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 2009 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 #include <sys/mutex.h> 30 #include <sys/stream.h> 31 #include <netinet/in.h> 32 #include <netinet/icmp6.h> 33 #include <inet/ip.h> 34 35 /* 36 * Internal definitions for the kernel implementation of the IPv6 37 * Neighbor Discovery Protocol (NDP). 38 */ 39 40 #ifdef __cplusplus 41 extern "C" { 42 #endif 43 44 #ifdef _KERNEL 45 #define NCE_TABLE_SIZE 256 46 /* NDP Cache Entry */ 47 typedef struct nce_s { 48 struct nce_s *nce_next; /* Hash chain next pointer */ 49 struct nce_s **nce_ptpn; /* Pointer to previous next */ 50 struct ill_s *nce_ill; /* Associated ill */ 51 uint16_t nce_flags; /* See below */ 52 uint16_t nce_state; /* See reachability states in if.h */ 53 int16_t nce_pcnt; /* Probe counter */ 54 uint16_t nce_rcnt; /* Retransmit counter */ 55 in6_addr_t nce_addr; /* address of the nighbor */ 56 in6_addr_t nce_mask; /* If not all ones, mask allows an */ 57 /* entry to respond to requests for a group of addresses, for */ 58 /* instantance multicast addresses */ 59 in6_addr_t nce_extract_mask; /* For mappings */ 60 uint32_t nce_ll_extract_start; /* For mappings */ 61 #define nce_first_mp_to_free nce_fp_mp 62 mblk_t *nce_fp_mp; /* link layer fast path mp */ 63 mblk_t *nce_res_mp; /* DL_UNITDATA_REQ */ 64 mblk_t *nce_qd_mp; /* Head outgoing queued packets */ 65 #define nce_last_mp_to_free nce_qd_mp 66 mblk_t *nce_timer_mp; /* NDP timer mblk */ 67 mblk_t *nce_mp; /* mblk we are in, last to be freed */ 68 uint64_t nce_last; /* Time last reachable in msec */ 69 uint32_t nce_refcnt; /* nce active usage count */ 70 kmutex_t nce_lock; /* See comments on top for what */ 71 /* this field protects */ 72 int nce_unsolicit_count; /* Unsolicited Adv count */ 73 struct nce_s *nce_fastpath; /* for fastpath list */ 74 timeout_id_t nce_timeout_id; 75 uchar_t nce_ipversion; /* IPv4(ARP)/IPv6(NDP) version */ 76 uint_t nce_defense_count; /* number of NDP conflicts */ 77 uint_t nce_defense_time; /* last time defended (secs) */ 78 uint64_t nce_init_time; /* time when it was set to ND_INITIAL */ 79 boolean_t nce_trace_disable; /* True when alloc fails */ 80 } nce_t; 81 82 /* 83 * The ndp_g_t structure contains protocol specific information needed 84 * to synchronize and manage neighbor cache entries for IPv4 and IPv6. 85 * There are 2 such structures, ips_ndp4 and ips_ndp6. 86 * ips_ndp6 contains the data structures needed for IPv6 Neighbor Discovery. 87 * ips_ndp4 has IPv4 link layer info in its nce_t structures 88 * Note that the nce_t is not currently used as the arp cache itself; 89 * it is used for the following purposes: 90 * - queue packets in nce_qd_mp while waiting for arp resolution to complete 91 * - nce_{res, fp}_mp are used to track DL_UNITDATA request/responses. 92 * - track state of ARP resolution in the nce_state; 93 * 94 * Locking notes: 95 * ndp_g_lock protects neighbor cache tables access and 96 * insertion/removal of cache entries into/from these tables. 97 * nce_lock protects nce_pcnt, nce_rcnt, nce_qd_mp nce_state, 98 * nce_res_mp, nce_refcnt and nce_last. 99 * nce_refcnt is incremented for every ire pointing to this nce and 100 * every time ndp_lookup() finds an nce. 101 * Should there be a need to obtain nce_lock and ndp_g_lock, ndp_g_lock is 102 * acquired first. 103 * To avoid becoming exclusive when deleting NCEs, ndp_walk() routine holds 104 * the ndp_g_lock (i.e global lock) and marks NCEs to be deleted with 105 * NCE_F_CONDEMNED. When all active users of such NCEs are gone the walk 106 * routine passes a list for deletion to nce_ire_delete_list(). 107 * 108 * When the link-layer address of some onlink host changes, ARP will send 109 * an AR_CN_ANNOUNCE message to ip so that stale neighbor-cache 110 * information will not get used. This message is processed in ip_arp_news() 111 * by walking the nce list, and updating as appropriate. The ndp_g_hw_change 112 * flag is set by ip_arp_news() to notify nce_t users that ip_arp_news() is 113 * in progress. 114 */ 115 typedef struct ndp_g_s { 116 kmutex_t ndp_g_lock; /* Lock protecting cache hash table */ 117 nce_t *nce_mask_entries; /* mask not all ones */ 118 nce_t *nce_hash_tbl[NCE_TABLE_SIZE]; 119 int ndp_g_walker; /* # of active thread walking hash list */ 120 boolean_t ndp_g_walker_cleanup; /* true implies defer deletion. */ 121 int ndp_g_hw_change; /* non-zero if nce flush in progress */ 122 } ndp_g_t; 123 124 #define NDP_HW_CHANGE_INCR(ndp) { \ 125 mutex_enter(&(ndp)->ndp_g_lock); \ 126 (ndp)->ndp_g_hw_change++; \ 127 mutex_exit(&(ndp)->ndp_g_lock); \ 128 } 129 130 #define NDP_HW_CHANGE_DECR(ndp) { \ 131 mutex_enter(&(ndp)->ndp_g_lock); \ 132 (ndp)->ndp_g_hw_change--; \ 133 mutex_exit(&(ndp)->ndp_g_lock); \ 134 } 135 136 /* nce_flags */ 137 #define NCE_F_PERMANENT 0x1 138 #define NCE_F_MAPPING 0x2 139 #define NCE_F_ISROUTER 0x4 140 /* unused 0x8 */ 141 #define NCE_F_NONUD 0x10 142 #define NCE_F_ANYCAST 0x20 143 #define NCE_F_CONDEMNED 0x40 144 #define NCE_F_UNSOL_ADV 0x80 145 #define NCE_F_BCAST 0x100 146 147 #define NCE_EXTERNAL_FLAGS_MASK \ 148 (NCE_F_PERMANENT | NCE_F_MAPPING | NCE_F_ISROUTER | NCE_F_NONUD | \ 149 NCE_F_ANYCAST | NCE_F_UNSOL_ADV) 150 151 /* State REACHABLE, STALE, DELAY or PROBE */ 152 #define NCE_ISREACHABLE(nce) \ 153 (((((nce)->nce_state) >= ND_REACHABLE) && \ 154 ((nce)->nce_state) <= ND_PROBE)) 155 156 /* NDP flags set in SOL/ADV requests */ 157 #define NDP_UNICAST 0x1 158 #define NDP_ISROUTER 0x2 159 #define NDP_SOLICITED 0x4 160 #define NDP_ORIDE 0x8 161 #define NDP_PROBE 0x10 162 163 /* Number of packets queued in NDP for a neighbor */ 164 #define ND_MAX_Q 4 165 166 167 #ifdef DEBUG 168 #define NCE_TRACE_REF(nce) nce_trace_ref(nce) 169 #define NCE_UNTRACE_REF(nce) nce_untrace_ref(nce) 170 #else 171 #define NCE_TRACE_REF(nce) 172 #define NCE_UNTRACE_REF(nce) 173 #endif 174 175 #define NCE_REFHOLD(nce) { \ 176 mutex_enter(&(nce)->nce_lock); \ 177 (nce)->nce_refcnt++; \ 178 ASSERT((nce)->nce_refcnt != 0); \ 179 NCE_TRACE_REF(nce); \ 180 mutex_exit(&(nce)->nce_lock); \ 181 } 182 183 #define NCE_REFHOLD_NOTR(nce) { \ 184 mutex_enter(&(nce)->nce_lock); \ 185 (nce)->nce_refcnt++; \ 186 ASSERT((nce)->nce_refcnt != 0); \ 187 mutex_exit(&(nce)->nce_lock); \ 188 } 189 190 #define NCE_REFHOLD_LOCKED(nce) { \ 191 ASSERT(MUTEX_HELD(&(nce)->nce_lock)); \ 192 (nce)->nce_refcnt++; \ 193 NCE_TRACE_REF(nce); \ 194 } 195 196 /* nce_inactive destroys the mutex thus no mutex_exit is needed */ 197 #define NCE_REFRELE(nce) { \ 198 mutex_enter(&(nce)->nce_lock); \ 199 NCE_UNTRACE_REF(nce); \ 200 ASSERT((nce)->nce_refcnt != 0); \ 201 if (--(nce)->nce_refcnt == 0) \ 202 ndp_inactive(nce); \ 203 else { \ 204 mutex_exit(&(nce)->nce_lock);\ 205 } \ 206 } 207 208 #define NCE_REFRELE_NOTR(nce) { \ 209 mutex_enter(&(nce)->nce_lock); \ 210 ASSERT((nce)->nce_refcnt != 0); \ 211 if (--(nce)->nce_refcnt == 0) \ 212 ndp_inactive(nce); \ 213 else { \ 214 mutex_exit(&(nce)->nce_lock);\ 215 } \ 216 } 217 218 #define NDP_RESTART_TIMER(nce, ms) { \ 219 ASSERT(!MUTEX_HELD(&(nce)->nce_lock)); \ 220 if ((nce)->nce_timeout_id != 0) { \ 221 /* Ok to untimeout bad id. we don't hold a lock. */ \ 222 (void) untimeout((nce)->nce_timeout_id); \ 223 } \ 224 mutex_enter(&(nce)->nce_lock); \ 225 /* Don't start the timer if the nce has been deleted */ \ 226 if (!((nce)->nce_flags & NCE_F_CONDEMNED)) \ 227 nce->nce_timeout_id = timeout(ndp_timer, nce, \ 228 MSEC_TO_TICK(ms) == 0 ? 1 : MSEC_TO_TICK(ms)); \ 229 mutex_exit(&(nce)->nce_lock); \ 230 } 231 232 /* Structure for ndp_cache_count() */ 233 typedef struct { 234 int ncc_total; /* Total number of NCEs */ 235 int ncc_host; /* NCE entries without R bit set */ 236 } ncc_cache_count_t; 237 238 /* 239 * Structure of ndp_cache_reclaim(). Each field is a fraction i.e. 1 means 240 * reclaim all, N means reclaim 1/Nth of all entries, 0 means reclaim none. 241 */ 242 typedef struct { 243 int ncr_host; /* Fraction for host entries */ 244 } nce_cache_reclaim_t; 245 246 /* 247 * Structure for nce_delete_hw_changed; specifies an IPv4 address to link-layer 248 * address mapping. Any route that has a cached copy of a mapping for that 249 * IPv4 address that doesn't match the given mapping must be purged. 250 */ 251 typedef struct { 252 ipaddr_t hwm_addr; /* IPv4 address */ 253 uint_t hwm_hwlen; /* Length of hardware address (may be 0) */ 254 uchar_t *hwm_hwaddr; /* Pointer to new hardware address, if any */ 255 } nce_hw_map_t; 256 257 /* When SAP is greater than zero address appears before SAP */ 258 #define NCE_LL_ADDR_OFFSET(ill) (((ill)->ill_sap_length) < 0 ? \ 259 (sizeof (dl_unitdata_req_t)) : \ 260 ((sizeof (dl_unitdata_req_t)) + (ABS((ill)->ill_sap_length)))) 261 262 #define NCE_LL_SAP_OFFSET(ill) (((ill)->ill_sap_length) < 0 ? \ 263 ((sizeof (dl_unitdata_req_t)) + ((ill)->ill_phys_addr_length)) : \ 264 (sizeof (dl_unitdata_req_t))) 265 266 #ifdef _BIG_ENDIAN 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 ASSERT(abs_sap_len <= sizeof (uint32_t)); \ 272 ASSERT((mp)->b_rptr + NCE_LL_SAP_OFFSET(ill) + \ 273 abs_sap_len <= ((mp)->b_wptr)); \ 274 bcopy((uint8_t *)&(ill)->ill_sap + sizeof (ill->ill_sap) - \ 275 abs_sap_len, \ 276 ((mp)->b_rptr + NCE_LL_SAP_OFFSET(ill)), \ 277 abs_sap_len); \ 278 } \ 279 } 280 #else 281 #define NCE_LL_SAP_COPY(ill, mp) \ 282 { \ 283 size_t abs_sap_len = ABS((ill)->ill_sap_length); \ 284 if (abs_sap_len > 0) { \ 285 uint32_t abs_sap_len = ABS((ill)->ill_sap_length); \ 286 ASSERT(abs_sap_len <= sizeof (uint32_t)); \ 287 ASSERT((mp)->b_rptr + NCE_LL_SAP_OFFSET(ill) + \ 288 abs_sap_len <= ((mp)->b_wptr)); \ 289 bcopy(&((ill)->ill_sap), \ 290 ((mp)->b_rptr + NCE_LL_SAP_OFFSET(ill)), \ 291 abs_sap_len); \ 292 } \ 293 } 294 #endif 295 296 /* 297 * Exclusive-or the 6 bytes that are likely to contain the MAC 298 * address. Assumes table_size does not exceed 256. 299 * Assumes EUI-64 format for good hashing. 300 */ 301 #define NCE_ADDR_HASH_V6(addr, table_size) \ 302 (((addr).s6_addr8[8] ^ (addr).s6_addr8[9] ^ \ 303 (addr).s6_addr8[10] ^ (addr).s6_addr8[13] ^ \ 304 (addr).s6_addr8[14] ^ (addr).s6_addr8[15]) % (table_size)) 305 306 /* NDP Cache Entry Hash Table */ 307 #define NCE_TABLE_SIZE 256 308 309 extern void ndp_cache_count(nce_t *, char *); 310 extern void ndp_cache_reclaim(nce_t *, char *); 311 extern void ndp_delete(nce_t *); 312 extern void ndp_delete_per_ill(nce_t *, uchar_t *); 313 extern void ndp_fastpath_flush(nce_t *, char *); 314 extern boolean_t ndp_fastpath_update(nce_t *, void *); 315 extern nd_opt_hdr_t *ndp_get_option(nd_opt_hdr_t *, int, int); 316 extern void ndp_inactive(nce_t *); 317 extern void ndp_input(ill_t *, mblk_t *, mblk_t *); 318 extern boolean_t ndp_lookup_ipaddr(in_addr_t, netstack_t *); 319 extern nce_t *ndp_lookup_v6(ill_t *, boolean_t, const in6_addr_t *, 320 boolean_t); 321 extern nce_t *ndp_lookup_v4(ill_t *, const in_addr_t *, boolean_t); 322 extern int ndp_mcastreq(ill_t *, const in6_addr_t *, uint32_t, uint32_t, 323 mblk_t *); 324 extern int ndp_noresolver(ill_t *, const in6_addr_t *); 325 extern void ndp_process(nce_t *, uchar_t *, uint32_t, boolean_t); 326 extern int ndp_query(ill_t *, lif_nd_req_t *); 327 extern int ndp_resolver(ill_t *, const in6_addr_t *, mblk_t *, zoneid_t); 328 extern int ndp_sioc_update(ill_t *, lif_nd_req_t *); 329 extern boolean_t ndp_verify_optlen(nd_opt_hdr_t *, int); 330 extern void ndp_timer(void *); 331 extern void ndp_walk(ill_t *, pfi_t, void *, ip_stack_t *); 332 extern void ndp_walk_common(ndp_g_t *, ill_t *, pfi_t, 333 void *, boolean_t); 334 extern boolean_t ndp_restart_dad(nce_t *); 335 extern void ndp_do_recovery(ipif_t *); 336 extern void nce_resolv_failed(nce_t *); 337 extern void arp_resolv_failed(nce_t *); 338 extern void nce_fastpath_list_add(nce_t *); 339 extern void nce_fastpath_list_delete(nce_t *); 340 extern void nce_fastpath_list_dispatch(ill_t *, 341 boolean_t (*)(nce_t *, void *), void *); 342 extern void nce_queue_mp_common(nce_t *, mblk_t *, boolean_t); 343 extern void nce_delete_hw_changed(nce_t *, void *); 344 extern void nce_fastpath(nce_t *); 345 extern int ndp_add_v6(ill_t *, uchar_t *, const in6_addr_t *, 346 const in6_addr_t *, const in6_addr_t *, uint32_t, uint16_t, uint16_t, 347 nce_t **); 348 extern int ndp_lookup_then_add_v6(ill_t *, boolean_t, uchar_t *, 349 const in6_addr_t *, const in6_addr_t *, const in6_addr_t *, uint32_t, 350 uint16_t, uint16_t, nce_t **); 351 extern int ndp_lookup_then_add_v4(ill_t *, 352 const in_addr_t *, uint16_t, nce_t **, nce_t *); 353 354 #ifdef DEBUG 355 extern void nce_trace_ref(nce_t *); 356 extern void nce_untrace_ref(nce_t *); 357 #endif 358 359 #endif /* _KERNEL */ 360 361 #ifdef __cplusplus 362 } 363 #endif 364 365 #endif /* _INET_IP_NDP_H */ 366