/* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright 2007 Sun Microsystems, Inc. All rights reserved. * Use is subject to license terms. */ /* Copyright (c) 1990 Mentat Inc. */ #ifndef _INET_IP_IRE_H #define _INET_IP_IRE_H #pragma ident "%Z%%M% %I% %E% SMI" #ifdef __cplusplus extern "C" { #endif #define IPV6_LL_PREFIXLEN 10 /* Number of bits in link-local pref */ #define IP_CACHE_TABLE_SIZE 256 #define IP_MASK_TABLE_SIZE (IP_ABITS + 1) /* 33 ptrs */ #define IP6_FTABLE_HASH_SIZE 32 /* size of each hash table in ptrs */ #define IP6_CACHE_TABLE_SIZE 256 #define IP6_MASK_TABLE_SIZE (IPV6_ABITS + 1) /* 129 ptrs */ /* * We use the common modulo hash function. In ip_ire_init(), we make * sure that the cache table size is always a power of 2. That's why * we can use & instead of %. Also note that we try hard to make sure * the lower bits of an address capture most info from the whole address. * The reason being that since our hash table is probably a lot smaller * than 2^32 buckets so the lower bits are the most important. */ #define IRE_ADDR_HASH(addr, table_size) \ (((addr) ^ ((addr) >> 8) ^ ((addr) >> 16) ^ ((addr) >> 24)) & \ ((table_size) - 1)) /* * To make a byte-order neutral hash for IPv6, just take all the * bytes in the bottom 32 bits into account. */ #define IRE_ADDR_HASH_V6(addr, table_size) \ IRE_ADDR_HASH((addr).s6_addr32[3], table_size) /* This assumes that the ftable size is a power of 2. */ #define IRE_ADDR_MASK_HASH_V6(addr, mask, table_size) \ ((((addr).s6_addr8[8] & (mask).s6_addr8[8]) ^ \ ((addr).s6_addr8[9] & (mask).s6_addr8[9]) ^ \ ((addr).s6_addr8[10] & (mask).s6_addr8[10]) ^ \ ((addr).s6_addr8[13] & (mask).s6_addr8[13]) ^ \ ((addr).s6_addr8[14] & (mask).s6_addr8[14]) ^ \ ((addr).s6_addr8[15] & (mask).s6_addr8[15])) & ((table_size) - 1)) /* * match parameter definitions for IRE lookup routines. */ #define MATCH_IRE_DSTONLY 0x0000 /* Match just the address */ #define MATCH_IRE_TYPE 0x0001 /* Match IRE type */ #define MATCH_IRE_SRC 0x0002 /* Match IRE source address */ #define MATCH_IRE_MASK 0x0004 /* Match IRE mask */ /* unused 0x0008 */ #define MATCH_IRE_GW 0x0010 /* Match IRE gateway */ #define MATCH_IRE_IPIF 0x0020 /* Match IRE ipif */ #define MATCH_IRE_RECURSIVE 0x0040 /* Do recursive lookup if necessary */ #define MATCH_IRE_DEFAULT 0x0080 /* Return default route if no route */ /* found. */ #define MATCH_IRE_RJ_BHOLE 0x0100 /* During lookup if we hit an ire */ /* with RTF_REJECT or RTF_BLACKHOLE, */ /* return the ire. No recursive */ /* lookup should be done. */ #define MATCH_IRE_IHANDLE 0x0200 /* Match IRE on ihandle */ #define MATCH_IRE_MARK_HIDDEN 0x0400 /* Match IRE ire_marks with */ /* IRE_MARK_HIDDEN. */ /* * MATCH_IRE_ILL is used whenever we want to specifically match an IRE * whose ire_ipif->ipif_ill or (ill_t *)ire_stq->q_ptr matches a given * ill. When MATCH_IRE_ILL is used to locate an IRE_CACHE, it implies * that the packet will not be load balanced. This is normally used * by in.mpathd to send out failure detection probes. * * MATCH_IRE_ILL_GROUP is used whenever we are not specific about which * interface (ill) the packet should be sent out. This implies that the * packets will be subjected to load balancing and it might go out on * any interface in the group. When there is only interface in the group, * MATCH_IRE_ILL_GROUP becomes MATCH_IRE_ILL. Most of the code uses * MATCH_IRE_ILL_GROUP and MATCH_IRE_ILL is used in very few cases where * we want to disable load balancing. * * MATCH_IRE_PARENT is used whenever we unconditionally want to get the * parent IRE (sire) while recursively searching IREs for an offsubnet * destination. With this flag, even if no IRE_CACHETABLE or IRE_INTERFACE * is found to help resolving IRE_OFFSUBNET in lookup routines, the * IRE_OFFSUBNET sire, if any, is returned to the caller. */ #define MATCH_IRE_ILL_GROUP 0x0800 /* Match IRE on ill or the ill_group. */ #define MATCH_IRE_ILL 0x1000 /* Match IRE on the ill only */ #define MATCH_IRE_PARENT 0x2000 /* Match parent ire, if any, */ /* even if ire is not matched. */ #define MATCH_IRE_ZONEONLY 0x4000 /* Match IREs in specified zone, ie */ /* don't match IRE_LOCALs from other */ /* zones or shared IREs */ #define MATCH_IRE_MARK_PRIVATE_ADDR 0x8000 /* Match IRE ire_marks with */ /* IRE_MARK_PRIVATE_ADDR. */ #define MATCH_IRE_SECATTR 0x10000 /* Match gateway security attributes */ #define MATCH_IRE_COMPLETE 0x20000 /* ire_ftable_lookup() can return */ /* IRE_CACHE entry only if it is */ /* ND_REACHABLE */ /* * Any ire to nce association is long term, and * the refhold and refrele may be done by different * threads. So all cases of making or breaking ire to * nce association should all effectively use the NOTR variants. * To understand the *effectively* part read on. * * ndp_lookup() and ndp_add_v4()/ndp_add_v6() implicitly do * NCE_REFHOLD. So wherever we make ire to nce association after * calling these functions, we effectively want to end up with * NCE_REFHOLD_NOTR. We call this macro to achieve this effect. This * macro changes a NCE_REFHOLD to a NCE_REFHOLD_NOTR. The macro's * NCE_REFRELE cancels off ndp_lookup[ndp_add]'s implicit NCE_REFHOLD, * and what you are left with is a NCE_REFHOLD_NOTR */ #define NCE_REFHOLD_TO_REFHOLD_NOTR(nce) { \ NCE_REFHOLD_NOTR(nce); \ NCE_REFRELE(nce); \ } /* * find the next ire_t entry in the ire_next chain starting at ire * that is not CONDEMNED. ire is set to NULL if we reach the end of the list. * Caller must hold the ire_bucket lock. */ #define IRE_FIND_NEXT_ORIGIN(ire) { \ while ((ire) != NULL && ((ire)->ire_marks & IRE_MARK_CONDEMNED))\ (ire) = (ire)->ire_next; \ } /* Structure for ire_cache_count() */ typedef struct { int icc_total; /* Total number of IRE_CACHE */ int icc_unused; /* # off/no PMTU unused since last reclaim */ int icc_offlink; /* # offlink without PMTU information */ int icc_pmtu; /* # offlink with PMTU information */ int icc_onlink; /* # onlink */ } ire_cache_count_t; /* * Structure for ire_cache_reclaim(). Each field is a fraction i.e. 1 meaning * reclaim all, N meaning reclaim 1/Nth of all entries, 0 meaning reclaim none. * * The comment below (and for other netstack_t references) refers * to the fact that we only do netstack_hold in particular cases, * such as the references from open streams (ill_t and conn_t's * pointers). Internally within IP we rely on IP's ability to cleanup e.g. * ire_t's when an ill goes away. */ typedef struct { int icr_unused; /* Fraction for unused since last reclaim */ int icr_offlink; /* Fraction for offlink without PMTU info */ int icr_pmtu; /* Fraction for offlink with PMTU info */ int icr_onlink; /* Fraction for onlink */ ip_stack_t *icr_ipst; /* Does not have a netstack_hold */ } ire_cache_reclaim_t; /* * We use atomics so that we get an accurate accounting on the ires. * Otherwise we can't determine leaks correctly. */ #define BUMP_IRE_STATS(ire_stats, x) atomic_add_64(&(ire_stats).x, 1) #ifdef _KERNEL struct ts_label_s; struct nce_s; extern ipaddr_t ip_plen_to_mask(uint_t); extern in6_addr_t *ip_plen_to_mask_v6(uint_t, in6_addr_t *); extern int ip_ire_advise(queue_t *, mblk_t *, cred_t *); extern int ip_ire_delete(queue_t *, mblk_t *, cred_t *); extern boolean_t ip_ire_clookup_and_delete(ipaddr_t, ipif_t *, ip_stack_t *); extern void ip_ire_clookup_and_delete_v6(const in6_addr_t *, ip_stack_t *); extern int ip_ire_report(queue_t *, mblk_t *, caddr_t, cred_t *); extern int ip_ire_report_v6(queue_t *, mblk_t *, caddr_t, cred_t *); extern void ire_report_ftable(ire_t *, char *); extern void ip_ire_req(queue_t *, mblk_t *); extern int ip_mask_to_plen(ipaddr_t); extern int ip_mask_to_plen_v6(const in6_addr_t *); extern ire_t *ipif_to_ire(const ipif_t *); extern ire_t *ipif_to_ire_v6(const ipif_t *); extern int ire_add(ire_t **, queue_t *, mblk_t *, ipsq_func_t, boolean_t); extern void ire_add_then_send(queue_t *, ire_t *, mblk_t *); extern int ire_add_v6(ire_t **, queue_t *, mblk_t *, ipsq_func_t); extern int ire_atomic_start(irb_t *irb_ptr, ire_t *ire, queue_t *q, mblk_t *mp, ipsq_func_t func); extern void ire_atomic_end(irb_t *irb_ptr, ire_t *ire); extern void ire_cache_count(ire_t *, char *); extern ire_t *ire_cache_lookup(ipaddr_t, zoneid_t, const struct ts_label_s *, ip_stack_t *); extern ire_t *ire_cache_lookup_v6(const in6_addr_t *, zoneid_t, const struct ts_label_s *, ip_stack_t *); extern void ire_cache_reclaim(ire_t *, char *); extern ire_t *ire_create_mp(uchar_t *, uchar_t *, uchar_t *, uchar_t *, uint_t, struct nce_s *, queue_t *, queue_t *, ushort_t, ipif_t *, ipaddr_t, uint32_t, uint32_t, uint32_t, const iulp_t *, tsol_gc_t *, tsol_gcgrp_t *, ip_stack_t *); extern ire_t *ire_create(uchar_t *, uchar_t *, uchar_t *, uchar_t *, uint_t *, struct nce_s *, queue_t *, queue_t *, ushort_t, ipif_t *, ipaddr_t, uint32_t, uint32_t, uint32_t, const iulp_t *, tsol_gc_t *, tsol_gcgrp_t *, ip_stack_t *); extern ire_t **ire_check_and_create_bcast(ipif_t *, ipaddr_t, ire_t **, int); extern ire_t **ire_create_bcast(ipif_t *, ipaddr_t, ire_t **); extern ire_t *ire_init(ire_t *, uchar_t *, uchar_t *, uchar_t *, uchar_t *, uint_t *, struct nce_s *, queue_t *, queue_t *, ushort_t, ipif_t *, ipaddr_t, uint32_t, uint32_t, uint32_t, const iulp_t *, tsol_gc_t *, tsol_gcgrp_t *, ip_stack_t *); extern boolean_t ire_init_common(ire_t *, uint_t *, struct nce_s *, queue_t *, queue_t *, ushort_t, ipif_t *, uint32_t, uint32_t, uint32_t, uchar_t, const iulp_t *, tsol_gc_t *, tsol_gcgrp_t *, ip_stack_t *); extern ire_t *ire_create_v6(const in6_addr_t *, const in6_addr_t *, const in6_addr_t *, const in6_addr_t *, uint_t *, struct nce_s *, queue_t *, queue_t *, ushort_t, ipif_t *, const in6_addr_t *, uint32_t, uint32_t, uint_t, const iulp_t *, tsol_gc_t *, tsol_gcgrp_t *, ip_stack_t *); extern ire_t *ire_create_mp_v6(const in6_addr_t *, const in6_addr_t *, const in6_addr_t *, const in6_addr_t *, struct nce_s *, queue_t *, queue_t *, ushort_t, ipif_t *, const in6_addr_t *, uint32_t, uint32_t, uint_t, const iulp_t *, tsol_gc_t *, tsol_gcgrp_t *, ip_stack_t *); extern void ire_clookup_delete_cache_gw(ipaddr_t, zoneid_t, ip_stack_t *); extern void ire_clookup_delete_cache_gw_v6(const in6_addr_t *, zoneid_t, ip_stack_t *); extern ire_t *ire_ctable_lookup(ipaddr_t, ipaddr_t, int, const ipif_t *, zoneid_t, const struct ts_label_s *, int, ip_stack_t *); extern ire_t *ire_ctable_lookup_v6(const in6_addr_t *, const in6_addr_t *, int, const ipif_t *, zoneid_t, const struct ts_label_s *, int, ip_stack_t *); extern void ire_delete(ire_t *); extern void ire_delete_cache_gw(ire_t *, char *); extern void ire_delete_cache_gw_v6(ire_t *, char *); extern void ire_delete_cache_v6(ire_t *, char *); extern void ire_delete_v6(ire_t *); extern void ire_expire(ire_t *, char *); extern void ire_flush_cache_v4(ire_t *, int); extern void ire_flush_cache_v6(ire_t *, int); extern ire_t *ire_ftable_lookup_v6(const in6_addr_t *, const in6_addr_t *, const in6_addr_t *, int, const ipif_t *, ire_t **, zoneid_t, uint32_t, const struct ts_label_s *, int, ip_stack_t *); extern ire_t *ire_ihandle_lookup_onlink(ire_t *); extern ire_t *ire_ihandle_lookup_offlink(ire_t *, ire_t *); extern ire_t *ire_ihandle_lookup_offlink_v6(ire_t *, ire_t *); extern boolean_t ire_local_same_ill_group(ire_t *, ire_t *); extern boolean_t ire_local_ok_across_zones(ire_t *, zoneid_t, void *, const struct ts_label_s *, ip_stack_t *); extern ire_t *ire_lookup_local(zoneid_t, ip_stack_t *); extern ire_t *ire_lookup_local_v6(zoneid_t, ip_stack_t *); extern ire_t *ire_lookup_multi(ipaddr_t, zoneid_t, ip_stack_t *); extern ire_t *ire_lookup_multi_v6(const in6_addr_t *, zoneid_t, ip_stack_t *); extern void ire_refrele(ire_t *); extern void ire_refrele_notr(ire_t *); extern ire_t *ire_route_lookup(ipaddr_t, ipaddr_t, ipaddr_t, int, const ipif_t *, ire_t **, zoneid_t, const struct ts_label_s *, int, ip_stack_t *); extern ire_t *ire_route_lookup_v6(const in6_addr_t *, const in6_addr_t *, const in6_addr_t *, int, const ipif_t *, ire_t **, zoneid_t, const struct ts_label_s *, int, ip_stack_t *); extern ill_t *ire_to_ill(const ire_t *); extern void ire_walk(pfv_t, void *, ip_stack_t *); extern void ire_walk_ill(uint_t, uint_t, pfv_t, void *, ill_t *); extern void ire_walk_ill_v4(uint_t, uint_t, pfv_t, void *, ill_t *); extern void ire_walk_ill_v6(uint_t, uint_t, pfv_t, void *, ill_t *); extern void ire_walk_v4(pfv_t, void *, zoneid_t, ip_stack_t *); extern void ire_walk_ill_tables(uint_t match_flags, uint_t ire_type, pfv_t func, void *arg, size_t ftbl_sz, size_t htbl_sz, irb_t **ipftbl, size_t ctbl_sz, irb_t *ipctbl, ill_t *ill, zoneid_t zoneid, ip_stack_t *); extern void ire_walk_v6(pfv_t, void *, zoneid_t, ip_stack_t *); extern boolean_t ire_multirt_lookup(ire_t **, ire_t **, uint32_t, const struct ts_label_s *, ip_stack_t *); extern boolean_t ire_multirt_need_resolve(ipaddr_t, const struct ts_label_s *, ip_stack_t *); extern boolean_t ire_multirt_lookup_v6(ire_t **, ire_t **, uint32_t, const struct ts_label_s *, ip_stack_t *); extern boolean_t ire_multirt_need_resolve_v6(const in6_addr_t *, const struct ts_label_s *, ip_stack_t *); extern ire_t *ipif_lookup_multi_ire(ipif_t *, ipaddr_t); extern ire_t *ipif_lookup_multi_ire_v6(ipif_t *, const in6_addr_t *); extern ire_t *ire_get_next_bcast_ire(ire_t *, ire_t *); extern ire_t *ire_get_next_default_ire(ire_t *, ire_t *); extern void ire_arpresolve(ire_t *, ill_t *); extern void ire_freemblk(ire_t *); extern boolean_t ire_match_args(ire_t *, ipaddr_t, ipaddr_t, ipaddr_t, int, const ipif_t *, zoneid_t, uint32_t, const struct ts_label_s *, int); extern int ire_nce_init(ire_t *, struct nce_s *); extern boolean_t ire_walk_ill_match(uint_t, uint_t, ire_t *, ill_t *, zoneid_t, ip_stack_t *); #endif /* _KERNEL */ #ifdef __cplusplus } #endif #endif /* _INET_IP_IRE_H */