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 26 #ifndef _INET_IP_NDP_H 27 #define _INET_IP_NDP_H 28 29 #pragma ident "%Z%%M% %I% %E% SMI" 30 31 #include <sys/mutex.h> 32 #include <sys/stream.h> 33 #include <netinet/in.h> 34 #include <netinet/icmp6.h> 35 #include <inet/ip.h> 36 37 /* 38 * Internal definitions for the kernel implementation of the IPv6 39 * Neighbor Discovery Protocol (NDP). 40 */ 41 42 #ifdef __cplusplus 43 extern "C" { 44 #endif 45 46 #ifdef _KERNEL 47 #define NCE_TABLE_SIZE 256 48 /* NDP Cache Entry */ 49 typedef struct nce_s { 50 struct nce_s *nce_next; /* Hash chain next pointer */ 51 struct nce_s **nce_ptpn; /* Pointer to previous next */ 52 struct ill_s *nce_ill; /* Associated ill */ 53 uint16_t nce_flags; /* See below */ 54 uint16_t nce_state; /* See reachability states in if.h */ 55 int16_t nce_pcnt; /* Probe counter */ 56 uint16_t nce_rcnt; /* Retransmit counter */ 57 in6_addr_t nce_addr; /* address of the nighbor */ 58 in6_addr_t nce_mask; /* If not all ones, mask allows an */ 59 /* entry to respond to requests for a group of addresses, for */ 60 /* instantance multicast addresses */ 61 in6_addr_t nce_extract_mask; /* For mappings */ 62 uint32_t nce_ll_extract_start; /* For mappings */ 63 #define nce_first_mp_to_free nce_fp_mp 64 mblk_t *nce_fp_mp; /* link layer fast path mp */ 65 mblk_t *nce_res_mp; /* DL_UNITDATA_REQ or link layer mp */ 66 mblk_t *nce_qd_mp; /* Head outgoing queued packets */ 67 #define nce_last_mp_to_free nce_qd_mp 68 mblk_t *nce_timer_mp; /* NDP timer mblk */ 69 mblk_t *nce_mp; /* mblk we are in, last to be freed */ 70 uint64_t nce_last; /* Time last reachable in msec */ 71 uint32_t nce_refcnt; /* nce active usage count */ 72 kmutex_t nce_lock; /* See comments on top for what */ 73 /* this field protects */ 74 int nce_unsolicit_count; /* Unsolicited Adv count */ 75 struct nce_s *nce_fastpath; /* for fastpath list */ 76 timeout_id_t nce_timeout_id; 77 uchar_t nce_ipversion; /* IPv4(ARP)/IPv6(NDP) version */ 78 uint_t nce_defense_count; /* number of NDP conflicts */ 79 uint_t nce_defense_time; /* last time defended (secs) */ 80 #ifdef NCE_DEBUG 81 th_trace_t *nce_trace[IP_TR_HASH_MAX]; 82 boolean_t nce_trace_disable; /* True when alloc fails */ 83 #endif 84 } nce_t; 85 86 /* 87 * The ndp_g_t structure contains protocol specific information needed 88 * to synchronize and manage neighbor cache entries for IPv4 and IPv6. 89 * There are 2 such structures, ndp4 and ndp6. 90 * ndp6 contains the data structures needed for IPv6 Neighbor Discovery. 91 * ndp4 has IPv4 link layer info in its nce_t structures 92 * Note that the nce_t is not currently used as the arp cache itself; 93 * it is used for the following purposes: 94 * - queue packets in nce_qd_mp while waiting for arp resolution to complete 95 * - nce_{res, fp}_mp are used to track DL_UNITDATA request/responses. 96 * - track state of ARP resolution in the nce_state; 97 * 98 * Locking notes: 99 * ndp_g_lock protects neighbor cache tables access and 100 * insertion/removal of cache entries into/from these tables. 101 * nce_lock protects nce_pcnt, nce_rcnt, nce_qd_mp nce_state, 102 * nce_res_mp, nce_refcnt and nce_last. 103 * nce_refcnt is incremented for every ire pointing to this nce and 104 * every time ndp_lookup() finds an nce. 105 * Should there be a need to obtain nce_lock and ndp_g_lock, ndp_g_lock is 106 * acquired first. 107 * To avoid becoming exclusive when deleting NCEs, ndp_walk() routine holds 108 * the ndp_g_lock (i.e global lock) and marks NCEs to be deleted with 109 * NCE_F_CONDEMNED. When all active users of such NCEs are gone the walk 110 * routine passes a list for deletion to nce_ire_delete_list(). 111 */ 112 typedef struct ndp_g_s { 113 kmutex_t ndp_g_lock; /* Lock protecting cache hash table */ 114 nce_t *nce_mask_entries; /* mask not all ones */ 115 nce_t *nce_hash_tbl[NCE_TABLE_SIZE]; 116 int ndp_g_walker; /* # of active thread walking hash list */ 117 boolean_t ndp_g_walker_cleanup; /* true implies defer deletion. */ 118 } ndp_g_t; 119 120 extern ndp_g_t ndp4, ndp6; 121 122 /* nce_flags */ 123 #define NCE_F_PERMANENT 0x1 124 #define NCE_F_MAPPING 0x2 125 #define NCE_F_ISROUTER 0x4 126 #define NCE_F_PROXY 0x8 127 #define NCE_F_NONUD 0x10 128 #define NCE_F_ANYCAST 0x20 129 #define NCE_F_CONDEMNED 0x40 130 #define NCE_F_UNSOL_ADV 0x80 131 #define NCE_F_BCAST 0x100 132 133 #define NCE_EXTERNAL_FLAGS_MASK \ 134 (NCE_F_PERMANENT | NCE_F_MAPPING | NCE_F_ISROUTER | NCE_F_NONUD | \ 135 NCE_F_ANYCAST | NCE_F_UNSOL_ADV) 136 137 /* State REACHABLE, STALE, DELAY or PROBE */ 138 #define NCE_ISREACHABLE(nce) \ 139 (((((nce)->nce_state) >= ND_REACHABLE) && \ 140 ((nce)->nce_state) <= ND_PROBE)) 141 142 /* NDP flags set in SOL/ADV requests */ 143 #define NDP_UNICAST 0x1 144 #define NDP_ISROUTER 0x2 145 #define NDP_SOLICITED 0x4 146 #define NDP_ORIDE 0x8 147 #define NDP_PROBE 0x10 148 149 /* Number of packets queued in NDP for a neighbor */ 150 #define ND_MAX_Q 4 151 152 153 #ifdef NCE_DEBUG 154 #define NCE_TRACE_REF(nce) nce_trace_ref(nce) 155 #define NCE_UNTRACE_REF(nce) nce_untrace_ref(nce) 156 #else 157 #define NCE_TRACE_REF(nce) 158 #define NCE_UNTRACE_REF(nce) 159 #endif 160 161 #define NCE_REFHOLD(nce) { \ 162 mutex_enter(&(nce)->nce_lock); \ 163 (nce)->nce_refcnt++; \ 164 ASSERT((nce)->nce_refcnt != 0); \ 165 NCE_TRACE_REF(nce); \ 166 mutex_exit(&(nce)->nce_lock); \ 167 } 168 169 #define NCE_REFHOLD_NOTR(nce) { \ 170 mutex_enter(&(nce)->nce_lock); \ 171 (nce)->nce_refcnt++; \ 172 ASSERT((nce)->nce_refcnt != 0); \ 173 mutex_exit(&(nce)->nce_lock); \ 174 } 175 176 #define NCE_REFHOLD_LOCKED(nce) { \ 177 ASSERT(MUTEX_HELD(&(nce)->nce_lock)); \ 178 (nce)->nce_refcnt++; \ 179 NCE_TRACE_REF(nce); \ 180 } 181 182 /* nce_inactive destroys the mutex thus no mutex_exit is needed */ 183 #define NCE_REFRELE(nce) { \ 184 mutex_enter(&(nce)->nce_lock); \ 185 NCE_UNTRACE_REF(nce); \ 186 ASSERT((nce)->nce_refcnt != 0); \ 187 if (--(nce)->nce_refcnt == 0) \ 188 ndp_inactive(nce); \ 189 else { \ 190 mutex_exit(&(nce)->nce_lock);\ 191 } \ 192 } 193 194 #define NCE_REFRELE_NOTR(nce) { \ 195 mutex_enter(&(nce)->nce_lock); \ 196 ASSERT((nce)->nce_refcnt != 0); \ 197 if (--(nce)->nce_refcnt == 0) \ 198 ndp_inactive(nce); \ 199 else { \ 200 mutex_exit(&(nce)->nce_lock);\ 201 } \ 202 } 203 204 #define NDP_RESTART_TIMER(nce, ms) { \ 205 ASSERT(!MUTEX_HELD(&(nce)->nce_lock)); \ 206 if ((nce)->nce_timeout_id != 0) { \ 207 /* Ok to untimeout bad id. we don't hold a lock. */ \ 208 (void) untimeout((nce)->nce_timeout_id); \ 209 } \ 210 mutex_enter(&(nce)->nce_lock); \ 211 /* Don't start the timer if the nce has been deleted */ \ 212 if (!((nce)->nce_flags & NCE_F_CONDEMNED)) \ 213 nce->nce_timeout_id = timeout(ndp_timer, nce, \ 214 MSEC_TO_TICK(ms) == 0 ? 1 : MSEC_TO_TICK(ms)); \ 215 mutex_exit(&(nce)->nce_lock); \ 216 } 217 218 /* Structure for ndp_cache_count() */ 219 typedef struct { 220 int ncc_total; /* Total number of NCEs */ 221 int ncc_host; /* NCE entries without R bit set */ 222 } ncc_cache_count_t; 223 224 /* 225 * Structure of ndp_cache_reclaim(). Each field is a fraction i.e. 1 means 226 * reclaim all, N means reclaim 1/Nth of all entries, 0 means reclaim none. 227 */ 228 typedef struct { 229 int ncr_host; /* Fraction for host entries */ 230 } nce_cache_reclaim_t; 231 232 /* 233 * Structure for nce_delete_hw_changed; specifies an IPv4 address to link-layer 234 * address mapping. Any route that has a cached copy of a mapping for that 235 * IPv4 address that doesn't match the given mapping must be purged. 236 */ 237 typedef struct { 238 ipaddr_t hwm_addr; /* IPv4 address */ 239 uint_t hwm_hwlen; /* Length of hardware address (may be 0) */ 240 uchar_t *hwm_hwaddr; /* Pointer to new hardware address, if any */ 241 } nce_hw_map_t; 242 243 /* When SAP is greater than zero address appears before SAP */ 244 #define NCE_LL_ADDR_OFFSET(ill) (((ill)->ill_sap_length) < 0 ? \ 245 (sizeof (dl_unitdata_req_t)) : \ 246 ((sizeof (dl_unitdata_req_t)) + (ABS((ill)->ill_sap_length)))) 247 248 #define NCE_LL_SAP_OFFSET(ill) (((ill)->ill_sap_length) < 0 ? \ 249 ((sizeof (dl_unitdata_req_t)) + ((ill)->ill_phys_addr_length)) : \ 250 (sizeof (dl_unitdata_req_t))) 251 252 #ifdef _BIG_ENDIAN 253 #define NCE_LL_SAP_COPY(ill, mp) \ 254 { \ 255 size_t abs_sap_len = ABS((ill)->ill_sap_length); \ 256 if (abs_sap_len > 0) { \ 257 ASSERT(abs_sap_len <= sizeof (uint32_t)); \ 258 ASSERT((mp)->b_rptr + NCE_LL_SAP_OFFSET(ill) + \ 259 abs_sap_len <= ((mp)->b_wptr)); \ 260 bcopy((uint8_t *)&(ill)->ill_sap + sizeof (ill->ill_sap) - \ 261 abs_sap_len, \ 262 ((mp)->b_rptr + NCE_LL_SAP_OFFSET(ill)), \ 263 abs_sap_len); \ 264 } \ 265 } 266 #else 267 #define NCE_LL_SAP_COPY(ill, mp) \ 268 { \ 269 size_t abs_sap_len = ABS((ill)->ill_sap_length); \ 270 if (abs_sap_len > 0) { \ 271 uint32_t abs_sap_len = ABS((ill)->ill_sap_length); \ 272 ASSERT(abs_sap_len <= sizeof (uint32_t)); \ 273 ASSERT((mp)->b_rptr + NCE_LL_SAP_OFFSET(ill) + \ 274 abs_sap_len <= ((mp)->b_wptr)); \ 275 bcopy(&((ill)->ill_sap), \ 276 ((mp)->b_rptr + NCE_LL_SAP_OFFSET(ill)), \ 277 abs_sap_len); \ 278 } \ 279 } 280 #endif 281 282 /* 283 * Exclusive-or the 6 bytes that are likely to contain the MAC 284 * address. Assumes table_size does not exceed 256. 285 * Assumes EUI-64 format for good hashing. 286 */ 287 #define NCE_ADDR_HASH_V6(addr, table_size) \ 288 (((addr).s6_addr8[8] ^ (addr).s6_addr8[9] ^ \ 289 (addr).s6_addr8[10] ^ (addr).s6_addr8[13] ^ \ 290 (addr).s6_addr8[14] ^ (addr).s6_addr8[15]) % (table_size)) 291 292 extern void ndp_cache_count(nce_t *, char *); 293 extern void ndp_cache_reclaim(nce_t *, char *); 294 extern void ndp_delete(nce_t *); 295 extern void ndp_delete_per_ill(nce_t *, uchar_t *); 296 extern void ndp_fastpath_flush(nce_t *, char *); 297 extern boolean_t ndp_fastpath_update(nce_t *, void *); 298 extern nd_opt_hdr_t *ndp_get_option(nd_opt_hdr_t *, int, int); 299 extern void ndp_inactive(nce_t *); 300 extern void ndp_input(ill_t *, mblk_t *, mblk_t *); 301 extern boolean_t ndp_lookup_ipaddr(in_addr_t); 302 extern nce_t *ndp_lookup_v6(ill_t *, const in6_addr_t *, boolean_t); 303 extern nce_t *ndp_lookup_v4(ill_t *, const in_addr_t *, boolean_t); 304 extern int ndp_lookup_then_add(ill_t *, uchar_t *, const void *, 305 const void *, const void *, uint32_t, uint16_t, 306 uint16_t, nce_t **, mblk_t *, mblk_t *); 307 extern int ndp_mcastreq(ill_t *, const in6_addr_t *, uint32_t, uint32_t, 308 mblk_t *); 309 extern int ndp_noresolver(ill_t *, const in6_addr_t *); 310 extern void ndp_process(nce_t *, uchar_t *, uint32_t, boolean_t); 311 extern int ndp_query(ill_t *, lif_nd_req_t *); 312 extern int ndp_report(queue_t *, mblk_t *, caddr_t, cred_t *); 313 extern int ndp_resolver(ill_t *, const in6_addr_t *, mblk_t *, zoneid_t); 314 extern int ndp_sioc_update(ill_t *, lif_nd_req_t *); 315 extern boolean_t ndp_verify_optlen(nd_opt_hdr_t *, int); 316 extern void ndp_timer(void *); 317 extern void ndp_walk(ill_t *, pfi_t, void *); 318 extern void ndp_walk_common(ndp_g_t *, ill_t *, pfi_t, 319 void *, boolean_t); 320 extern int ndp_add(ill_t *, uchar_t *, const void *, 321 const void *, const void *, 322 uint32_t, uint16_t, uint16_t, nce_t **, mblk_t *, mblk_t *); 323 extern boolean_t ndp_restart_dad(nce_t *); 324 extern void ndp_do_recovery(ipif_t *); 325 extern void nce_resolv_failed(nce_t *); 326 extern void arp_resolv_failed(nce_t *); 327 extern void nce_fastpath_list_add(nce_t *); 328 extern void nce_fastpath_list_delete(nce_t *); 329 extern void nce_fastpath_list_dispatch(ill_t *, 330 boolean_t (*)(nce_t *, void *), void *); 331 extern void nce_queue_mp_common(nce_t *, mblk_t *, boolean_t); 332 extern void ndp_flush_qd_mp(nce_t *); 333 extern nce_t *nce_reinit(nce_t *); 334 extern void nce_delete_hw_changed(nce_t *, void *); 335 extern void nce_fastpath(nce_t *); 336 337 #ifdef NCE_DEBUG 338 extern void nce_trace_inactive(nce_t *); 339 extern void nce_trace_ref(nce_t *); 340 extern void nce_untrace_ref(nce_t *); 341 extern int nce_thread_exit(nce_t *, caddr_t); 342 #endif 343 344 #endif /* _KERNEL */ 345 346 #ifdef __cplusplus 347 } 348 #endif 349 350 #endif /* _INET_IP_NDP_H */ 351