushort_t, ipif_t *, 261 const in6_addr_t *, uint32_t, uint32_t, uint_t, const iulp_t *, 262 tsol_gc_t *, tsol_gcgrp_t *, ip_stack_t *); 263 264 extern ire_t *ire_create_mp_v6(const in6_addr_t *, const in6_addr_t *, 265 const in6_addr_t *, const in6_addr_t *, struct nce_s *, queue_t *, 266 queue_t *, ushort_t, ipif_t *, 267 const in6_addr_t *, uint32_t, uint32_t, uint_t, const iulp_t *, 268 tsol_gc_t *, tsol_gcgrp_t *, ip_stack_t *); 269 270 271 extern void ire_clookup_delete_cache_gw(ipaddr_t, zoneid_t, 272 ip_stack_t *); 273 extern void ire_clookup_delete_cache_gw_v6(const in6_addr_t *, zoneid_t, 274 ip_stack_t *); 275 276 extern ire_t *ire_ctable_lookup(ipaddr_t, ipaddr_t, int, const ipif_t *, 277 zoneid_t, const struct ts_label_s *, int, ip_stack_t *); 278 279 extern ire_t *ire_ctable_lookup_v6(const in6_addr_t *, const in6_addr_t *, 280 int, const ipif_t *, zoneid_t, const struct ts_label_s *, int, 281 ip_stack_t *); 282 283 extern void ire_delete(ire_t *); 284 extern void ire_delete_cache_gw(ire_t *, char *); 285 extern void ire_delete_cache_gw_v6(ire_t *, char *); 286 extern void ire_delete_cache_v6(ire_t *, char *); 287 extern void ire_delete_v6(ire_t *); 288 289 extern void ire_expire(ire_t *, char *); 290 291 extern void ire_flush_cache_v4(ire_t *, int); 292 extern void ire_flush_cache_v6(ire_t *, int); 293 294 extern ire_t *ire_ftable_lookup_v6(const in6_addr_t *, const in6_addr_t *, 295 const in6_addr_t *, int, const ipif_t *, ire_t **, zoneid_t, 296 uint32_t, const struct ts_label_s *, int, ip_stack_t *); 297 298 extern ire_t *ire_ihandle_lookup_onlink(ire_t *); 299 extern ire_t *ire_ihandle_lookup_offlink(ire_t *, ire_t *); 300 extern ire_t *ire_ihandle_lookup_offlink_v6(ire_t *, ire_t *); 301 302 extern boolean_t ire_local_same_ill_group(ire_t *, ire_t *); 303 extern boolean_t ire_local_ok_across_zones(ire_t *, zoneid_t, void *, 304 const struct ts_label_s *, ip_stack_t *); 305 306 extern ire_t *ire_lookup_local(zoneid_t, ip_stack_t *); 307 extern ire_t *ire_lookup_local_v6(zoneid_t, ip_stack_t *); 308 309 extern ire_t *ire_lookup_multi(ipaddr_t, zoneid_t, ip_stack_t *); 310 extern ire_t *ire_lookup_multi_v6(const in6_addr_t *, zoneid_t, 311 ip_stack_t *); 312 313 extern void ire_refrele(ire_t *); 314 extern void ire_refrele_notr(ire_t *); 315 extern ire_t *ire_route_lookup(ipaddr_t, ipaddr_t, ipaddr_t, int, 316 const ipif_t *, ire_t **, zoneid_t, const struct ts_label_s *, int, 317 ip_stack_t *); 318 319 extern ire_t *ire_route_lookup_v6(const in6_addr_t *, const in6_addr_t *, 320 const in6_addr_t *, int, const ipif_t *, ire_t **, zoneid_t, 321 const struct ts_label_s *, int, ip_stack_t *); 322 323 extern ill_t *ire_to_ill(const ire_t *); 324 325 extern void ire_walk(pfv_t, void *, ip_stack_t *); 326 extern void ire_walk_ill(uint_t, uint_t, pfv_t, void *, ill_t *); 327 extern void ire_walk_ill_v4(uint_t, uint_t, pfv_t, void *, ill_t *); 328 extern void ire_walk_ill_v6(uint_t, uint_t, pfv_t, void *, ill_t *); 329 extern void ire_walk_v4(pfv_t, void *, zoneid_t, ip_stack_t *); 330 extern void ire_walk_ill_tables(uint_t match_flags, uint_t ire_type, 331 pfv_t func, void *arg, size_t ftbl_sz, size_t htbl_sz, 332 irb_t **ipftbl, size_t ctbl_sz, irb_t *ipctbl, ill_t *ill, 333 zoneid_t zoneid, ip_stack_t *); 334 extern void ire_walk_v6(pfv_t, void *, zoneid_t, ip_stack_t *); 335 336 extern boolean_t ire_multirt_lookup(ire_t **, ire_t **, uint32_t, 337 const struct ts_label_s *, ip_stack_t *); 338 extern boolean_t ire_multirt_need_resolve(ipaddr_t, 339 const struct ts_label_s *, ip_stack_t *); 340 extern boolean_t ire_multirt_lookup_v6(ire_t **, ire_t **, uint32_t, 341 const struct ts_label_s *, ip_stack_t *); 342 extern boolean_t ire_multirt_need_resolve_v6(const in6_addr_t *, 343 const struct ts_label_s *, ip_stack_t *); 344 345 extern ire_t *ipif_lookup_multi_ire(ipif_t *, ipaddr_t); 346 extern ire_t *ipif_lookup_multi_ire_v6(ipif_t *, const in6_addr_t *); 347 348 extern ire_t *ire_get_next_bcast_ire(ire_t *, ire_t *); 349 extern ire_t *ire_get_next_default_ire(ire_t *, ire_t *); 350 351 extern void ire_arpresolve(ire_t *, ill_t *); 352 extern void ire_freemblk(ire_t *); 353 extern boolean_t ire_match_args(ire_t *, ipaddr_t, ipaddr_t, ipaddr_t, 354 int, const ipif_t *, zoneid_t, uint32_t, const struct ts_label_s *, int); 355 extern int ire_nce_init(ire_t *, struct nce_s *); 356 extern boolean_t ire_walk_ill_match(uint_t, uint_t, ire_t *, ill_t *, 357 zoneid_t, ip_stack_t *); 358 359 #endif /* _KERNEL */ 360 361 #ifdef __cplusplus 362 } 363 #endif 364 365 #endif /* _INET_IP_IRE_H */ 366