1*0Sstevel@tonic-gate /* 2*0Sstevel@tonic-gate * CDDL HEADER START 3*0Sstevel@tonic-gate * 4*0Sstevel@tonic-gate * The contents of this file are subject to the terms of the 5*0Sstevel@tonic-gate * Common Development and Distribution License, Version 1.0 only 6*0Sstevel@tonic-gate * (the "License"). You may not use this file except in compliance 7*0Sstevel@tonic-gate * with the License. 8*0Sstevel@tonic-gate * 9*0Sstevel@tonic-gate * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE 10*0Sstevel@tonic-gate * or http://www.opensolaris.org/os/licensing. 11*0Sstevel@tonic-gate * See the License for the specific language governing permissions 12*0Sstevel@tonic-gate * and limitations under the License. 13*0Sstevel@tonic-gate * 14*0Sstevel@tonic-gate * When distributing Covered Code, include this CDDL HEADER in each 15*0Sstevel@tonic-gate * file and include the License file at usr/src/OPENSOLARIS.LICENSE. 16*0Sstevel@tonic-gate * If applicable, add the following below this CDDL HEADER, with the 17*0Sstevel@tonic-gate * fields enclosed by brackets "[]" replaced with your own identifying 18*0Sstevel@tonic-gate * information: Portions Copyright [yyyy] [name of copyright owner] 19*0Sstevel@tonic-gate * 20*0Sstevel@tonic-gate * CDDL HEADER END 21*0Sstevel@tonic-gate */ 22*0Sstevel@tonic-gate /* 23*0Sstevel@tonic-gate * Copyright 2005 Sun Microsystems, Inc. All rights reserved. 24*0Sstevel@tonic-gate * Use is subject to license terms. 25*0Sstevel@tonic-gate */ 26*0Sstevel@tonic-gate /* 27*0Sstevel@tonic-gate * Copyright (c) 1990 Mentat Inc. 28*0Sstevel@tonic-gate */ 29*0Sstevel@tonic-gate 30*0Sstevel@tonic-gate #pragma ident "%Z%%M% %I% %E% SMI" 31*0Sstevel@tonic-gate 32*0Sstevel@tonic-gate /* 33*0Sstevel@tonic-gate * This file contains the interface control functions for IPv6. 34*0Sstevel@tonic-gate */ 35*0Sstevel@tonic-gate 36*0Sstevel@tonic-gate #include <sys/types.h> 37*0Sstevel@tonic-gate #include <sys/sysmacros.h> 38*0Sstevel@tonic-gate #include <sys/stream.h> 39*0Sstevel@tonic-gate #include <sys/dlpi.h> 40*0Sstevel@tonic-gate #include <sys/stropts.h> 41*0Sstevel@tonic-gate #include <sys/strlog.h> 42*0Sstevel@tonic-gate #include <sys/ddi.h> 43*0Sstevel@tonic-gate #include <sys/cmn_err.h> 44*0Sstevel@tonic-gate #include <sys/kstat.h> 45*0Sstevel@tonic-gate #include <sys/debug.h> 46*0Sstevel@tonic-gate #include <sys/zone.h> 47*0Sstevel@tonic-gate 48*0Sstevel@tonic-gate #include <sys/systm.h> 49*0Sstevel@tonic-gate #include <sys/param.h> 50*0Sstevel@tonic-gate #include <sys/socket.h> 51*0Sstevel@tonic-gate #define _SUN_TPI_VERSION 2 52*0Sstevel@tonic-gate #include <sys/tihdr.h> 53*0Sstevel@tonic-gate #include <sys/isa_defs.h> 54*0Sstevel@tonic-gate #include <net/if.h> 55*0Sstevel@tonic-gate #include <net/if_types.h> 56*0Sstevel@tonic-gate #include <net/if_dl.h> 57*0Sstevel@tonic-gate #include <net/route.h> 58*0Sstevel@tonic-gate #include <sys/sockio.h> 59*0Sstevel@tonic-gate #include <netinet/in.h> 60*0Sstevel@tonic-gate #include <netinet/igmp_var.h> 61*0Sstevel@tonic-gate #include <netinet/ip6.h> 62*0Sstevel@tonic-gate #include <netinet/icmp6.h> 63*0Sstevel@tonic-gate #include <netinet/in.h> 64*0Sstevel@tonic-gate 65*0Sstevel@tonic-gate #include <inet/common.h> 66*0Sstevel@tonic-gate #include <inet/mi.h> 67*0Sstevel@tonic-gate #include <inet/nd.h> 68*0Sstevel@tonic-gate #include <inet/mib2.h> 69*0Sstevel@tonic-gate #include <inet/arp.h> 70*0Sstevel@tonic-gate #include <inet/ip.h> 71*0Sstevel@tonic-gate #include <inet/ip6.h> 72*0Sstevel@tonic-gate #include <inet/ip_multi.h> 73*0Sstevel@tonic-gate #include <inet/ip_ire.h> 74*0Sstevel@tonic-gate #include <inet/ip_rts.h> 75*0Sstevel@tonic-gate #include <inet/ip_ndp.h> 76*0Sstevel@tonic-gate #include <inet/ip_if.h> 77*0Sstevel@tonic-gate #include <inet/ip6_asp.h> 78*0Sstevel@tonic-gate #include <inet/tun.h> 79*0Sstevel@tonic-gate #include <inet/ipclassifier.h> 80*0Sstevel@tonic-gate #include <inet/sctp_ip.h> 81*0Sstevel@tonic-gate 82*0Sstevel@tonic-gate #include <netinet/igmp.h> 83*0Sstevel@tonic-gate #include <netinet/ip_mroute.h> 84*0Sstevel@tonic-gate 85*0Sstevel@tonic-gate static in6_addr_t ipv6_ll_template = 86*0Sstevel@tonic-gate {(uint32_t)V6_LINKLOCAL, 0x0, 0x0, 0x0}; 87*0Sstevel@tonic-gate 88*0Sstevel@tonic-gate static ipif_t * 89*0Sstevel@tonic-gate ipif_lookup_interface_v6(const in6_addr_t *if_addr, const in6_addr_t *dst, 90*0Sstevel@tonic-gate queue_t *q, mblk_t *mp, ipsq_func_t func, int *error); 91*0Sstevel@tonic-gate 92*0Sstevel@tonic-gate /* 93*0Sstevel@tonic-gate * ipif_lookup_group_v6 94*0Sstevel@tonic-gate */ 95*0Sstevel@tonic-gate ipif_t * 96*0Sstevel@tonic-gate ipif_lookup_group_v6(const in6_addr_t *group, zoneid_t zoneid) 97*0Sstevel@tonic-gate { 98*0Sstevel@tonic-gate ire_t *ire; 99*0Sstevel@tonic-gate ipif_t *ipif; 100*0Sstevel@tonic-gate 101*0Sstevel@tonic-gate ire = ire_lookup_multi_v6(group, zoneid); 102*0Sstevel@tonic-gate if (ire == NULL) 103*0Sstevel@tonic-gate return (NULL); 104*0Sstevel@tonic-gate ipif = ire->ire_ipif; 105*0Sstevel@tonic-gate ipif_refhold(ipif); 106*0Sstevel@tonic-gate ire_refrele(ire); 107*0Sstevel@tonic-gate return (ipif); 108*0Sstevel@tonic-gate } 109*0Sstevel@tonic-gate 110*0Sstevel@tonic-gate /* 111*0Sstevel@tonic-gate * ill_lookup_group_v6 112*0Sstevel@tonic-gate */ 113*0Sstevel@tonic-gate ill_t * 114*0Sstevel@tonic-gate ill_lookup_group_v6(const in6_addr_t *group, zoneid_t zoneid) 115*0Sstevel@tonic-gate { 116*0Sstevel@tonic-gate ire_t *ire; 117*0Sstevel@tonic-gate ill_t *ill; 118*0Sstevel@tonic-gate 119*0Sstevel@tonic-gate ire = ire_lookup_multi_v6(group, zoneid); 120*0Sstevel@tonic-gate if (ire == NULL) 121*0Sstevel@tonic-gate return (NULL); 122*0Sstevel@tonic-gate ill = ire->ire_ipif->ipif_ill; 123*0Sstevel@tonic-gate ill_refhold(ill); 124*0Sstevel@tonic-gate ire_refrele(ire); 125*0Sstevel@tonic-gate return (ill); 126*0Sstevel@tonic-gate } 127*0Sstevel@tonic-gate 128*0Sstevel@tonic-gate /* 129*0Sstevel@tonic-gate * Look for an ipif with the specified interface address and destination. 130*0Sstevel@tonic-gate * The destination address is used only for matching point-to-point interfaces. 131*0Sstevel@tonic-gate */ 132*0Sstevel@tonic-gate static ipif_t * 133*0Sstevel@tonic-gate ipif_lookup_interface_v6(const in6_addr_t *if_addr, const in6_addr_t *dst, 134*0Sstevel@tonic-gate queue_t *q, mblk_t *mp, ipsq_func_t func, int *error) 135*0Sstevel@tonic-gate { 136*0Sstevel@tonic-gate ipif_t *ipif; 137*0Sstevel@tonic-gate ill_t *ill; 138*0Sstevel@tonic-gate ipsq_t *ipsq; 139*0Sstevel@tonic-gate ill_walk_context_t ctx; 140*0Sstevel@tonic-gate 141*0Sstevel@tonic-gate if (error != NULL) 142*0Sstevel@tonic-gate *error = 0; 143*0Sstevel@tonic-gate 144*0Sstevel@tonic-gate /* 145*0Sstevel@tonic-gate * First match all the point-to-point interfaces 146*0Sstevel@tonic-gate * before looking at non-point-to-point interfaces. 147*0Sstevel@tonic-gate * This is done to avoid returning non-point-to-point 148*0Sstevel@tonic-gate * ipif instead of unnumbered point-to-point ipif. 149*0Sstevel@tonic-gate */ 150*0Sstevel@tonic-gate rw_enter(&ill_g_lock, RW_READER); 151*0Sstevel@tonic-gate ill = ILL_START_WALK_V6(&ctx); 152*0Sstevel@tonic-gate for (; ill != NULL; ill = ill_next(&ctx, ill)) { 153*0Sstevel@tonic-gate GRAB_CONN_LOCK(q); 154*0Sstevel@tonic-gate mutex_enter(&ill->ill_lock); 155*0Sstevel@tonic-gate for (ipif = ill->ill_ipif; ipif; ipif = ipif->ipif_next) { 156*0Sstevel@tonic-gate /* Allow the ipif to be down */ 157*0Sstevel@tonic-gate if ((ipif->ipif_flags & IPIF_POINTOPOINT) && 158*0Sstevel@tonic-gate (IN6_ARE_ADDR_EQUAL(&ipif->ipif_v6lcl_addr, 159*0Sstevel@tonic-gate if_addr)) && 160*0Sstevel@tonic-gate (IN6_ARE_ADDR_EQUAL(&ipif->ipif_v6pp_dst_addr, 161*0Sstevel@tonic-gate dst))) { 162*0Sstevel@tonic-gate if (IPIF_CAN_LOOKUP(ipif)) { 163*0Sstevel@tonic-gate ipif_refhold_locked(ipif); 164*0Sstevel@tonic-gate mutex_exit(&ill->ill_lock); 165*0Sstevel@tonic-gate RELEASE_CONN_LOCK(q); 166*0Sstevel@tonic-gate rw_exit(&ill_g_lock); 167*0Sstevel@tonic-gate return (ipif); 168*0Sstevel@tonic-gate } else if (IPIF_CAN_WAIT(ipif, q)) { 169*0Sstevel@tonic-gate ipsq = ill->ill_phyint->phyint_ipsq; 170*0Sstevel@tonic-gate mutex_enter(&ipsq->ipsq_lock); 171*0Sstevel@tonic-gate mutex_exit(&ill->ill_lock); 172*0Sstevel@tonic-gate rw_exit(&ill_g_lock); 173*0Sstevel@tonic-gate ipsq_enq(ipsq, q, mp, func, NEW_OP, 174*0Sstevel@tonic-gate ill); 175*0Sstevel@tonic-gate mutex_exit(&ipsq->ipsq_lock); 176*0Sstevel@tonic-gate RELEASE_CONN_LOCK(q); 177*0Sstevel@tonic-gate *error = EINPROGRESS; 178*0Sstevel@tonic-gate return (NULL); 179*0Sstevel@tonic-gate } 180*0Sstevel@tonic-gate } 181*0Sstevel@tonic-gate } 182*0Sstevel@tonic-gate mutex_exit(&ill->ill_lock); 183*0Sstevel@tonic-gate RELEASE_CONN_LOCK(q); 184*0Sstevel@tonic-gate } 185*0Sstevel@tonic-gate rw_exit(&ill_g_lock); 186*0Sstevel@tonic-gate /* lookup the ipif based on interface address */ 187*0Sstevel@tonic-gate ipif = ipif_lookup_addr_v6(if_addr, NULL, ALL_ZONES, q, mp, func, 188*0Sstevel@tonic-gate error); 189*0Sstevel@tonic-gate ASSERT(ipif == NULL || ipif->ipif_isv6); 190*0Sstevel@tonic-gate return (ipif); 191*0Sstevel@tonic-gate } 192*0Sstevel@tonic-gate 193*0Sstevel@tonic-gate /* 194*0Sstevel@tonic-gate * Look for an ipif with the specified address. For point-point links 195*0Sstevel@tonic-gate * we look for matches on either the destination address and the local 196*0Sstevel@tonic-gate * address, but we ignore the check on the local address if IPIF_UNNUMBERED 197*0Sstevel@tonic-gate * is set. 198*0Sstevel@tonic-gate * Matches on a specific ill if match_ill is set. 199*0Sstevel@tonic-gate */ 200*0Sstevel@tonic-gate /* ARGSUSED */ 201*0Sstevel@tonic-gate ipif_t * 202*0Sstevel@tonic-gate ipif_lookup_addr_v6(const in6_addr_t *addr, ill_t *match_ill, zoneid_t zoneid, 203*0Sstevel@tonic-gate queue_t *q, mblk_t *mp, ipsq_func_t func, int *error) 204*0Sstevel@tonic-gate { 205*0Sstevel@tonic-gate ipif_t *ipif; 206*0Sstevel@tonic-gate ill_t *ill; 207*0Sstevel@tonic-gate boolean_t ptp = B_FALSE; 208*0Sstevel@tonic-gate ipsq_t *ipsq; 209*0Sstevel@tonic-gate ill_walk_context_t ctx; 210*0Sstevel@tonic-gate 211*0Sstevel@tonic-gate if (error != NULL) 212*0Sstevel@tonic-gate *error = 0; 213*0Sstevel@tonic-gate 214*0Sstevel@tonic-gate rw_enter(&ill_g_lock, RW_READER); 215*0Sstevel@tonic-gate /* 216*0Sstevel@tonic-gate * Repeat twice, first based on local addresses and 217*0Sstevel@tonic-gate * next time for pointopoint. 218*0Sstevel@tonic-gate */ 219*0Sstevel@tonic-gate repeat: 220*0Sstevel@tonic-gate ill = ILL_START_WALK_V6(&ctx); 221*0Sstevel@tonic-gate for (; ill != NULL; ill = ill_next(&ctx, ill)) { 222*0Sstevel@tonic-gate if (match_ill != NULL && ill != match_ill) { 223*0Sstevel@tonic-gate continue; 224*0Sstevel@tonic-gate } 225*0Sstevel@tonic-gate GRAB_CONN_LOCK(q); 226*0Sstevel@tonic-gate mutex_enter(&ill->ill_lock); 227*0Sstevel@tonic-gate for (ipif = ill->ill_ipif; ipif; ipif = ipif->ipif_next) { 228*0Sstevel@tonic-gate if (zoneid != ALL_ZONES && ipif->ipif_zoneid != zoneid) 229*0Sstevel@tonic-gate continue; 230*0Sstevel@tonic-gate /* Allow the ipif to be down */ 231*0Sstevel@tonic-gate if ((!ptp && (IN6_ARE_ADDR_EQUAL( 232*0Sstevel@tonic-gate &ipif->ipif_v6lcl_addr, addr) && 233*0Sstevel@tonic-gate (ipif->ipif_flags & IPIF_UNNUMBERED) == 0)) || 234*0Sstevel@tonic-gate (ptp && (ipif->ipif_flags & IPIF_POINTOPOINT) && 235*0Sstevel@tonic-gate IN6_ARE_ADDR_EQUAL(&ipif->ipif_v6pp_dst_addr, 236*0Sstevel@tonic-gate addr))) { 237*0Sstevel@tonic-gate if (IPIF_CAN_LOOKUP(ipif)) { 238*0Sstevel@tonic-gate ipif_refhold_locked(ipif); 239*0Sstevel@tonic-gate mutex_exit(&ill->ill_lock); 240*0Sstevel@tonic-gate RELEASE_CONN_LOCK(q); 241*0Sstevel@tonic-gate rw_exit(&ill_g_lock); 242*0Sstevel@tonic-gate return (ipif); 243*0Sstevel@tonic-gate } else if (IPIF_CAN_WAIT(ipif, q)) { 244*0Sstevel@tonic-gate ipsq = ill->ill_phyint->phyint_ipsq; 245*0Sstevel@tonic-gate mutex_enter(&ipsq->ipsq_lock); 246*0Sstevel@tonic-gate mutex_exit(&ill->ill_lock); 247*0Sstevel@tonic-gate rw_exit(&ill_g_lock); 248*0Sstevel@tonic-gate ipsq_enq(ipsq, q, mp, func, NEW_OP, 249*0Sstevel@tonic-gate ill); 250*0Sstevel@tonic-gate mutex_exit(&ipsq->ipsq_lock); 251*0Sstevel@tonic-gate RELEASE_CONN_LOCK(q); 252*0Sstevel@tonic-gate *error = EINPROGRESS; 253*0Sstevel@tonic-gate return (NULL); 254*0Sstevel@tonic-gate } 255*0Sstevel@tonic-gate } 256*0Sstevel@tonic-gate } 257*0Sstevel@tonic-gate mutex_exit(&ill->ill_lock); 258*0Sstevel@tonic-gate RELEASE_CONN_LOCK(q); 259*0Sstevel@tonic-gate } 260*0Sstevel@tonic-gate 261*0Sstevel@tonic-gate /* Repeat once more if needed */ 262*0Sstevel@tonic-gate if (ptp) { 263*0Sstevel@tonic-gate rw_exit(&ill_g_lock); 264*0Sstevel@tonic-gate if (error != NULL) 265*0Sstevel@tonic-gate *error = ENXIO; 266*0Sstevel@tonic-gate return (NULL); 267*0Sstevel@tonic-gate } 268*0Sstevel@tonic-gate ptp = B_TRUE; 269*0Sstevel@tonic-gate goto repeat; 270*0Sstevel@tonic-gate } 271*0Sstevel@tonic-gate 272*0Sstevel@tonic-gate /* 273*0Sstevel@tonic-gate * Perform various checks to verify that an address would make sense as a local 274*0Sstevel@tonic-gate * interface address. This is currently only called when an attempt is made 275*0Sstevel@tonic-gate * to set a local address. 276*0Sstevel@tonic-gate * 277*0Sstevel@tonic-gate * Does not allow a v4-mapped address, an address that equals the subnet 278*0Sstevel@tonic-gate * anycast address, ... a multicast address, ... 279*0Sstevel@tonic-gate */ 280*0Sstevel@tonic-gate boolean_t 281*0Sstevel@tonic-gate ip_local_addr_ok_v6(const in6_addr_t *addr, const in6_addr_t *subnet_mask) 282*0Sstevel@tonic-gate { 283*0Sstevel@tonic-gate in6_addr_t subnet; 284*0Sstevel@tonic-gate 285*0Sstevel@tonic-gate if (IN6_IS_ADDR_UNSPECIFIED(addr)) 286*0Sstevel@tonic-gate return (B_TRUE); /* Allow all zeros */ 287*0Sstevel@tonic-gate 288*0Sstevel@tonic-gate /* 289*0Sstevel@tonic-gate * Don't allow all zeroes or host part, but allow 290*0Sstevel@tonic-gate * all ones netmask. 291*0Sstevel@tonic-gate */ 292*0Sstevel@tonic-gate V6_MASK_COPY(*addr, *subnet_mask, subnet); 293*0Sstevel@tonic-gate if (IN6_IS_ADDR_V4MAPPED(addr) || 294*0Sstevel@tonic-gate (IN6_ARE_ADDR_EQUAL(addr, &subnet) && 295*0Sstevel@tonic-gate !IN6_ARE_ADDR_EQUAL(subnet_mask, &ipv6_all_ones)) || 296*0Sstevel@tonic-gate (IN6_IS_ADDR_V4COMPAT(addr) && CLASSD(V4_PART_OF_V6((*addr)))) || 297*0Sstevel@tonic-gate IN6_IS_ADDR_MULTICAST(addr)) 298*0Sstevel@tonic-gate return (B_FALSE); 299*0Sstevel@tonic-gate 300*0Sstevel@tonic-gate return (B_TRUE); 301*0Sstevel@tonic-gate } 302*0Sstevel@tonic-gate 303*0Sstevel@tonic-gate /* 304*0Sstevel@tonic-gate * Perform various checks to verify that an address would make sense as a 305*0Sstevel@tonic-gate * remote/subnet interface address. 306*0Sstevel@tonic-gate */ 307*0Sstevel@tonic-gate boolean_t 308*0Sstevel@tonic-gate ip_remote_addr_ok_v6(const in6_addr_t *addr, const in6_addr_t *subnet_mask) 309*0Sstevel@tonic-gate { 310*0Sstevel@tonic-gate in6_addr_t subnet; 311*0Sstevel@tonic-gate 312*0Sstevel@tonic-gate if (IN6_IS_ADDR_UNSPECIFIED(addr)) 313*0Sstevel@tonic-gate return (B_TRUE); /* Allow all zeros */ 314*0Sstevel@tonic-gate 315*0Sstevel@tonic-gate V6_MASK_COPY(*addr, *subnet_mask, subnet); 316*0Sstevel@tonic-gate if (IN6_IS_ADDR_V4MAPPED(addr) || 317*0Sstevel@tonic-gate (IN6_ARE_ADDR_EQUAL(addr, &subnet) && 318*0Sstevel@tonic-gate !IN6_ARE_ADDR_EQUAL(subnet_mask, &ipv6_all_ones)) || 319*0Sstevel@tonic-gate IN6_IS_ADDR_MULTICAST(addr) || 320*0Sstevel@tonic-gate (IN6_IS_ADDR_V4COMPAT(addr) && CLASSD(V4_PART_OF_V6((*addr))))) 321*0Sstevel@tonic-gate return (B_FALSE); 322*0Sstevel@tonic-gate 323*0Sstevel@tonic-gate return (B_TRUE); 324*0Sstevel@tonic-gate } 325*0Sstevel@tonic-gate 326*0Sstevel@tonic-gate /* 327*0Sstevel@tonic-gate * ip_rt_add_v6 is called to add an IPv6 route to the forwarding table. 328*0Sstevel@tonic-gate * ipif_arg is passed in to associate it with the correct interface 329*0Sstevel@tonic-gate * (for link-local destinations and gateways). 330*0Sstevel@tonic-gate */ 331*0Sstevel@tonic-gate /* ARGSUSED1 */ 332*0Sstevel@tonic-gate int 333*0Sstevel@tonic-gate ip_rt_add_v6(const in6_addr_t *dst_addr, const in6_addr_t *mask, 334*0Sstevel@tonic-gate const in6_addr_t *gw_addr, const in6_addr_t *src_addr, int flags, 335*0Sstevel@tonic-gate ipif_t *ipif_arg, ire_t **ire_arg, queue_t *q, mblk_t *mp, ipsq_func_t func) 336*0Sstevel@tonic-gate { 337*0Sstevel@tonic-gate ire_t *ire; 338*0Sstevel@tonic-gate ire_t *gw_ire = NULL; 339*0Sstevel@tonic-gate ipif_t *ipif; 340*0Sstevel@tonic-gate boolean_t ipif_refheld = B_FALSE; 341*0Sstevel@tonic-gate uint_t type; 342*0Sstevel@tonic-gate int match_flags = MATCH_IRE_TYPE; 343*0Sstevel@tonic-gate int error; 344*0Sstevel@tonic-gate 345*0Sstevel@tonic-gate if (ire_arg != NULL) 346*0Sstevel@tonic-gate *ire_arg = NULL; 347*0Sstevel@tonic-gate 348*0Sstevel@tonic-gate /* 349*0Sstevel@tonic-gate * Prevent routes with a zero gateway from being created (since 350*0Sstevel@tonic-gate * interfaces can currently be plumbed and brought up with no assigned 351*0Sstevel@tonic-gate * address). 352*0Sstevel@tonic-gate */ 353*0Sstevel@tonic-gate if (IN6_IS_ADDR_UNSPECIFIED(gw_addr)) 354*0Sstevel@tonic-gate return (ENETUNREACH); 355*0Sstevel@tonic-gate 356*0Sstevel@tonic-gate /* 357*0Sstevel@tonic-gate * If this is the case of RTF_HOST being set, then we set the netmask 358*0Sstevel@tonic-gate * to all ones (regardless if one was supplied). 359*0Sstevel@tonic-gate */ 360*0Sstevel@tonic-gate if (flags & RTF_HOST) 361*0Sstevel@tonic-gate mask = &ipv6_all_ones; 362*0Sstevel@tonic-gate 363*0Sstevel@tonic-gate /* 364*0Sstevel@tonic-gate * Get the ipif, if any, corresponding to the gw_addr 365*0Sstevel@tonic-gate */ 366*0Sstevel@tonic-gate ipif = ipif_lookup_interface_v6(gw_addr, dst_addr, q, mp, func, 367*0Sstevel@tonic-gate &error); 368*0Sstevel@tonic-gate if (ipif != NULL) 369*0Sstevel@tonic-gate ipif_refheld = B_TRUE; 370*0Sstevel@tonic-gate else if (error == EINPROGRESS) { 371*0Sstevel@tonic-gate ip1dbg(("ip_rt_add_v6: null and EINPROGRESS")); 372*0Sstevel@tonic-gate return (error); 373*0Sstevel@tonic-gate } 374*0Sstevel@tonic-gate 375*0Sstevel@tonic-gate /* 376*0Sstevel@tonic-gate * GateD will attempt to create routes with a loopback interface 377*0Sstevel@tonic-gate * address as the gateway and with RTF_GATEWAY set. We allow 378*0Sstevel@tonic-gate * these routes to be added, but create them as interface routes 379*0Sstevel@tonic-gate * since the gateway is an interface address. 380*0Sstevel@tonic-gate */ 381*0Sstevel@tonic-gate if ((ipif != NULL) && (ipif->ipif_ire_type == IRE_LOOPBACK)) 382*0Sstevel@tonic-gate flags &= ~RTF_GATEWAY; 383*0Sstevel@tonic-gate 384*0Sstevel@tonic-gate /* 385*0Sstevel@tonic-gate * Traditionally, interface routes are ones where RTF_GATEWAY isn't set 386*0Sstevel@tonic-gate * and the gateway address provided is one of the system's interface 387*0Sstevel@tonic-gate * addresses. By using the routing socket interface and supplying an 388*0Sstevel@tonic-gate * RTA_IFP sockaddr with an interface index, an alternate method of 389*0Sstevel@tonic-gate * specifying an interface route to be created is available which uses 390*0Sstevel@tonic-gate * the interface index that specifies the outgoing interface rather than 391*0Sstevel@tonic-gate * the address of an outgoing interface (which may not be able to 392*0Sstevel@tonic-gate * uniquely identify an interface). When coupled with the RTF_GATEWAY 393*0Sstevel@tonic-gate * flag, routes can be specified which not only specify the next-hop to 394*0Sstevel@tonic-gate * be used when routing to a certain prefix, but also which outgoing 395*0Sstevel@tonic-gate * interface should be used. 396*0Sstevel@tonic-gate * 397*0Sstevel@tonic-gate * Previously, interfaces would have unique addresses assigned to them 398*0Sstevel@tonic-gate * and so the address assigned to a particular interface could be used 399*0Sstevel@tonic-gate * to identify a particular interface. One exception to this was the 400*0Sstevel@tonic-gate * case of an unnumbered interface (where IPIF_UNNUMBERED was set). 401*0Sstevel@tonic-gate * 402*0Sstevel@tonic-gate * With the advent of IPv6 and its link-local addresses, this 403*0Sstevel@tonic-gate * restriction was relaxed and interfaces could share addresses between 404*0Sstevel@tonic-gate * themselves. In fact, typically all of the link-local interfaces on 405*0Sstevel@tonic-gate * an IPv6 node or router will have the same link-local address. In 406*0Sstevel@tonic-gate * order to differentiate between these interfaces, the use of an 407*0Sstevel@tonic-gate * interface index is necessary and this index can be carried inside a 408*0Sstevel@tonic-gate * RTA_IFP sockaddr (which is actually a sockaddr_dl). One restriction 409*0Sstevel@tonic-gate * of using the interface index, however, is that all of the ipif's that 410*0Sstevel@tonic-gate * are part of an ill have the same index and so the RTA_IFP sockaddr 411*0Sstevel@tonic-gate * cannot be used to differentiate between ipif's (or logical 412*0Sstevel@tonic-gate * interfaces) that belong to the same ill (physical interface). 413*0Sstevel@tonic-gate * 414*0Sstevel@tonic-gate * For example, in the following case involving IPv4 interfaces and 415*0Sstevel@tonic-gate * logical interfaces 416*0Sstevel@tonic-gate * 417*0Sstevel@tonic-gate * 192.0.2.32 255.255.255.224 192.0.2.33 U if0 418*0Sstevel@tonic-gate * 192.0.2.32 255.255.255.224 192.0.2.34 U if0:1 419*0Sstevel@tonic-gate * 192.0.2.32 255.255.255.224 192.0.2.35 U if0:2 420*0Sstevel@tonic-gate * 421*0Sstevel@tonic-gate * the ipif's corresponding to each of these interface routes can be 422*0Sstevel@tonic-gate * uniquely identified by the "gateway" (actually interface address). 423*0Sstevel@tonic-gate * 424*0Sstevel@tonic-gate * In this case involving multiple IPv6 default routes to a particular 425*0Sstevel@tonic-gate * link-local gateway, the use of RTA_IFP is necessary to specify which 426*0Sstevel@tonic-gate * default route is of interest: 427*0Sstevel@tonic-gate * 428*0Sstevel@tonic-gate * default fe80::123:4567:89ab:cdef U if0 429*0Sstevel@tonic-gate * default fe80::123:4567:89ab:cdef U if1 430*0Sstevel@tonic-gate */ 431*0Sstevel@tonic-gate 432*0Sstevel@tonic-gate /* RTF_GATEWAY not set */ 433*0Sstevel@tonic-gate if (!(flags & RTF_GATEWAY)) { 434*0Sstevel@tonic-gate queue_t *stq; 435*0Sstevel@tonic-gate 436*0Sstevel@tonic-gate /* 437*0Sstevel@tonic-gate * As the interface index specified with the RTA_IFP sockaddr is 438*0Sstevel@tonic-gate * the same for all ipif's off of an ill, the matching logic 439*0Sstevel@tonic-gate * below uses MATCH_IRE_ILL if such an index was specified. 440*0Sstevel@tonic-gate * This means that routes sharing the same prefix when added 441*0Sstevel@tonic-gate * using a RTA_IFP sockaddr must have distinct interface 442*0Sstevel@tonic-gate * indices (namely, they must be on distinct ill's). 443*0Sstevel@tonic-gate * 444*0Sstevel@tonic-gate * On the other hand, since the gateway address will usually be 445*0Sstevel@tonic-gate * different for each ipif on the system, the matching logic 446*0Sstevel@tonic-gate * uses MATCH_IRE_IPIF in the case of a traditional interface 447*0Sstevel@tonic-gate * route. This means that interface routes for the same prefix 448*0Sstevel@tonic-gate * can be created if they belong to distinct ipif's and if a 449*0Sstevel@tonic-gate * RTA_IFP sockaddr is not present. 450*0Sstevel@tonic-gate */ 451*0Sstevel@tonic-gate if (ipif_arg != NULL) { 452*0Sstevel@tonic-gate if (ipif_refheld) { 453*0Sstevel@tonic-gate ipif_refrele(ipif); 454*0Sstevel@tonic-gate ipif_refheld = B_FALSE; 455*0Sstevel@tonic-gate } 456*0Sstevel@tonic-gate ipif = ipif_arg; 457*0Sstevel@tonic-gate match_flags |= MATCH_IRE_ILL; 458*0Sstevel@tonic-gate } else { 459*0Sstevel@tonic-gate /* 460*0Sstevel@tonic-gate * Check the ipif corresponding to the gw_addr 461*0Sstevel@tonic-gate */ 462*0Sstevel@tonic-gate if (ipif == NULL) 463*0Sstevel@tonic-gate return (ENETUNREACH); 464*0Sstevel@tonic-gate match_flags |= MATCH_IRE_IPIF; 465*0Sstevel@tonic-gate } 466*0Sstevel@tonic-gate 467*0Sstevel@tonic-gate ASSERT(ipif != NULL); 468*0Sstevel@tonic-gate /* 469*0Sstevel@tonic-gate * We check for an existing entry at this point. 470*0Sstevel@tonic-gate */ 471*0Sstevel@tonic-gate match_flags |= MATCH_IRE_MASK; 472*0Sstevel@tonic-gate ire = ire_ftable_lookup_v6(dst_addr, mask, 0, IRE_INTERFACE, 473*0Sstevel@tonic-gate ipif, NULL, ALL_ZONES, 0, match_flags); 474*0Sstevel@tonic-gate if (ire != NULL) { 475*0Sstevel@tonic-gate ire_refrele(ire); 476*0Sstevel@tonic-gate if (ipif_refheld) 477*0Sstevel@tonic-gate ipif_refrele(ipif); 478*0Sstevel@tonic-gate return (EEXIST); 479*0Sstevel@tonic-gate } 480*0Sstevel@tonic-gate 481*0Sstevel@tonic-gate stq = (ipif->ipif_net_type == IRE_IF_RESOLVER) 482*0Sstevel@tonic-gate ? ipif->ipif_rq : ipif->ipif_wq; 483*0Sstevel@tonic-gate 484*0Sstevel@tonic-gate /* 485*0Sstevel@tonic-gate * Create a copy of the IRE_LOOPBACK, IRE_IF_NORESOLVER or 486*0Sstevel@tonic-gate * IRE_IF_RESOLVER with the modified address and netmask. 487*0Sstevel@tonic-gate */ 488*0Sstevel@tonic-gate ire = ire_create_v6( 489*0Sstevel@tonic-gate dst_addr, 490*0Sstevel@tonic-gate mask, 491*0Sstevel@tonic-gate &ipif->ipif_v6src_addr, 492*0Sstevel@tonic-gate NULL, 493*0Sstevel@tonic-gate &ipif->ipif_mtu, 494*0Sstevel@tonic-gate NULL, 495*0Sstevel@tonic-gate NULL, 496*0Sstevel@tonic-gate stq, 497*0Sstevel@tonic-gate ipif->ipif_net_type, 498*0Sstevel@tonic-gate ipif->ipif_resolver_mp, 499*0Sstevel@tonic-gate ipif, 500*0Sstevel@tonic-gate NULL, 501*0Sstevel@tonic-gate 0, 502*0Sstevel@tonic-gate 0, 503*0Sstevel@tonic-gate flags, 504*0Sstevel@tonic-gate &ire_uinfo_null); 505*0Sstevel@tonic-gate if (ire == NULL) { 506*0Sstevel@tonic-gate if (ipif_refheld) 507*0Sstevel@tonic-gate ipif_refrele(ipif); 508*0Sstevel@tonic-gate return (ENOMEM); 509*0Sstevel@tonic-gate } 510*0Sstevel@tonic-gate 511*0Sstevel@tonic-gate /* 512*0Sstevel@tonic-gate * Some software (for example, GateD and Sun Cluster) attempts 513*0Sstevel@tonic-gate * to create (what amount to) IRE_PREFIX routes with the 514*0Sstevel@tonic-gate * loopback address as the gateway. This is primarily done to 515*0Sstevel@tonic-gate * set up prefixes with the RTF_REJECT flag set (for example, 516*0Sstevel@tonic-gate * when generating aggregate routes.) 517*0Sstevel@tonic-gate * 518*0Sstevel@tonic-gate * If the IRE type (as defined by ipif->ipif_net_type) is 519*0Sstevel@tonic-gate * IRE_LOOPBACK, then we map the request into a 520*0Sstevel@tonic-gate * IRE_IF_NORESOLVER. 521*0Sstevel@tonic-gate * 522*0Sstevel@tonic-gate * Needless to say, the real IRE_LOOPBACK is NOT created by this 523*0Sstevel@tonic-gate * routine, but rather using ire_create_v6() directly. 524*0Sstevel@tonic-gate */ 525*0Sstevel@tonic-gate if (ipif->ipif_net_type == IRE_LOOPBACK) 526*0Sstevel@tonic-gate ire->ire_type = IRE_IF_NORESOLVER; 527*0Sstevel@tonic-gate error = ire_add(&ire, q, mp, func); 528*0Sstevel@tonic-gate if (error == 0) 529*0Sstevel@tonic-gate goto save_ire; 530*0Sstevel@tonic-gate /* 531*0Sstevel@tonic-gate * In the result of failure, ire_add() will have already 532*0Sstevel@tonic-gate * deleted the ire in question, so there is no need to 533*0Sstevel@tonic-gate * do that here. 534*0Sstevel@tonic-gate */ 535*0Sstevel@tonic-gate if (ipif_refheld) 536*0Sstevel@tonic-gate ipif_refrele(ipif); 537*0Sstevel@tonic-gate return (error); 538*0Sstevel@tonic-gate } 539*0Sstevel@tonic-gate if (ipif_refheld) { 540*0Sstevel@tonic-gate ipif_refrele(ipif); 541*0Sstevel@tonic-gate ipif_refheld = B_FALSE; 542*0Sstevel@tonic-gate } 543*0Sstevel@tonic-gate 544*0Sstevel@tonic-gate /* 545*0Sstevel@tonic-gate * Get an interface IRE for the specified gateway. 546*0Sstevel@tonic-gate * If we don't have an IRE_IF_NORESOLVER or IRE_IF_RESOLVER for the 547*0Sstevel@tonic-gate * gateway, it is currently unreachable and we fail the request 548*0Sstevel@tonic-gate * accordingly. 549*0Sstevel@tonic-gate */ 550*0Sstevel@tonic-gate ipif = ipif_arg; 551*0Sstevel@tonic-gate if (ipif_arg != NULL) 552*0Sstevel@tonic-gate match_flags |= MATCH_IRE_ILL; 553*0Sstevel@tonic-gate gw_ire = ire_ftable_lookup_v6(gw_addr, 0, 0, IRE_INTERFACE, ipif_arg, 554*0Sstevel@tonic-gate NULL, ALL_ZONES, 0, match_flags); 555*0Sstevel@tonic-gate if (gw_ire == NULL) 556*0Sstevel@tonic-gate return (ENETUNREACH); 557*0Sstevel@tonic-gate 558*0Sstevel@tonic-gate /* 559*0Sstevel@tonic-gate * We create one of three types of IREs as a result of this request 560*0Sstevel@tonic-gate * based on the netmask. A netmask of all ones (which is automatically 561*0Sstevel@tonic-gate * assumed when RTF_HOST is set) results in an IRE_HOST being created. 562*0Sstevel@tonic-gate * An all zeroes netmask implies a default route so an IRE_DEFAULT is 563*0Sstevel@tonic-gate * created. Otherwise, an IRE_PREFIX route is created for the 564*0Sstevel@tonic-gate * destination prefix. 565*0Sstevel@tonic-gate */ 566*0Sstevel@tonic-gate if (IN6_ARE_ADDR_EQUAL(mask, &ipv6_all_ones)) 567*0Sstevel@tonic-gate type = IRE_HOST; 568*0Sstevel@tonic-gate else if (IN6_IS_ADDR_UNSPECIFIED(mask)) 569*0Sstevel@tonic-gate type = IRE_DEFAULT; 570*0Sstevel@tonic-gate else 571*0Sstevel@tonic-gate type = IRE_PREFIX; 572*0Sstevel@tonic-gate 573*0Sstevel@tonic-gate /* check for a duplicate entry */ 574*0Sstevel@tonic-gate ire = ire_ftable_lookup_v6(dst_addr, mask, gw_addr, type, ipif_arg, 575*0Sstevel@tonic-gate NULL, ALL_ZONES, 0, match_flags | MATCH_IRE_MASK | MATCH_IRE_GW); 576*0Sstevel@tonic-gate if (ire != NULL) { 577*0Sstevel@tonic-gate ire_refrele(gw_ire); 578*0Sstevel@tonic-gate ire_refrele(ire); 579*0Sstevel@tonic-gate return (EEXIST); 580*0Sstevel@tonic-gate } 581*0Sstevel@tonic-gate 582*0Sstevel@tonic-gate /* Create the IRE. */ 583*0Sstevel@tonic-gate ire = ire_create_v6( 584*0Sstevel@tonic-gate dst_addr, /* dest address */ 585*0Sstevel@tonic-gate mask, /* mask */ 586*0Sstevel@tonic-gate /* src address assigned by the caller? */ 587*0Sstevel@tonic-gate (((flags & RTF_SETSRC) && !IN6_IS_ADDR_UNSPECIFIED(src_addr)) ? 588*0Sstevel@tonic-gate src_addr : NULL), 589*0Sstevel@tonic-gate gw_addr, /* gateway address */ 590*0Sstevel@tonic-gate &gw_ire->ire_max_frag, 591*0Sstevel@tonic-gate NULL, /* no Fast Path header */ 592*0Sstevel@tonic-gate NULL, /* no recv-from queue */ 593*0Sstevel@tonic-gate NULL, /* no send-to queue */ 594*0Sstevel@tonic-gate (ushort_t)type, /* IRE type */ 595*0Sstevel@tonic-gate NULL, 596*0Sstevel@tonic-gate ipif_arg, 597*0Sstevel@tonic-gate NULL, 598*0Sstevel@tonic-gate 0, 599*0Sstevel@tonic-gate 0, 600*0Sstevel@tonic-gate flags, 601*0Sstevel@tonic-gate &gw_ire->ire_uinfo); /* Inherit ULP info from gw */ 602*0Sstevel@tonic-gate if (ire == NULL) { 603*0Sstevel@tonic-gate ire_refrele(gw_ire); 604*0Sstevel@tonic-gate return (ENOMEM); 605*0Sstevel@tonic-gate } 606*0Sstevel@tonic-gate 607*0Sstevel@tonic-gate /* 608*0Sstevel@tonic-gate * POLICY: should we allow an RTF_HOST with address INADDR_ANY? 609*0Sstevel@tonic-gate * SUN/OS socket stuff does but do we really want to allow ::0 ? 610*0Sstevel@tonic-gate */ 611*0Sstevel@tonic-gate 612*0Sstevel@tonic-gate /* Add the new IRE. */ 613*0Sstevel@tonic-gate error = ire_add(&ire, q, mp, func); 614*0Sstevel@tonic-gate /* 615*0Sstevel@tonic-gate * In the result of failure, ire_add() will have already 616*0Sstevel@tonic-gate * deleted the ire in question, so there is no need to 617*0Sstevel@tonic-gate * do that here. 618*0Sstevel@tonic-gate */ 619*0Sstevel@tonic-gate if (error != 0) { 620*0Sstevel@tonic-gate ire_refrele(gw_ire); 621*0Sstevel@tonic-gate return (error); 622*0Sstevel@tonic-gate } 623*0Sstevel@tonic-gate 624*0Sstevel@tonic-gate if (flags & RTF_MULTIRT) { 625*0Sstevel@tonic-gate /* 626*0Sstevel@tonic-gate * Invoke the CGTP (multirouting) filtering module 627*0Sstevel@tonic-gate * to add the dst address in the filtering database. 628*0Sstevel@tonic-gate * Replicated inbound packets coming from that address 629*0Sstevel@tonic-gate * will be filtered to discard the duplicates. 630*0Sstevel@tonic-gate * It is not necessary to call the CGTP filter hook 631*0Sstevel@tonic-gate * when the dst address is a multicast, because an 632*0Sstevel@tonic-gate * IP source address cannot be a multicast. 633*0Sstevel@tonic-gate */ 634*0Sstevel@tonic-gate if ((ip_cgtp_filter_ops != NULL) && 635*0Sstevel@tonic-gate !IN6_IS_ADDR_MULTICAST(&(ire->ire_addr_v6))) { 636*0Sstevel@tonic-gate int res = ip_cgtp_filter_ops->cfo_add_dest_v6( 637*0Sstevel@tonic-gate &ire->ire_addr_v6, 638*0Sstevel@tonic-gate &ire->ire_gateway_addr_v6, 639*0Sstevel@tonic-gate &ire->ire_src_addr_v6, 640*0Sstevel@tonic-gate &gw_ire->ire_src_addr_v6); 641*0Sstevel@tonic-gate if (res != 0) { 642*0Sstevel@tonic-gate ire_refrele(gw_ire); 643*0Sstevel@tonic-gate ire_delete(ire); 644*0Sstevel@tonic-gate return (res); 645*0Sstevel@tonic-gate } 646*0Sstevel@tonic-gate } 647*0Sstevel@tonic-gate } 648*0Sstevel@tonic-gate 649*0Sstevel@tonic-gate save_ire: 650*0Sstevel@tonic-gate if (gw_ire != NULL) { 651*0Sstevel@tonic-gate ire_refrele(gw_ire); 652*0Sstevel@tonic-gate } 653*0Sstevel@tonic-gate if (ipif != NULL) { 654*0Sstevel@tonic-gate mblk_t *save_mp; 655*0Sstevel@tonic-gate 656*0Sstevel@tonic-gate /* 657*0Sstevel@tonic-gate * Save enough information so that we can recreate the IRE if 658*0Sstevel@tonic-gate * the interface goes down and then up. The metrics associated 659*0Sstevel@tonic-gate * with the route will be saved as well when rts_setmetrics() is 660*0Sstevel@tonic-gate * called after the IRE has been created. In the case where 661*0Sstevel@tonic-gate * memory cannot be allocated, none of this information will be 662*0Sstevel@tonic-gate * saved. 663*0Sstevel@tonic-gate */ 664*0Sstevel@tonic-gate save_mp = allocb(sizeof (ifrt_t), BPRI_MED); 665*0Sstevel@tonic-gate if (save_mp != NULL) { 666*0Sstevel@tonic-gate ifrt_t *ifrt; 667*0Sstevel@tonic-gate 668*0Sstevel@tonic-gate save_mp->b_wptr += sizeof (ifrt_t); 669*0Sstevel@tonic-gate ifrt = (ifrt_t *)save_mp->b_rptr; 670*0Sstevel@tonic-gate bzero(ifrt, sizeof (ifrt_t)); 671*0Sstevel@tonic-gate ifrt->ifrt_type = ire->ire_type; 672*0Sstevel@tonic-gate ifrt->ifrt_v6addr = ire->ire_addr_v6; 673*0Sstevel@tonic-gate mutex_enter(&ire->ire_lock); 674*0Sstevel@tonic-gate ifrt->ifrt_v6gateway_addr = ire->ire_gateway_addr_v6; 675*0Sstevel@tonic-gate ifrt->ifrt_v6src_addr = ire->ire_src_addr_v6; 676*0Sstevel@tonic-gate mutex_exit(&ire->ire_lock); 677*0Sstevel@tonic-gate ifrt->ifrt_v6mask = ire->ire_mask_v6; 678*0Sstevel@tonic-gate ifrt->ifrt_flags = ire->ire_flags; 679*0Sstevel@tonic-gate ifrt->ifrt_max_frag = ire->ire_max_frag; 680*0Sstevel@tonic-gate mutex_enter(&ipif->ipif_saved_ire_lock); 681*0Sstevel@tonic-gate save_mp->b_cont = ipif->ipif_saved_ire_mp; 682*0Sstevel@tonic-gate ipif->ipif_saved_ire_mp = save_mp; 683*0Sstevel@tonic-gate ipif->ipif_saved_ire_cnt++; 684*0Sstevel@tonic-gate mutex_exit(&ipif->ipif_saved_ire_lock); 685*0Sstevel@tonic-gate } 686*0Sstevel@tonic-gate } 687*0Sstevel@tonic-gate if (ire_arg != NULL) { 688*0Sstevel@tonic-gate /* 689*0Sstevel@tonic-gate * Store the ire that was successfully added into where ire_arg 690*0Sstevel@tonic-gate * points to so that callers don't have to look it up 691*0Sstevel@tonic-gate * themselves (but they are responsible for ire_refrele()ing 692*0Sstevel@tonic-gate * the ire when they are finished with it). 693*0Sstevel@tonic-gate */ 694*0Sstevel@tonic-gate *ire_arg = ire; 695*0Sstevel@tonic-gate } else { 696*0Sstevel@tonic-gate ire_refrele(ire); /* Held in ire_add */ 697*0Sstevel@tonic-gate } 698*0Sstevel@tonic-gate if (ipif_refheld) 699*0Sstevel@tonic-gate ipif_refrele(ipif); 700*0Sstevel@tonic-gate return (0); 701*0Sstevel@tonic-gate } 702*0Sstevel@tonic-gate 703*0Sstevel@tonic-gate /* 704*0Sstevel@tonic-gate * ip_rt_delete_v6 is called to delete an IPv6 route. 705*0Sstevel@tonic-gate * ipif_arg is passed in to associate it with the correct interface 706*0Sstevel@tonic-gate * (for link-local destinations and gateways). 707*0Sstevel@tonic-gate */ 708*0Sstevel@tonic-gate /* ARGSUSED4 */ 709*0Sstevel@tonic-gate int 710*0Sstevel@tonic-gate ip_rt_delete_v6(const in6_addr_t *dst_addr, const in6_addr_t *mask, 711*0Sstevel@tonic-gate const in6_addr_t *gw_addr, uint_t rtm_addrs, int flags, ipif_t *ipif_arg, 712*0Sstevel@tonic-gate queue_t *q, mblk_t *mp, ipsq_func_t func) 713*0Sstevel@tonic-gate { 714*0Sstevel@tonic-gate ire_t *ire = NULL; 715*0Sstevel@tonic-gate ipif_t *ipif; 716*0Sstevel@tonic-gate uint_t type; 717*0Sstevel@tonic-gate uint_t match_flags = MATCH_IRE_TYPE; 718*0Sstevel@tonic-gate int err = 0; 719*0Sstevel@tonic-gate boolean_t ipif_refheld = B_FALSE; 720*0Sstevel@tonic-gate 721*0Sstevel@tonic-gate /* 722*0Sstevel@tonic-gate * If this is the case of RTF_HOST being set, then we set the netmask 723*0Sstevel@tonic-gate * to all ones. Otherwise, we use the netmask if one was supplied. 724*0Sstevel@tonic-gate */ 725*0Sstevel@tonic-gate if (flags & RTF_HOST) { 726*0Sstevel@tonic-gate mask = &ipv6_all_ones; 727*0Sstevel@tonic-gate match_flags |= MATCH_IRE_MASK; 728*0Sstevel@tonic-gate } else if (rtm_addrs & RTA_NETMASK) { 729*0Sstevel@tonic-gate match_flags |= MATCH_IRE_MASK; 730*0Sstevel@tonic-gate } 731*0Sstevel@tonic-gate 732*0Sstevel@tonic-gate /* 733*0Sstevel@tonic-gate * Note that RTF_GATEWAY is never set on a delete, therefore 734*0Sstevel@tonic-gate * we check if the gateway address is one of our interfaces first, 735*0Sstevel@tonic-gate * and fall back on RTF_GATEWAY routes. 736*0Sstevel@tonic-gate * 737*0Sstevel@tonic-gate * This makes it possible to delete an original 738*0Sstevel@tonic-gate * IRE_IF_NORESOLVER/IRE_IF_RESOLVER - consistent with SunOS 4.1. 739*0Sstevel@tonic-gate * 740*0Sstevel@tonic-gate * As the interface index specified with the RTA_IFP sockaddr is the 741*0Sstevel@tonic-gate * same for all ipif's off of an ill, the matching logic below uses 742*0Sstevel@tonic-gate * MATCH_IRE_ILL if such an index was specified. This means a route 743*0Sstevel@tonic-gate * sharing the same prefix and interface index as the the route 744*0Sstevel@tonic-gate * intended to be deleted might be deleted instead if a RTA_IFP sockaddr 745*0Sstevel@tonic-gate * is specified in the request. 746*0Sstevel@tonic-gate * 747*0Sstevel@tonic-gate * On the other hand, since the gateway address will usually be 748*0Sstevel@tonic-gate * different for each ipif on the system, the matching logic 749*0Sstevel@tonic-gate * uses MATCH_IRE_IPIF in the case of a traditional interface 750*0Sstevel@tonic-gate * route. This means that interface routes for the same prefix can be 751*0Sstevel@tonic-gate * uniquely identified if they belong to distinct ipif's and if a 752*0Sstevel@tonic-gate * RTA_IFP sockaddr is not present. 753*0Sstevel@tonic-gate * 754*0Sstevel@tonic-gate * For more detail on specifying routes by gateway address and by 755*0Sstevel@tonic-gate * interface index, see the comments in ip_rt_add_v6(). 756*0Sstevel@tonic-gate */ 757*0Sstevel@tonic-gate ipif = ipif_lookup_interface_v6(gw_addr, dst_addr, q, mp, func, &err); 758*0Sstevel@tonic-gate if (ipif != NULL) { 759*0Sstevel@tonic-gate ipif_refheld = B_TRUE; 760*0Sstevel@tonic-gate if (ipif_arg != NULL) { 761*0Sstevel@tonic-gate ipif_refrele(ipif); 762*0Sstevel@tonic-gate ipif_refheld = B_FALSE; 763*0Sstevel@tonic-gate ipif = ipif_arg; 764*0Sstevel@tonic-gate match_flags |= MATCH_IRE_ILL; 765*0Sstevel@tonic-gate } else { 766*0Sstevel@tonic-gate match_flags |= MATCH_IRE_IPIF; 767*0Sstevel@tonic-gate } 768*0Sstevel@tonic-gate 769*0Sstevel@tonic-gate if (ipif->ipif_ire_type == IRE_LOOPBACK) 770*0Sstevel@tonic-gate ire = ire_ctable_lookup_v6(dst_addr, 0, IRE_LOOPBACK, 771*0Sstevel@tonic-gate ipif, ALL_ZONES, match_flags); 772*0Sstevel@tonic-gate if (ire == NULL) 773*0Sstevel@tonic-gate ire = ire_ftable_lookup_v6(dst_addr, mask, 0, 774*0Sstevel@tonic-gate IRE_INTERFACE, ipif, NULL, ALL_ZONES, 0, 775*0Sstevel@tonic-gate match_flags); 776*0Sstevel@tonic-gate } else if (err == EINPROGRESS) { 777*0Sstevel@tonic-gate return (err); 778*0Sstevel@tonic-gate } else { 779*0Sstevel@tonic-gate err = 0; 780*0Sstevel@tonic-gate } 781*0Sstevel@tonic-gate if (ire == NULL) { 782*0Sstevel@tonic-gate /* 783*0Sstevel@tonic-gate * At this point, the gateway address is not one of our own 784*0Sstevel@tonic-gate * addresses or a matching interface route was not found. We 785*0Sstevel@tonic-gate * set the IRE type to lookup based on whether 786*0Sstevel@tonic-gate * this is a host route, a default route or just a prefix. 787*0Sstevel@tonic-gate * 788*0Sstevel@tonic-gate * If an ipif_arg was passed in, then the lookup is based on an 789*0Sstevel@tonic-gate * interface index so MATCH_IRE_ILL is added to match_flags. 790*0Sstevel@tonic-gate * In any case, MATCH_IRE_IPIF is cleared and MATCH_IRE_GW is 791*0Sstevel@tonic-gate * set as the route being looked up is not a traditional 792*0Sstevel@tonic-gate * interface route. 793*0Sstevel@tonic-gate */ 794*0Sstevel@tonic-gate match_flags &= ~MATCH_IRE_IPIF; 795*0Sstevel@tonic-gate match_flags |= MATCH_IRE_GW; 796*0Sstevel@tonic-gate if (ipif_arg != NULL) 797*0Sstevel@tonic-gate match_flags |= MATCH_IRE_ILL; 798*0Sstevel@tonic-gate if (IN6_ARE_ADDR_EQUAL(mask, &ipv6_all_ones)) 799*0Sstevel@tonic-gate type = IRE_HOST; 800*0Sstevel@tonic-gate else if (IN6_IS_ADDR_UNSPECIFIED(mask)) 801*0Sstevel@tonic-gate type = IRE_DEFAULT; 802*0Sstevel@tonic-gate else 803*0Sstevel@tonic-gate type = IRE_PREFIX; 804*0Sstevel@tonic-gate ire = ire_ftable_lookup_v6(dst_addr, mask, gw_addr, type, 805*0Sstevel@tonic-gate ipif_arg, NULL, ALL_ZONES, 0, match_flags); 806*0Sstevel@tonic-gate if (ire == NULL && type == IRE_HOST) { 807*0Sstevel@tonic-gate ire = ire_ftable_lookup_v6(dst_addr, mask, gw_addr, 808*0Sstevel@tonic-gate IRE_HOST_REDIRECT, ipif_arg, NULL, ALL_ZONES, 0, 809*0Sstevel@tonic-gate match_flags); 810*0Sstevel@tonic-gate } 811*0Sstevel@tonic-gate } 812*0Sstevel@tonic-gate 813*0Sstevel@tonic-gate if (ipif_refheld) { 814*0Sstevel@tonic-gate ipif_refrele(ipif); 815*0Sstevel@tonic-gate ipif_refheld = B_FALSE; 816*0Sstevel@tonic-gate } 817*0Sstevel@tonic-gate if (ire == NULL) 818*0Sstevel@tonic-gate return (ESRCH); 819*0Sstevel@tonic-gate 820*0Sstevel@tonic-gate if (ire->ire_flags & RTF_MULTIRT) { 821*0Sstevel@tonic-gate /* 822*0Sstevel@tonic-gate * Invoke the CGTP (multirouting) filtering module 823*0Sstevel@tonic-gate * to remove the dst address from the filtering database. 824*0Sstevel@tonic-gate * Packets coming from that address will no longer be 825*0Sstevel@tonic-gate * filtered to remove duplicates. 826*0Sstevel@tonic-gate */ 827*0Sstevel@tonic-gate if (ip_cgtp_filter_ops != NULL) { 828*0Sstevel@tonic-gate err = ip_cgtp_filter_ops->cfo_del_dest_v6( 829*0Sstevel@tonic-gate &ire->ire_addr_v6, &ire->ire_gateway_addr_v6); 830*0Sstevel@tonic-gate } 831*0Sstevel@tonic-gate } 832*0Sstevel@tonic-gate 833*0Sstevel@tonic-gate ipif = ire->ire_ipif; 834*0Sstevel@tonic-gate if (ipif != NULL) { 835*0Sstevel@tonic-gate mblk_t **mpp; 836*0Sstevel@tonic-gate mblk_t *mp; 837*0Sstevel@tonic-gate ifrt_t *ifrt; 838*0Sstevel@tonic-gate in6_addr_t gw_addr_v6; 839*0Sstevel@tonic-gate 840*0Sstevel@tonic-gate /* Remove from ipif_saved_ire_mp list if it is there */ 841*0Sstevel@tonic-gate mutex_enter(&ire->ire_lock); 842*0Sstevel@tonic-gate gw_addr_v6 = ire->ire_gateway_addr_v6; 843*0Sstevel@tonic-gate mutex_exit(&ire->ire_lock); 844*0Sstevel@tonic-gate mutex_enter(&ipif->ipif_saved_ire_lock); 845*0Sstevel@tonic-gate for (mpp = &ipif->ipif_saved_ire_mp; *mpp != NULL; 846*0Sstevel@tonic-gate mpp = &(*mpp)->b_cont) { 847*0Sstevel@tonic-gate /* 848*0Sstevel@tonic-gate * On a given ipif, the triple of address, gateway and 849*0Sstevel@tonic-gate * mask is unique for each saved IRE (in the case of 850*0Sstevel@tonic-gate * ordinary interface routes, the gateway address is 851*0Sstevel@tonic-gate * all-zeroes). 852*0Sstevel@tonic-gate */ 853*0Sstevel@tonic-gate mp = *mpp; 854*0Sstevel@tonic-gate ifrt = (ifrt_t *)mp->b_rptr; 855*0Sstevel@tonic-gate if (IN6_ARE_ADDR_EQUAL(&ifrt->ifrt_v6addr, 856*0Sstevel@tonic-gate &ire->ire_addr_v6) && 857*0Sstevel@tonic-gate IN6_ARE_ADDR_EQUAL(&ifrt->ifrt_v6gateway_addr, 858*0Sstevel@tonic-gate &gw_addr_v6) && 859*0Sstevel@tonic-gate IN6_ARE_ADDR_EQUAL(&ifrt->ifrt_v6mask, 860*0Sstevel@tonic-gate &ire->ire_mask_v6)) { 861*0Sstevel@tonic-gate *mpp = mp->b_cont; 862*0Sstevel@tonic-gate ipif->ipif_saved_ire_cnt--; 863*0Sstevel@tonic-gate freeb(mp); 864*0Sstevel@tonic-gate break; 865*0Sstevel@tonic-gate } 866*0Sstevel@tonic-gate } 867*0Sstevel@tonic-gate mutex_exit(&ipif->ipif_saved_ire_lock); 868*0Sstevel@tonic-gate } 869*0Sstevel@tonic-gate ire_delete(ire); 870*0Sstevel@tonic-gate ire_refrele(ire); 871*0Sstevel@tonic-gate return (err); 872*0Sstevel@tonic-gate } 873*0Sstevel@tonic-gate 874*0Sstevel@tonic-gate /* 875*0Sstevel@tonic-gate * Derive a token from the link layer address. 876*0Sstevel@tonic-gate */ 877*0Sstevel@tonic-gate boolean_t 878*0Sstevel@tonic-gate ill_setdefaulttoken(ill_t *ill) 879*0Sstevel@tonic-gate { 880*0Sstevel@tonic-gate int i; 881*0Sstevel@tonic-gate in6_addr_t v6addr, v6mask; 882*0Sstevel@tonic-gate 883*0Sstevel@tonic-gate /* 884*0Sstevel@tonic-gate * Though we execute on the ipsq, we need to hold the ill_lock 885*0Sstevel@tonic-gate * to prevent readers from seeing partially updated values 886*0Sstevel@tonic-gate * while we do the update. 887*0Sstevel@tonic-gate */ 888*0Sstevel@tonic-gate mutex_enter(&ill->ill_lock); 889*0Sstevel@tonic-gate if (!MEDIA_V6INTFID(ill->ill_media, ill->ill_phys_addr_length, 890*0Sstevel@tonic-gate ill->ill_phys_addr, &v6addr)) { 891*0Sstevel@tonic-gate mutex_exit(&ill->ill_lock); 892*0Sstevel@tonic-gate return (B_FALSE); 893*0Sstevel@tonic-gate } 894*0Sstevel@tonic-gate 895*0Sstevel@tonic-gate (void) ip_plen_to_mask_v6(IPV6_TOKEN_LEN, &v6mask); 896*0Sstevel@tonic-gate 897*0Sstevel@tonic-gate for (i = 0; i < 4; i++) 898*0Sstevel@tonic-gate v6mask.s6_addr32[i] = v6mask.s6_addr32[i] ^ 899*0Sstevel@tonic-gate (uint32_t)0xffffffff; 900*0Sstevel@tonic-gate 901*0Sstevel@tonic-gate V6_MASK_COPY(v6addr, v6mask, ill->ill_token); 902*0Sstevel@tonic-gate ill->ill_token_length = IPV6_TOKEN_LEN; 903*0Sstevel@tonic-gate mutex_exit(&ill->ill_lock); 904*0Sstevel@tonic-gate return (B_TRUE); 905*0Sstevel@tonic-gate } 906*0Sstevel@tonic-gate 907*0Sstevel@tonic-gate /* 908*0Sstevel@tonic-gate * Create a link-local address from a token. 909*0Sstevel@tonic-gate */ 910*0Sstevel@tonic-gate static void 911*0Sstevel@tonic-gate ipif_get_linklocal(in6_addr_t *dest, const in6_addr_t *token) 912*0Sstevel@tonic-gate { 913*0Sstevel@tonic-gate int i; 914*0Sstevel@tonic-gate 915*0Sstevel@tonic-gate for (i = 0; i < 4; i++) { 916*0Sstevel@tonic-gate dest->s6_addr32[i] = 917*0Sstevel@tonic-gate token->s6_addr32[i] | ipv6_ll_template.s6_addr32[i]; 918*0Sstevel@tonic-gate } 919*0Sstevel@tonic-gate } 920*0Sstevel@tonic-gate 921*0Sstevel@tonic-gate /* 922*0Sstevel@tonic-gate * Set a nice default address for either automatic tunnels tsrc/96 or 923*0Sstevel@tonic-gate * 6to4 tunnels 2002:<tsrc>::1/64 924*0Sstevel@tonic-gate */ 925*0Sstevel@tonic-gate static void 926*0Sstevel@tonic-gate ipif_set_tun_auto_addr(ipif_t *ipif, struct iftun_req *ta) 927*0Sstevel@tonic-gate { 928*0Sstevel@tonic-gate sin6_t sin6; 929*0Sstevel@tonic-gate sin_t *sin; 930*0Sstevel@tonic-gate ill_t *ill = ipif->ipif_ill; 931*0Sstevel@tonic-gate tun_t *tp = (tun_t *)ill->ill_wq->q_next->q_ptr; 932*0Sstevel@tonic-gate 933*0Sstevel@tonic-gate if (ta->ifta_saddr.ss_family != AF_INET || 934*0Sstevel@tonic-gate (ipif->ipif_flags & IPIF_UP) || !ipif->ipif_isv6 || 935*0Sstevel@tonic-gate (ta->ifta_flags & IFTUN_SRC) == 0) 936*0Sstevel@tonic-gate return; 937*0Sstevel@tonic-gate 938*0Sstevel@tonic-gate /* 939*0Sstevel@tonic-gate * Check the tunnel type by examining q_next->q_ptr 940*0Sstevel@tonic-gate */ 941*0Sstevel@tonic-gate if (tp->tun_flags & TUN_AUTOMATIC) { 942*0Sstevel@tonic-gate /* this is an automatic tunnel */ 943*0Sstevel@tonic-gate (void) ip_plen_to_mask_v6(IPV6_ABITS - IP_ABITS, 944*0Sstevel@tonic-gate &ipif->ipif_v6net_mask); 945*0Sstevel@tonic-gate bzero(&sin6, sizeof (sin6_t)); 946*0Sstevel@tonic-gate sin = (sin_t *)&ta->ifta_saddr; 947*0Sstevel@tonic-gate V4_PART_OF_V6(sin6.sin6_addr) = sin->sin_addr.s_addr; 948*0Sstevel@tonic-gate sin6.sin6_family = AF_INET6; 949*0Sstevel@tonic-gate (void) ip_sioctl_addr(ipif, (sin_t *)&sin6, 950*0Sstevel@tonic-gate NULL, NULL, NULL, NULL); 951*0Sstevel@tonic-gate } else if (tp->tun_flags & TUN_6TO4) { 952*0Sstevel@tonic-gate /* this is a 6to4 tunnel */ 953*0Sstevel@tonic-gate (void) ip_plen_to_mask_v6(IPV6_PREFIX_LEN, 954*0Sstevel@tonic-gate &ipif->ipif_v6net_mask); 955*0Sstevel@tonic-gate sin = (sin_t *)&ta->ifta_saddr; 956*0Sstevel@tonic-gate /* create a 6to4 address from the IPv4 tsrc */ 957*0Sstevel@tonic-gate IN6_V4ADDR_TO_6TO4(&sin->sin_addr, &sin6.sin6_addr); 958*0Sstevel@tonic-gate sin6.sin6_family = AF_INET6; 959*0Sstevel@tonic-gate (void) ip_sioctl_addr(ipif, (sin_t *)&sin6, 960*0Sstevel@tonic-gate NULL, NULL, NULL, NULL); 961*0Sstevel@tonic-gate } else { 962*0Sstevel@tonic-gate ip1dbg(("ipif_set_tun_auto_addr: Unknown tunnel type")); 963*0Sstevel@tonic-gate return; 964*0Sstevel@tonic-gate } 965*0Sstevel@tonic-gate } 966*0Sstevel@tonic-gate 967*0Sstevel@tonic-gate /* 968*0Sstevel@tonic-gate * Set link local for ipif_id 0 of a configured tunnel based on the 969*0Sstevel@tonic-gate * tsrc or tdst parameter 970*0Sstevel@tonic-gate * For tunnels over IPv4 use the IPv4 address prepended with 32 zeros as 971*0Sstevel@tonic-gate * the token. 972*0Sstevel@tonic-gate * For tunnels over IPv6 use the low-order 64 bits of the "inner" IPv6 address 973*0Sstevel@tonic-gate * as the token for the "outer" link. 974*0Sstevel@tonic-gate */ 975*0Sstevel@tonic-gate void 976*0Sstevel@tonic-gate ipif_set_tun_llink(ill_t *ill, struct iftun_req *ta) 977*0Sstevel@tonic-gate { 978*0Sstevel@tonic-gate ipif_t *ipif; 979*0Sstevel@tonic-gate sin_t *sin; 980*0Sstevel@tonic-gate in6_addr_t *s6addr; 981*0Sstevel@tonic-gate 982*0Sstevel@tonic-gate ASSERT(IAM_WRITER_ILL(ill)); 983*0Sstevel@tonic-gate 984*0Sstevel@tonic-gate /* The first ipif must be id zero. */ 985*0Sstevel@tonic-gate ipif = ill->ill_ipif; 986*0Sstevel@tonic-gate ASSERT(ipif->ipif_id == 0); 987*0Sstevel@tonic-gate 988*0Sstevel@tonic-gate /* no link local for automatic tunnels */ 989*0Sstevel@tonic-gate if (!(ipif->ipif_flags & IPIF_POINTOPOINT)) { 990*0Sstevel@tonic-gate ipif_set_tun_auto_addr(ipif, ta); 991*0Sstevel@tonic-gate return; 992*0Sstevel@tonic-gate } 993*0Sstevel@tonic-gate 994*0Sstevel@tonic-gate if ((ta->ifta_flags & IFTUN_DST) && 995*0Sstevel@tonic-gate IN6_IS_ADDR_UNSPECIFIED(&ipif->ipif_v6pp_dst_addr)) { 996*0Sstevel@tonic-gate sin6_t sin6; 997*0Sstevel@tonic-gate 998*0Sstevel@tonic-gate ASSERT(!(ipif->ipif_flags & IPIF_UP)); 999*0Sstevel@tonic-gate bzero(&sin6, sizeof (sin6_t)); 1000*0Sstevel@tonic-gate if ((ta->ifta_saddr.ss_family == AF_INET)) { 1001*0Sstevel@tonic-gate sin = (sin_t *)&ta->ifta_daddr; 1002*0Sstevel@tonic-gate V4_PART_OF_V6(sin6.sin6_addr) = 1003*0Sstevel@tonic-gate sin->sin_addr.s_addr; 1004*0Sstevel@tonic-gate } else { 1005*0Sstevel@tonic-gate s6addr = 1006*0Sstevel@tonic-gate &((sin6_t *)&ta->ifta_daddr)->sin6_addr; 1007*0Sstevel@tonic-gate sin6.sin6_addr.s6_addr32[3] = s6addr->s6_addr32[3]; 1008*0Sstevel@tonic-gate sin6.sin6_addr.s6_addr32[2] = s6addr->s6_addr32[2]; 1009*0Sstevel@tonic-gate } 1010*0Sstevel@tonic-gate ipif_get_linklocal(&ipif->ipif_v6pp_dst_addr, 1011*0Sstevel@tonic-gate &sin6.sin6_addr); 1012*0Sstevel@tonic-gate ipif->ipif_v6subnet = ipif->ipif_v6pp_dst_addr; 1013*0Sstevel@tonic-gate } 1014*0Sstevel@tonic-gate if ((ta->ifta_flags & IFTUN_SRC)) { 1015*0Sstevel@tonic-gate ASSERT(!(ipif->ipif_flags & IPIF_UP)); 1016*0Sstevel@tonic-gate 1017*0Sstevel@tonic-gate /* Set the token if it isn't already set */ 1018*0Sstevel@tonic-gate if (IN6_IS_ADDR_UNSPECIFIED(&ill->ill_token)) { 1019*0Sstevel@tonic-gate if ((ta->ifta_saddr.ss_family == AF_INET)) { 1020*0Sstevel@tonic-gate sin = (sin_t *)&ta->ifta_saddr; 1021*0Sstevel@tonic-gate V4_PART_OF_V6(ill->ill_token) = 1022*0Sstevel@tonic-gate sin->sin_addr.s_addr; 1023*0Sstevel@tonic-gate } else { 1024*0Sstevel@tonic-gate s6addr = 1025*0Sstevel@tonic-gate &((sin6_t *)&ta->ifta_saddr)->sin6_addr; 1026*0Sstevel@tonic-gate ill->ill_token.s6_addr32[3] = 1027*0Sstevel@tonic-gate s6addr->s6_addr32[3]; 1028*0Sstevel@tonic-gate ill->ill_token.s6_addr32[2] = 1029*0Sstevel@tonic-gate s6addr->s6_addr32[2]; 1030*0Sstevel@tonic-gate } 1031*0Sstevel@tonic-gate ill->ill_token_length = IPV6_TOKEN_LEN; 1032*0Sstevel@tonic-gate } 1033*0Sstevel@tonic-gate /* 1034*0Sstevel@tonic-gate * Attempt to set the link local address if it isn't set. 1035*0Sstevel@tonic-gate */ 1036*0Sstevel@tonic-gate if (IN6_IS_ADDR_UNSPECIFIED(&ipif->ipif_v6lcl_addr)) 1037*0Sstevel@tonic-gate (void) ipif_setlinklocal(ipif); 1038*0Sstevel@tonic-gate } 1039*0Sstevel@tonic-gate } 1040*0Sstevel@tonic-gate 1041*0Sstevel@tonic-gate /* 1042*0Sstevel@tonic-gate * Is it not possible to set the link local address? 1043*0Sstevel@tonic-gate * The address can be set if the token is set, and the token 1044*0Sstevel@tonic-gate * isn't too long. 1045*0Sstevel@tonic-gate * Return B_TRUE if the address can't be set, or B_FALSE if it can. 1046*0Sstevel@tonic-gate */ 1047*0Sstevel@tonic-gate boolean_t 1048*0Sstevel@tonic-gate ipif_cant_setlinklocal(ipif_t *ipif) 1049*0Sstevel@tonic-gate { 1050*0Sstevel@tonic-gate ill_t *ill = ipif->ipif_ill; 1051*0Sstevel@tonic-gate 1052*0Sstevel@tonic-gate if (IN6_IS_ADDR_UNSPECIFIED(&ill->ill_token) || 1053*0Sstevel@tonic-gate ill->ill_token_length > IPV6_ABITS - IPV6_LL_PREFIXLEN) 1054*0Sstevel@tonic-gate return (B_TRUE); 1055*0Sstevel@tonic-gate 1056*0Sstevel@tonic-gate return (B_FALSE); 1057*0Sstevel@tonic-gate } 1058*0Sstevel@tonic-gate 1059*0Sstevel@tonic-gate /* 1060*0Sstevel@tonic-gate * Generate a link-local address from the token. 1061*0Sstevel@tonic-gate * Return zero if the address was set, or non-zero if it couldn't be set. 1062*0Sstevel@tonic-gate */ 1063*0Sstevel@tonic-gate int 1064*0Sstevel@tonic-gate ipif_setlinklocal(ipif_t *ipif) 1065*0Sstevel@tonic-gate { 1066*0Sstevel@tonic-gate ill_t *ill = ipif->ipif_ill; 1067*0Sstevel@tonic-gate 1068*0Sstevel@tonic-gate ASSERT(IAM_WRITER_ILL(ill)); 1069*0Sstevel@tonic-gate 1070*0Sstevel@tonic-gate if (ipif_cant_setlinklocal(ipif)) 1071*0Sstevel@tonic-gate return (-1); 1072*0Sstevel@tonic-gate 1073*0Sstevel@tonic-gate ipif_get_linklocal(&ipif->ipif_v6lcl_addr, &ill->ill_token); 1074*0Sstevel@tonic-gate (void) ip_plen_to_mask_v6(IPV6_LL_PREFIXLEN, &ipif->ipif_v6net_mask); 1075*0Sstevel@tonic-gate V6_MASK_COPY(ipif->ipif_v6lcl_addr, ipif->ipif_v6net_mask, 1076*0Sstevel@tonic-gate ipif->ipif_v6subnet); 1077*0Sstevel@tonic-gate 1078*0Sstevel@tonic-gate if (ipif->ipif_flags & IPIF_NOLOCAL) { 1079*0Sstevel@tonic-gate ipif->ipif_v6src_addr = ipv6_all_zeros; 1080*0Sstevel@tonic-gate } else { 1081*0Sstevel@tonic-gate ipif->ipif_v6src_addr = ipif->ipif_v6lcl_addr; 1082*0Sstevel@tonic-gate } 1083*0Sstevel@tonic-gate return (0); 1084*0Sstevel@tonic-gate } 1085*0Sstevel@tonic-gate 1086*0Sstevel@tonic-gate /* 1087*0Sstevel@tonic-gate * This function sets up the multicast mappings in NDP. 1088*0Sstevel@tonic-gate * Unlike ARP, there are no mapping_mps here. We delete the 1089*0Sstevel@tonic-gate * mapping nces and add a new one. 1090*0Sstevel@tonic-gate * 1091*0Sstevel@tonic-gate * Returns non-zero on error and 0 on success. 1092*0Sstevel@tonic-gate */ 1093*0Sstevel@tonic-gate int 1094*0Sstevel@tonic-gate ipif_ndp_setup_multicast(ipif_t *ipif, nce_t **ret_nce) 1095*0Sstevel@tonic-gate { 1096*0Sstevel@tonic-gate ill_t *ill = ipif->ipif_ill; 1097*0Sstevel@tonic-gate in6_addr_t v6_mcast_addr = {(uint32_t)V6_MCAST, 0, 0, 0}; 1098*0Sstevel@tonic-gate in6_addr_t v6_mcast_mask = {(uint32_t)V6_MCAST, 0, 0, 0}; 1099*0Sstevel@tonic-gate in6_addr_t v6_extract_mask; 1100*0Sstevel@tonic-gate uchar_t *phys_addr, *bphys_addr, *alloc_phys; 1101*0Sstevel@tonic-gate nce_t *mnce = NULL; 1102*0Sstevel@tonic-gate int err = 0; 1103*0Sstevel@tonic-gate phyint_t *phyi = ill->ill_phyint; 1104*0Sstevel@tonic-gate uint32_t hw_extract_start; 1105*0Sstevel@tonic-gate dl_unitdata_req_t *dlur; 1106*0Sstevel@tonic-gate 1107*0Sstevel@tonic-gate if (ret_nce != NULL) 1108*0Sstevel@tonic-gate *ret_nce = NULL; 1109*0Sstevel@tonic-gate /* 1110*0Sstevel@tonic-gate * Delete the mapping nce. Normally these should not exist 1111*0Sstevel@tonic-gate * as a previous ipif_down -> ipif_ndp_down should have deleted 1112*0Sstevel@tonic-gate * all the nces. But they can exist if ip_rput_dlpi_writer 1113*0Sstevel@tonic-gate * calls this when PHYI_MULTI_BCAST is set. 1114*0Sstevel@tonic-gate */ 1115*0Sstevel@tonic-gate mnce = ndp_lookup(ill, &v6_mcast_addr, B_FALSE); 1116*0Sstevel@tonic-gate if (mnce != NULL) { 1117*0Sstevel@tonic-gate ndp_delete(mnce); 1118*0Sstevel@tonic-gate NCE_REFRELE(mnce); 1119*0Sstevel@tonic-gate mnce = NULL; 1120*0Sstevel@tonic-gate } 1121*0Sstevel@tonic-gate 1122*0Sstevel@tonic-gate /* 1123*0Sstevel@tonic-gate * Get media specific v6 mapping information. Note that 1124*0Sstevel@tonic-gate * nd_lla_len can be 0 for tunnels. 1125*0Sstevel@tonic-gate */ 1126*0Sstevel@tonic-gate alloc_phys = kmem_alloc(ill->ill_nd_lla_len, KM_NOSLEEP); 1127*0Sstevel@tonic-gate if ((alloc_phys == NULL) && (ill->ill_nd_lla_len != 0)) 1128*0Sstevel@tonic-gate return (ENOMEM); 1129*0Sstevel@tonic-gate /* 1130*0Sstevel@tonic-gate * Determine the broadcast address. 1131*0Sstevel@tonic-gate */ 1132*0Sstevel@tonic-gate dlur = (dl_unitdata_req_t *)ill->ill_bcast_mp->b_rptr; 1133*0Sstevel@tonic-gate if (ill->ill_sap_length < 0) 1134*0Sstevel@tonic-gate bphys_addr = (uchar_t *)dlur + dlur->dl_dest_addr_offset; 1135*0Sstevel@tonic-gate else 1136*0Sstevel@tonic-gate bphys_addr = (uchar_t *)dlur + 1137*0Sstevel@tonic-gate dlur->dl_dest_addr_offset + ill->ill_sap_length; 1138*0Sstevel@tonic-gate 1139*0Sstevel@tonic-gate /* 1140*0Sstevel@tonic-gate * Check PHYI_MULTI_BCAST and possible length of physical 1141*0Sstevel@tonic-gate * address to determine if we use the mapping or the 1142*0Sstevel@tonic-gate * broadcast address. 1143*0Sstevel@tonic-gate */ 1144*0Sstevel@tonic-gate if ((phyi->phyint_flags & PHYI_MULTI_BCAST) || 1145*0Sstevel@tonic-gate (!MEDIA_V6MINFO(ill->ill_media, ill->ill_nd_lla_len, 1146*0Sstevel@tonic-gate bphys_addr, alloc_phys, &hw_extract_start, 1147*0Sstevel@tonic-gate &v6_extract_mask))) { 1148*0Sstevel@tonic-gate if (ill->ill_phys_addr_length > IP_MAX_HW_LEN) { 1149*0Sstevel@tonic-gate kmem_free(alloc_phys, ill->ill_nd_lla_len); 1150*0Sstevel@tonic-gate return (E2BIG); 1151*0Sstevel@tonic-gate } 1152*0Sstevel@tonic-gate /* Use the link-layer broadcast address for MULTI_BCAST */ 1153*0Sstevel@tonic-gate phys_addr = bphys_addr; 1154*0Sstevel@tonic-gate bzero(&v6_extract_mask, sizeof (v6_extract_mask)); 1155*0Sstevel@tonic-gate hw_extract_start = ill->ill_nd_lla_len; 1156*0Sstevel@tonic-gate } else { 1157*0Sstevel@tonic-gate phys_addr = alloc_phys; 1158*0Sstevel@tonic-gate } 1159*0Sstevel@tonic-gate if ((ipif->ipif_flags & IPIF_BROADCAST) || 1160*0Sstevel@tonic-gate (ill->ill_flags & ILLF_MULTICAST) || 1161*0Sstevel@tonic-gate (phyi->phyint_flags & PHYI_MULTI_BCAST)) { 1162*0Sstevel@tonic-gate mutex_enter(&ndp_g_lock); 1163*0Sstevel@tonic-gate err = ndp_add(ill, 1164*0Sstevel@tonic-gate phys_addr, 1165*0Sstevel@tonic-gate &v6_mcast_addr, /* v6 address */ 1166*0Sstevel@tonic-gate &v6_mcast_mask, /* v6 mask */ 1167*0Sstevel@tonic-gate &v6_extract_mask, 1168*0Sstevel@tonic-gate hw_extract_start, 1169*0Sstevel@tonic-gate NCE_F_MAPPING | NCE_F_PERMANENT | NCE_F_NONUD, 1170*0Sstevel@tonic-gate ND_REACHABLE, 1171*0Sstevel@tonic-gate &mnce); 1172*0Sstevel@tonic-gate mutex_exit(&ndp_g_lock); 1173*0Sstevel@tonic-gate if (err == 0) { 1174*0Sstevel@tonic-gate if (ret_nce != NULL) { 1175*0Sstevel@tonic-gate *ret_nce = mnce; 1176*0Sstevel@tonic-gate } else { 1177*0Sstevel@tonic-gate NCE_REFRELE(mnce); 1178*0Sstevel@tonic-gate } 1179*0Sstevel@tonic-gate } 1180*0Sstevel@tonic-gate } 1181*0Sstevel@tonic-gate kmem_free(alloc_phys, ill->ill_nd_lla_len); 1182*0Sstevel@tonic-gate return (err); 1183*0Sstevel@tonic-gate } 1184*0Sstevel@tonic-gate 1185*0Sstevel@tonic-gate /* 1186*0Sstevel@tonic-gate * Get the resolver set up for a new interface address. (Always called 1187*0Sstevel@tonic-gate * as writer.) 1188*0Sstevel@tonic-gate */ 1189*0Sstevel@tonic-gate int 1190*0Sstevel@tonic-gate ipif_ndp_up(ipif_t *ipif, const in6_addr_t *addr, boolean_t macaddr_change) 1191*0Sstevel@tonic-gate { 1192*0Sstevel@tonic-gate ill_t *ill = ipif->ipif_ill; 1193*0Sstevel@tonic-gate int err = 0; 1194*0Sstevel@tonic-gate nce_t *nce = NULL; 1195*0Sstevel@tonic-gate nce_t *mnce = NULL; 1196*0Sstevel@tonic-gate 1197*0Sstevel@tonic-gate ip1dbg(("ipif_ndp_up(%s:%u)\n", 1198*0Sstevel@tonic-gate ipif->ipif_ill->ill_name, ipif->ipif_id)); 1199*0Sstevel@tonic-gate 1200*0Sstevel@tonic-gate /* 1201*0Sstevel@tonic-gate * ND not supported on XRESOLV interfaces. If ND support (multicast) 1202*0Sstevel@tonic-gate * added later, take out this check. 1203*0Sstevel@tonic-gate */ 1204*0Sstevel@tonic-gate if (ill->ill_flags & ILLF_XRESOLV) 1205*0Sstevel@tonic-gate return (0); 1206*0Sstevel@tonic-gate 1207*0Sstevel@tonic-gate if (IN6_IS_ADDR_UNSPECIFIED(addr) || 1208*0Sstevel@tonic-gate (!(ill->ill_net_type & IRE_INTERFACE))) 1209*0Sstevel@tonic-gate return (0); 1210*0Sstevel@tonic-gate 1211*0Sstevel@tonic-gate /* 1212*0Sstevel@tonic-gate * Need to setup multicast mapping only when the first 1213*0Sstevel@tonic-gate * interface is coming UP. 1214*0Sstevel@tonic-gate */ 1215*0Sstevel@tonic-gate if (ill->ill_ipif_up_count == 0 && 1216*0Sstevel@tonic-gate (ill->ill_flags & ILLF_MULTICAST)) { 1217*0Sstevel@tonic-gate /* 1218*0Sstevel@tonic-gate * We set the multicast before setting up the mapping for 1219*0Sstevel@tonic-gate * local address because ipif_ndp_setup_multicast does 1220*0Sstevel@tonic-gate * ndp_walk to delete nces which will delete the mapping 1221*0Sstevel@tonic-gate * for local address also if we added the mapping for 1222*0Sstevel@tonic-gate * local address first. 1223*0Sstevel@tonic-gate */ 1224*0Sstevel@tonic-gate err = ipif_ndp_setup_multicast(ipif, &mnce); 1225*0Sstevel@tonic-gate if (err != 0) 1226*0Sstevel@tonic-gate return (err); 1227*0Sstevel@tonic-gate } 1228*0Sstevel@tonic-gate 1229*0Sstevel@tonic-gate if ((ipif->ipif_flags & (IPIF_UNNUMBERED|IPIF_NOLOCAL)) == 0) { 1230*0Sstevel@tonic-gate uint16_t flags; 1231*0Sstevel@tonic-gate uchar_t *hw_addr = NULL; 1232*0Sstevel@tonic-gate 1233*0Sstevel@tonic-gate /* Permanent entries don't need NUD */ 1234*0Sstevel@tonic-gate flags = NCE_F_PERMANENT; 1235*0Sstevel@tonic-gate flags |= NCE_F_NONUD; 1236*0Sstevel@tonic-gate if (ill->ill_flags & ILLF_ROUTER) 1237*0Sstevel@tonic-gate flags |= NCE_F_ISROUTER; 1238*0Sstevel@tonic-gate 1239*0Sstevel@tonic-gate if (ipif->ipif_flags & IPIF_ANYCAST) 1240*0Sstevel@tonic-gate flags |= NCE_F_ANYCAST; 1241*0Sstevel@tonic-gate 1242*0Sstevel@tonic-gate if (ill->ill_net_type == IRE_IF_RESOLVER) { 1243*0Sstevel@tonic-gate hw_addr = ill->ill_nd_lla; 1244*0Sstevel@tonic-gate 1245*0Sstevel@tonic-gate if (ill->ill_move_in_progress || macaddr_change) { 1246*0Sstevel@tonic-gate /* 1247*0Sstevel@tonic-gate * Addresses are failing over to this ill. 1248*0Sstevel@tonic-gate * Don't wait for NUD to see this change. 1249*0Sstevel@tonic-gate * Publish our new link-layer address. 1250*0Sstevel@tonic-gate */ 1251*0Sstevel@tonic-gate flags |= NCE_F_UNSOL_ADV; 1252*0Sstevel@tonic-gate } 1253*0Sstevel@tonic-gate } 1254*0Sstevel@tonic-gate err = ndp_lookup_then_add(ill, 1255*0Sstevel@tonic-gate hw_addr, 1256*0Sstevel@tonic-gate addr, 1257*0Sstevel@tonic-gate &ipv6_all_ones, 1258*0Sstevel@tonic-gate &ipv6_all_zeros, 1259*0Sstevel@tonic-gate 0, 1260*0Sstevel@tonic-gate flags, 1261*0Sstevel@tonic-gate ND_REACHABLE, 1262*0Sstevel@tonic-gate &nce); 1263*0Sstevel@tonic-gate switch (err) { 1264*0Sstevel@tonic-gate case 0: 1265*0Sstevel@tonic-gate ip1dbg(("ipif_ndp_up: NCE created for %s\n", 1266*0Sstevel@tonic-gate ill->ill_name)); 1267*0Sstevel@tonic-gate break; 1268*0Sstevel@tonic-gate case EEXIST: 1269*0Sstevel@tonic-gate NCE_REFRELE(nce); 1270*0Sstevel@tonic-gate ip1dbg(("ipif_ndp_up: NCE already exists for %s\n", 1271*0Sstevel@tonic-gate ill->ill_name)); 1272*0Sstevel@tonic-gate if (mnce != NULL) { 1273*0Sstevel@tonic-gate ndp_delete(mnce); 1274*0Sstevel@tonic-gate NCE_REFRELE(mnce); 1275*0Sstevel@tonic-gate } 1276*0Sstevel@tonic-gate return (err); 1277*0Sstevel@tonic-gate default: 1278*0Sstevel@tonic-gate ip1dbg(("ipif_ndp_up: NCE creation failed %s\n", 1279*0Sstevel@tonic-gate ill->ill_name)); 1280*0Sstevel@tonic-gate if (mnce != NULL) { 1281*0Sstevel@tonic-gate ndp_delete(mnce); 1282*0Sstevel@tonic-gate NCE_REFRELE(mnce); 1283*0Sstevel@tonic-gate } 1284*0Sstevel@tonic-gate return (err); 1285*0Sstevel@tonic-gate } 1286*0Sstevel@tonic-gate } 1287*0Sstevel@tonic-gate if (nce != NULL) 1288*0Sstevel@tonic-gate NCE_REFRELE(nce); 1289*0Sstevel@tonic-gate if (mnce != NULL) 1290*0Sstevel@tonic-gate NCE_REFRELE(mnce); 1291*0Sstevel@tonic-gate return (0); 1292*0Sstevel@tonic-gate } 1293*0Sstevel@tonic-gate 1294*0Sstevel@tonic-gate /* Remove all cache entries for this logical interface */ 1295*0Sstevel@tonic-gate void 1296*0Sstevel@tonic-gate ipif_ndp_down(ipif_t *ipif) 1297*0Sstevel@tonic-gate { 1298*0Sstevel@tonic-gate nce_t *nce; 1299*0Sstevel@tonic-gate 1300*0Sstevel@tonic-gate nce = ndp_lookup(ipif->ipif_ill, &ipif->ipif_v6lcl_addr, B_FALSE); 1301*0Sstevel@tonic-gate if (nce != NULL) { 1302*0Sstevel@tonic-gate ndp_delete(nce); 1303*0Sstevel@tonic-gate NCE_REFRELE(nce); 1304*0Sstevel@tonic-gate } 1305*0Sstevel@tonic-gate /* 1306*0Sstevel@tonic-gate * Remove mapping and all other nces dependent on this ill 1307*0Sstevel@tonic-gate * when the last ipif is going away. 1308*0Sstevel@tonic-gate */ 1309*0Sstevel@tonic-gate if (ipif->ipif_ill->ill_ipif_up_count == 0) { 1310*0Sstevel@tonic-gate ndp_walk(ipif->ipif_ill, (pfi_t)ndp_delete_per_ill, 1311*0Sstevel@tonic-gate (uchar_t *)ipif->ipif_ill); 1312*0Sstevel@tonic-gate } 1313*0Sstevel@tonic-gate } 1314*0Sstevel@tonic-gate 1315*0Sstevel@tonic-gate /* 1316*0Sstevel@tonic-gate * Used when an interface comes up to recreate any extra routes on this 1317*0Sstevel@tonic-gate * interface. 1318*0Sstevel@tonic-gate */ 1319*0Sstevel@tonic-gate static ire_t ** 1320*0Sstevel@tonic-gate ipif_recover_ire_v6(ipif_t *ipif) 1321*0Sstevel@tonic-gate { 1322*0Sstevel@tonic-gate mblk_t *mp; 1323*0Sstevel@tonic-gate ire_t **ipif_saved_irep; 1324*0Sstevel@tonic-gate ire_t **irep; 1325*0Sstevel@tonic-gate 1326*0Sstevel@tonic-gate ip1dbg(("ipif_recover_ire_v6(%s:%u)", ipif->ipif_ill->ill_name, 1327*0Sstevel@tonic-gate ipif->ipif_id)); 1328*0Sstevel@tonic-gate 1329*0Sstevel@tonic-gate ASSERT(ipif->ipif_isv6); 1330*0Sstevel@tonic-gate 1331*0Sstevel@tonic-gate mutex_enter(&ipif->ipif_saved_ire_lock); 1332*0Sstevel@tonic-gate ipif_saved_irep = (ire_t **)kmem_zalloc(sizeof (ire_t *) * 1333*0Sstevel@tonic-gate ipif->ipif_saved_ire_cnt, KM_NOSLEEP); 1334*0Sstevel@tonic-gate if (ipif_saved_irep == NULL) { 1335*0Sstevel@tonic-gate mutex_exit(&ipif->ipif_saved_ire_lock); 1336*0Sstevel@tonic-gate return (NULL); 1337*0Sstevel@tonic-gate } 1338*0Sstevel@tonic-gate 1339*0Sstevel@tonic-gate irep = ipif_saved_irep; 1340*0Sstevel@tonic-gate 1341*0Sstevel@tonic-gate for (mp = ipif->ipif_saved_ire_mp; mp != NULL; mp = mp->b_cont) { 1342*0Sstevel@tonic-gate ire_t *ire; 1343*0Sstevel@tonic-gate queue_t *rfq; 1344*0Sstevel@tonic-gate queue_t *stq; 1345*0Sstevel@tonic-gate ifrt_t *ifrt; 1346*0Sstevel@tonic-gate in6_addr_t *src_addr; 1347*0Sstevel@tonic-gate in6_addr_t *gateway_addr; 1348*0Sstevel@tonic-gate mblk_t *resolver_mp; 1349*0Sstevel@tonic-gate char buf[INET6_ADDRSTRLEN]; 1350*0Sstevel@tonic-gate ushort_t type; 1351*0Sstevel@tonic-gate 1352*0Sstevel@tonic-gate /* 1353*0Sstevel@tonic-gate * When the ire was initially created and then added in 1354*0Sstevel@tonic-gate * ip_rt_add_v6(), it was created either using 1355*0Sstevel@tonic-gate * ipif->ipif_net_type in the case of a traditional interface 1356*0Sstevel@tonic-gate * route, or as one of the IRE_OFFSUBNET types (with the 1357*0Sstevel@tonic-gate * exception of IRE_HOST_REDIRECT which is created by 1358*0Sstevel@tonic-gate * icmp_redirect_v6() and which we don't need to save or 1359*0Sstevel@tonic-gate * recover). In the case where ipif->ipif_net_type was 1360*0Sstevel@tonic-gate * IRE_LOOPBACK, ip_rt_add_v6() will update the ire_type to 1361*0Sstevel@tonic-gate * IRE_IF_NORESOLVER before calling ire_add_v6() to satisfy 1362*0Sstevel@tonic-gate * software like GateD and Sun Cluster which creates routes 1363*0Sstevel@tonic-gate * using the the loopback interface's address as a gateway. 1364*0Sstevel@tonic-gate * 1365*0Sstevel@tonic-gate * As ifrt->ifrt_type reflects the already updated ire_type and 1366*0Sstevel@tonic-gate * since ire_create_v6() expects that IRE_IF_NORESOLVER will 1367*0Sstevel@tonic-gate * have a valid ire_dlureq_mp field (which doesn't make sense 1368*0Sstevel@tonic-gate * for a IRE_LOOPBACK), ire_create_v6() will be called in the 1369*0Sstevel@tonic-gate * same way here as in ip_rt_add_v6(), namely using 1370*0Sstevel@tonic-gate * ipif->ipif_net_type when the route looks like a traditional 1371*0Sstevel@tonic-gate * interface route (where ifrt->ifrt_type & IRE_INTERFACE is 1372*0Sstevel@tonic-gate * true) and otherwise using the saved ifrt->ifrt_type. This 1373*0Sstevel@tonic-gate * means that in the case where ipif->ipif_net_type is 1374*0Sstevel@tonic-gate * IRE_LOOPBACK, the ire created by ire_create_v6() will be an 1375*0Sstevel@tonic-gate * IRE_LOOPBACK, it will then be turned into an 1376*0Sstevel@tonic-gate * IRE_IF_NORESOLVER and then added by ire_add_v6(). 1377*0Sstevel@tonic-gate */ 1378*0Sstevel@tonic-gate ifrt = (ifrt_t *)mp->b_rptr; 1379*0Sstevel@tonic-gate if (ifrt->ifrt_type & IRE_INTERFACE) { 1380*0Sstevel@tonic-gate rfq = NULL; 1381*0Sstevel@tonic-gate stq = (ipif->ipif_net_type == IRE_IF_RESOLVER) 1382*0Sstevel@tonic-gate ? ipif->ipif_rq : ipif->ipif_wq; 1383*0Sstevel@tonic-gate src_addr = (ifrt->ifrt_flags & RTF_SETSRC) 1384*0Sstevel@tonic-gate ? &ifrt->ifrt_v6src_addr 1385*0Sstevel@tonic-gate : &ipif->ipif_v6src_addr; 1386*0Sstevel@tonic-gate gateway_addr = NULL; 1387*0Sstevel@tonic-gate resolver_mp = ipif->ipif_resolver_mp; 1388*0Sstevel@tonic-gate type = ipif->ipif_net_type; 1389*0Sstevel@tonic-gate } else { 1390*0Sstevel@tonic-gate rfq = NULL; 1391*0Sstevel@tonic-gate stq = NULL; 1392*0Sstevel@tonic-gate src_addr = (ifrt->ifrt_flags & RTF_SETSRC) 1393*0Sstevel@tonic-gate ? &ifrt->ifrt_v6src_addr : NULL; 1394*0Sstevel@tonic-gate gateway_addr = &ifrt->ifrt_v6gateway_addr; 1395*0Sstevel@tonic-gate resolver_mp = NULL; 1396*0Sstevel@tonic-gate type = ifrt->ifrt_type; 1397*0Sstevel@tonic-gate } 1398*0Sstevel@tonic-gate 1399*0Sstevel@tonic-gate /* 1400*0Sstevel@tonic-gate * Create a copy of the IRE with the saved address and netmask. 1401*0Sstevel@tonic-gate */ 1402*0Sstevel@tonic-gate ip1dbg(("ipif_recover_ire_v6: creating IRE %s (%d) for %s/%d\n", 1403*0Sstevel@tonic-gate ip_nv_lookup(ire_nv_tbl, ifrt->ifrt_type), ifrt->ifrt_type, 1404*0Sstevel@tonic-gate inet_ntop(AF_INET6, &ifrt->ifrt_v6addr, buf, sizeof (buf)), 1405*0Sstevel@tonic-gate ip_mask_to_plen_v6(&ifrt->ifrt_v6mask))); 1406*0Sstevel@tonic-gate ire = ire_create_v6( 1407*0Sstevel@tonic-gate &ifrt->ifrt_v6addr, 1408*0Sstevel@tonic-gate &ifrt->ifrt_v6mask, 1409*0Sstevel@tonic-gate src_addr, 1410*0Sstevel@tonic-gate gateway_addr, 1411*0Sstevel@tonic-gate &ifrt->ifrt_max_frag, 1412*0Sstevel@tonic-gate NULL, 1413*0Sstevel@tonic-gate rfq, 1414*0Sstevel@tonic-gate stq, 1415*0Sstevel@tonic-gate type, 1416*0Sstevel@tonic-gate resolver_mp, 1417*0Sstevel@tonic-gate ipif, 1418*0Sstevel@tonic-gate NULL, 1419*0Sstevel@tonic-gate 0, 1420*0Sstevel@tonic-gate 0, 1421*0Sstevel@tonic-gate ifrt->ifrt_flags, 1422*0Sstevel@tonic-gate &ifrt->ifrt_iulp_info); 1423*0Sstevel@tonic-gate if (ire == NULL) { 1424*0Sstevel@tonic-gate mutex_exit(&ipif->ipif_saved_ire_lock); 1425*0Sstevel@tonic-gate kmem_free(ipif_saved_irep, 1426*0Sstevel@tonic-gate ipif->ipif_saved_ire_cnt * sizeof (ire_t *)); 1427*0Sstevel@tonic-gate return (NULL); 1428*0Sstevel@tonic-gate } 1429*0Sstevel@tonic-gate 1430*0Sstevel@tonic-gate /* 1431*0Sstevel@tonic-gate * Some software (for example, GateD and Sun Cluster) attempts 1432*0Sstevel@tonic-gate * to create (what amount to) IRE_PREFIX routes with the 1433*0Sstevel@tonic-gate * loopback address as the gateway. This is primarily done to 1434*0Sstevel@tonic-gate * set up prefixes with the RTF_REJECT flag set (for example, 1435*0Sstevel@tonic-gate * when generating aggregate routes.) 1436*0Sstevel@tonic-gate * 1437*0Sstevel@tonic-gate * If the IRE type (as defined by ipif->ipif_net_type) is 1438*0Sstevel@tonic-gate * IRE_LOOPBACK, then we map the request into a 1439*0Sstevel@tonic-gate * IRE_IF_NORESOLVER. 1440*0Sstevel@tonic-gate */ 1441*0Sstevel@tonic-gate if (ipif->ipif_net_type == IRE_LOOPBACK) 1442*0Sstevel@tonic-gate ire->ire_type = IRE_IF_NORESOLVER; 1443*0Sstevel@tonic-gate /* 1444*0Sstevel@tonic-gate * ire held by ire_add, will be refreled' in ipif_up_done 1445*0Sstevel@tonic-gate * towards the end 1446*0Sstevel@tonic-gate */ 1447*0Sstevel@tonic-gate (void) ire_add(&ire, NULL, NULL, NULL); 1448*0Sstevel@tonic-gate *irep = ire; 1449*0Sstevel@tonic-gate irep++; 1450*0Sstevel@tonic-gate ip1dbg(("ipif_recover_ire_v6: added ire %p\n", (void *)ire)); 1451*0Sstevel@tonic-gate } 1452*0Sstevel@tonic-gate mutex_exit(&ipif->ipif_saved_ire_lock); 1453*0Sstevel@tonic-gate return (ipif_saved_irep); 1454*0Sstevel@tonic-gate } 1455*0Sstevel@tonic-gate 1456*0Sstevel@tonic-gate /* 1457*0Sstevel@tonic-gate * Return the scope of the given IPv6 address. If the address is an 1458*0Sstevel@tonic-gate * IPv4 mapped IPv6 address, return the scope of the corresponding 1459*0Sstevel@tonic-gate * IPv4 address. 1460*0Sstevel@tonic-gate */ 1461*0Sstevel@tonic-gate in6addr_scope_t 1462*0Sstevel@tonic-gate ip_addr_scope_v6(const in6_addr_t *addr) 1463*0Sstevel@tonic-gate { 1464*0Sstevel@tonic-gate static in6_addr_t ipv6loopback = IN6ADDR_LOOPBACK_INIT; 1465*0Sstevel@tonic-gate 1466*0Sstevel@tonic-gate if (IN6_IS_ADDR_V4MAPPED(addr)) { 1467*0Sstevel@tonic-gate in_addr_t v4addr_h = ntohl(V4_PART_OF_V6((*addr))); 1468*0Sstevel@tonic-gate if ((v4addr_h >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET || 1469*0Sstevel@tonic-gate (v4addr_h & IN_AUTOCONF_MASK) == IN_AUTOCONF_NET) 1470*0Sstevel@tonic-gate return (IP6_SCOPE_LINKLOCAL); 1471*0Sstevel@tonic-gate if ((v4addr_h & IN_PRIVATE8_MASK) == IN_PRIVATE8_NET || 1472*0Sstevel@tonic-gate (v4addr_h & IN_PRIVATE12_MASK) == IN_PRIVATE12_NET || 1473*0Sstevel@tonic-gate (v4addr_h & IN_PRIVATE16_MASK) == IN_PRIVATE16_NET) 1474*0Sstevel@tonic-gate return (IP6_SCOPE_SITELOCAL); 1475*0Sstevel@tonic-gate return (IP6_SCOPE_GLOBAL); 1476*0Sstevel@tonic-gate } 1477*0Sstevel@tonic-gate 1478*0Sstevel@tonic-gate if (IN6_IS_ADDR_MULTICAST(addr)) 1479*0Sstevel@tonic-gate return (IN6_ADDR_MC_SCOPE(addr)); 1480*0Sstevel@tonic-gate 1481*0Sstevel@tonic-gate /* link-local and loopback addresses are of link-local scope */ 1482*0Sstevel@tonic-gate if (IN6_IS_ADDR_LINKLOCAL(addr) || 1483*0Sstevel@tonic-gate IN6_ARE_ADDR_EQUAL(addr, &ipv6loopback)) 1484*0Sstevel@tonic-gate return (IP6_SCOPE_LINKLOCAL); 1485*0Sstevel@tonic-gate if (IN6_IS_ADDR_SITELOCAL(addr)) 1486*0Sstevel@tonic-gate return (IP6_SCOPE_SITELOCAL); 1487*0Sstevel@tonic-gate return (IP6_SCOPE_GLOBAL); 1488*0Sstevel@tonic-gate } 1489*0Sstevel@tonic-gate 1490*0Sstevel@tonic-gate 1491*0Sstevel@tonic-gate /* 1492*0Sstevel@tonic-gate * Calculates the xor of a1 and a2, and stores the result in res. 1493*0Sstevel@tonic-gate */ 1494*0Sstevel@tonic-gate static void 1495*0Sstevel@tonic-gate ip_addr_xor_v6(const in6_addr_t *a1, const in6_addr_t *a2, in6_addr_t *res) 1496*0Sstevel@tonic-gate { 1497*0Sstevel@tonic-gate int i; 1498*0Sstevel@tonic-gate 1499*0Sstevel@tonic-gate for (i = 0; i < 4; i++) 1500*0Sstevel@tonic-gate res->s6_addr32[i] = a1->s6_addr32[i] ^ a2->s6_addr32[i]; 1501*0Sstevel@tonic-gate } 1502*0Sstevel@tonic-gate 1503*0Sstevel@tonic-gate #define IPIF_VALID_IPV6_SOURCE(ipif) \ 1504*0Sstevel@tonic-gate (((ipif)->ipif_flags & IPIF_UP) && \ 1505*0Sstevel@tonic-gate !((ipif)->ipif_flags & (IPIF_NOLOCAL|IPIF_ANYCAST))) 1506*0Sstevel@tonic-gate 1507*0Sstevel@tonic-gate /* source address candidate */ 1508*0Sstevel@tonic-gate typedef struct candidate { 1509*0Sstevel@tonic-gate ipif_t *cand_ipif; 1510*0Sstevel@tonic-gate /* The properties of this candidate */ 1511*0Sstevel@tonic-gate boolean_t cand_isdst; 1512*0Sstevel@tonic-gate boolean_t cand_isdst_set; 1513*0Sstevel@tonic-gate in6addr_scope_t cand_scope; 1514*0Sstevel@tonic-gate boolean_t cand_scope_set; 1515*0Sstevel@tonic-gate boolean_t cand_isdeprecated; 1516*0Sstevel@tonic-gate boolean_t cand_isdeprecated_set; 1517*0Sstevel@tonic-gate boolean_t cand_ispreferred; 1518*0Sstevel@tonic-gate boolean_t cand_ispreferred_set; 1519*0Sstevel@tonic-gate boolean_t cand_matchedinterface; 1520*0Sstevel@tonic-gate boolean_t cand_matchedinterface_set; 1521*0Sstevel@tonic-gate boolean_t cand_matchedlabel; 1522*0Sstevel@tonic-gate boolean_t cand_matchedlabel_set; 1523*0Sstevel@tonic-gate boolean_t cand_istmp; 1524*0Sstevel@tonic-gate boolean_t cand_istmp_set; 1525*0Sstevel@tonic-gate in6_addr_t cand_xor; 1526*0Sstevel@tonic-gate boolean_t cand_xor_set; 1527*0Sstevel@tonic-gate } cand_t; 1528*0Sstevel@tonic-gate #define cand_srcaddr cand_ipif->ipif_v6lcl_addr 1529*0Sstevel@tonic-gate #define cand_flags cand_ipif->ipif_flags 1530*0Sstevel@tonic-gate #define cand_ill cand_ipif->ipif_ill 1531*0Sstevel@tonic-gate 1532*0Sstevel@tonic-gate /* information about the destination for source address selection */ 1533*0Sstevel@tonic-gate typedef struct dstinfo { 1534*0Sstevel@tonic-gate const in6_addr_t *dst_addr; 1535*0Sstevel@tonic-gate ill_t *dst_ill; 1536*0Sstevel@tonic-gate boolean_t dst_restrict_ill; 1537*0Sstevel@tonic-gate boolean_t dst_prefer_src_tmp; 1538*0Sstevel@tonic-gate in6addr_scope_t dst_scope; 1539*0Sstevel@tonic-gate char *dst_label; 1540*0Sstevel@tonic-gate } dstinfo_t; 1541*0Sstevel@tonic-gate 1542*0Sstevel@tonic-gate /* 1543*0Sstevel@tonic-gate * The following functions are rules used to select a source address in 1544*0Sstevel@tonic-gate * ipif_select_source_v6(). Each rule compares a current candidate (cc) 1545*0Sstevel@tonic-gate * against the best candidate (bc). Each rule has three possible outcomes; 1546*0Sstevel@tonic-gate * the candidate is preferred over the best candidate (CAND_PREFER), the 1547*0Sstevel@tonic-gate * candidate is not preferred over the best candidate (CAND_AVOID), or the 1548*0Sstevel@tonic-gate * candidate is of equal value as the best candidate (CAND_TIE). 1549*0Sstevel@tonic-gate * 1550*0Sstevel@tonic-gate * These rules are part of a greater "Default Address Selection for IPv6" 1551*0Sstevel@tonic-gate * sheme, which is standards based work coming out of the IETF ipv6 working 1552*0Sstevel@tonic-gate * group. The IETF document defines both IPv6 source address selection and 1553*0Sstevel@tonic-gate * destination address ordering. The rules defined here implement the IPv6 1554*0Sstevel@tonic-gate * source address selection. Destination address ordering is done by 1555*0Sstevel@tonic-gate * libnsl, and uses a similar set of rules to implement the sorting. 1556*0Sstevel@tonic-gate */ 1557*0Sstevel@tonic-gate typedef enum {CAND_AVOID, CAND_TIE, CAND_PREFER} rule_res_t; 1558*0Sstevel@tonic-gate typedef rule_res_t (*rulef_t)(cand_t *, cand_t *, const dstinfo_t *); 1559*0Sstevel@tonic-gate 1560*0Sstevel@tonic-gate /* Prefer an address if it is equal to the destination address. */ 1561*0Sstevel@tonic-gate static rule_res_t 1562*0Sstevel@tonic-gate rule_isdst(cand_t *bc, cand_t *cc, const dstinfo_t *dstinfo) 1563*0Sstevel@tonic-gate { 1564*0Sstevel@tonic-gate if (!bc->cand_isdst_set) { 1565*0Sstevel@tonic-gate bc->cand_isdst = 1566*0Sstevel@tonic-gate IN6_ARE_ADDR_EQUAL(&bc->cand_srcaddr, dstinfo->dst_addr); 1567*0Sstevel@tonic-gate bc->cand_isdst_set = B_TRUE; 1568*0Sstevel@tonic-gate } 1569*0Sstevel@tonic-gate 1570*0Sstevel@tonic-gate cc->cand_isdst = 1571*0Sstevel@tonic-gate IN6_ARE_ADDR_EQUAL(&cc->cand_srcaddr, dstinfo->dst_addr); 1572*0Sstevel@tonic-gate cc->cand_isdst_set = B_TRUE; 1573*0Sstevel@tonic-gate 1574*0Sstevel@tonic-gate if (cc->cand_isdst == bc->cand_isdst) 1575*0Sstevel@tonic-gate return (CAND_TIE); 1576*0Sstevel@tonic-gate else if (cc->cand_isdst) 1577*0Sstevel@tonic-gate return (CAND_PREFER); 1578*0Sstevel@tonic-gate else 1579*0Sstevel@tonic-gate return (CAND_AVOID); 1580*0Sstevel@tonic-gate } 1581*0Sstevel@tonic-gate 1582*0Sstevel@tonic-gate /* 1583*0Sstevel@tonic-gate * Prefer addresses that are of closest scope to the destination. Always 1584*0Sstevel@tonic-gate * prefer addresses that are of greater scope than the destination over 1585*0Sstevel@tonic-gate * those that are of lesser scope than the destination. 1586*0Sstevel@tonic-gate */ 1587*0Sstevel@tonic-gate static rule_res_t 1588*0Sstevel@tonic-gate rule_scope(cand_t *bc, cand_t *cc, const dstinfo_t *dstinfo) 1589*0Sstevel@tonic-gate { 1590*0Sstevel@tonic-gate if (!bc->cand_scope_set) { 1591*0Sstevel@tonic-gate bc->cand_scope = ip_addr_scope_v6(&bc->cand_srcaddr); 1592*0Sstevel@tonic-gate bc->cand_scope_set = B_TRUE; 1593*0Sstevel@tonic-gate } 1594*0Sstevel@tonic-gate 1595*0Sstevel@tonic-gate cc->cand_scope = ip_addr_scope_v6(&cc->cand_srcaddr); 1596*0Sstevel@tonic-gate cc->cand_scope_set = B_TRUE; 1597*0Sstevel@tonic-gate 1598*0Sstevel@tonic-gate if (cc->cand_scope < bc->cand_scope) { 1599*0Sstevel@tonic-gate if (cc->cand_scope < dstinfo->dst_scope) 1600*0Sstevel@tonic-gate return (CAND_AVOID); 1601*0Sstevel@tonic-gate else 1602*0Sstevel@tonic-gate return (CAND_PREFER); 1603*0Sstevel@tonic-gate } else if (bc->cand_scope < cc->cand_scope) { 1604*0Sstevel@tonic-gate if (bc->cand_scope < dstinfo->dst_scope) 1605*0Sstevel@tonic-gate return (CAND_PREFER); 1606*0Sstevel@tonic-gate else 1607*0Sstevel@tonic-gate return (CAND_AVOID); 1608*0Sstevel@tonic-gate } else { 1609*0Sstevel@tonic-gate return (CAND_TIE); 1610*0Sstevel@tonic-gate } 1611*0Sstevel@tonic-gate } 1612*0Sstevel@tonic-gate 1613*0Sstevel@tonic-gate /* 1614*0Sstevel@tonic-gate * Prefer non-deprecated source addresses. 1615*0Sstevel@tonic-gate */ 1616*0Sstevel@tonic-gate /* ARGSUSED2 */ 1617*0Sstevel@tonic-gate static rule_res_t 1618*0Sstevel@tonic-gate rule_deprecated(cand_t *bc, cand_t *cc, const dstinfo_t *dstinfo) 1619*0Sstevel@tonic-gate { 1620*0Sstevel@tonic-gate if (!bc->cand_isdeprecated_set) { 1621*0Sstevel@tonic-gate bc->cand_isdeprecated = 1622*0Sstevel@tonic-gate ((bc->cand_flags & IPIF_DEPRECATED) != 0); 1623*0Sstevel@tonic-gate bc->cand_isdeprecated_set = B_TRUE; 1624*0Sstevel@tonic-gate } 1625*0Sstevel@tonic-gate 1626*0Sstevel@tonic-gate cc->cand_isdeprecated = ((cc->cand_flags & IPIF_DEPRECATED) != 0); 1627*0Sstevel@tonic-gate cc->cand_isdeprecated_set = B_TRUE; 1628*0Sstevel@tonic-gate 1629*0Sstevel@tonic-gate if (bc->cand_isdeprecated == cc->cand_isdeprecated) 1630*0Sstevel@tonic-gate return (CAND_TIE); 1631*0Sstevel@tonic-gate else if (cc->cand_isdeprecated) 1632*0Sstevel@tonic-gate return (CAND_AVOID); 1633*0Sstevel@tonic-gate else 1634*0Sstevel@tonic-gate return (CAND_PREFER); 1635*0Sstevel@tonic-gate } 1636*0Sstevel@tonic-gate 1637*0Sstevel@tonic-gate /* 1638*0Sstevel@tonic-gate * Prefer source addresses that have the IPIF_PREFERRED flag set. This 1639*0Sstevel@tonic-gate * rule must be before rule_interface because the flag could be set on any 1640*0Sstevel@tonic-gate * interface, not just the interface being used for outgoing packets (for 1641*0Sstevel@tonic-gate * example, the IFF_PREFERRED could be set on an address assigned to the 1642*0Sstevel@tonic-gate * loopback interface). 1643*0Sstevel@tonic-gate */ 1644*0Sstevel@tonic-gate /* ARGSUSED2 */ 1645*0Sstevel@tonic-gate static rule_res_t 1646*0Sstevel@tonic-gate rule_preferred(cand_t *bc, cand_t *cc, const dstinfo_t *dstinfo) 1647*0Sstevel@tonic-gate { 1648*0Sstevel@tonic-gate if (!bc->cand_ispreferred_set) { 1649*0Sstevel@tonic-gate bc->cand_ispreferred = ((bc->cand_flags & IPIF_PREFERRED) != 0); 1650*0Sstevel@tonic-gate bc->cand_ispreferred_set = B_TRUE; 1651*0Sstevel@tonic-gate } 1652*0Sstevel@tonic-gate 1653*0Sstevel@tonic-gate cc->cand_ispreferred = ((cc->cand_flags & IPIF_PREFERRED) != 0); 1654*0Sstevel@tonic-gate cc->cand_ispreferred_set = B_TRUE; 1655*0Sstevel@tonic-gate 1656*0Sstevel@tonic-gate if (bc->cand_ispreferred == cc->cand_ispreferred) 1657*0Sstevel@tonic-gate return (CAND_TIE); 1658*0Sstevel@tonic-gate else if (cc->cand_ispreferred) 1659*0Sstevel@tonic-gate return (CAND_PREFER); 1660*0Sstevel@tonic-gate else 1661*0Sstevel@tonic-gate return (CAND_AVOID); 1662*0Sstevel@tonic-gate } 1663*0Sstevel@tonic-gate 1664*0Sstevel@tonic-gate /* 1665*0Sstevel@tonic-gate * Prefer source addresses that are assigned to the outgoing interface, or 1666*0Sstevel@tonic-gate * to an interface that is in the same IPMP group as the outgoing 1667*0Sstevel@tonic-gate * interface. 1668*0Sstevel@tonic-gate */ 1669*0Sstevel@tonic-gate static rule_res_t 1670*0Sstevel@tonic-gate rule_interface(cand_t *bc, cand_t *cc, const dstinfo_t *dstinfo) 1671*0Sstevel@tonic-gate { 1672*0Sstevel@tonic-gate ill_t *dstill = dstinfo->dst_ill; 1673*0Sstevel@tonic-gate 1674*0Sstevel@tonic-gate /* 1675*0Sstevel@tonic-gate * If dstinfo->dst_restrict_ill is set, this rule is unnecessary 1676*0Sstevel@tonic-gate * since we know all candidates will be on the same link. 1677*0Sstevel@tonic-gate */ 1678*0Sstevel@tonic-gate if (dstinfo->dst_restrict_ill) 1679*0Sstevel@tonic-gate return (CAND_TIE); 1680*0Sstevel@tonic-gate 1681*0Sstevel@tonic-gate if (!bc->cand_matchedinterface_set) { 1682*0Sstevel@tonic-gate bc->cand_matchedinterface = (bc->cand_ill == dstill || 1683*0Sstevel@tonic-gate (dstill->ill_group != NULL && 1684*0Sstevel@tonic-gate dstill->ill_group == bc->cand_ill->ill_group)); 1685*0Sstevel@tonic-gate bc->cand_matchedinterface_set = B_TRUE; 1686*0Sstevel@tonic-gate } 1687*0Sstevel@tonic-gate 1688*0Sstevel@tonic-gate cc->cand_matchedinterface = (cc->cand_ill == dstill || 1689*0Sstevel@tonic-gate (dstill->ill_group != NULL && 1690*0Sstevel@tonic-gate dstill->ill_group == cc->cand_ill->ill_group)); 1691*0Sstevel@tonic-gate cc->cand_matchedinterface_set = B_TRUE; 1692*0Sstevel@tonic-gate 1693*0Sstevel@tonic-gate if (bc->cand_matchedinterface == cc->cand_matchedinterface) 1694*0Sstevel@tonic-gate return (CAND_TIE); 1695*0Sstevel@tonic-gate else if (cc->cand_matchedinterface) 1696*0Sstevel@tonic-gate return (CAND_PREFER); 1697*0Sstevel@tonic-gate else 1698*0Sstevel@tonic-gate return (CAND_AVOID); 1699*0Sstevel@tonic-gate } 1700*0Sstevel@tonic-gate 1701*0Sstevel@tonic-gate /* 1702*0Sstevel@tonic-gate * Prefer source addresses whose label matches the destination's label. 1703*0Sstevel@tonic-gate */ 1704*0Sstevel@tonic-gate static rule_res_t 1705*0Sstevel@tonic-gate rule_label(cand_t *bc, cand_t *cc, const dstinfo_t *dstinfo) 1706*0Sstevel@tonic-gate { 1707*0Sstevel@tonic-gate char *label; 1708*0Sstevel@tonic-gate 1709*0Sstevel@tonic-gate if (!bc->cand_matchedlabel_set) { 1710*0Sstevel@tonic-gate label = ip6_asp_lookup(&bc->cand_srcaddr, NULL); 1711*0Sstevel@tonic-gate bc->cand_matchedlabel = 1712*0Sstevel@tonic-gate ip6_asp_labelcmp(label, dstinfo->dst_label); 1713*0Sstevel@tonic-gate bc->cand_matchedlabel_set = B_TRUE; 1714*0Sstevel@tonic-gate } 1715*0Sstevel@tonic-gate 1716*0Sstevel@tonic-gate label = ip6_asp_lookup(&cc->cand_srcaddr, NULL); 1717*0Sstevel@tonic-gate cc->cand_matchedlabel = ip6_asp_labelcmp(label, dstinfo->dst_label); 1718*0Sstevel@tonic-gate cc->cand_matchedlabel_set = B_TRUE; 1719*0Sstevel@tonic-gate 1720*0Sstevel@tonic-gate if (bc->cand_matchedlabel == cc->cand_matchedlabel) 1721*0Sstevel@tonic-gate return (CAND_TIE); 1722*0Sstevel@tonic-gate else if (cc->cand_matchedlabel) 1723*0Sstevel@tonic-gate return (CAND_PREFER); 1724*0Sstevel@tonic-gate else 1725*0Sstevel@tonic-gate return (CAND_AVOID); 1726*0Sstevel@tonic-gate } 1727*0Sstevel@tonic-gate 1728*0Sstevel@tonic-gate /* 1729*0Sstevel@tonic-gate * Prefer public addresses over temporary ones. An application can reverse 1730*0Sstevel@tonic-gate * the logic of this rule and prefer temporary addresses by using the 1731*0Sstevel@tonic-gate * IPV6_SRC_PREFERENCES socket option. 1732*0Sstevel@tonic-gate */ 1733*0Sstevel@tonic-gate static rule_res_t 1734*0Sstevel@tonic-gate rule_temporary(cand_t *bc, cand_t *cc, const dstinfo_t *dstinfo) 1735*0Sstevel@tonic-gate { 1736*0Sstevel@tonic-gate if (!bc->cand_istmp_set) { 1737*0Sstevel@tonic-gate bc->cand_istmp = ((bc->cand_flags & IPIF_TEMPORARY) != 0); 1738*0Sstevel@tonic-gate bc->cand_istmp_set = B_TRUE; 1739*0Sstevel@tonic-gate } 1740*0Sstevel@tonic-gate 1741*0Sstevel@tonic-gate cc->cand_istmp = ((cc->cand_flags & IPIF_TEMPORARY) != 0); 1742*0Sstevel@tonic-gate cc->cand_istmp_set = B_TRUE; 1743*0Sstevel@tonic-gate 1744*0Sstevel@tonic-gate if (bc->cand_istmp == cc->cand_istmp) 1745*0Sstevel@tonic-gate return (CAND_TIE); 1746*0Sstevel@tonic-gate 1747*0Sstevel@tonic-gate if (dstinfo->dst_prefer_src_tmp && cc->cand_istmp) 1748*0Sstevel@tonic-gate return (CAND_PREFER); 1749*0Sstevel@tonic-gate else if (!dstinfo->dst_prefer_src_tmp && !cc->cand_istmp) 1750*0Sstevel@tonic-gate return (CAND_PREFER); 1751*0Sstevel@tonic-gate else 1752*0Sstevel@tonic-gate return (CAND_AVOID); 1753*0Sstevel@tonic-gate } 1754*0Sstevel@tonic-gate 1755*0Sstevel@tonic-gate /* 1756*0Sstevel@tonic-gate * Prefer source addresses with longer matching prefix with the 1757*0Sstevel@tonic-gate * destination. Since this is the last rule, it must not produce a tie. 1758*0Sstevel@tonic-gate * We do the longest matching prefix calculation and the tie break in one 1759*0Sstevel@tonic-gate * calculation by doing an xor of both addresses with the destination, and 1760*0Sstevel@tonic-gate * pick the address with the smallest xor value. That way, we're both 1761*0Sstevel@tonic-gate * picking the address with the longest matching prefix, and breaking the 1762*0Sstevel@tonic-gate * tie if they happen to have both have equal mathing prefixes. 1763*0Sstevel@tonic-gate */ 1764*0Sstevel@tonic-gate static rule_res_t 1765*0Sstevel@tonic-gate rule_prefix(cand_t *bc, cand_t *cc, const dstinfo_t *dstinfo) 1766*0Sstevel@tonic-gate { 1767*0Sstevel@tonic-gate int i; 1768*0Sstevel@tonic-gate 1769*0Sstevel@tonic-gate if (!bc->cand_xor_set) { 1770*0Sstevel@tonic-gate ip_addr_xor_v6(&bc->cand_srcaddr, 1771*0Sstevel@tonic-gate dstinfo->dst_addr, &bc->cand_xor); 1772*0Sstevel@tonic-gate bc->cand_xor_set = B_TRUE; 1773*0Sstevel@tonic-gate } 1774*0Sstevel@tonic-gate 1775*0Sstevel@tonic-gate ip_addr_xor_v6(&cc->cand_srcaddr, dstinfo->dst_addr, &cc->cand_xor); 1776*0Sstevel@tonic-gate cc->cand_xor_set = B_TRUE; 1777*0Sstevel@tonic-gate 1778*0Sstevel@tonic-gate for (i = 0; i < 4; i++) { 1779*0Sstevel@tonic-gate if (bc->cand_xor.s6_addr32[i] != cc->cand_xor.s6_addr32[i]) 1780*0Sstevel@tonic-gate break; 1781*0Sstevel@tonic-gate } 1782*0Sstevel@tonic-gate 1783*0Sstevel@tonic-gate if (cc->cand_xor.s6_addr32[i] < bc->cand_xor.s6_addr32[i]) 1784*0Sstevel@tonic-gate return (CAND_PREFER); 1785*0Sstevel@tonic-gate else 1786*0Sstevel@tonic-gate return (CAND_AVOID); 1787*0Sstevel@tonic-gate } 1788*0Sstevel@tonic-gate 1789*0Sstevel@tonic-gate /* 1790*0Sstevel@tonic-gate * Determine the best source address given a destination address and a 1791*0Sstevel@tonic-gate * destination ill. If no suitable source address is found, it returns 1792*0Sstevel@tonic-gate * NULL. If there is a usable address pointed to by the usesrc 1793*0Sstevel@tonic-gate * (i.e ill_usesrc_ifindex != 0) then return that first since it is more 1794*0Sstevel@tonic-gate * fine grained (i.e per interface) 1795*0Sstevel@tonic-gate * 1796*0Sstevel@tonic-gate * This implementation is based on the "Default Address Selection for IPv6" 1797*0Sstevel@tonic-gate * specification produced by the IETF IPv6 working group. It has been 1798*0Sstevel@tonic-gate * implemented so that the list of addresses is only traversed once (the 1799*0Sstevel@tonic-gate * specification's algorithm could traverse the list of addresses once for 1800*0Sstevel@tonic-gate * every rule). 1801*0Sstevel@tonic-gate * 1802*0Sstevel@tonic-gate * The restrict_ill argument restricts the algorithm to chose a source 1803*0Sstevel@tonic-gate * address that is assigned to the destination ill or an ill in the same 1804*0Sstevel@tonic-gate * IPMP group as the destination ill. This is used when the destination 1805*0Sstevel@tonic-gate * address is a link-local or multicast address, and when 1806*0Sstevel@tonic-gate * ipv6_strict_dst_multihoming is turned on. 1807*0Sstevel@tonic-gate * 1808*0Sstevel@tonic-gate * src_prefs is the caller's set of source address preferences. If source 1809*0Sstevel@tonic-gate * address selection is being called to determine the source address of a 1810*0Sstevel@tonic-gate * connected socket (from ip_bind_connected_v6()), then the preferences are 1811*0Sstevel@tonic-gate * taken from conn_src_preferences. These preferences can be set on a 1812*0Sstevel@tonic-gate * per-socket basis using the IPV6_SRC_PREFERENCES socket option. The only 1813*0Sstevel@tonic-gate * preference currently implemented is for rfc3041 temporary addresses. 1814*0Sstevel@tonic-gate */ 1815*0Sstevel@tonic-gate ipif_t * 1816*0Sstevel@tonic-gate ipif_select_source_v6(ill_t *dstill, const in6_addr_t *dst, 1817*0Sstevel@tonic-gate boolean_t restrict_ill, uint32_t src_prefs, zoneid_t zoneid) 1818*0Sstevel@tonic-gate { 1819*0Sstevel@tonic-gate dstinfo_t dstinfo; 1820*0Sstevel@tonic-gate char dstr[INET6_ADDRSTRLEN]; 1821*0Sstevel@tonic-gate char sstr[INET6_ADDRSTRLEN]; 1822*0Sstevel@tonic-gate ipif_t *ipif; 1823*0Sstevel@tonic-gate ill_t *ill, *usesrc_ill = NULL; 1824*0Sstevel@tonic-gate ill_walk_context_t ctx; 1825*0Sstevel@tonic-gate cand_t best_c; /* The best candidate */ 1826*0Sstevel@tonic-gate cand_t curr_c; /* The current candidate */ 1827*0Sstevel@tonic-gate uint_t index; 1828*0Sstevel@tonic-gate boolean_t first_candidate = B_TRUE; 1829*0Sstevel@tonic-gate rule_res_t rule_result; 1830*0Sstevel@tonic-gate phyint_t *phyi; 1831*0Sstevel@tonic-gate 1832*0Sstevel@tonic-gate /* 1833*0Sstevel@tonic-gate * The list of ordering rules. They are applied in the order they 1834*0Sstevel@tonic-gate * appear in the list. 1835*0Sstevel@tonic-gate * 1836*0Sstevel@tonic-gate * XXX rule_mipv6 will need to be implemented (the specification's 1837*0Sstevel@tonic-gate * rules 4) if a mobile IPv6 node is ever implemented. 1838*0Sstevel@tonic-gate */ 1839*0Sstevel@tonic-gate rulef_t rules[] = { 1840*0Sstevel@tonic-gate rule_isdst, 1841*0Sstevel@tonic-gate rule_scope, 1842*0Sstevel@tonic-gate rule_deprecated, 1843*0Sstevel@tonic-gate rule_preferred, 1844*0Sstevel@tonic-gate rule_interface, 1845*0Sstevel@tonic-gate rule_label, 1846*0Sstevel@tonic-gate rule_temporary, 1847*0Sstevel@tonic-gate rule_prefix, 1848*0Sstevel@tonic-gate NULL 1849*0Sstevel@tonic-gate }; 1850*0Sstevel@tonic-gate 1851*0Sstevel@tonic-gate ASSERT(dstill->ill_isv6); 1852*0Sstevel@tonic-gate ASSERT(!IN6_IS_ADDR_V4MAPPED(dst)); 1853*0Sstevel@tonic-gate 1854*0Sstevel@tonic-gate /* 1855*0Sstevel@tonic-gate * Check if there is a usable src address pointed to by the 1856*0Sstevel@tonic-gate * usesrc ifindex. This has higher precedence since it is 1857*0Sstevel@tonic-gate * finer grained (i.e per interface) v/s being system wide. 1858*0Sstevel@tonic-gate */ 1859*0Sstevel@tonic-gate if (dstill->ill_usesrc_ifindex != 0) { 1860*0Sstevel@tonic-gate if ((usesrc_ill = 1861*0Sstevel@tonic-gate ill_lookup_on_ifindex(dstill->ill_usesrc_ifindex, B_TRUE, 1862*0Sstevel@tonic-gate NULL, NULL, NULL, NULL)) != NULL) { 1863*0Sstevel@tonic-gate dstinfo.dst_ill = usesrc_ill; 1864*0Sstevel@tonic-gate } else { 1865*0Sstevel@tonic-gate return (NULL); 1866*0Sstevel@tonic-gate } 1867*0Sstevel@tonic-gate } else { 1868*0Sstevel@tonic-gate dstinfo.dst_ill = dstill; 1869*0Sstevel@tonic-gate } 1870*0Sstevel@tonic-gate 1871*0Sstevel@tonic-gate dstinfo.dst_addr = dst; 1872*0Sstevel@tonic-gate dstinfo.dst_scope = ip_addr_scope_v6(dst); 1873*0Sstevel@tonic-gate dstinfo.dst_label = ip6_asp_lookup(dst, NULL); 1874*0Sstevel@tonic-gate dstinfo.dst_prefer_src_tmp = ((src_prefs & IPV6_PREFER_SRC_TMP) != 0); 1875*0Sstevel@tonic-gate 1876*0Sstevel@tonic-gate rw_enter(&ill_g_lock, RW_READER); 1877*0Sstevel@tonic-gate /* 1878*0Sstevel@tonic-gate * Section three of the I-D states that for multicast and 1879*0Sstevel@tonic-gate * link-local destinations, the candidate set must be restricted to 1880*0Sstevel@tonic-gate * an interface that is on the same link as the outgoing interface. 1881*0Sstevel@tonic-gate * Also, when ipv6_strict_dst_multihoming is turned on, always 1882*0Sstevel@tonic-gate * restrict the source address to the destination link as doing 1883*0Sstevel@tonic-gate * otherwise will almost certainly cause problems. 1884*0Sstevel@tonic-gate */ 1885*0Sstevel@tonic-gate if (IN6_IS_ADDR_LINKLOCAL(dst) || IN6_IS_ADDR_MULTICAST(dst) || 1886*0Sstevel@tonic-gate ipv6_strict_dst_multihoming || usesrc_ill != NULL) 1887*0Sstevel@tonic-gate dstinfo.dst_restrict_ill = B_TRUE; 1888*0Sstevel@tonic-gate else 1889*0Sstevel@tonic-gate dstinfo.dst_restrict_ill = restrict_ill; 1890*0Sstevel@tonic-gate 1891*0Sstevel@tonic-gate bzero(&best_c, sizeof (cand_t)); 1892*0Sstevel@tonic-gate 1893*0Sstevel@tonic-gate /* 1894*0Sstevel@tonic-gate * Take a pass through the list of IPv6 interfaces to chose the 1895*0Sstevel@tonic-gate * best possible source address. If restrict_ill is true, we only 1896*0Sstevel@tonic-gate * iterate through the ill's that are in the same IPMP group as the 1897*0Sstevel@tonic-gate * destination's outgoing ill. If restrict_ill is false, we walk 1898*0Sstevel@tonic-gate * the entire list of IPv6 ill's. 1899*0Sstevel@tonic-gate */ 1900*0Sstevel@tonic-gate if (dstinfo.dst_restrict_ill) { 1901*0Sstevel@tonic-gate if (dstinfo.dst_ill->ill_group != NULL) { 1902*0Sstevel@tonic-gate /* 1903*0Sstevel@tonic-gate * Try to avoid FAILED/OFFLINE ills. Global and 1904*0Sstevel@tonic-gate * site local addresses will failover and are not 1905*0Sstevel@tonic-gate * an issue even if we select them. (i.e. this is 1906*0Sstevel@tonic-gate * a race where we hit this path before in.mpathd 1907*0Sstevel@tonic-gate * moves them. But link local addresses don't move. 1908*0Sstevel@tonic-gate * This creates a problem for NUD. If NUD ends up 1909*0Sstevel@tonic-gate * (nce_xmit) using the src addr from a failed 1910*0Sstevel@tonic-gate * interface NUD will fail and end up deleting the nce 1911*0Sstevel@tonic-gate * This will cause performance issues where ires 1912*0Sstevel@tonic-gate * are frequently created and deleted every few secs. 1913*0Sstevel@tonic-gate */ 1914*0Sstevel@tonic-gate for (ill = dstinfo.dst_ill->ill_group->illgrp_ill; 1915*0Sstevel@tonic-gate ill != NULL; ill = ill->ill_group_next) { 1916*0Sstevel@tonic-gate phyi = ill->ill_phyint; 1917*0Sstevel@tonic-gate if (!(phyi->phyint_flags & 1918*0Sstevel@tonic-gate (PHYI_OFFLINE | PHYI_FAILED))) 1919*0Sstevel@tonic-gate break; 1920*0Sstevel@tonic-gate } 1921*0Sstevel@tonic-gate if (ill == NULL) 1922*0Sstevel@tonic-gate ill = dstinfo.dst_ill->ill_group->illgrp_ill; 1923*0Sstevel@tonic-gate } else { 1924*0Sstevel@tonic-gate ill = dstinfo.dst_ill; 1925*0Sstevel@tonic-gate } 1926*0Sstevel@tonic-gate } else { 1927*0Sstevel@tonic-gate ill = ILL_START_WALK_V6(&ctx); 1928*0Sstevel@tonic-gate } 1929*0Sstevel@tonic-gate 1930*0Sstevel@tonic-gate while (ill != NULL) { 1931*0Sstevel@tonic-gate ASSERT(ill->ill_isv6); 1932*0Sstevel@tonic-gate 1933*0Sstevel@tonic-gate for (ipif = ill->ill_ipif; ipif != NULL; 1934*0Sstevel@tonic-gate ipif = ipif->ipif_next) { 1935*0Sstevel@tonic-gate 1936*0Sstevel@tonic-gate if (!IPIF_VALID_IPV6_SOURCE(ipif)) 1937*0Sstevel@tonic-gate continue; 1938*0Sstevel@tonic-gate 1939*0Sstevel@tonic-gate if (zoneid != ALL_ZONES && ipif->ipif_zoneid != zoneid) 1940*0Sstevel@tonic-gate continue; 1941*0Sstevel@tonic-gate 1942*0Sstevel@tonic-gate if (first_candidate) { 1943*0Sstevel@tonic-gate /* 1944*0Sstevel@tonic-gate * This is first valid address in the list. 1945*0Sstevel@tonic-gate * It is automatically the best candidate 1946*0Sstevel@tonic-gate * so far. 1947*0Sstevel@tonic-gate */ 1948*0Sstevel@tonic-gate best_c.cand_ipif = ipif; 1949*0Sstevel@tonic-gate first_candidate = B_FALSE; 1950*0Sstevel@tonic-gate continue; 1951*0Sstevel@tonic-gate } 1952*0Sstevel@tonic-gate 1953*0Sstevel@tonic-gate bzero(&curr_c, sizeof (cand_t)); 1954*0Sstevel@tonic-gate curr_c.cand_ipif = ipif; 1955*0Sstevel@tonic-gate 1956*0Sstevel@tonic-gate /* 1957*0Sstevel@tonic-gate * Compare this current candidate (curr_c) with the 1958*0Sstevel@tonic-gate * best candidate (best_c) by applying the 1959*0Sstevel@tonic-gate * comparison rules in order until one breaks the 1960*0Sstevel@tonic-gate * tie. 1961*0Sstevel@tonic-gate */ 1962*0Sstevel@tonic-gate for (index = 0; rules[index] != NULL; index++) { 1963*0Sstevel@tonic-gate /* Apply a comparison rule. */ 1964*0Sstevel@tonic-gate rule_result = 1965*0Sstevel@tonic-gate (rules[index])(&best_c, &curr_c, &dstinfo); 1966*0Sstevel@tonic-gate if (rule_result == CAND_AVOID) { 1967*0Sstevel@tonic-gate /* 1968*0Sstevel@tonic-gate * The best candidate is still the 1969*0Sstevel@tonic-gate * best candidate. Forget about 1970*0Sstevel@tonic-gate * this current candidate and go on 1971*0Sstevel@tonic-gate * to the next one. 1972*0Sstevel@tonic-gate */ 1973*0Sstevel@tonic-gate break; 1974*0Sstevel@tonic-gate } else if (rule_result == CAND_PREFER) { 1975*0Sstevel@tonic-gate /* 1976*0Sstevel@tonic-gate * This candidate is prefered. It 1977*0Sstevel@tonic-gate * becomes the best candidate so 1978*0Sstevel@tonic-gate * far. Go on to the next address. 1979*0Sstevel@tonic-gate */ 1980*0Sstevel@tonic-gate best_c = curr_c; 1981*0Sstevel@tonic-gate break; 1982*0Sstevel@tonic-gate } 1983*0Sstevel@tonic-gate /* We have a tie, apply the next rule. */ 1984*0Sstevel@tonic-gate } 1985*0Sstevel@tonic-gate 1986*0Sstevel@tonic-gate /* 1987*0Sstevel@tonic-gate * The last rule must be a tie breaker rule and 1988*0Sstevel@tonic-gate * must never produce a tie. At this point, the 1989*0Sstevel@tonic-gate * candidate should have either been rejected, or 1990*0Sstevel@tonic-gate * have been prefered as the best candidate so far. 1991*0Sstevel@tonic-gate */ 1992*0Sstevel@tonic-gate ASSERT(rule_result != CAND_TIE); 1993*0Sstevel@tonic-gate } 1994*0Sstevel@tonic-gate 1995*0Sstevel@tonic-gate /* 1996*0Sstevel@tonic-gate * We may be walking the linked-list of ill's in an 1997*0Sstevel@tonic-gate * IPMP group or traversing the IPv6 ill avl tree. If it is a 1998*0Sstevel@tonic-gate * usesrc ILL then it can't be part of IPMP group and we 1999*0Sstevel@tonic-gate * will exit the while loop. 2000*0Sstevel@tonic-gate */ 2001*0Sstevel@tonic-gate if (dstinfo.dst_restrict_ill) 2002*0Sstevel@tonic-gate ill = ill->ill_group_next; 2003*0Sstevel@tonic-gate else 2004*0Sstevel@tonic-gate ill = ill_next(&ctx, ill); 2005*0Sstevel@tonic-gate } 2006*0Sstevel@tonic-gate 2007*0Sstevel@tonic-gate ipif = best_c.cand_ipif; 2008*0Sstevel@tonic-gate ip1dbg(("ipif_select_source_v6(%s, %s) -> %s\n", 2009*0Sstevel@tonic-gate dstinfo.dst_ill->ill_name, 2010*0Sstevel@tonic-gate inet_ntop(AF_INET6, dstinfo.dst_addr, dstr, sizeof (dstr)), 2011*0Sstevel@tonic-gate (ipif == NULL ? "NULL" : 2012*0Sstevel@tonic-gate inet_ntop(AF_INET6, &ipif->ipif_v6lcl_addr, sstr, sizeof (sstr))))); 2013*0Sstevel@tonic-gate 2014*0Sstevel@tonic-gate if (usesrc_ill != NULL) 2015*0Sstevel@tonic-gate ill_refrele(usesrc_ill); 2016*0Sstevel@tonic-gate 2017*0Sstevel@tonic-gate if (ipif == NULL) { 2018*0Sstevel@tonic-gate rw_exit(&ill_g_lock); 2019*0Sstevel@tonic-gate return (NULL); 2020*0Sstevel@tonic-gate } 2021*0Sstevel@tonic-gate 2022*0Sstevel@tonic-gate mutex_enter(&ipif->ipif_ill->ill_lock); 2023*0Sstevel@tonic-gate if (IPIF_CAN_LOOKUP(ipif)) { 2024*0Sstevel@tonic-gate ipif_refhold_locked(ipif); 2025*0Sstevel@tonic-gate mutex_exit(&ipif->ipif_ill->ill_lock); 2026*0Sstevel@tonic-gate rw_exit(&ill_g_lock); 2027*0Sstevel@tonic-gate return (ipif); 2028*0Sstevel@tonic-gate } 2029*0Sstevel@tonic-gate mutex_exit(&ipif->ipif_ill->ill_lock); 2030*0Sstevel@tonic-gate rw_exit(&ill_g_lock); 2031*0Sstevel@tonic-gate ip1dbg(("ipif_select_source_v6 cannot lookup ipif %p" 2032*0Sstevel@tonic-gate " returning null \n", (void *)ipif)); 2033*0Sstevel@tonic-gate 2034*0Sstevel@tonic-gate return (NULL); 2035*0Sstevel@tonic-gate } 2036*0Sstevel@tonic-gate 2037*0Sstevel@tonic-gate /* 2038*0Sstevel@tonic-gate * If old_ipif is not NULL, see if ipif was derived from old 2039*0Sstevel@tonic-gate * ipif and if so, recreate the interface route by re-doing 2040*0Sstevel@tonic-gate * source address selection. This happens when ipif_down -> 2041*0Sstevel@tonic-gate * ipif_update_other_ipifs calls us. 2042*0Sstevel@tonic-gate * 2043*0Sstevel@tonic-gate * If old_ipif is NULL, just redo the source address selection 2044*0Sstevel@tonic-gate * if needed. This happens when illgrp_insert or ipif_up_done_v6 2045*0Sstevel@tonic-gate * calls us. 2046*0Sstevel@tonic-gate */ 2047*0Sstevel@tonic-gate void 2048*0Sstevel@tonic-gate ipif_recreate_interface_routes_v6(ipif_t *old_ipif, ipif_t *ipif) 2049*0Sstevel@tonic-gate { 2050*0Sstevel@tonic-gate ire_t *ire; 2051*0Sstevel@tonic-gate ire_t *ipif_ire; 2052*0Sstevel@tonic-gate queue_t *stq; 2053*0Sstevel@tonic-gate ill_t *ill; 2054*0Sstevel@tonic-gate ipif_t *nipif = NULL; 2055*0Sstevel@tonic-gate boolean_t nipif_refheld = B_FALSE; 2056*0Sstevel@tonic-gate boolean_t ip6_asp_table_held = B_FALSE; 2057*0Sstevel@tonic-gate 2058*0Sstevel@tonic-gate ill = ipif->ipif_ill; 2059*0Sstevel@tonic-gate 2060*0Sstevel@tonic-gate if (!(ipif->ipif_flags & 2061*0Sstevel@tonic-gate (IPIF_NOLOCAL|IPIF_ANYCAST|IPIF_DEPRECATED))) { 2062*0Sstevel@tonic-gate /* 2063*0Sstevel@tonic-gate * Can't possibly have borrowed the source 2064*0Sstevel@tonic-gate * from old_ipif. 2065*0Sstevel@tonic-gate */ 2066*0Sstevel@tonic-gate return; 2067*0Sstevel@tonic-gate } 2068*0Sstevel@tonic-gate 2069*0Sstevel@tonic-gate /* 2070*0Sstevel@tonic-gate * Is there any work to be done? No work if the address 2071*0Sstevel@tonic-gate * is INADDR_ANY, loopback or NOLOCAL or ANYCAST ( 2072*0Sstevel@tonic-gate * ipif_select_source_v6() does not borrow addresses from 2073*0Sstevel@tonic-gate * NOLOCAL and ANYCAST interfaces). 2074*0Sstevel@tonic-gate */ 2075*0Sstevel@tonic-gate if ((old_ipif != NULL) && 2076*0Sstevel@tonic-gate ((IN6_IS_ADDR_UNSPECIFIED(&old_ipif->ipif_v6lcl_addr)) || 2077*0Sstevel@tonic-gate (old_ipif->ipif_ill->ill_wq == NULL) || 2078*0Sstevel@tonic-gate (old_ipif->ipif_flags & 2079*0Sstevel@tonic-gate (IPIF_NOLOCAL|IPIF_ANYCAST)))) { 2080*0Sstevel@tonic-gate return; 2081*0Sstevel@tonic-gate } 2082*0Sstevel@tonic-gate 2083*0Sstevel@tonic-gate /* 2084*0Sstevel@tonic-gate * Perform the same checks as when creating the 2085*0Sstevel@tonic-gate * IRE_INTERFACE in ipif_up_done_v6. 2086*0Sstevel@tonic-gate */ 2087*0Sstevel@tonic-gate if (!(ipif->ipif_flags & IPIF_UP)) 2088*0Sstevel@tonic-gate return; 2089*0Sstevel@tonic-gate 2090*0Sstevel@tonic-gate if ((ipif->ipif_flags & IPIF_NOXMIT)) 2091*0Sstevel@tonic-gate return; 2092*0Sstevel@tonic-gate 2093*0Sstevel@tonic-gate if (IN6_IS_ADDR_UNSPECIFIED(&ipif->ipif_v6subnet) && 2094*0Sstevel@tonic-gate IN6_IS_ADDR_UNSPECIFIED(&ipif->ipif_v6net_mask)) 2095*0Sstevel@tonic-gate return; 2096*0Sstevel@tonic-gate 2097*0Sstevel@tonic-gate /* 2098*0Sstevel@tonic-gate * We know that ipif uses some other source for its 2099*0Sstevel@tonic-gate * IRE_INTERFACE. Is it using the source of this 2100*0Sstevel@tonic-gate * old_ipif? 2101*0Sstevel@tonic-gate */ 2102*0Sstevel@tonic-gate ipif_ire = ipif_to_ire_v6(ipif); 2103*0Sstevel@tonic-gate if (ipif_ire == NULL) 2104*0Sstevel@tonic-gate return; 2105*0Sstevel@tonic-gate 2106*0Sstevel@tonic-gate if (old_ipif != NULL && 2107*0Sstevel@tonic-gate !IN6_ARE_ADDR_EQUAL(&old_ipif->ipif_v6lcl_addr, 2108*0Sstevel@tonic-gate &ipif_ire->ire_src_addr_v6)) { 2109*0Sstevel@tonic-gate ire_refrele(ipif_ire); 2110*0Sstevel@tonic-gate return; 2111*0Sstevel@tonic-gate } 2112*0Sstevel@tonic-gate 2113*0Sstevel@tonic-gate if (ip_debug > 2) { 2114*0Sstevel@tonic-gate /* ip1dbg */ 2115*0Sstevel@tonic-gate pr_addr_dbg("ipif_recreate_interface_routes_v6: deleting IRE" 2116*0Sstevel@tonic-gate " for src %s\n", AF_INET6, &ipif_ire->ire_src_addr_v6); 2117*0Sstevel@tonic-gate } 2118*0Sstevel@tonic-gate 2119*0Sstevel@tonic-gate stq = ipif_ire->ire_stq; 2120*0Sstevel@tonic-gate 2121*0Sstevel@tonic-gate /* 2122*0Sstevel@tonic-gate * Can't use our source address. Select a different source address 2123*0Sstevel@tonic-gate * for the IRE_INTERFACE. We restrict interface route source 2124*0Sstevel@tonic-gate * address selection to ipif's assigned to the same link as the 2125*0Sstevel@tonic-gate * interface. 2126*0Sstevel@tonic-gate */ 2127*0Sstevel@tonic-gate if (ip6_asp_can_lookup()) { 2128*0Sstevel@tonic-gate ip6_asp_table_held = B_TRUE; 2129*0Sstevel@tonic-gate nipif = ipif_select_source_v6(ill, &ipif->ipif_v6subnet, 2130*0Sstevel@tonic-gate B_TRUE, IPV6_PREFER_SRC_DEFAULT, ipif->ipif_zoneid); 2131*0Sstevel@tonic-gate } 2132*0Sstevel@tonic-gate if (nipif == NULL) { 2133*0Sstevel@tonic-gate /* Last resort - all ipif's have IPIF_NOLOCAL */ 2134*0Sstevel@tonic-gate nipif = ipif; 2135*0Sstevel@tonic-gate } else { 2136*0Sstevel@tonic-gate nipif_refheld = B_TRUE; 2137*0Sstevel@tonic-gate } 2138*0Sstevel@tonic-gate 2139*0Sstevel@tonic-gate ire = ire_create_v6( 2140*0Sstevel@tonic-gate &ipif->ipif_v6subnet, /* dest pref */ 2141*0Sstevel@tonic-gate &ipif->ipif_v6net_mask, /* mask */ 2142*0Sstevel@tonic-gate &nipif->ipif_v6src_addr, /* src addr */ 2143*0Sstevel@tonic-gate NULL, /* no gateway */ 2144*0Sstevel@tonic-gate &ipif->ipif_mtu, /* max frag */ 2145*0Sstevel@tonic-gate NULL, /* no Fast path header */ 2146*0Sstevel@tonic-gate NULL, /* no recv from queue */ 2147*0Sstevel@tonic-gate stq, /* send-to queue */ 2148*0Sstevel@tonic-gate ill->ill_net_type, /* IF_[NO]RESOLVER */ 2149*0Sstevel@tonic-gate ill->ill_resolver_mp, /* xmit header */ 2150*0Sstevel@tonic-gate ipif, 2151*0Sstevel@tonic-gate NULL, 2152*0Sstevel@tonic-gate 0, 2153*0Sstevel@tonic-gate 0, 2154*0Sstevel@tonic-gate 0, 2155*0Sstevel@tonic-gate &ire_uinfo_null); 2156*0Sstevel@tonic-gate 2157*0Sstevel@tonic-gate if (ire != NULL) { 2158*0Sstevel@tonic-gate ire_t *ret_ire; 2159*0Sstevel@tonic-gate int error; 2160*0Sstevel@tonic-gate 2161*0Sstevel@tonic-gate /* 2162*0Sstevel@tonic-gate * We don't need ipif_ire anymore. We need to delete 2163*0Sstevel@tonic-gate * before we add so that ire_add does not detect 2164*0Sstevel@tonic-gate * duplicates. 2165*0Sstevel@tonic-gate */ 2166*0Sstevel@tonic-gate ire_delete(ipif_ire); 2167*0Sstevel@tonic-gate ret_ire = ire; 2168*0Sstevel@tonic-gate error = ire_add(&ret_ire, NULL, NULL, NULL); 2169*0Sstevel@tonic-gate ASSERT(error == 0); 2170*0Sstevel@tonic-gate ASSERT(ret_ire == ire); 2171*0Sstevel@tonic-gate if (ret_ire != NULL) { 2172*0Sstevel@tonic-gate /* Held in ire_add */ 2173*0Sstevel@tonic-gate ire_refrele(ret_ire); 2174*0Sstevel@tonic-gate } 2175*0Sstevel@tonic-gate } 2176*0Sstevel@tonic-gate /* 2177*0Sstevel@tonic-gate * Either we are falling through from above or could not 2178*0Sstevel@tonic-gate * allocate a replacement. 2179*0Sstevel@tonic-gate */ 2180*0Sstevel@tonic-gate ire_refrele(ipif_ire); 2181*0Sstevel@tonic-gate if (ip6_asp_table_held) 2182*0Sstevel@tonic-gate ip6_asp_table_refrele(); 2183*0Sstevel@tonic-gate if (nipif_refheld) 2184*0Sstevel@tonic-gate ipif_refrele(nipif); 2185*0Sstevel@tonic-gate } 2186*0Sstevel@tonic-gate 2187*0Sstevel@tonic-gate /* 2188*0Sstevel@tonic-gate * This old_ipif is going away. 2189*0Sstevel@tonic-gate * 2190*0Sstevel@tonic-gate * Determine if any other ipif's are using our address as 2191*0Sstevel@tonic-gate * ipif_v6lcl_addr (due to those being IPIF_NOLOCAL, IPIF_ANYCAST, or 2192*0Sstevel@tonic-gate * IPIF_DEPRECATED). 2193*0Sstevel@tonic-gate * Find the IRE_INTERFACE for such ipif's and recreate them 2194*0Sstevel@tonic-gate * to use an different source address following the rules in 2195*0Sstevel@tonic-gate * ipif_up_done_v6. 2196*0Sstevel@tonic-gate * 2197*0Sstevel@tonic-gate * This function takes an illgrp as an argument so that illgrp_delete 2198*0Sstevel@tonic-gate * can call this to update source address even after deleting the 2199*0Sstevel@tonic-gate * old_ipif->ipif_ill from the ill group. 2200*0Sstevel@tonic-gate */ 2201*0Sstevel@tonic-gate void 2202*0Sstevel@tonic-gate ipif_update_other_ipifs_v6(ipif_t *old_ipif, ill_group_t *illgrp) 2203*0Sstevel@tonic-gate { 2204*0Sstevel@tonic-gate ipif_t *ipif; 2205*0Sstevel@tonic-gate ill_t *ill; 2206*0Sstevel@tonic-gate char buf[INET6_ADDRSTRLEN]; 2207*0Sstevel@tonic-gate 2208*0Sstevel@tonic-gate ASSERT(IAM_WRITER_IPIF(old_ipif)); 2209*0Sstevel@tonic-gate 2210*0Sstevel@tonic-gate ill = old_ipif->ipif_ill; 2211*0Sstevel@tonic-gate 2212*0Sstevel@tonic-gate ip1dbg(("ipif_update_other_ipifs_v6(%s, %s)\n", 2213*0Sstevel@tonic-gate ill->ill_name, 2214*0Sstevel@tonic-gate inet_ntop(AF_INET6, &old_ipif->ipif_v6lcl_addr, 2215*0Sstevel@tonic-gate buf, sizeof (buf)))); 2216*0Sstevel@tonic-gate 2217*0Sstevel@tonic-gate /* 2218*0Sstevel@tonic-gate * If this part of a group, look at all ills as ipif_select_source 2219*0Sstevel@tonic-gate * borrows a source address across all the ills in the group. 2220*0Sstevel@tonic-gate */ 2221*0Sstevel@tonic-gate if (illgrp != NULL) 2222*0Sstevel@tonic-gate ill = illgrp->illgrp_ill; 2223*0Sstevel@tonic-gate 2224*0Sstevel@tonic-gate /* Don't need a lock since this is a writer */ 2225*0Sstevel@tonic-gate for (; ill != NULL; ill = ill->ill_group_next) { 2226*0Sstevel@tonic-gate for (ipif = ill->ill_ipif; ipif != NULL; 2227*0Sstevel@tonic-gate ipif = ipif->ipif_next) { 2228*0Sstevel@tonic-gate 2229*0Sstevel@tonic-gate if (ipif == old_ipif) 2230*0Sstevel@tonic-gate continue; 2231*0Sstevel@tonic-gate 2232*0Sstevel@tonic-gate ipif_recreate_interface_routes_v6(old_ipif, ipif); 2233*0Sstevel@tonic-gate } 2234*0Sstevel@tonic-gate } 2235*0Sstevel@tonic-gate } 2236*0Sstevel@tonic-gate 2237*0Sstevel@tonic-gate /* 2238*0Sstevel@tonic-gate * Perform an attach and bind to get phys addr plus info_req for 2239*0Sstevel@tonic-gate * the physical device. 2240*0Sstevel@tonic-gate * q and mp represents an ioctl which will be queued waiting for 2241*0Sstevel@tonic-gate * completion of the DLPI message exchange. 2242*0Sstevel@tonic-gate * MUST be called on an ill queue. Can not set conn_pending_ill for that 2243*0Sstevel@tonic-gate * reason thus the DL_PHYS_ADDR_ACK code does not assume ill_pending_q. 2244*0Sstevel@tonic-gate * 2245*0Sstevel@tonic-gate * Returns EINPROGRESS when mp has been consumed by queueing it on 2246*0Sstevel@tonic-gate * ill_pending_mp and the ioctl will complete in ip_rput. 2247*0Sstevel@tonic-gate */ 2248*0Sstevel@tonic-gate int 2249*0Sstevel@tonic-gate ill_dl_phys(ill_t *ill, ipif_t *ipif, mblk_t *mp, queue_t *q) 2250*0Sstevel@tonic-gate { 2251*0Sstevel@tonic-gate mblk_t *v6token_mp = NULL; 2252*0Sstevel@tonic-gate mblk_t *v6lla_mp = NULL; 2253*0Sstevel@tonic-gate mblk_t *phys_mp = NULL; 2254*0Sstevel@tonic-gate mblk_t *info_mp = NULL; 2255*0Sstevel@tonic-gate mblk_t *attach_mp = NULL; 2256*0Sstevel@tonic-gate mblk_t *detach_mp = NULL; 2257*0Sstevel@tonic-gate mblk_t *bind_mp = NULL; 2258*0Sstevel@tonic-gate mblk_t *unbind_mp = NULL; 2259*0Sstevel@tonic-gate mblk_t *notify_mp = NULL; 2260*0Sstevel@tonic-gate 2261*0Sstevel@tonic-gate ip1dbg(("ill_dl_phys(%s:%u)\n", ill->ill_name, ipif->ipif_id)); 2262*0Sstevel@tonic-gate ASSERT(ill->ill_dlpi_style_set); 2263*0Sstevel@tonic-gate ASSERT(WR(q)->q_next != NULL); 2264*0Sstevel@tonic-gate 2265*0Sstevel@tonic-gate if (ill->ill_isv6) { 2266*0Sstevel@tonic-gate v6token_mp = ip_dlpi_alloc(sizeof (dl_phys_addr_req_t) + 2267*0Sstevel@tonic-gate sizeof (t_scalar_t), DL_PHYS_ADDR_REQ); 2268*0Sstevel@tonic-gate if (v6token_mp == NULL) 2269*0Sstevel@tonic-gate goto bad; 2270*0Sstevel@tonic-gate ((dl_phys_addr_req_t *)v6token_mp->b_rptr)->dl_addr_type = 2271*0Sstevel@tonic-gate DL_IPV6_TOKEN; 2272*0Sstevel@tonic-gate 2273*0Sstevel@tonic-gate v6lla_mp = ip_dlpi_alloc(sizeof (dl_phys_addr_req_t) + 2274*0Sstevel@tonic-gate sizeof (t_scalar_t), DL_PHYS_ADDR_REQ); 2275*0Sstevel@tonic-gate if (v6lla_mp == NULL) 2276*0Sstevel@tonic-gate goto bad; 2277*0Sstevel@tonic-gate ((dl_phys_addr_req_t *)v6lla_mp->b_rptr)->dl_addr_type = 2278*0Sstevel@tonic-gate DL_IPV6_LINK_LAYER_ADDR; 2279*0Sstevel@tonic-gate } 2280*0Sstevel@tonic-gate 2281*0Sstevel@tonic-gate /* 2282*0Sstevel@tonic-gate * Allocate a DL_NOTIFY_REQ and set the notifications we want. 2283*0Sstevel@tonic-gate */ 2284*0Sstevel@tonic-gate notify_mp = ip_dlpi_alloc(sizeof (dl_notify_req_t) + sizeof (long), 2285*0Sstevel@tonic-gate DL_NOTIFY_REQ); 2286*0Sstevel@tonic-gate if (notify_mp == NULL) 2287*0Sstevel@tonic-gate goto bad; 2288*0Sstevel@tonic-gate ((dl_notify_req_t *)notify_mp->b_rptr)->dl_notifications = 2289*0Sstevel@tonic-gate (DL_NOTE_PHYS_ADDR | DL_NOTE_SDU_SIZE | DL_NOTE_FASTPATH_FLUSH | 2290*0Sstevel@tonic-gate DL_NOTE_LINK_UP | DL_NOTE_LINK_DOWN | DL_NOTE_CAPAB_RENEG); 2291*0Sstevel@tonic-gate 2292*0Sstevel@tonic-gate phys_mp = ip_dlpi_alloc(sizeof (dl_phys_addr_req_t) + 2293*0Sstevel@tonic-gate sizeof (t_scalar_t), DL_PHYS_ADDR_REQ); 2294*0Sstevel@tonic-gate if (phys_mp == NULL) 2295*0Sstevel@tonic-gate goto bad; 2296*0Sstevel@tonic-gate ((dl_phys_addr_req_t *)phys_mp->b_rptr)->dl_addr_type = 2297*0Sstevel@tonic-gate DL_CURR_PHYS_ADDR; 2298*0Sstevel@tonic-gate 2299*0Sstevel@tonic-gate info_mp = ip_dlpi_alloc( 2300*0Sstevel@tonic-gate sizeof (dl_info_req_t) + sizeof (dl_info_ack_t), 2301*0Sstevel@tonic-gate DL_INFO_REQ); 2302*0Sstevel@tonic-gate if (info_mp == NULL) 2303*0Sstevel@tonic-gate goto bad; 2304*0Sstevel@tonic-gate 2305*0Sstevel@tonic-gate bind_mp = ip_dlpi_alloc(sizeof (dl_bind_req_t) + sizeof (long), 2306*0Sstevel@tonic-gate DL_BIND_REQ); 2307*0Sstevel@tonic-gate if (bind_mp == NULL) 2308*0Sstevel@tonic-gate goto bad; 2309*0Sstevel@tonic-gate ((dl_bind_req_t *)bind_mp->b_rptr)->dl_sap = ill->ill_sap; 2310*0Sstevel@tonic-gate ((dl_bind_req_t *)bind_mp->b_rptr)->dl_service_mode = DL_CLDLS; 2311*0Sstevel@tonic-gate 2312*0Sstevel@tonic-gate unbind_mp = ip_dlpi_alloc(sizeof (dl_unbind_req_t), DL_UNBIND_REQ); 2313*0Sstevel@tonic-gate if (unbind_mp == NULL) 2314*0Sstevel@tonic-gate goto bad; 2315*0Sstevel@tonic-gate 2316*0Sstevel@tonic-gate /* If we need to attach/detach, pre-alloc and initialize the mblks */ 2317*0Sstevel@tonic-gate if (ill->ill_needs_attach) { 2318*0Sstevel@tonic-gate attach_mp = ip_dlpi_alloc(sizeof (dl_attach_req_t), 2319*0Sstevel@tonic-gate DL_ATTACH_REQ); 2320*0Sstevel@tonic-gate if (attach_mp == NULL) 2321*0Sstevel@tonic-gate goto bad; 2322*0Sstevel@tonic-gate ((dl_attach_req_t *)attach_mp->b_rptr)->dl_ppa = ill->ill_ppa; 2323*0Sstevel@tonic-gate 2324*0Sstevel@tonic-gate detach_mp = ip_dlpi_alloc(sizeof (dl_detach_req_t), 2325*0Sstevel@tonic-gate DL_DETACH_REQ); 2326*0Sstevel@tonic-gate if (detach_mp == NULL) 2327*0Sstevel@tonic-gate goto bad; 2328*0Sstevel@tonic-gate } 2329*0Sstevel@tonic-gate 2330*0Sstevel@tonic-gate /* 2331*0Sstevel@tonic-gate * Here we are going to delay the ioctl ack until after 2332*0Sstevel@tonic-gate * ACKs from DL_PHYS_ADDR_REQ. So need to save the 2333*0Sstevel@tonic-gate * original ioctl message before sending the requests 2334*0Sstevel@tonic-gate */ 2335*0Sstevel@tonic-gate mutex_enter(&ill->ill_lock); 2336*0Sstevel@tonic-gate /* ipsq_pending_mp_add won't fail since we pass in a NULL connp */ 2337*0Sstevel@tonic-gate (void) ipsq_pending_mp_add(NULL, ipif, ill->ill_wq, mp, 0); 2338*0Sstevel@tonic-gate /* 2339*0Sstevel@tonic-gate * Set ill_phys_addr_pend to zero. It will be set to the addr_type of 2340*0Sstevel@tonic-gate * the DL_PHYS_ADDR_REQ in ill_dlpi_send() and ill_dlpi_done(). It will 2341*0Sstevel@tonic-gate * be used to track which DL_PHYS_ADDR_REQ is being ACK'd/NAK'd. 2342*0Sstevel@tonic-gate */ 2343*0Sstevel@tonic-gate ill->ill_phys_addr_pend = 0; 2344*0Sstevel@tonic-gate mutex_exit(&ill->ill_lock); 2345*0Sstevel@tonic-gate 2346*0Sstevel@tonic-gate if (attach_mp != NULL) { 2347*0Sstevel@tonic-gate ip1dbg(("ill_dl_phys: attach\n")); 2348*0Sstevel@tonic-gate ill_dlpi_send(ill, attach_mp); 2349*0Sstevel@tonic-gate } 2350*0Sstevel@tonic-gate ill_dlpi_send(ill, bind_mp); 2351*0Sstevel@tonic-gate ill_dlpi_send(ill, info_mp); 2352*0Sstevel@tonic-gate if (ill->ill_isv6) { 2353*0Sstevel@tonic-gate ill_dlpi_send(ill, v6token_mp); 2354*0Sstevel@tonic-gate ill_dlpi_send(ill, v6lla_mp); 2355*0Sstevel@tonic-gate } 2356*0Sstevel@tonic-gate ill_dlpi_send(ill, phys_mp); 2357*0Sstevel@tonic-gate ill_dlpi_send(ill, notify_mp); 2358*0Sstevel@tonic-gate ill_dlpi_send(ill, unbind_mp); 2359*0Sstevel@tonic-gate 2360*0Sstevel@tonic-gate /* 2361*0Sstevel@tonic-gate * Save the DL_DETACH_REQ (if there is one) for use in ill_delete(). 2362*0Sstevel@tonic-gate */ 2363*0Sstevel@tonic-gate ASSERT(ill->ill_detach_mp == NULL); 2364*0Sstevel@tonic-gate ill->ill_detach_mp = detach_mp; 2365*0Sstevel@tonic-gate 2366*0Sstevel@tonic-gate /* 2367*0Sstevel@tonic-gate * This operation will complete in ip_rput_dlpi_writer with either 2368*0Sstevel@tonic-gate * a DL_PHYS_ADDR_ACK or DL_ERROR_ACK. 2369*0Sstevel@tonic-gate */ 2370*0Sstevel@tonic-gate return (EINPROGRESS); 2371*0Sstevel@tonic-gate bad: 2372*0Sstevel@tonic-gate if (v6token_mp != NULL) 2373*0Sstevel@tonic-gate freemsg(v6token_mp); 2374*0Sstevel@tonic-gate if (v6lla_mp != NULL) 2375*0Sstevel@tonic-gate freemsg(v6lla_mp); 2376*0Sstevel@tonic-gate if (phys_mp != NULL) 2377*0Sstevel@tonic-gate freemsg(phys_mp); 2378*0Sstevel@tonic-gate if (info_mp != NULL) 2379*0Sstevel@tonic-gate freemsg(info_mp); 2380*0Sstevel@tonic-gate if (attach_mp != NULL) 2381*0Sstevel@tonic-gate freemsg(attach_mp); 2382*0Sstevel@tonic-gate if (detach_mp != NULL) 2383*0Sstevel@tonic-gate freemsg(detach_mp); 2384*0Sstevel@tonic-gate if (bind_mp != NULL) 2385*0Sstevel@tonic-gate freemsg(bind_mp); 2386*0Sstevel@tonic-gate if (unbind_mp != NULL) 2387*0Sstevel@tonic-gate freemsg(unbind_mp); 2388*0Sstevel@tonic-gate if (notify_mp != NULL) 2389*0Sstevel@tonic-gate freemsg(notify_mp); 2390*0Sstevel@tonic-gate return (ENOMEM); 2391*0Sstevel@tonic-gate } 2392*0Sstevel@tonic-gate 2393*0Sstevel@tonic-gate uint_t ip_loopback_mtu_v6plus = IP_LOOPBACK_MTU + IPV6_HDR_LEN + 20; 2394*0Sstevel@tonic-gate 2395*0Sstevel@tonic-gate /* 2396*0Sstevel@tonic-gate * DLPI is up. 2397*0Sstevel@tonic-gate * Create all the IREs associated with an interface bring up multicast. 2398*0Sstevel@tonic-gate * Set the interface flag and finish other initialization 2399*0Sstevel@tonic-gate * that potentially had to be differed to after DL_BIND_ACK. 2400*0Sstevel@tonic-gate */ 2401*0Sstevel@tonic-gate int 2402*0Sstevel@tonic-gate ipif_up_done_v6(ipif_t *ipif) 2403*0Sstevel@tonic-gate { 2404*0Sstevel@tonic-gate ire_t *ire_array[20]; 2405*0Sstevel@tonic-gate ire_t **irep = ire_array; 2406*0Sstevel@tonic-gate ire_t **irep1; 2407*0Sstevel@tonic-gate ill_t *ill = ipif->ipif_ill; 2408*0Sstevel@tonic-gate queue_t *stq; 2409*0Sstevel@tonic-gate in6_addr_t v6addr; 2410*0Sstevel@tonic-gate in6_addr_t route_mask; 2411*0Sstevel@tonic-gate ipif_t *src_ipif = NULL; 2412*0Sstevel@tonic-gate ipif_t *tmp_ipif; 2413*0Sstevel@tonic-gate boolean_t flush_ire_cache = B_TRUE; 2414*0Sstevel@tonic-gate int err; 2415*0Sstevel@tonic-gate char buf[INET6_ADDRSTRLEN]; 2416*0Sstevel@tonic-gate phyint_t *phyi; 2417*0Sstevel@tonic-gate ire_t **ipif_saved_irep = NULL; 2418*0Sstevel@tonic-gate int ipif_saved_ire_cnt; 2419*0Sstevel@tonic-gate int cnt; 2420*0Sstevel@tonic-gate boolean_t src_ipif_held = B_FALSE; 2421*0Sstevel@tonic-gate boolean_t ire_added = B_FALSE; 2422*0Sstevel@tonic-gate boolean_t loopback = B_FALSE; 2423*0Sstevel@tonic-gate boolean_t ip6_asp_table_held = B_FALSE; 2424*0Sstevel@tonic-gate 2425*0Sstevel@tonic-gate ip1dbg(("ipif_up_done_v6(%s:%u)\n", 2426*0Sstevel@tonic-gate ipif->ipif_ill->ill_name, ipif->ipif_id)); 2427*0Sstevel@tonic-gate 2428*0Sstevel@tonic-gate /* Check if this is a loopback interface */ 2429*0Sstevel@tonic-gate if (ipif->ipif_ill->ill_wq == NULL) 2430*0Sstevel@tonic-gate loopback = B_TRUE; 2431*0Sstevel@tonic-gate 2432*0Sstevel@tonic-gate ASSERT(ipif->ipif_isv6); 2433*0Sstevel@tonic-gate ASSERT(!MUTEX_HELD(&ipif->ipif_ill->ill_lock)); 2434*0Sstevel@tonic-gate 2435*0Sstevel@tonic-gate /* 2436*0Sstevel@tonic-gate * If all other interfaces for this ill are down or DEPRECATED, 2437*0Sstevel@tonic-gate * or otherwise unsuitable for source address selection, remove 2438*0Sstevel@tonic-gate * any IRE_CACHE entries for this ill to make sure source 2439*0Sstevel@tonic-gate * address selection gets to take this new ipif into account. 2440*0Sstevel@tonic-gate * No need to hold ill_lock while traversing the ipif list since 2441*0Sstevel@tonic-gate * we are writer 2442*0Sstevel@tonic-gate */ 2443*0Sstevel@tonic-gate for (tmp_ipif = ill->ill_ipif; tmp_ipif; 2444*0Sstevel@tonic-gate tmp_ipif = tmp_ipif->ipif_next) { 2445*0Sstevel@tonic-gate if (((tmp_ipif->ipif_flags & 2446*0Sstevel@tonic-gate (IPIF_NOXMIT|IPIF_ANYCAST|IPIF_NOLOCAL|IPIF_DEPRECATED)) || 2447*0Sstevel@tonic-gate !(tmp_ipif->ipif_flags & IPIF_UP)) || 2448*0Sstevel@tonic-gate (tmp_ipif == ipif)) 2449*0Sstevel@tonic-gate continue; 2450*0Sstevel@tonic-gate /* first useable pre-existing interface */ 2451*0Sstevel@tonic-gate flush_ire_cache = B_FALSE; 2452*0Sstevel@tonic-gate break; 2453*0Sstevel@tonic-gate } 2454*0Sstevel@tonic-gate if (flush_ire_cache) 2455*0Sstevel@tonic-gate ire_walk_ill_v6(MATCH_IRE_ILL_GROUP | MATCH_IRE_TYPE, 2456*0Sstevel@tonic-gate IRE_CACHE, ill_ipif_cache_delete, (char *)ill, ill); 2457*0Sstevel@tonic-gate 2458*0Sstevel@tonic-gate /* 2459*0Sstevel@tonic-gate * Figure out which way the send-to queue should go. Only 2460*0Sstevel@tonic-gate * IRE_IF_RESOLVER or IRE_IF_NORESOLVER should show up here. 2461*0Sstevel@tonic-gate */ 2462*0Sstevel@tonic-gate switch (ill->ill_net_type) { 2463*0Sstevel@tonic-gate case IRE_IF_RESOLVER: 2464*0Sstevel@tonic-gate stq = ill->ill_rq; 2465*0Sstevel@tonic-gate break; 2466*0Sstevel@tonic-gate case IRE_IF_NORESOLVER: 2467*0Sstevel@tonic-gate case IRE_LOOPBACK: 2468*0Sstevel@tonic-gate stq = ill->ill_wq; 2469*0Sstevel@tonic-gate break; 2470*0Sstevel@tonic-gate default: 2471*0Sstevel@tonic-gate return (EINVAL); 2472*0Sstevel@tonic-gate } 2473*0Sstevel@tonic-gate 2474*0Sstevel@tonic-gate if (ill->ill_phyint->phyint_flags & PHYI_LOOPBACK) { 2475*0Sstevel@tonic-gate /* 2476*0Sstevel@tonic-gate * lo0:1 and subsequent ipifs were marked IRE_LOCAL in 2477*0Sstevel@tonic-gate * ipif_lookup_on_name(), but in the case of zones we can have 2478*0Sstevel@tonic-gate * several loopback addresses on lo0. So all the interfaces with 2479*0Sstevel@tonic-gate * loopback addresses need to be marked IRE_LOOPBACK. 2480*0Sstevel@tonic-gate */ 2481*0Sstevel@tonic-gate if (IN6_ARE_ADDR_EQUAL(&ipif->ipif_v6lcl_addr, &ipv6_loopback)) 2482*0Sstevel@tonic-gate ipif->ipif_ire_type = IRE_LOOPBACK; 2483*0Sstevel@tonic-gate else 2484*0Sstevel@tonic-gate ipif->ipif_ire_type = IRE_LOCAL; 2485*0Sstevel@tonic-gate } 2486*0Sstevel@tonic-gate 2487*0Sstevel@tonic-gate if (ipif->ipif_flags & (IPIF_NOLOCAL|IPIF_ANYCAST|IPIF_DEPRECATED)) { 2488*0Sstevel@tonic-gate /* 2489*0Sstevel@tonic-gate * Can't use our source address. Select a different 2490*0Sstevel@tonic-gate * source address for the IRE_INTERFACE and IRE_LOCAL 2491*0Sstevel@tonic-gate */ 2492*0Sstevel@tonic-gate if (ip6_asp_can_lookup()) { 2493*0Sstevel@tonic-gate ip6_asp_table_held = B_TRUE; 2494*0Sstevel@tonic-gate src_ipif = ipif_select_source_v6(ipif->ipif_ill, 2495*0Sstevel@tonic-gate &ipif->ipif_v6subnet, B_FALSE, 2496*0Sstevel@tonic-gate IPV6_PREFER_SRC_DEFAULT, ipif->ipif_zoneid); 2497*0Sstevel@tonic-gate } 2498*0Sstevel@tonic-gate if (src_ipif == NULL) 2499*0Sstevel@tonic-gate src_ipif = ipif; /* Last resort */ 2500*0Sstevel@tonic-gate else 2501*0Sstevel@tonic-gate src_ipif_held = B_TRUE; 2502*0Sstevel@tonic-gate } else { 2503*0Sstevel@tonic-gate src_ipif = ipif; 2504*0Sstevel@tonic-gate } 2505*0Sstevel@tonic-gate 2506*0Sstevel@tonic-gate if (!IN6_IS_ADDR_UNSPECIFIED(&ipif->ipif_v6lcl_addr) && 2507*0Sstevel@tonic-gate !(ipif->ipif_flags & IPIF_NOLOCAL)) { 2508*0Sstevel@tonic-gate /* Register the source address for __sin6_src_id */ 2509*0Sstevel@tonic-gate err = ip_srcid_insert(&ipif->ipif_v6lcl_addr, 2510*0Sstevel@tonic-gate ipif->ipif_zoneid); 2511*0Sstevel@tonic-gate if (err != 0) { 2512*0Sstevel@tonic-gate ip0dbg(("ipif_up_done_v6: srcid_insert %d\n", err)); 2513*0Sstevel@tonic-gate if (src_ipif_held) 2514*0Sstevel@tonic-gate ipif_refrele(src_ipif); 2515*0Sstevel@tonic-gate if (ip6_asp_table_held) 2516*0Sstevel@tonic-gate ip6_asp_table_refrele(); 2517*0Sstevel@tonic-gate return (err); 2518*0Sstevel@tonic-gate } 2519*0Sstevel@tonic-gate /* 2520*0Sstevel@tonic-gate * If the interface address is set, create the LOCAL 2521*0Sstevel@tonic-gate * or LOOPBACK IRE. 2522*0Sstevel@tonic-gate */ 2523*0Sstevel@tonic-gate ip1dbg(("ipif_up_done_v6: creating IRE %d for %s\n", 2524*0Sstevel@tonic-gate ipif->ipif_ire_type, 2525*0Sstevel@tonic-gate inet_ntop(AF_INET6, &ipif->ipif_v6lcl_addr, 2526*0Sstevel@tonic-gate buf, sizeof (buf)))); 2527*0Sstevel@tonic-gate 2528*0Sstevel@tonic-gate *irep++ = ire_create_v6( 2529*0Sstevel@tonic-gate &ipif->ipif_v6lcl_addr, /* dest address */ 2530*0Sstevel@tonic-gate &ipv6_all_ones, /* mask */ 2531*0Sstevel@tonic-gate &src_ipif->ipif_v6src_addr, /* source address */ 2532*0Sstevel@tonic-gate NULL, /* no gateway */ 2533*0Sstevel@tonic-gate &ip_loopback_mtu_v6plus, /* max frag size */ 2534*0Sstevel@tonic-gate NULL, 2535*0Sstevel@tonic-gate ipif->ipif_rq, /* recv-from queue */ 2536*0Sstevel@tonic-gate NULL, /* no send-to queue */ 2537*0Sstevel@tonic-gate ipif->ipif_ire_type, /* LOCAL or LOOPBACK */ 2538*0Sstevel@tonic-gate NULL, 2539*0Sstevel@tonic-gate ipif, /* interface */ 2540*0Sstevel@tonic-gate NULL, 2541*0Sstevel@tonic-gate 0, 2542*0Sstevel@tonic-gate 0, 2543*0Sstevel@tonic-gate (ipif->ipif_flags & IPIF_PRIVATE) ? RTF_PRIVATE : 0, 2544*0Sstevel@tonic-gate &ire_uinfo_null); 2545*0Sstevel@tonic-gate } 2546*0Sstevel@tonic-gate 2547*0Sstevel@tonic-gate /* 2548*0Sstevel@tonic-gate * Set up the IRE_IF_RESOLVER or IRE_IF_NORESOLVER, as appropriate. 2549*0Sstevel@tonic-gate * Note that atun interfaces have an all-zero ipif_v6subnet. 2550*0Sstevel@tonic-gate * Thus we allow a zero subnet as long as the mask is non-zero. 2551*0Sstevel@tonic-gate */ 2552*0Sstevel@tonic-gate if (stq != NULL && !(ipif->ipif_flags & IPIF_NOXMIT) && 2553*0Sstevel@tonic-gate !(IN6_IS_ADDR_UNSPECIFIED(&ipif->ipif_v6subnet) && 2554*0Sstevel@tonic-gate IN6_IS_ADDR_UNSPECIFIED(&ipif->ipif_v6net_mask))) { 2555*0Sstevel@tonic-gate /* ipif_v6subnet is ipif_v6pp_dst_addr for pt-pt */ 2556*0Sstevel@tonic-gate v6addr = ipif->ipif_v6subnet; 2557*0Sstevel@tonic-gate 2558*0Sstevel@tonic-gate if (ipif->ipif_flags & IPIF_POINTOPOINT) { 2559*0Sstevel@tonic-gate route_mask = ipv6_all_ones; 2560*0Sstevel@tonic-gate } else { 2561*0Sstevel@tonic-gate route_mask = ipif->ipif_v6net_mask; 2562*0Sstevel@tonic-gate } 2563*0Sstevel@tonic-gate 2564*0Sstevel@tonic-gate ip1dbg(("ipif_up_done_v6: creating if IRE %d for %s\n", 2565*0Sstevel@tonic-gate ill->ill_net_type, 2566*0Sstevel@tonic-gate inet_ntop(AF_INET6, &v6addr, buf, sizeof (buf)))); 2567*0Sstevel@tonic-gate 2568*0Sstevel@tonic-gate *irep++ = ire_create_v6( 2569*0Sstevel@tonic-gate &v6addr, /* dest pref */ 2570*0Sstevel@tonic-gate &route_mask, /* mask */ 2571*0Sstevel@tonic-gate &src_ipif->ipif_v6src_addr, /* src addr */ 2572*0Sstevel@tonic-gate NULL, /* no gateway */ 2573*0Sstevel@tonic-gate &ipif->ipif_mtu, /* max frag */ 2574*0Sstevel@tonic-gate NULL, /* no Fast path header */ 2575*0Sstevel@tonic-gate NULL, /* no recv from queue */ 2576*0Sstevel@tonic-gate stq, /* send-to queue */ 2577*0Sstevel@tonic-gate ill->ill_net_type, /* IF_[NO]RESOLVER */ 2578*0Sstevel@tonic-gate ill->ill_resolver_mp, /* xmit header */ 2579*0Sstevel@tonic-gate ipif, 2580*0Sstevel@tonic-gate NULL, 2581*0Sstevel@tonic-gate 0, 2582*0Sstevel@tonic-gate 0, 2583*0Sstevel@tonic-gate (ipif->ipif_flags & IPIF_PRIVATE) ? RTF_PRIVATE : 0, 2584*0Sstevel@tonic-gate &ire_uinfo_null); 2585*0Sstevel@tonic-gate } 2586*0Sstevel@tonic-gate 2587*0Sstevel@tonic-gate /* 2588*0Sstevel@tonic-gate * Setup 2002::/16 route, if this interface is a 6to4 tunnel 2589*0Sstevel@tonic-gate */ 2590*0Sstevel@tonic-gate if (IN6_IS_ADDR_6TO4(&ipif->ipif_v6lcl_addr) && 2591*0Sstevel@tonic-gate (ill->ill_is_6to4tun)) { 2592*0Sstevel@tonic-gate /* 2593*0Sstevel@tonic-gate * Destination address is 2002::/16 2594*0Sstevel@tonic-gate */ 2595*0Sstevel@tonic-gate #ifdef _BIG_ENDIAN 2596*0Sstevel@tonic-gate const in6_addr_t prefix_addr = { 0x20020000U, 0, 0, 0 }; 2597*0Sstevel@tonic-gate const in6_addr_t prefix_mask = { 0xffff0000U, 0, 0, 0 }; 2598*0Sstevel@tonic-gate #else 2599*0Sstevel@tonic-gate const in6_addr_t prefix_addr = { 0x00000220U, 0, 0, 0 }; 2600*0Sstevel@tonic-gate const in6_addr_t prefix_mask = { 0x0000ffffU, 0, 0, 0 }; 2601*0Sstevel@tonic-gate #endif /* _BIG_ENDIAN */ 2602*0Sstevel@tonic-gate char buf2[INET6_ADDRSTRLEN]; 2603*0Sstevel@tonic-gate ire_t *isdup; 2604*0Sstevel@tonic-gate in6_addr_t *first_addr = &ill->ill_ipif->ipif_v6lcl_addr; 2605*0Sstevel@tonic-gate 2606*0Sstevel@tonic-gate /* 2607*0Sstevel@tonic-gate * check to see if this route has already been added for 2608*0Sstevel@tonic-gate * this tunnel interface. 2609*0Sstevel@tonic-gate */ 2610*0Sstevel@tonic-gate isdup = ire_ftable_lookup_v6(first_addr, &prefix_mask, 0, 2611*0Sstevel@tonic-gate IRE_IF_NORESOLVER, ill->ill_ipif, NULL, ALL_ZONES, 0, 2612*0Sstevel@tonic-gate (MATCH_IRE_SRC | MATCH_IRE_MASK)); 2613*0Sstevel@tonic-gate 2614*0Sstevel@tonic-gate if (isdup == NULL) { 2615*0Sstevel@tonic-gate ip1dbg(("ipif_up_done_v6: creating if IRE %d for %s", 2616*0Sstevel@tonic-gate IRE_IF_NORESOLVER, inet_ntop(AF_INET6, &v6addr, 2617*0Sstevel@tonic-gate buf2, sizeof (buf2)))); 2618*0Sstevel@tonic-gate 2619*0Sstevel@tonic-gate *irep++ = ire_create_v6( 2620*0Sstevel@tonic-gate &prefix_addr, /* 2002:: */ 2621*0Sstevel@tonic-gate &prefix_mask, /* ffff:: */ 2622*0Sstevel@tonic-gate &ipif->ipif_v6lcl_addr, /* src addr */ 2623*0Sstevel@tonic-gate NULL, /* gateway */ 2624*0Sstevel@tonic-gate &ipif->ipif_mtu, /* max_frag */ 2625*0Sstevel@tonic-gate NULL, /* no Fast Path hdr */ 2626*0Sstevel@tonic-gate NULL, /* no rfq */ 2627*0Sstevel@tonic-gate ill->ill_wq, /* stq */ 2628*0Sstevel@tonic-gate IRE_IF_NORESOLVER, /* type */ 2629*0Sstevel@tonic-gate ill->ill_resolver_mp, /* dlureq_mp */ 2630*0Sstevel@tonic-gate ipif, /* interface */ 2631*0Sstevel@tonic-gate NULL, /* v6cmask */ 2632*0Sstevel@tonic-gate 0, 2633*0Sstevel@tonic-gate 0, 2634*0Sstevel@tonic-gate RTF_UP, 2635*0Sstevel@tonic-gate &ire_uinfo_null); 2636*0Sstevel@tonic-gate } else { 2637*0Sstevel@tonic-gate ire_refrele(isdup); 2638*0Sstevel@tonic-gate } 2639*0Sstevel@tonic-gate } 2640*0Sstevel@tonic-gate 2641*0Sstevel@tonic-gate /* If an earlier ire_create failed, get out now */ 2642*0Sstevel@tonic-gate for (irep1 = irep; irep1 > ire_array; ) { 2643*0Sstevel@tonic-gate irep1--; 2644*0Sstevel@tonic-gate if (*irep1 == NULL) { 2645*0Sstevel@tonic-gate ip1dbg(("ipif_up_done_v6: NULL ire found in" 2646*0Sstevel@tonic-gate " ire_array\n")); 2647*0Sstevel@tonic-gate err = ENOMEM; 2648*0Sstevel@tonic-gate goto bad; 2649*0Sstevel@tonic-gate } 2650*0Sstevel@tonic-gate } 2651*0Sstevel@tonic-gate 2652*0Sstevel@tonic-gate ASSERT(!MUTEX_HELD(&ipif->ipif_ill->ill_lock)); 2653*0Sstevel@tonic-gate 2654*0Sstevel@tonic-gate /* 2655*0Sstevel@tonic-gate * Need to atomically check for ip_addr_availablity_check 2656*0Sstevel@tonic-gate * now under ill_g_lock, and if it fails got bad, and remove 2657*0Sstevel@tonic-gate * from group also 2658*0Sstevel@tonic-gate */ 2659*0Sstevel@tonic-gate rw_enter(&ill_g_lock, RW_READER); 2660*0Sstevel@tonic-gate mutex_enter(&ip_addr_avail_lock); 2661*0Sstevel@tonic-gate ill->ill_ipif_up_count++; 2662*0Sstevel@tonic-gate ipif->ipif_flags |= IPIF_UP; 2663*0Sstevel@tonic-gate err = ip_addr_availability_check(ipif); 2664*0Sstevel@tonic-gate mutex_exit(&ip_addr_avail_lock); 2665*0Sstevel@tonic-gate rw_exit(&ill_g_lock); 2666*0Sstevel@tonic-gate 2667*0Sstevel@tonic-gate if (err != 0) { 2668*0Sstevel@tonic-gate /* 2669*0Sstevel@tonic-gate * Our address may already be up on the same ill. In this case, 2670*0Sstevel@tonic-gate * the external resolver entry for our ipif replaced the one for 2671*0Sstevel@tonic-gate * the other ipif. So we don't want to delete it (otherwise the 2672*0Sstevel@tonic-gate * other ipif would be unable to send packets). 2673*0Sstevel@tonic-gate * ip_addr_availability_check() identifies this case for us and 2674*0Sstevel@tonic-gate * returns EADDRINUSE; we need to turn it into EADDRNOTAVAIL 2675*0Sstevel@tonic-gate * which is the expected error code. 2676*0Sstevel@tonic-gate */ 2677*0Sstevel@tonic-gate if (err == EADDRINUSE) { 2678*0Sstevel@tonic-gate if (ipif->ipif_ill->ill_flags & ILLF_XRESOLV) { 2679*0Sstevel@tonic-gate freemsg(ipif->ipif_arp_del_mp); 2680*0Sstevel@tonic-gate ipif->ipif_arp_del_mp = NULL; 2681*0Sstevel@tonic-gate } 2682*0Sstevel@tonic-gate err = EADDRNOTAVAIL; 2683*0Sstevel@tonic-gate } 2684*0Sstevel@tonic-gate ill->ill_ipif_up_count--; 2685*0Sstevel@tonic-gate ipif->ipif_flags &= ~IPIF_UP; 2686*0Sstevel@tonic-gate goto bad; 2687*0Sstevel@tonic-gate } 2688*0Sstevel@tonic-gate 2689*0Sstevel@tonic-gate /* 2690*0Sstevel@tonic-gate * Add in all newly created IREs. We want to add before 2691*0Sstevel@tonic-gate * we call ifgrp_insert which wants to know whether 2692*0Sstevel@tonic-gate * IRE_IF_RESOLVER exists or not. 2693*0Sstevel@tonic-gate * 2694*0Sstevel@tonic-gate * NOTE : We refrele the ire though we may branch to "bad" 2695*0Sstevel@tonic-gate * later on where we do ire_delete. This is okay 2696*0Sstevel@tonic-gate * because nobody can delete it as we are running 2697*0Sstevel@tonic-gate * exclusively. 2698*0Sstevel@tonic-gate */ 2699*0Sstevel@tonic-gate for (irep1 = irep; irep1 > ire_array; ) { 2700*0Sstevel@tonic-gate irep1--; 2701*0Sstevel@tonic-gate /* Shouldn't be adding any bcast ire's */ 2702*0Sstevel@tonic-gate ASSERT((*irep1)->ire_type != IRE_BROADCAST); 2703*0Sstevel@tonic-gate ASSERT(!MUTEX_HELD(&ipif->ipif_ill->ill_lock)); 2704*0Sstevel@tonic-gate /* 2705*0Sstevel@tonic-gate * refheld by ire_add. refele towards the end of the func 2706*0Sstevel@tonic-gate */ 2707*0Sstevel@tonic-gate (void) ire_add(irep1, NULL, NULL, NULL); 2708*0Sstevel@tonic-gate } 2709*0Sstevel@tonic-gate if (ip6_asp_table_held) { 2710*0Sstevel@tonic-gate ip6_asp_table_refrele(); 2711*0Sstevel@tonic-gate ip6_asp_table_held = B_FALSE; 2712*0Sstevel@tonic-gate } 2713*0Sstevel@tonic-gate ire_added = B_TRUE; 2714*0Sstevel@tonic-gate 2715*0Sstevel@tonic-gate /* 2716*0Sstevel@tonic-gate * Form groups if possible. 2717*0Sstevel@tonic-gate * 2718*0Sstevel@tonic-gate * If we are supposed to be in a ill_group with a name, insert it 2719*0Sstevel@tonic-gate * now as we know that at least one ipif is UP. Otherwise form 2720*0Sstevel@tonic-gate * nameless groups. 2721*0Sstevel@tonic-gate * 2722*0Sstevel@tonic-gate * If ip_enable_group_ifs is set and ipif address is not ::0, insert 2723*0Sstevel@tonic-gate * this ipif into the appropriate interface group, or create a 2724*0Sstevel@tonic-gate * new one. If this is already in a nameless group, we try to form 2725*0Sstevel@tonic-gate * a bigger group looking at other ills potentially sharing this 2726*0Sstevel@tonic-gate * ipif's prefix. 2727*0Sstevel@tonic-gate */ 2728*0Sstevel@tonic-gate phyi = ill->ill_phyint; 2729*0Sstevel@tonic-gate if (phyi->phyint_groupname_len != 0) { 2730*0Sstevel@tonic-gate ASSERT(phyi->phyint_groupname != NULL); 2731*0Sstevel@tonic-gate if (ill->ill_ipif_up_count == 1) { 2732*0Sstevel@tonic-gate ASSERT(ill->ill_group == NULL); 2733*0Sstevel@tonic-gate err = illgrp_insert(&illgrp_head_v6, ill, 2734*0Sstevel@tonic-gate phyi->phyint_groupname, NULL, B_TRUE); 2735*0Sstevel@tonic-gate if (err != 0) { 2736*0Sstevel@tonic-gate ip1dbg(("ipif_up_done_v6: illgrp allocation " 2737*0Sstevel@tonic-gate "failed, error %d\n", err)); 2738*0Sstevel@tonic-gate goto bad; 2739*0Sstevel@tonic-gate } 2740*0Sstevel@tonic-gate } 2741*0Sstevel@tonic-gate ASSERT(ill->ill_group != NULL); 2742*0Sstevel@tonic-gate } 2743*0Sstevel@tonic-gate 2744*0Sstevel@tonic-gate /* Recover any additional IRE_IF_[NO]RESOLVER entries for this ipif */ 2745*0Sstevel@tonic-gate ipif_saved_ire_cnt = ipif->ipif_saved_ire_cnt; 2746*0Sstevel@tonic-gate ipif_saved_irep = ipif_recover_ire_v6(ipif); 2747*0Sstevel@tonic-gate 2748*0Sstevel@tonic-gate if (ipif->ipif_ipif_up_count == 1 && !loopback) { 2749*0Sstevel@tonic-gate /* 2750*0Sstevel@tonic-gate * Need to recover all multicast memberships in the driver. 2751*0Sstevel@tonic-gate * This had to be deferred until we had attached. 2752*0Sstevel@tonic-gate */ 2753*0Sstevel@tonic-gate ill_recover_multicast(ill); 2754*0Sstevel@tonic-gate } 2755*0Sstevel@tonic-gate /* Join the allhosts multicast address and the solicited node MC */ 2756*0Sstevel@tonic-gate ipif_multicast_up(ipif); 2757*0Sstevel@tonic-gate 2758*0Sstevel@tonic-gate if (!loopback) { 2759*0Sstevel@tonic-gate /* 2760*0Sstevel@tonic-gate * See whether anybody else would benefit from the 2761*0Sstevel@tonic-gate * new ipif that we added. We call this always rather 2762*0Sstevel@tonic-gate * than while adding a non-IPIF_NOLOCAL/DEPRECATED/ANYCAST 2763*0Sstevel@tonic-gate * ipif for the benefit of illgrp_insert (done above) 2764*0Sstevel@tonic-gate * which does not do source address selection as it does 2765*0Sstevel@tonic-gate * not want to re-create interface routes that we are 2766*0Sstevel@tonic-gate * having reference to it here. 2767*0Sstevel@tonic-gate */ 2768*0Sstevel@tonic-gate ill_update_source_selection(ill); 2769*0Sstevel@tonic-gate } 2770*0Sstevel@tonic-gate 2771*0Sstevel@tonic-gate for (irep1 = irep; irep1 > ire_array; ) { 2772*0Sstevel@tonic-gate irep1--; 2773*0Sstevel@tonic-gate if (*irep1 != NULL) { 2774*0Sstevel@tonic-gate /* was held in ire_add */ 2775*0Sstevel@tonic-gate ire_refrele(*irep1); 2776*0Sstevel@tonic-gate } 2777*0Sstevel@tonic-gate } 2778*0Sstevel@tonic-gate 2779*0Sstevel@tonic-gate cnt = ipif_saved_ire_cnt; 2780*0Sstevel@tonic-gate for (irep1 = ipif_saved_irep; cnt > 0; irep1++, cnt--) { 2781*0Sstevel@tonic-gate if (*irep1 != NULL) { 2782*0Sstevel@tonic-gate /* was held in ire_add */ 2783*0Sstevel@tonic-gate ire_refrele(*irep1); 2784*0Sstevel@tonic-gate } 2785*0Sstevel@tonic-gate } 2786*0Sstevel@tonic-gate 2787*0Sstevel@tonic-gate 2788*0Sstevel@tonic-gate ip_rts_ifmsg(ipif); 2789*0Sstevel@tonic-gate ip_rts_newaddrmsg(RTM_ADD, 0, ipif); 2790*0Sstevel@tonic-gate if (ipif_saved_irep != NULL) { 2791*0Sstevel@tonic-gate kmem_free(ipif_saved_irep, 2792*0Sstevel@tonic-gate ipif_saved_ire_cnt * sizeof (ire_t *)); 2793*0Sstevel@tonic-gate } 2794*0Sstevel@tonic-gate 2795*0Sstevel@tonic-gate if (src_ipif_held) 2796*0Sstevel@tonic-gate ipif_refrele(src_ipif); 2797*0Sstevel@tonic-gate sctp_update_ipif(ipif, SCTP_IPIF_UP); 2798*0Sstevel@tonic-gate return (0); 2799*0Sstevel@tonic-gate 2800*0Sstevel@tonic-gate bad: 2801*0Sstevel@tonic-gate if (ip6_asp_table_held) 2802*0Sstevel@tonic-gate ip6_asp_table_refrele(); 2803*0Sstevel@tonic-gate /* 2804*0Sstevel@tonic-gate * We don't have to bother removing from ill groups because 2805*0Sstevel@tonic-gate * 2806*0Sstevel@tonic-gate * 1) For groups with names, we insert only when the first ipif 2807*0Sstevel@tonic-gate * comes up. In that case if it fails, it will not be in any 2808*0Sstevel@tonic-gate * group. So, we need not try to remove for that case. 2809*0Sstevel@tonic-gate * 2810*0Sstevel@tonic-gate * 2) For groups without names, either we tried to insert ipif_ill 2811*0Sstevel@tonic-gate * in a group as singleton or found some other group to become 2812*0Sstevel@tonic-gate * a bigger group. For the former, if it fails we don't have 2813*0Sstevel@tonic-gate * anything to do as ipif_ill is not in the group and for the 2814*0Sstevel@tonic-gate * latter, there are no failures in illgrp_insert/illgrp_delete 2815*0Sstevel@tonic-gate * (ENOMEM can't occur for this. Check ifgrp_insert). 2816*0Sstevel@tonic-gate */ 2817*0Sstevel@tonic-gate 2818*0Sstevel@tonic-gate while (irep > ire_array) { 2819*0Sstevel@tonic-gate irep--; 2820*0Sstevel@tonic-gate if (*irep != NULL) { 2821*0Sstevel@tonic-gate ire_delete(*irep); 2822*0Sstevel@tonic-gate if (ire_added) 2823*0Sstevel@tonic-gate ire_refrele(*irep); 2824*0Sstevel@tonic-gate } 2825*0Sstevel@tonic-gate 2826*0Sstevel@tonic-gate } 2827*0Sstevel@tonic-gate (void) ip_srcid_remove(&ipif->ipif_v6lcl_addr, ipif->ipif_zoneid); 2828*0Sstevel@tonic-gate 2829*0Sstevel@tonic-gate if (ipif_saved_irep != NULL) { 2830*0Sstevel@tonic-gate kmem_free(ipif_saved_irep, 2831*0Sstevel@tonic-gate ipif_saved_ire_cnt * sizeof (ire_t *)); 2832*0Sstevel@tonic-gate } 2833*0Sstevel@tonic-gate if (src_ipif_held) 2834*0Sstevel@tonic-gate ipif_refrele(src_ipif); 2835*0Sstevel@tonic-gate 2836*0Sstevel@tonic-gate ipif_ndp_down(ipif); 2837*0Sstevel@tonic-gate if (ipif->ipif_ill->ill_flags & ILLF_XRESOLV) 2838*0Sstevel@tonic-gate ipif_arp_down(ipif); 2839*0Sstevel@tonic-gate 2840*0Sstevel@tonic-gate return (err); 2841*0Sstevel@tonic-gate } 2842*0Sstevel@tonic-gate 2843*0Sstevel@tonic-gate /* 2844*0Sstevel@tonic-gate * Delete an ND entry and the corresponding IRE_CACHE entry if it exists. 2845*0Sstevel@tonic-gate */ 2846*0Sstevel@tonic-gate /* ARGSUSED */ 2847*0Sstevel@tonic-gate int 2848*0Sstevel@tonic-gate ip_siocdelndp_v6(ipif_t *ipif, sin_t *dummy_sin, queue_t *q, mblk_t *mp, 2849*0Sstevel@tonic-gate ip_ioctl_cmd_t *ipip, void *dummy_ifreq) 2850*0Sstevel@tonic-gate { 2851*0Sstevel@tonic-gate in6_addr_t addr; 2852*0Sstevel@tonic-gate sin6_t *sin6; 2853*0Sstevel@tonic-gate nce_t *nce; 2854*0Sstevel@tonic-gate struct lifreq *lifr; 2855*0Sstevel@tonic-gate lif_nd_req_t *lnr; 2856*0Sstevel@tonic-gate mblk_t *mp1; 2857*0Sstevel@tonic-gate 2858*0Sstevel@tonic-gate mp1 = mp->b_cont->b_cont; 2859*0Sstevel@tonic-gate lifr = (struct lifreq *)mp1->b_rptr; 2860*0Sstevel@tonic-gate lnr = &lifr->lifr_nd; 2861*0Sstevel@tonic-gate /* Only allow for logical unit zero i.e. not on "le0:17" */ 2862*0Sstevel@tonic-gate if (ipif->ipif_id != 0) 2863*0Sstevel@tonic-gate return (EINVAL); 2864*0Sstevel@tonic-gate 2865*0Sstevel@tonic-gate if (!ipif->ipif_isv6) 2866*0Sstevel@tonic-gate return (EINVAL); 2867*0Sstevel@tonic-gate 2868*0Sstevel@tonic-gate if (lnr->lnr_addr.ss_family != AF_INET6) 2869*0Sstevel@tonic-gate return (EAFNOSUPPORT); 2870*0Sstevel@tonic-gate 2871*0Sstevel@tonic-gate sin6 = (sin6_t *)&lnr->lnr_addr; 2872*0Sstevel@tonic-gate addr = sin6->sin6_addr; 2873*0Sstevel@tonic-gate nce = ndp_lookup(ipif->ipif_ill, &addr, B_FALSE); 2874*0Sstevel@tonic-gate if (nce == NULL) 2875*0Sstevel@tonic-gate return (ESRCH); 2876*0Sstevel@tonic-gate ndp_delete(nce); 2877*0Sstevel@tonic-gate NCE_REFRELE(nce); 2878*0Sstevel@tonic-gate return (0); 2879*0Sstevel@tonic-gate } 2880*0Sstevel@tonic-gate 2881*0Sstevel@tonic-gate /* 2882*0Sstevel@tonic-gate * Return nbr cache info. 2883*0Sstevel@tonic-gate */ 2884*0Sstevel@tonic-gate /* ARGSUSED */ 2885*0Sstevel@tonic-gate int 2886*0Sstevel@tonic-gate ip_siocqueryndp_v6(ipif_t *ipif, sin_t *dummy_sin, queue_t *q, mblk_t *mp, 2887*0Sstevel@tonic-gate ip_ioctl_cmd_t *ipip, void *dummy_ifreq) 2888*0Sstevel@tonic-gate { 2889*0Sstevel@tonic-gate ill_t *ill = ipif->ipif_ill; 2890*0Sstevel@tonic-gate struct lifreq *lifr; 2891*0Sstevel@tonic-gate lif_nd_req_t *lnr; 2892*0Sstevel@tonic-gate 2893*0Sstevel@tonic-gate lifr = (struct lifreq *)mp->b_cont->b_cont->b_rptr; 2894*0Sstevel@tonic-gate lnr = &lifr->lifr_nd; 2895*0Sstevel@tonic-gate /* Only allow for logical unit zero i.e. not on "le0:17" */ 2896*0Sstevel@tonic-gate if (ipif->ipif_id != 0) 2897*0Sstevel@tonic-gate return (EINVAL); 2898*0Sstevel@tonic-gate 2899*0Sstevel@tonic-gate if (!ipif->ipif_isv6) 2900*0Sstevel@tonic-gate return (EINVAL); 2901*0Sstevel@tonic-gate 2902*0Sstevel@tonic-gate if (lnr->lnr_addr.ss_family != AF_INET6) 2903*0Sstevel@tonic-gate return (EAFNOSUPPORT); 2904*0Sstevel@tonic-gate 2905*0Sstevel@tonic-gate if (ill->ill_phys_addr_length > sizeof (lnr->lnr_hdw_addr)) 2906*0Sstevel@tonic-gate return (EINVAL); 2907*0Sstevel@tonic-gate 2908*0Sstevel@tonic-gate return (ndp_query(ill, lnr)); 2909*0Sstevel@tonic-gate } 2910*0Sstevel@tonic-gate 2911*0Sstevel@tonic-gate /* 2912*0Sstevel@tonic-gate * Perform an update of the nd entry for the specified address. 2913*0Sstevel@tonic-gate */ 2914*0Sstevel@tonic-gate /* ARGSUSED */ 2915*0Sstevel@tonic-gate int 2916*0Sstevel@tonic-gate ip_siocsetndp_v6(ipif_t *ipif, sin_t *dummy_sin, queue_t *q, mblk_t *mp, 2917*0Sstevel@tonic-gate ip_ioctl_cmd_t *ipip, void *dummy_ifreq) 2918*0Sstevel@tonic-gate { 2919*0Sstevel@tonic-gate ill_t *ill = ipif->ipif_ill; 2920*0Sstevel@tonic-gate struct lifreq *lifr; 2921*0Sstevel@tonic-gate lif_nd_req_t *lnr; 2922*0Sstevel@tonic-gate 2923*0Sstevel@tonic-gate lifr = (struct lifreq *)mp->b_cont->b_cont->b_rptr; 2924*0Sstevel@tonic-gate lnr = &lifr->lifr_nd; 2925*0Sstevel@tonic-gate /* Only allow for logical unit zero i.e. not on "le0:17" */ 2926*0Sstevel@tonic-gate if (ipif->ipif_id != 0) 2927*0Sstevel@tonic-gate return (EINVAL); 2928*0Sstevel@tonic-gate 2929*0Sstevel@tonic-gate if (!ipif->ipif_isv6) 2930*0Sstevel@tonic-gate return (EINVAL); 2931*0Sstevel@tonic-gate 2932*0Sstevel@tonic-gate if (lnr->lnr_addr.ss_family != AF_INET6) 2933*0Sstevel@tonic-gate return (EAFNOSUPPORT); 2934*0Sstevel@tonic-gate 2935*0Sstevel@tonic-gate return (ndp_sioc_update(ill, lnr)); 2936*0Sstevel@tonic-gate } 2937