Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files. =================================================================== RCS file: /ftp/cvs/cvsroot/src/sys/netinet6/ip6_input.c,v rcsdiff: /ftp/cvs/cvsroot/src/sys/netinet6/ip6_input.c,v: warning: Unknown phrases like `commitid ...;' are present. retrieving revision 1.139 retrieving revision 1.180 diff -u -p -r1.139 -r1.180 --- src/sys/netinet6/ip6_input.c 2012/06/23 03:14:04 1.139 +++ src/sys/netinet6/ip6_input.c 2017/07/06 17:14:35 1.180 @@ -1,4 +1,4 @@ -/* $NetBSD: ip6_input.c,v 1.139 2012/06/23 03:14:04 christos Exp $ */ +/* $NetBSD: ip6_input.c,v 1.180 2017/07/06 17:14:35 christos Exp $ */ /* $KAME: ip6_input.c,v 1.188 2001/03/29 05:34:31 itojun Exp $ */ /* @@ -62,18 +62,18 @@ */ #include -__KERNEL_RCSID(0, "$NetBSD: ip6_input.c,v 1.139 2012/06/23 03:14:04 christos Exp $"); +__KERNEL_RCSID(0, "$NetBSD: ip6_input.c,v 1.180 2017/07/06 17:14:35 christos Exp $"); +#ifdef _KERNEL_OPT #include "opt_gateway.h" #include "opt_inet.h" #include "opt_inet6.h" #include "opt_ipsec.h" -#include "opt_pfil_hooks.h" -#include "opt_compat_netbsd.h" +#include "opt_net_mpsafe.h" +#endif #include #include -#include #include #include #include @@ -86,24 +86,24 @@ __KERNEL_RCSID(0, "$NetBSD: ip6_input.c, #include #include #include +#include #include #include #include #include -#include -#ifdef PFIL_HOOKS +#include #include -#endif #include #include #ifdef INET #include +#include #include #endif /* INET */ #include -#include +#include #include #include #include @@ -113,48 +113,35 @@ __KERNEL_RCSID(0, "$NetBSD: ip6_input.c, #include #include -#ifdef FAST_IPSEC +#ifdef IPSEC #include #include #include -#endif /* FAST_IPSEC */ - -#ifdef COMPAT_50 -#include -#include -#endif +#endif /* IPSEC */ #include #include "faith.h" -#include "gif.h" - -#if NGIF > 0 -#include -#endif #include extern struct domain inet6domain; u_char ip6_protox[IPPROTO_MAX]; -static int ip6qmaxlen = IFQ_MAXLEN; -struct in6_ifaddr *in6_ifaddr; -struct ifqueue ip6intrq; - -extern callout_t in6_tmpaddrtimer_ch; +pktqueue_t *ip6_pktq __read_mostly; int ip6_forward_srcrt; /* XXX */ int ip6_sourcecheck; /* XXX */ int ip6_sourcecheck_interval; /* XXX */ -#ifdef PFIL_HOOKS -struct pfil_head inet6_pfil_hook; -#endif +pfil_head_t *inet6_pfil_hook; percpu_t *ip6stat_percpu; -static void ip6_init2(void *); +percpu_t *ip6_forward_rt_percpu __cacheline_aligned; + +static void ip6_init2(void); +static void ip6intr(void *); static struct m_tag *ip6_setdstifaddr(struct mbuf *, const struct in6_ifaddr *); static int ip6_process_hopopts(struct mbuf *, u_int8_t *, int, u_int32_t *, @@ -162,6 +149,14 @@ static int ip6_process_hopopts(struct mb static struct mbuf *ip6_pullexthdr(struct mbuf *, size_t, int); static void sysctl_net_inet6_ip6_setup(struct sysctllog **); +#ifdef NET_MPSAFE +#define SOFTNET_LOCK() mutex_enter(softnet_lock) +#define SOFTNET_UNLOCK() mutex_exit(softnet_lock) +#else +#define SOFTNET_LOCK() KASSERT(mutex_owned(softnet_lock)) +#define SOFTNET_UNLOCK() KASSERT(mutex_owned(softnet_lock)) +#endif + /* * IP6 initialization: fill in IP6 protocol switch table. * All protocols not implemented in kernel go to raw IP6 protocol handler. @@ -172,6 +167,8 @@ ip6_init(void) const struct ip6protosw *pr; int i; + in6_init(); + sysctl_net_inet6_ip6_setup(NULL); pr = (const struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW); if (pr == 0) @@ -183,39 +180,32 @@ ip6_init(void) if (pr->pr_domain->dom_family == PF_INET6 && pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) ip6_protox[pr->pr_protocol] = pr - inet6sw; - ip6intrq.ifq_maxlen = ip6qmaxlen; + + ip6_pktq = pktq_create(IFQ_MAXLEN, ip6intr, NULL); + KASSERT(ip6_pktq != NULL); + scope6_init(); addrsel_policy_init(); nd6_init(); frag6_init(); ip6_desync_factor = cprng_fast32() % MAX_TEMP_DESYNC_FACTOR; - ip6_init2(NULL); + ip6_init2(); #ifdef GATEWAY ip6flow_init(ip6_hashsize); #endif - -#ifdef PFIL_HOOKS /* Register our Packet Filter hook. */ - inet6_pfil_hook.ph_type = PFIL_TYPE_AF; - inet6_pfil_hook.ph_af = AF_INET6; - i = pfil_head_register(&inet6_pfil_hook); - if (i != 0) - printf("ip6_init: WARNING: unable to register pfil hook, " - "error %d\n", i); -#endif /* PFIL_HOOKS */ + inet6_pfil_hook = pfil_head_create(PFIL_TYPE_AF, (void *)AF_INET6); + KASSERT(inet6_pfil_hook != NULL); ip6stat_percpu = percpu_alloc(sizeof(uint64_t) * IP6_NSTATS); + ip6_forward_rt_percpu = percpu_alloc(sizeof(struct route)); } static void -ip6_init2(void *dummy) +ip6_init2(void) { - /* nd6_timer_init */ - callout_init(&nd6_timer_ch, CALLOUT_MPSAFE); - callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL); - /* timer for regeneranation of temporary addresses randomize ID */ callout_init(&in6_tmpaddrtimer_ch, CALLOUT_MPSAFE); callout_reset(&in6_tmpaddrtimer_ch, @@ -227,35 +217,40 @@ ip6_init2(void *dummy) /* * IP6 input interrupt handling. Just pass the packet to ip6_input. */ -void -ip6intr(void) +static void +ip6intr(void *arg __unused) { - int s; struct mbuf *m; +#ifndef NET_MPSAFE mutex_enter(softnet_lock); - KERNEL_LOCK(1, NULL); - for (;;) { - s = splnet(); - IF_DEQUEUE(&ip6intrq, m); - splx(s); - if (m == 0) - break; - /* drop the packet if IPv6 operation is disabled on the IF */ - if ((ND_IFINFO(m->m_pkthdr.rcvif)->flags & ND6_IFF_IFDISABLED)) { +#endif + while ((m = pktq_dequeue(ip6_pktq)) != NULL) { + struct psref psref; + struct ifnet *rcvif = m_get_rcvif_psref(m, &psref); + + if (rcvif == NULL) { m_freem(m); - break; + continue; + } + /* + * Drop the packet if IPv6 is disabled on the interface. + */ + if ((ND_IFINFO(rcvif)->flags & ND6_IFF_IFDISABLED)) { + m_put_rcvif_psref(rcvif, &psref); + m_freem(m); + continue; } - ip6_input(m); + ip6_input(m, rcvif); + m_put_rcvif_psref(rcvif, &psref); } - KERNEL_UNLOCK_ONE(NULL); +#ifndef NET_MPSAFE mutex_exit(softnet_lock); +#endif } -extern struct route ip6_forward_rt; - void -ip6_input(struct mbuf *m) +ip6_input(struct mbuf *m, struct ifnet *rcvif) { struct ip6_hdr *ip6; int hit, off = sizeof(struct ip6_hdr), nest; @@ -264,17 +259,12 @@ ip6_input(struct mbuf *m) int nxt, ours = 0, rh_present = 0; struct ifnet *deliverifp = NULL; int srcrt = 0; - const struct rtentry *rt; + struct rtentry *rt = NULL; union { struct sockaddr dst; struct sockaddr_in6 dst6; } u; -#ifdef FAST_IPSEC - struct m_tag *mtag; - struct tdb_ident *tdbi; - struct secpolicy *sp; - int s, error; -#endif + struct route *ro; /* * make sure we don't have onion peering information into m_tag. @@ -292,19 +282,18 @@ ip6_input(struct mbuf *m) } else { #define M2MMAX 32 if (m->m_next) { - if (m->m_flags & M_LOOP) { + if (m->m_flags & M_LOOP) /*XXX*/ IP6_STATINC(IP6_STAT_M2M + lo0ifp->if_index); - } else if (m->m_pkthdr.rcvif->if_index < M2MMAX) { - IP6_STATINC(IP6_STAT_M2M + - m->m_pkthdr.rcvif->if_index); - } else + else if (rcvif->if_index < M2MMAX) + IP6_STATINC(IP6_STAT_M2M + rcvif->if_index); + else IP6_STATINC(IP6_STAT_M2M); } else IP6_STATINC(IP6_STAT_M1); #undef M2MMAX } - in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive); + in6_ifstat_inc(rcvif, ifs6_in_receive); IP6_STATINC(IP6_STAT_TOTAL); /* @@ -314,19 +303,17 @@ ip6_input(struct mbuf *m) * IPv6 header is in the first mbuf of the chain. */ if (IP6_HDR_ALIGNED_P(mtod(m, void *)) == 0) { - struct ifnet *inifp = m->m_pkthdr.rcvif; if ((m = m_copyup(m, sizeof(struct ip6_hdr), (max_linkhdr + 3) & ~3)) == NULL) { /* XXXJRT new stat, please */ IP6_STATINC(IP6_STAT_TOOSMALL); - in6_ifstat_inc(inifp, ifs6_in_hdrerr); + in6_ifstat_inc(rcvif, ifs6_in_hdrerr); return; } } else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) { - struct ifnet *inifp = m->m_pkthdr.rcvif; if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) { IP6_STATINC(IP6_STAT_TOOSMALL); - in6_ifstat_inc(inifp, ifs6_in_hdrerr); + in6_ifstat_inc(rcvif, ifs6_in_hdrerr); return; } } @@ -335,7 +322,7 @@ ip6_input(struct mbuf *m) if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { IP6_STATINC(IP6_STAT_BADVERS); - in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); + in6_ifstat_inc(rcvif, ifs6_in_hdrerr); goto bad; } @@ -345,7 +332,6 @@ ip6_input(struct mbuf *m) */ m->m_flags |= M_CANFASTFWD; -#ifdef PFIL_HOOKS /* * Run through list of hooks for input packets. If there are any * filters which require that additional packets in the flow are @@ -357,8 +343,8 @@ ip6_input(struct mbuf *m) * let ipfilter look at packet on the wire, * not the decapsulated packet. */ -#if defined(FAST_IPSEC) - if (!ipsec_indone(m)) +#if defined(IPSEC) + if (!ipsec_used || !ipsec_indone(m)) #else if (1) #endif @@ -366,22 +352,25 @@ ip6_input(struct mbuf *m) struct in6_addr odst; odst = ip6->ip6_dst; - if (pfil_run_hooks(&inet6_pfil_hook, &m, m->m_pkthdr.rcvif, - PFIL_IN) != 0) + if (pfil_run_hooks(inet6_pfil_hook, &m, rcvif, PFIL_IN) != 0) return; if (m == NULL) return; ip6 = mtod(m, struct ip6_hdr *); srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst); } -#endif /* PFIL_HOOKS */ IP6_STATINC(IP6_STAT_NXTHIST + ip6->ip6_nxt); #ifdef ALTQ - if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) { - /* packet is dropped by traffic conditioner */ - return; + if (altq_input != NULL) { + SOFTNET_LOCK(); + if ((*altq_input)(m, AF_INET6) == 0) { + SOFTNET_UNLOCK(); + /* packet is dropped by traffic conditioner */ + return; + } + SOFTNET_UNLOCK(); } #endif @@ -394,7 +383,7 @@ ip6_input(struct mbuf *m) * XXX: "badscope" is not very suitable for a multicast source. */ IP6_STATINC(IP6_STAT_BADSCOPE); - in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); + in6_ifstat_inc(rcvif, ifs6_in_addrerr); goto bad; } /* @@ -412,7 +401,7 @@ ip6_input(struct mbuf *m) if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { IP6_STATINC(IP6_STAT_BADSCOPE); - in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); + in6_ifstat_inc(rcvif, ifs6_in_addrerr); goto bad; } #if 0 @@ -426,7 +415,7 @@ ip6_input(struct mbuf *m) if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) || IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) { IP6_STATINC(IP6_STAT_BADSCOPE); - in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); + in6_ifstat_inc(rcvif, ifs6_in_addrerr); goto bad; } #endif @@ -449,35 +438,36 @@ ip6_input(struct mbuf *m) IP6_STATINC(IP6_STAT_BADSCOPE); /* XXX */ goto bad; } - if (in6_setscope(&ip6->ip6_src, m->m_pkthdr.rcvif, NULL) || - in6_setscope(&ip6->ip6_dst, m->m_pkthdr.rcvif, NULL)) { + if (in6_setscope(&ip6->ip6_src, rcvif, NULL) || + in6_setscope(&ip6->ip6_dst, rcvif, NULL)) { IP6_STATINC(IP6_STAT_BADSCOPE); goto bad; } + ro = percpu_getref(ip6_forward_rt_percpu); /* * Multicast check */ if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { - struct in6_multi *in6m = 0; + bool ingroup; - in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast); + in6_ifstat_inc(rcvif, ifs6_in_mcast); /* * See if we belong to the destination multicast group on the * arrival interface. */ - IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m); - if (in6m) + ingroup = in6_multi_group(&ip6->ip6_dst, rcvif); + if (ingroup) ours = 1; else if (!ip6_mrouter) { uint64_t *ip6s = IP6_STAT_GETREF(); ip6s[IP6_STAT_NOTMEMBER]++; ip6s[IP6_STAT_CANTFORWARD]++; IP6_STAT_PUTREF(); - in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); - goto bad; + in6_ifstat_inc(rcvif, ifs6_in_discard); + goto bad_unref; } - deliverifp = m->m_pkthdr.rcvif; + deliverifp = rcvif; goto hbhcheck; } @@ -486,7 +476,7 @@ ip6_input(struct mbuf *m) /* * Unicast check */ - rt = rtcache_lookup2(&ip6_forward_rt, &u.dst, 1, &hit); + rt = rtcache_lookup2(ro, &u.dst, 1, &hit); if (hit) IP6_STATINC(IP6_STAT_FORWARD_CACHEHIT); else @@ -505,7 +495,6 @@ ip6_input(struct mbuf *m) */ if (rt != NULL && (rt->rt_flags & (RTF_HOST|RTF_GATEWAY)) == RTF_HOST && - !(rt->rt_flags & RTF_CLONED) && #if 0 /* * The check below is redundant since the comparison of @@ -522,19 +511,20 @@ ip6_input(struct mbuf *m) * packets to a tentative, duplicated, or somehow invalid * address must not be accepted. */ - if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) { + if (!(ia6->ia6_flags & (IN6_IFF_NOTREADY | IN6_IFF_DETACHED))) { /* this address is ready */ ours = 1; deliverifp = ia6->ia_ifp; /* correct? */ goto hbhcheck; } else { /* address is not ready, so discard the packet. */ - nd6log((LOG_INFO, - "ip6_input: packet to an unready address %s->%s\n", - ip6_sprintf(&ip6->ip6_src), - ip6_sprintf(&ip6->ip6_dst))); + char ip6bufs[INET6_ADDRSTRLEN]; + char ip6bufd[INET6_ADDRSTRLEN]; + nd6log(LOG_INFO, "packet to an unready address %s->%s\n", + IN6_PRINT(ip6bufs, &ip6->ip6_src), + IN6_PRINT(ip6bufd, &ip6->ip6_dst)); - goto bad; + goto bad_unref; } } @@ -561,9 +551,7 @@ ip6_input(struct mbuf *m) * working right. */ struct ifaddr *ifa; - IFADDR_FOREACH(ifa, m->m_pkthdr.rcvif) { - if (ifa->ifa_addr == NULL) - continue; /* just for safety */ + IFADDR_READER_FOREACH(ifa, rcvif) { if (ifa->ifa_addr->sa_family != AF_INET6) continue; if (IN6_ARE_ADDR_EQUAL(IFA_IN6(ifa), &ip6->ip6_dst)) { @@ -581,8 +569,8 @@ ip6_input(struct mbuf *m) */ if (!ip6_forwarding) { IP6_STATINC(IP6_STAT_CANTFORWARD); - in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); - goto bad; + in6_ifstat_inc(rcvif, ifs6_in_discard); + goto bad_unref; } hbhcheck: @@ -594,8 +582,10 @@ ip6_input(struct mbuf *m) */ if (deliverifp && ip6_getdstifaddr(m) == NULL) { struct in6_ifaddr *ia6; + int s = pserialize_read_enter(); ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst); + /* Depends on ip6_setdstifaddr never sleep */ if (ia6 != NULL && ip6_setdstifaddr(m, ia6) == NULL) { /* * XXX maybe we should drop the packet here, @@ -603,6 +593,7 @@ ip6_input(struct mbuf *m) * to the upper layers. */ } + pserialize_read_exit(s); } /* @@ -616,8 +607,10 @@ ip6_input(struct mbuf *m) if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) { #if 0 /*touches NULL pointer*/ - in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); + in6_ifstat_inc(rcvif, ifs6_in_discard); #endif + rtcache_unref(rt, ro); + percpu_putref(ip6_forward_rt_percpu); return; /* m have already been freed */ } @@ -636,17 +629,21 @@ ip6_input(struct mbuf *m) * (non-zero) payload length to the variable plen. */ IP6_STATINC(IP6_STAT_BADOPTIONS); - in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard); - in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); + in6_ifstat_inc(rcvif, ifs6_in_discard); + in6_ifstat_inc(rcvif, ifs6_in_hdrerr); icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_HEADER, (char *)&ip6->ip6_plen - (char *)ip6); + rtcache_unref(rt, ro); + percpu_putref(ip6_forward_rt_percpu); return; } IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), sizeof(struct ip6_hbh)); if (hbh == NULL) { IP6_STATINC(IP6_STAT_TOOSHORT); + rtcache_unref(rt, ro); + percpu_putref(ip6_forward_rt_percpu); return; } KASSERT(IP6_HDR_ALIGNED_P(hbh)); @@ -669,8 +666,8 @@ ip6_input(struct mbuf *m) */ if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) { IP6_STATINC(IP6_STAT_TOOSHORT); - in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated); - goto bad; + in6_ifstat_inc(rcvif, ifs6_in_truncated); + goto bad_unref; } if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) { if (m->m_len == m->m_pkthdr.len) { @@ -692,16 +689,25 @@ ip6_input(struct mbuf *m) * ip6_mforward() returns a non-zero value, the packet * must be discarded, else it may be accepted below. */ - if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) { - IP6_STATINC(IP6_STAT_CANTFORWARD); - m_freem(m); - return; - } - if (!ours) { - m_freem(m); - return; + if (ip6_mrouter != NULL) { + int error; + + SOFTNET_LOCK(); + error = ip6_mforward(ip6, rcvif, m); + SOFTNET_UNLOCK(); + + if (error != 0) { + rtcache_unref(rt, ro); + percpu_putref(ip6_forward_rt_percpu); + IP6_STATINC(IP6_STAT_CANTFORWARD); + goto bad; + } } + if (!ours) + goto bad_unref; } else if (!ours) { + rtcache_unref(rt, ro); + percpu_putref(ip6_forward_rt_percpu); ip6_forward(m, srcrt); return; } @@ -720,8 +726,8 @@ ip6_input(struct mbuf *m) if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { IP6_STATINC(IP6_STAT_BADSCOPE); - in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr); - goto bad; + in6_ifstat_inc(rcvif, ifs6_in_addrerr); + goto bad_unref; } /* @@ -730,20 +736,28 @@ ip6_input(struct mbuf *m) #ifdef IFA_STATS if (deliverifp != NULL) { struct in6_ifaddr *ia6; + int s = pserialize_read_enter(); ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst); if (ia6) ia6->ia_ifa.ifa_data.ifad_inbytes += m->m_pkthdr.len; + pserialize_read_exit(s); } #endif IP6_STATINC(IP6_STAT_DELIVERED); in6_ifstat_inc(deliverifp, ifs6_in_deliver); nest = 0; + if (rt != NULL) { + rtcache_unref(rt, ro); + rt = NULL; + } + percpu_putref(ip6_forward_rt_percpu); + rh_present = 0; while (nxt != IPPROTO_DONE) { if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) { IP6_STATINC(IP6_STAT_TOOMANYHDR); - in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); + in6_ifstat_inc(rcvif, ifs6_in_hdrerr); goto bad; } @@ -753,64 +767,50 @@ ip6_input(struct mbuf *m) */ if (m->m_pkthdr.len < off) { IP6_STATINC(IP6_STAT_TOOSHORT); - in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated); + in6_ifstat_inc(rcvif, ifs6_in_truncated); goto bad; } if (nxt == IPPROTO_ROUTING) { if (rh_present++) { - in6_ifstat_inc(m->m_pkthdr.rcvif, - ifs6_in_hdrerr); + in6_ifstat_inc(rcvif, ifs6_in_hdrerr); IP6_STATINC(IP6_STAT_BADOPTIONS); goto bad; } } -#ifdef FAST_IPSEC - /* - * enforce IPsec policy checking if we are seeing last header. - * note that we do not visit this with protocols with pcb layer - * code - like udp/tcp/raw ip. - */ - if ((inet6sw[ip_protox[nxt]].pr_flags & PR_LASTHDR) != 0) { - /* - * Check if the packet has already had IPsec processing - * done. If so, then just pass it along. This tag gets - * set during AH, ESP, etc. input handling, before the - * packet is returned to the ip input queue for delivery. - */ - mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL); - s = splsoftnet(); - if (mtag != NULL) { - tdbi = (struct tdb_ident *)(mtag + 1); - sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND); - } else { - sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND, - IP_FORWARDING, &error); - } - if (sp != NULL) { +#ifdef IPSEC + if (ipsec_used) { /* - * Check security policy against packet attributes. + * enforce IPsec policy checking if we are seeing last + * header. note that we do not visit this with + * protocols with pcb layer code - like udp/tcp/raw ip. */ - error = ipsec_in_reject(sp, m); - KEY_FREESP(&sp); - } else { - /* XXX error stat??? */ - error = EINVAL; - DPRINTF(("ip6_input: no SP, packet discarded\n"));/*XXX*/ + if ((inet6sw[ip_protox[nxt]].pr_flags + & PR_LASTHDR) != 0) { + int error; + + SOFTNET_LOCK(); + error = ipsec6_input(m); + SOFTNET_UNLOCK(); + if (error) + goto bad; + } } - splx(s); - if (error) - goto bad; - } -#endif /* FAST_IPSEC */ - +#endif /* IPSEC */ + SOFTNET_LOCK(); nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt); + SOFTNET_UNLOCK(); } return; + + bad_unref: + rtcache_unref(rt, ro); + percpu_putref(ip6_forward_rt_percpu); bad: m_freem(m); + return; } /* @@ -820,17 +820,21 @@ static struct m_tag * ip6_setdstifaddr(struct mbuf *m, const struct in6_ifaddr *ia) { struct m_tag *mtag; + struct ip6aux *ip6a; mtag = ip6_addaux(m); - if (mtag != NULL) { - struct ip6aux *ip6a; + if (mtag == NULL) + return NULL; - ip6a = (struct ip6aux *)(mtag + 1); - in6_setscope(&ip6a->ip6a_src, ia->ia_ifp, &ip6a->ip6a_scope_id); - ip6a->ip6a_src = ia->ia_addr.sin6_addr; - ip6a->ip6a_flags = ia->ia6_flags; + ip6a = (struct ip6aux *)(mtag + 1); + if (in6_setscope(&ip6a->ip6a_src, ia->ia_ifp, &ip6a->ip6a_scope_id)) { + IP6_STATINC(IP6_STAT_BADSCOPE); + return NULL; } - return mtag; /* NULL if failed to set */ + + ip6a->ip6a_src = ia->ia_addr.sin6_addr; + ip6a->ip6a_flags = ia->ia6_flags; + return mtag; } const struct ip6aux * @@ -1075,33 +1079,15 @@ void ip6_savecontrol(struct in6pcb *in6p, struct mbuf **mp, struct ip6_hdr *ip6, struct mbuf *m) { + struct socket *so = in6p->in6p_socket; #ifdef RFC2292 #define IS2292(x, y) ((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y)) #else #define IS2292(x, y) (y) #endif - if (in6p->in6p_socket->so_options & SO_TIMESTAMP -#ifdef SO_OTIMESTAMP - || in6p->in6p_socket->so_options & SO_OTIMESTAMP -#endif - ) { - struct timeval tv; - - microtime(&tv); -#ifdef SO_OTIMESTAMP - if (in6p->in6p_socket->so_options & SO_OTIMESTAMP) { - struct timeval50 tv50; - timeval_to_timeval50(&tv, &tv50); - *mp = sbcreatecontrol((void *) &tv50, sizeof(tv50), - SCM_OTIMESTAMP, SOL_SOCKET); - } else -#endif - *mp = sbcreatecontrol((void *) &tv, sizeof(tv), - SCM_TIMESTAMP, SOL_SOCKET); - if (*mp) - mp = &(*mp)->m_next; - } + if (SOOPT_TIMESTAMP(so->so_options)) + mp = sbsavetimestamp(so->so_options, m, mp); /* some OSes call this logic with IPv4 packet, for SO_TIMESTAMP */ if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) @@ -1113,10 +1099,8 @@ ip6_savecontrol(struct in6pcb *in6p, str memcpy(&pi6.ipi6_addr, &ip6->ip6_dst, sizeof(struct in6_addr)); in6_clearscope(&pi6.ipi6_addr); /* XXX */ - pi6.ipi6_ifindex = m->m_pkthdr.rcvif ? - m->m_pkthdr.rcvif->if_index : 0; - *mp = sbcreatecontrol((void *) &pi6, - sizeof(struct in6_pktinfo), + pi6.ipi6_ifindex = m->m_pkthdr.rcvif_index; + *mp = sbcreatecontrol(&pi6, sizeof(pi6), IS2292(IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6); if (*mp) mp = &(*mp)->m_next; @@ -1125,7 +1109,7 @@ ip6_savecontrol(struct in6pcb *in6p, str if (in6p->in6p_flags & IN6P_HOPLIMIT) { int hlim = ip6->ip6_hlim & 0xff; - *mp = sbcreatecontrol((void *) &hlim, sizeof(int), + *mp = sbcreatecontrol(&hlim, sizeof(hlim), IS2292(IPV6_2292HOPLIMIT, IPV6_HOPLIMIT), IPPROTO_IPV6); if (*mp) mp = &(*mp)->m_next; @@ -1139,7 +1123,7 @@ ip6_savecontrol(struct in6pcb *in6p, str flowinfo >>= 20; tclass = flowinfo & 0xff; - *mp = sbcreatecontrol((void *)&tclass, sizeof(tclass), + *mp = sbcreatecontrol(&tclass, sizeof(tclass), IPV6_TCLASS, IPPROTO_IPV6); if (*mp) @@ -1187,7 +1171,7 @@ ip6_savecontrol(struct in6pcb *in6p, str * be removed before returning in the RFC 2292. * Note: this constraint is removed in RFC3542. */ - *mp = sbcreatecontrol((void *)hbh, hbhlen, + *mp = sbcreatecontrol(hbh, hbhlen, IS2292(IPV6_2292HOPOPTS, IPV6_HOPOPTS), IPPROTO_IPV6); if (*mp) @@ -1249,7 +1233,7 @@ ip6_savecontrol(struct in6pcb *in6p, str if (!(in6p->in6p_flags & IN6P_DSTOPTS)) break; - *mp = sbcreatecontrol((void *)ip6e, elen, + *mp = sbcreatecontrol(ip6e, elen, IS2292(IPV6_2292DSTOPTS, IPV6_DSTOPTS), IPPROTO_IPV6); if (*mp) @@ -1260,7 +1244,7 @@ ip6_savecontrol(struct in6pcb *in6p, str if (!(in6p->in6p_flags & IN6P_RTHDR)) break; - *mp = sbcreatecontrol((void *)ip6e, elen, + *mp = sbcreatecontrol(ip6e, elen, IS2292(IPV6_2292RTHDR, IPV6_RTHDR), IPPROTO_IPV6); if (*mp) @@ -1310,10 +1294,7 @@ ip6_notify_pmtu(struct in6pcb *in6p, con if (mtu == NULL) return; -#ifdef DIAGNOSTIC - if (so == NULL) /* I believe this is impossible */ - panic("ip6_notify_pmtu: socket is NULL"); -#endif + KASSERT(so != NULL); memset(&mtuctl, 0, sizeof(mtuctl)); /* zero-clear for safety */ mtuctl.ip6m_mtu = *mtu; @@ -1321,7 +1302,7 @@ ip6_notify_pmtu(struct in6pcb *in6p, con if (sa6_recoverscope(&mtuctl.ip6m_addr)) return; - if ((m_mtu = sbcreatecontrol((void *)&mtuctl, sizeof(mtuctl), + if ((m_mtu = sbcreatecontrol(&mtuctl, sizeof(mtuctl), IPV6_PATHMTU, IPPROTO_IPV6)) == NULL) return; @@ -1576,66 +1557,6 @@ ip6_delaux(struct mbuf *m) m_tag_delete(m, mtag); } -#ifdef GATEWAY -/* - * sysctl helper routine for net.inet.ip6.maxflows. Since - * we could reduce this value, call ip6flow_reap(); - */ -static int -sysctl_net_inet6_ip6_maxflows(SYSCTLFN_ARGS) -{ - int error; - - error = sysctl_lookup(SYSCTLFN_CALL(rnode)); - if (error || newp == NULL) - return (error); - - mutex_enter(softnet_lock); - KERNEL_LOCK(1, NULL); - - ip6flow_reap(0); - - KERNEL_UNLOCK_ONE(NULL); - mutex_exit(softnet_lock); - - return (0); -} - -static int -sysctl_net_inet6_ip6_hashsize(SYSCTLFN_ARGS) -{ - int error, tmp; - struct sysctlnode node; - - node = *rnode; - tmp = ip6_hashsize; - node.sysctl_data = &tmp; - error = sysctl_lookup(SYSCTLFN_CALL(&node)); - if (error || newp == NULL) - return (error); - - if ((tmp & (tmp - 1)) == 0 && tmp != 0) { - /* - * Can only fail due to malloc() - */ - mutex_enter(softnet_lock); - KERNEL_LOCK(1, NULL); - - error = ip6flow_invalidate_all(tmp); - - KERNEL_UNLOCK_ONE(NULL); - mutex_exit(softnet_lock); - } else { - /* - * EINVAL if not a power of 2 - */ - error = EINVAL; - } - - return error; -} -#endif /* GATEWAY */ - /* * System control for IP6 */ @@ -1649,6 +1570,8 @@ const u_char inet6ctlerrmap[PRC_NCMDS] = ENOPROTOOPT }; +extern int sysctl_net_inet6_addrctlpolicy(SYSCTLFN_ARGS); + static int sysctl_net_inet6_ip6_stats(SYSCTLFN_ARGS) { @@ -1667,11 +1590,6 @@ sysctl_net_inet6_ip6_setup(struct sysctl sysctl_createv(clog, 0, NULL, NULL, CTLFLAG_PERMANENT, - CTLTYPE_NODE, "net", NULL, - NULL, 0, NULL, 0, - CTL_NET, CTL_EOL); - sysctl_createv(clog, 0, NULL, NULL, - CTLFLAG_PERMANENT, CTLTYPE_NODE, "inet6", SYSCTL_DESCR("PF_INET6 related settings"), NULL, 0, NULL, 0, @@ -1819,15 +1737,6 @@ sysctl_net_inet6_ip6_setup(struct sysctl NULL, 0, &ip6_defmcasthlim, 0, CTL_NET, PF_INET6, IPPROTO_IPV6, IPV6CTL_DEFMCASTHLIM, CTL_EOL); -#if NGIF > 0 - sysctl_createv(clog, 0, NULL, NULL, - CTLFLAG_PERMANENT|CTLFLAG_READWRITE, - CTLTYPE_INT, "gifhlim", - SYSCTL_DESCR("Default hop limit for a gif tunnel datagram"), - NULL, 0, &ip6_gif_hlim, 0, - CTL_NET, PF_INET6, IPPROTO_IPV6, - IPV6CTL_GIF_HLIM, CTL_EOL); -#endif /* NGIF */ sysctl_createv(clog, 0, NULL, NULL, CTLFLAG_PERMANENT, CTLTYPE_STRING, "kame_version", @@ -1894,6 +1803,23 @@ sysctl_net_inet6_ip6_setup(struct sysctl #endif /* IPNOPRIVPORTS */ sysctl_createv(clog, 0, NULL, NULL, CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "auto_linklocal", + SYSCTL_DESCR("Default value of per-interface flag for " + "adding an IPv6 link-local address to " + "interfaces when attached"), + NULL, 0, &ip6_auto_linklocal, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_AUTO_LINKLOCAL, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READONLY, + CTLTYPE_STRUCT, "addctlpolicy", + SYSCTL_DESCR("Return the current address control" + " policy"), + sysctl_net_inet6_addrctlpolicy, 0, NULL, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + IPV6CTL_ADDRCTLPOLICY, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, CTLTYPE_INT, "use_tempaddr", SYSCTL_DESCR("Use temporary address"), NULL, 0, &ip6_use_tempaddr, 0, @@ -1901,6 +1827,14 @@ sysctl_net_inet6_ip6_setup(struct sysctl CTL_CREATE, CTL_EOL); sysctl_createv(clog, 0, NULL, NULL, CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "prefer_tempaddr", + SYSCTL_DESCR("Prefer temporary address as source " + "address"), + NULL, 0, &ip6_prefer_tempaddr, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + CTL_CREATE, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, CTLTYPE_INT, "temppltime", SYSCTL_DESCR("preferred lifetime of a temporary address"), NULL, 0, &ip6_temp_preferred_lifetime, 0, @@ -1941,41 +1875,31 @@ sysctl_net_inet6_ip6_setup(struct sysctl NULL, 0, &ip6_mcast_pmtu, 0, CTL_NET, PF_INET6, IPPROTO_IPV6, CTL_CREATE, CTL_EOL); -#ifdef GATEWAY - sysctl_createv(clog, 0, NULL, NULL, - CTLFLAG_PERMANENT|CTLFLAG_READWRITE, - CTLTYPE_INT, "maxflows", - SYSCTL_DESCR("Number of flows for fast forwarding (IPv6)"), - sysctl_net_inet6_ip6_maxflows, 0, &ip6_maxflows, 0, - CTL_NET, PF_INET6, IPPROTO_IPV6, - CTL_CREATE, CTL_EOL); - sysctl_createv(clog, 0, NULL, NULL, - CTLFLAG_PERMANENT|CTLFLAG_READWRITE, - CTLTYPE_INT, "hashsize", - SYSCTL_DESCR("Size of hash table for fast forwarding (IPv6)"), - sysctl_net_inet6_ip6_hashsize, 0, &ip6_hashsize, 0, - CTL_NET, PF_INET6, IPPROTO_IPV6, - CTL_CREATE, CTL_EOL); -#endif /* anonportalgo RFC6056 subtree */ - const struct sysctlnode *rfc6056_node; - sysctl_createv(clog, 0, NULL, &rfc6056_node, + const struct sysctlnode *portalgo_node; + sysctl_createv(clog, 0, NULL, &portalgo_node, CTLFLAG_PERMANENT, CTLTYPE_NODE, "anonportalgo", SYSCTL_DESCR("Anonymous port algorithm selection (RFC 6056)"), NULL, 0, NULL, 0, CTL_NET, PF_INET6, IPPROTO_IPV6, CTL_CREATE, CTL_EOL); - sysctl_createv(clog, 0, &rfc6056_node, NULL, + sysctl_createv(clog, 0, &portalgo_node, NULL, CTLFLAG_PERMANENT, CTLTYPE_STRING, "available", SYSCTL_DESCR("available algorithms"), - sysctl_rfc6056_available, 0, NULL, RFC6056_MAXLEN, + sysctl_portalgo_available, 0, NULL, PORTALGO_MAXLEN, CTL_CREATE, CTL_EOL); - sysctl_createv(clog, 0, &rfc6056_node, NULL, + sysctl_createv(clog, 0, &portalgo_node, NULL, CTLFLAG_PERMANENT|CTLFLAG_READWRITE, CTLTYPE_STRING, "selected", SYSCTL_DESCR("selected algorithm"), - sysctl_rfc6056_selected6, 0, NULL, RFC6056_MAXLEN, + sysctl_portalgo_selected6, 0, NULL, PORTALGO_MAXLEN, + CTL_CREATE, CTL_EOL); + sysctl_createv(clog, 0, &portalgo_node, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_STRUCT, "reserve", + SYSCTL_DESCR("bitmap of reserved ports"), + sysctl_portalgo_reserve6, 0, NULL, 0, CTL_CREATE, CTL_EOL); sysctl_createv(clog, 0, NULL, NULL, CTLFLAG_PERMANENT|CTLFLAG_READWRITE,