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 2008 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 #ifdef __cplusplus 31 extern "C" { 32 #endif 33 34 #define IPV6_LL_PREFIXLEN 10 /* Number of bits in link-local pref */ 35 36 #define IP_CACHE_TABLE_SIZE 256 37 #define IP_MASK_TABLE_SIZE (IP_ABITS + 1) /* 33 ptrs */ 38 39 #define IP6_FTABLE_HASH_SIZE 32 /* size of each hash table in ptrs */ 40 #define IP6_CACHE_TABLE_SIZE 256 41 #define IP6_MASK_TABLE_SIZE (IPV6_ABITS + 1) /* 129 ptrs */ 42 43 /* 44 * We use the common modulo hash function. In ip_ire_init(), we make 45 * sure that the cache table size is always a power of 2. That's why 46 * we can use & instead of %. Also note that we try hard to make sure 47 * the lower bits of an address capture most info from the whole address. 48 * The reason being that since our hash table is probably a lot smaller 49 * than 2^32 buckets so the lower bits are the most important. 50 */ 51 #define IRE_ADDR_HASH(addr, table_size) \ 52 (((addr) ^ ((addr) >> 8) ^ ((addr) >> 16) ^ ((addr) >> 24)) & \ 53 ((table_size) - 1)) 54 55 /* 56 * To make a byte-order neutral hash for IPv6, just take all the 57 * bytes in the bottom 32 bits into account. 58 */ 59 #define IRE_ADDR_HASH_V6(addr, table_size) \ 60 IRE_ADDR_HASH((addr).s6_addr32[3], table_size) 61 62 /* This assumes that the ftable size is a power of 2. */ 63 #define IRE_ADDR_MASK_HASH_V6(addr, mask, table_size) \ 64 ((((addr).s6_addr8[8] & (mask).s6_addr8[8]) ^ \ 65 ((addr).s6_addr8[9] & (mask).s6_addr8[9]) ^ \ 66 ((addr).s6_addr8[10] & (mask).s6_addr8[10]) ^ \ 67 ((addr).s6_addr8[13] & (mask).s6_addr8[13]) ^ \ 68 ((addr).s6_addr8[14] & (mask).s6_addr8[14]) ^ \ 69 ((addr).s6_addr8[15] & (mask).s6_addr8[15])) & ((table_size) - 1)) 70 71 /* 72 * match parameter definitions for IRE lookup routines. 73 */ 74 #define MATCH_IRE_DSTONLY 0x0000 /* Match just the address */ 75 #define MATCH_IRE_TYPE 0x0001 /* Match IRE type */ 76 #define MATCH_IRE_SRC 0x0002 /* Match IRE source address */ 77 #define MATCH_IRE_MASK 0x0004 /* Match IRE mask */ 78 #define MATCH_IRE_WQ 0x0008 /* Match IRE ire_stq to write queue */ 79 #define MATCH_IRE_GW 0x0010 /* Match IRE gateway */ 80 #define MATCH_IRE_IPIF 0x0020 /* Match IRE ipif */ 81 #define MATCH_IRE_RECURSIVE 0x0040 /* Do recursive lookup if necessary */ 82 #define MATCH_IRE_DEFAULT 0x0080 /* Return default route if no route */ 83 /* found. */ 84 #define MATCH_IRE_RJ_BHOLE 0x0100 /* During lookup if we hit an ire */ 85 /* with RTF_REJECT or RTF_BLACKHOLE, */ 86 /* return the ire. No recursive */ 87 /* lookup should be done. */ 88 #define MATCH_IRE_IHANDLE 0x0200 /* Match IRE on ihandle */ 89 #define MATCH_IRE_MARK_HIDDEN 0x0400 /* Match IRE ire_marks with */ 90 /* IRE_MARK_HIDDEN. */ 91 /* 92 * MATCH_IRE_ILL is used whenever we want to specifically match an IRE 93 * whose ire_ipif->ipif_ill or (ill_t *)ire_stq->q_ptr matches a given 94 * ill. When MATCH_IRE_ILL is used to locate an IRE_CACHE, it implies 95 * that the packet will not be load balanced. This is normally used 96 * by in.mpathd to send out failure detection probes. 97 * 98 * MATCH_IRE_ILL_GROUP is used whenever we are not specific about which 99 * interface (ill) the packet should be sent out. This implies that the 100 * packets will be subjected to load balancing and it might go out on 101 * any interface in the group. When there is only interface in the group, 102 * MATCH_IRE_ILL_GROUP becomes MATCH_IRE_ILL. Most of the code uses 103 * MATCH_IRE_ILL_GROUP and MATCH_IRE_ILL is used in very few cases where 104 * we want to disable load balancing. 105 * 106 * MATCH_IRE_PARENT is used whenever we unconditionally want to get the 107 * parent IRE (sire) while recursively searching IREs for an offsubnet 108 * destination. With this flag, even if no IRE_CACHETABLE or IRE_INTERFACE 109 * is found to help resolving IRE_OFFSUBNET in lookup routines, the 110 * IRE_OFFSUBNET sire, if any, is returned to the caller. 111 */ 112 #define MATCH_IRE_ILL_GROUP 0x0800 /* Match IRE on ill or the ill_group. */ 113 #define MATCH_IRE_ILL 0x1000 /* Match IRE on the ill only */ 114 115 #define MATCH_IRE_PARENT 0x2000 /* Match parent ire, if any, */ 116 /* even if ire is not matched. */ 117 #define MATCH_IRE_ZONEONLY 0x4000 /* Match IREs in specified zone, ie */ 118 /* don't match IRE_LOCALs from other */ 119 /* zones or shared IREs */ 120 #define MATCH_IRE_MARK_PRIVATE_ADDR 0x8000 /* Match IRE ire_marks with */ 121 /* IRE_MARK_PRIVATE_ADDR. */ 122 #define MATCH_IRE_SECATTR 0x10000 /* Match gateway security attributes */ 123 #define MATCH_IRE_COMPLETE 0x20000 /* ire_ftable_lookup() can return */ 124 /* IRE_CACHE entry only if it is */ 125 /* ND_REACHABLE */ 126 127 /* 128 * Any ire to nce association is long term, and 129 * the refhold and refrele may be done by different 130 * threads. So all cases of making or breaking ire to 131 * nce association should all effectively use the NOTR variants. 132 * To understand the *effectively* part read on. 133 * 134 * ndp_lookup() and ndp_add_v4()/ndp_add_v6() implicitly do 135 * NCE_REFHOLD. So wherever we make ire to nce association after 136 * calling these functions, we effectively want to end up with 137 * NCE_REFHOLD_NOTR. We call this macro to achieve this effect. This 138 * macro changes a NCE_REFHOLD to a NCE_REFHOLD_NOTR. The macro's 139 * NCE_REFRELE cancels off ndp_lookup[ndp_add]'s implicit NCE_REFHOLD, 140 * and what you are left with is a NCE_REFHOLD_NOTR 141 */ 142 #define NCE_REFHOLD_TO_REFHOLD_NOTR(nce) { \ 143 NCE_REFHOLD_NOTR(nce); \ 144 NCE_REFRELE(nce); \ 145 } 146 147 /* 148 * find the next ire_t entry in the ire_next chain starting at ire 149 * that is not CONDEMNED. ire is set to NULL if we reach the end of the list. 150 * Caller must hold the ire_bucket lock. 151 */ 152 153 #define IRE_FIND_NEXT_ORIGIN(ire) { \ 154 while ((ire) != NULL && ((ire)->ire_marks & IRE_MARK_CONDEMNED))\ 155 (ire) = (ire)->ire_next; \ 156 } 157 158 159 /* Structure for ire_cache_count() */ 160 typedef struct { 161 int icc_total; /* Total number of IRE_CACHE */ 162 int icc_unused; /* # off/no PMTU unused since last reclaim */ 163 int icc_offlink; /* # offlink without PMTU information */ 164 int icc_pmtu; /* # offlink with PMTU information */ 165 int icc_onlink; /* # onlink */ 166 } ire_cache_count_t; 167 168 /* 169 * Structure for ire_cache_reclaim(). Each field is a fraction i.e. 1 meaning 170 * reclaim all, N meaning reclaim 1/Nth of all entries, 0 meaning reclaim none. 171 * 172 * The comment below (and for other netstack_t references) refers 173 * to the fact that we only do netstack_hold in particular cases, 174 * such as the references from open streams (ill_t and conn_t's 175 * pointers). Internally within IP we rely on IP's ability to cleanup e.g. 176 * ire_t's when an ill goes away. 177 */ 178 typedef struct { 179 int icr_unused; /* Fraction for unused since last reclaim */ 180 int icr_offlink; /* Fraction for offlink without PMTU info */ 181 int icr_pmtu; /* Fraction for offlink with PMTU info */ 182 int icr_onlink; /* Fraction for onlink */ 183 ip_stack_t *icr_ipst; /* Does not have a netstack_hold */ 184 } ire_cache_reclaim_t; 185 186 /* 187 * We use atomics so that we get an accurate accounting on the ires. 188 * Otherwise we can't determine leaks correctly. 189 */ 190 #define BUMP_IRE_STATS(ire_stats, x) atomic_add_64(&(ire_stats).x, 1) 191 192 #ifdef _KERNEL 193 /* 194 * Structure for passing args for the IRE cache lookup functions. 195 */ 196 typedef struct ire_ctable_args_s { 197 void *ict_addr; 198 void *ict_gateway; 199 int ict_type; 200 const ipif_t *ict_ipif; 201 zoneid_t ict_zoneid; 202 const ts_label_t *ict_tsl; 203 int ict_flags; 204 ip_stack_t *ict_ipst; 205 queue_t *ict_wq; 206 } ire_ctable_args_t; 207 208 struct ts_label_s; 209 struct nce_s; 210 211 extern ipaddr_t ip_plen_to_mask(uint_t); 212 extern in6_addr_t *ip_plen_to_mask_v6(uint_t, in6_addr_t *); 213 214 extern int ip_ire_advise(queue_t *, mblk_t *, cred_t *); 215 extern int ip_ire_delete(queue_t *, mblk_t *, cred_t *); 216 extern boolean_t ip_ire_clookup_and_delete(ipaddr_t, ipif_t *, ip_stack_t *); 217 extern void ip_ire_clookup_and_delete_v6(const in6_addr_t *, 218 ip_stack_t *); 219 220 extern void ip_ire_req(queue_t *, mblk_t *); 221 222 extern int ip_mask_to_plen(ipaddr_t); 223 extern int ip_mask_to_plen_v6(const in6_addr_t *); 224 225 extern ire_t *ipif_to_ire(const ipif_t *); 226 extern ire_t *ipif_to_ire_v6(const ipif_t *); 227 228 extern int ire_add(ire_t **, queue_t *, mblk_t *, ipsq_func_t, boolean_t); 229 extern void ire_add_then_send(queue_t *, ire_t *, mblk_t *); 230 extern int ire_add_v6(ire_t **, queue_t *, mblk_t *, ipsq_func_t); 231 extern int ire_atomic_start(irb_t *irb_ptr, ire_t *ire, queue_t *q, 232 mblk_t *mp, ipsq_func_t func); 233 extern void ire_atomic_end(irb_t *irb_ptr, ire_t *ire); 234 235 extern void ire_cache_count(ire_t *, char *); 236 extern ire_t *ire_cache_lookup(ipaddr_t, zoneid_t, 237 const struct ts_label_s *, ip_stack_t *); 238 extern ire_t *ire_cache_lookup_simple(ipaddr_t, ip_stack_t *); 239 extern ire_t *ire_cache_lookup_v6(const in6_addr_t *, zoneid_t, 240 const struct ts_label_s *, ip_stack_t *); 241 extern void ire_cache_reclaim(ire_t *, char *); 242 243 extern ire_t *ire_create_mp(uchar_t *, uchar_t *, uchar_t *, uchar_t *, 244 uint_t, struct nce_s *, queue_t *, queue_t *, ushort_t, ipif_t *, ipaddr_t, 245 uint32_t, uint32_t, uint32_t, const iulp_t *, tsol_gc_t *, tsol_gcgrp_t *, 246 ip_stack_t *); 247 extern ire_t *ire_create(uchar_t *, uchar_t *, uchar_t *, uchar_t *, 248 uint_t *, struct nce_s *, queue_t *, queue_t *, ushort_t, ipif_t *, 249 ipaddr_t, uint32_t, uint32_t, uint32_t, const iulp_t *, tsol_gc_t *, 250 tsol_gcgrp_t *, ip_stack_t *); 251 252 extern ire_t **ire_check_and_create_bcast(ipif_t *, ipaddr_t, 253 ire_t **, int); 254 extern ire_t **ire_create_bcast(ipif_t *, ipaddr_t, ire_t **); 255 extern ire_t *ire_init(ire_t *, uchar_t *, uchar_t *, uchar_t *, uchar_t *, 256 uint_t *, struct nce_s *, queue_t *, queue_t *, ushort_t, ipif_t *, 257 ipaddr_t, uint32_t, uint32_t, uint32_t, const iulp_t *, tsol_gc_t *, 258 tsol_gcgrp_t *, ip_stack_t *); 259 260 extern boolean_t ire_init_common(ire_t *, uint_t *, struct nce_s *, queue_t *, 261 queue_t *, ushort_t, ipif_t *, uint32_t, uint32_t, uint32_t, uchar_t, 262 const iulp_t *, tsol_gc_t *, tsol_gcgrp_t *, ip_stack_t *); 263 264 extern ire_t *ire_create_v6(const in6_addr_t *, const in6_addr_t *, 265 const in6_addr_t *, const in6_addr_t *, uint_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 extern ire_t *ire_create_mp_v6(const in6_addr_t *, const in6_addr_t *, 271 const in6_addr_t *, const in6_addr_t *, struct nce_s *, queue_t *, 272 queue_t *, ushort_t, ipif_t *, 273 const in6_addr_t *, uint32_t, uint32_t, uint_t, const iulp_t *, 274 tsol_gc_t *, tsol_gcgrp_t *, ip_stack_t *); 275 276 277 extern void ire_clookup_delete_cache_gw(ipaddr_t, zoneid_t, 278 ip_stack_t *); 279 extern void ire_clookup_delete_cache_gw_v6(const in6_addr_t *, zoneid_t, 280 ip_stack_t *); 281 282 extern ire_t *ire_ctable_lookup(ipaddr_t, ipaddr_t, int, const ipif_t *, 283 zoneid_t, const struct ts_label_s *, int, ip_stack_t *); 284 285 extern ire_t *ire_ctable_lookup_v6(const in6_addr_t *, const in6_addr_t *, 286 int, const ipif_t *, zoneid_t, const struct ts_label_s *, int, 287 ip_stack_t *); 288 289 extern void ire_delete(ire_t *); 290 extern void ire_delete_cache_gw(ire_t *, char *); 291 extern void ire_delete_cache_gw_v6(ire_t *, char *); 292 extern void ire_delete_cache_v6(ire_t *, char *); 293 extern void ire_delete_v6(ire_t *); 294 295 extern void ire_expire(ire_t *, char *); 296 297 extern void ire_flush_cache_v4(ire_t *, int); 298 extern void ire_flush_cache_v6(ire_t *, int); 299 300 extern ire_t *ire_ftable_lookup_v6(const in6_addr_t *, const in6_addr_t *, 301 const in6_addr_t *, int, const ipif_t *, ire_t **, zoneid_t, 302 uint32_t, const struct ts_label_s *, int, ip_stack_t *); 303 304 extern ire_t *ire_ihandle_lookup_onlink(ire_t *); 305 extern ire_t *ire_ihandle_lookup_offlink(ire_t *, ire_t *); 306 extern ire_t *ire_ihandle_lookup_offlink_v6(ire_t *, ire_t *); 307 308 extern boolean_t ire_local_same_ill_group(ire_t *, ire_t *); 309 extern boolean_t ire_local_ok_across_zones(ire_t *, zoneid_t, void *, 310 const struct ts_label_s *, ip_stack_t *); 311 312 extern ire_t *ire_lookup_local(zoneid_t, ip_stack_t *); 313 extern ire_t *ire_lookup_local_v6(zoneid_t, ip_stack_t *); 314 315 extern ire_t *ire_lookup_multi(ipaddr_t, zoneid_t, ip_stack_t *); 316 extern ire_t *ire_lookup_multi_v6(const in6_addr_t *, zoneid_t, 317 ip_stack_t *); 318 319 extern void ire_refrele(ire_t *); 320 extern void ire_refrele_notr(ire_t *); 321 extern ire_t *ire_route_lookup(ipaddr_t, ipaddr_t, ipaddr_t, int, 322 const ipif_t *, ire_t **, zoneid_t, const struct ts_label_s *, int, 323 ip_stack_t *); 324 325 extern ire_t *ire_route_lookup_v6(const in6_addr_t *, const in6_addr_t *, 326 const in6_addr_t *, int, const ipif_t *, ire_t **, zoneid_t, 327 const struct ts_label_s *, int, ip_stack_t *); 328 329 extern ill_t *ire_to_ill(const ire_t *); 330 331 extern void ire_walk(pfv_t, void *, ip_stack_t *); 332 extern void ire_walk_ill(uint_t, uint_t, pfv_t, void *, ill_t *); 333 extern void ire_walk_ill_v4(uint_t, uint_t, pfv_t, void *, ill_t *); 334 extern void ire_walk_ill_v6(uint_t, uint_t, pfv_t, void *, ill_t *); 335 extern void ire_walk_v4(pfv_t, void *, zoneid_t, ip_stack_t *); 336 extern void ire_walk_ill_tables(uint_t match_flags, uint_t ire_type, 337 pfv_t func, void *arg, size_t ftbl_sz, size_t htbl_sz, 338 irb_t **ipftbl, size_t ctbl_sz, irb_t *ipctbl, ill_t *ill, 339 zoneid_t zoneid, ip_stack_t *); 340 extern void ire_walk_v6(pfv_t, void *, zoneid_t, ip_stack_t *); 341 342 extern boolean_t ire_multirt_lookup(ire_t **, ire_t **, uint32_t, 343 const struct ts_label_s *, ip_stack_t *); 344 extern boolean_t ire_multirt_need_resolve(ipaddr_t, 345 const struct ts_label_s *, ip_stack_t *); 346 extern boolean_t ire_multirt_lookup_v6(ire_t **, ire_t **, uint32_t, 347 const struct ts_label_s *, ip_stack_t *); 348 extern boolean_t ire_multirt_need_resolve_v6(const in6_addr_t *, 349 const struct ts_label_s *, ip_stack_t *); 350 351 extern ire_t *ipif_lookup_multi_ire(ipif_t *, ipaddr_t); 352 extern ire_t *ipif_lookup_multi_ire_v6(ipif_t *, const in6_addr_t *); 353 354 extern ire_t *ire_get_next_bcast_ire(ire_t *, ire_t *); 355 extern ire_t *ire_get_next_default_ire(ire_t *, ire_t *); 356 357 extern void ire_arpresolve(ire_t *, ill_t *); 358 extern void ire_freemblk(ire_t *); 359 extern boolean_t ire_match_args(ire_t *, ipaddr_t, ipaddr_t, ipaddr_t, 360 int, const ipif_t *, zoneid_t, uint32_t, const struct ts_label_s *, int, 361 queue_t *); 362 extern int ire_nce_init(ire_t *, struct nce_s *); 363 extern boolean_t ire_walk_ill_match(uint_t, uint_t, ire_t *, ill_t *, 364 zoneid_t, ip_stack_t *); 365 extern ire_t *ire_arpresolve_lookup(ipaddr_t, ipaddr_t, ipif_t *, zoneid_t, 366 ip_stack_t *, queue_t *); 367 368 #endif /* _KERNEL */ 369 370 #ifdef __cplusplus 371 } 372 #endif 373 374 #endif /* _INET_IP_IRE_H */ 375