Index: projects/routing/sys/net/rt_nhops.c =================================================================== --- projects/routing/sys/net/rt_nhops.c (revision 287051) +++ projects/routing/sys/net/rt_nhops.c (revision 287052) @@ -1,631 +1,635 @@ /*- * Copyright (c) 2014 * Alexander V. Chernikov * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ /* * Temporary file. In future it should be split between net/route.c * and per-AF files like netinet/in_rmx.c | netinet6/in6_rmx.c */ #include "opt_inet.h" #include "opt_inet6.h" #include "opt_route.h" #include "opt_mpath.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef RADIX_MPATH #include #endif #include #include #include #include #include #include #include #include #include struct fwd_info { fib_lookup_t *lookup; void *state; }; #define FWD_FSM_NONE 0 #define FWD_FSM_INIT 1 #define FWD_FSM_FWD 2 struct fwd_control { int fwd_state; /* FSM */ struct fwd_module *fm; }; #if 0 static struct fwd_info *fwd_db[FWD_SIZE]; static struct fwd_control *fwd_ctl[FWD_SIZE]; static TAILQ_HEAD(fwd_module_list, fwd_module) modulehead = TAILQ_HEAD_INITIALIZER(modulehead); static struct fwd_module_list fwd_modules[FWD_SIZE]; static uint8_t fwd_map_af[] = { AF_INET, AF_INET6, }; static struct rwlock fwd_lock; #define FWD_LOCK_INIT() rw_init(&fwd_lock, "fwd_lock") #define FWD_RLOCK() rw_rlock(&fwd_lock) #define FWD_RUNLOCK() rw_runlock(&fwd_lock) #define FWD_WLOCK() rw_wlock(&fwd_lock) #define FWD_WUNLOCK() rw_wunlock(&fwd_lock) int fwd_attach_fib(struct fwd_module *fm, u_int fib); int fwd_destroy_fib(struct fwd_module *fm, u_int fib); #endif #ifdef INET static void fib4_rte_to_nh_extended(struct rtentry *rte, struct in_addr dst, struct nhop4_extended *pnh4); static void fib4_rte_to_nh_basic(struct rtentry *rte, struct in_addr dst, struct nhop4_basic *pnh4); #endif #ifdef INET static void fib6_rte_to_nh_basic(struct rtentry *rte, struct in6_addr dst, struct nhop6_basic *pnh6); #endif MALLOC_DEFINE(M_RTFIB, "rtfib", "routing fwd"); /* * Per-AF fast routines returning minimal needed info. * It is not safe to dereference any pointers since it * may end up with use-after-free case. * Typically it may be used to check if outgoing * interface matches or to calculate proper MTU. * * Note that returned interface pointer is logical one, * e.g. actual transmit ifp may be different. * Difference may be triggered by * 1) loopback routes installed for interface addresses. * e.g. for address 10.0.0.1 with prefix /24 bound to * interface ix0, "logical" interface will be "ix0", * while "trasmit" interface will be "lo0" since this is * loopback route. You should consider using other * functions if you need "transmit" interface or both. * * * Returns 0 on match, error code overwise. */ #define NHOP_FLAGS_MASK (RTF_REJECT|RTF_BLACKHOLE) //#define NHOP_DIRECT #define RNTORT(p) ((struct rtentry *)(p)) /* * Copies proper nexthop data based on @nh_src nexthop. * * For non-ECMP nexthop function simply copies @nh_src. * For ECMP nexthops flowid is used to select proper * nexthop. * */ static inline void fib_choose_prepend(uint32_t fibnum, struct nhop_data *nh_src, uint32_t flowid, struct nhop_data *nh, int af) { struct nhop_multi *nh_multi; int idx; if ((nh_src->nh_flags & NH_FLAGS_RECURSE) != 0) { /* * Recursive nexthop. Choose direct nexthop * based on flowid. */ nh_multi = (struct nhop_multi *)nh_src; idx = nh_multi->nh_nhops[flowid % nh_multi->nh_count]; #if 0 KASSERT((fibnum < rt_numfibs), ("fib4_lookup_prepend§: bad fibnum")); rnh = rt_tables_get_rnh(fibnum, AF_INET); //nh_src = &rnh->nhops[i]; #endif } *nh = *nh_src; /* TODO: Do some light-weight refcounting on egress ifp's */ } static inline void fib_free_nh(uint32_t fibnum, struct nhop_data *nh, int af) { /* TODO: Do some light-weight refcounting on egress ifp's */ } #ifdef INET void fib4_free_nh(uint32_t fibnum, struct nhop_data *nh) { fib_free_nh(fibnum, nh, AF_INET); } void fib4_choose_prepend(uint32_t fibnum, struct nhop_data *nh_src, uint32_t flowid, struct nhop_data *nh, struct nhop4_extended *nh_ext) { fib_choose_prepend(fibnum, nh_src, flowid, nh, AF_INET); if (nh_ext == NULL) return; nh_ext->nh_ifp = NH_LIFP(nh); nh_ext->nh_mtu = nh->nh_mtu; nh_ext->nh_flags = nh->nh_flags; #if 0 /* TODO: copy source/gw address from extended nexthop data */ nh_ext->nh_addr = ; nh_ext->nh_src= ; #endif } /* * Function performs lookup in IPv4 table fib @fibnum. * * In case of successful lookup @nh header is filled with * appropriate interface info and full L2 header to prepend. * * If no valid ARP record is present, NH_FLAGS_L2_INCOMPLETE flag * is set and gateway address is stored into nh->d.gw4 * * If @nh_ext is not NULL, additional nexthop data is stored there. * * Returns 0 on success. * */ int fib4_lookup_prepend(uint32_t fibnum, struct in_addr dst, struct mbuf *m, struct nhop_data *nh, struct nhop4_extended *nh_ext) { struct radix_node_head *rnh; struct radix_node *rn; struct sockaddr_in *gw_sa, sin; struct ifnet *lifp; struct in_addr gw; struct ether_header *eh; int error, flags; //uint32_t flowid; struct rtentry *rte; KASSERT((fibnum < rt_numfibs), ("fib4_lookup_prepend: bad fibnum")); rnh = rt_tables_get_rnh(fibnum, AF_INET); if (rnh == NULL) return (EHOSTUNREACH); /* Prepare lookup key */ memset(&sin, 0, sizeof(sin)); sin.sin_len = sizeof(struct sockaddr_in); sin.sin_addr = dst; RADIX_NODE_HEAD_RLOCK(rnh); rn = rnh->rnh_matchaddr((void *)&sin, rnh); rte = RNTORT(rn); if (rn == NULL || ((rn->rn_flags & RNF_ROOT) != 0) || RT_LINK_IS_UP(rte->rt_ifp) == 0) { RADIX_NODE_HEAD_RUNLOCK(rnh); return (EHOSTUNREACH); } /* * Currently we fill in @nh ourselves. * In near future rte will have nhop index to copy from. */ /* Calculate L3 info */ flags = 0; nh->nh_mtu = min(rte->rt_mtu, rte->rt_ifp->if_mtu); if (rte->rt_flags & RTF_GATEWAY) { gw_sa = (struct sockaddr_in *)rte->rt_gateway; gw = gw_sa->sin_addr; } else gw = dst; /* Set flags */ flags = rte->rt_flags & NHOP_FLAGS_MASK; gw_sa = (struct sockaddr_in *)rt_key(rte); if (gw_sa->sin_addr.s_addr == 0) flags |= NHOP_DEFAULT; /* * TODO: nh L2/L3 resolve. * Currently all we have is rte ifp. * Simply use it. */ lifp = rte->rt_ifp; /* Save both logical and transmit interface indexes */ nh->lifp_idx = lifp->if_index; nh->i.ifp_idx = nh->lifp_idx; if (nh_ext != NULL) { /* Fill in extended info */ fib4_rte_to_nh_extended(rte, dst, nh_ext); } RADIX_NODE_HEAD_RUNLOCK(rnh); nh->nh_flags = flags; /* * Try to lookup L2 info. * Do this using separate LLE locks. * TODO: move this under radix lock. */ if (lifp->if_type == IFT_ETHER) { eh = (struct ether_header *)nh->d.data; /* * Fill in ethernet header. * It should be already presented if we're * sending data via known gateway. */ error = arpresolve_fast(lifp, gw, m ? m->m_flags : 0, eh->ether_dhost); if (error == 0) { memcpy(&eh->ether_shost, IF_LLADDR(lifp), ETHER_ADDR_LEN); eh->ether_type = htons(ETHERTYPE_IP); nh->nh_count = ETHER_HDR_LEN; return (0); } } /* Notify caller that no L2 info is linked */ nh->nh_count = 0; nh->nh_flags |= NH_FLAGS_L2_INCOMPLETE; /* ..And save gateway address */ nh->d.gw4 = gw; return (0); } int fib4_sendmbuf(struct ifnet *ifp, struct mbuf *m, struct nhop_data *nh, struct in_addr dst) { int error; if (nh != NULL && (nh->nh_flags & NH_FLAGS_L2_INCOMPLETE) == 0) { /* * Fast path case. Most packets should * be sent from here. * TODO: Make special ifnet * 'if_output_frame' handler for that. */ struct route_compat rc; struct ether_header *eh; rc.ro_flags = AF_INET << 8 | RT_NHOP; rc.ro_nh = nh; M_PREPEND(m, nh->nh_count, M_NOWAIT); if (m == NULL) return (ENOBUFS); eh = mtod(m, struct ether_header *); memcpy(eh, nh->d.data, nh->nh_count); error = (*ifp->if_output)(ifp, m, NULL, (struct route *)&rc); } else { struct sockaddr_in gw_out; memset(&gw_out, 0, sizeof(gw_out)); gw_out.sin_len = sizeof(gw_out); gw_out.sin_family = AF_INET; gw_out.sin_addr = nh ? nh->d.gw4 : dst; error = (*ifp->if_output)(ifp, m, (const struct sockaddr *)&gw_out, NULL); } return (error); } static void fib4_rte_to_nh_extended(struct rtentry *rte, struct in_addr dst, struct nhop4_extended *pnh4) { struct sockaddr_in *gw; struct in_ifaddr *ia; pnh4->nh_ifp = rte->rt_ifp; pnh4->nh_mtu = min(rte->rt_mtu, rte->rt_ifp->if_mtu); if (rte->rt_flags & RTF_GATEWAY) { gw = (struct sockaddr_in *)rte->rt_gateway; pnh4->nh_addr = gw->sin_addr; } else pnh4->nh_addr = dst; ia = ifatoia(rte->rt_ifa); pnh4->nh_src = IA_SIN(ia)->sin_addr; /* Set flags */ pnh4->nh_flags = rte->rt_flags & NHOP_FLAGS_MASK; + if (rte->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) + pnh4->nh_flags |= NHOP_REDIRECT; gw = (struct sockaddr_in *)rt_key(rte); if (gw->sin_addr.s_addr == 0) pnh4->nh_flags |= NHOP_DEFAULT; } static void fib4_rte_to_nh_basic(struct rtentry *rte, struct in_addr dst, struct nhop4_basic *pnh4) { struct sockaddr_in *gw; pnh4->nh_ifp = rte->rt_ifa->ifa_ifp; pnh4->nh_mtu = min(rte->rt_mtu, rte->rt_ifp->if_mtu); if (rte->rt_flags & RTF_GATEWAY) { gw = (struct sockaddr_in *)rte->rt_gateway; pnh4->nh_addr = gw->sin_addr; } else pnh4->nh_addr = dst; /* Set flags */ pnh4->nh_flags = rte->rt_flags & NHOP_FLAGS_MASK; + if (rte->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) + pnh4->nh_flags |= NHOP_REDIRECT; gw = (struct sockaddr_in *)rt_key(rte); if (gw->sin_addr.s_addr == 0) pnh4->nh_flags |= NHOP_DEFAULT; } int fib4_lookup_nh_basic(uint32_t fibnum, struct in_addr dst, uint32_t flowid, struct nhop4_basic *pnh4) { struct radix_node_head *rnh; struct radix_node *rn; struct sockaddr_in sin; struct rtentry *rte; KASSERT((fibnum < rt_numfibs), ("fib4_lookup_nh_basic: bad fibnum")); rnh = rt_tables_get_rnh(fibnum, AF_INET); if (rnh == NULL) return (ENOENT); /* Prepare lookup key */ memset(&sin, 0, sizeof(sin)); sin.sin_len = sizeof(struct sockaddr_in); sin.sin_addr = dst; RADIX_NODE_HEAD_RLOCK(rnh); rn = rnh->rnh_matchaddr((void *)&sin, rnh); if (rn != NULL && ((rn->rn_flags & RNF_ROOT) == 0)) { rte = RNTORT(rn); /* Ensure route & ifp is UP */ if (RT_LINK_IS_UP(rte->rt_ifp)) { fib4_rte_to_nh_basic(rte, dst, pnh4); RADIX_NODE_HEAD_RUNLOCK(rnh); return (0); } } RADIX_NODE_HEAD_RUNLOCK(rnh); return (ENOENT); } int fib4_lookup_nh_extended(uint32_t fibnum, struct in_addr dst, uint32_t flowid, struct nhop4_extended *pnh4) { struct radix_node_head *rnh; struct radix_node *rn; struct sockaddr_in sin; struct rtentry *rte; KASSERT((fibnum < rt_numfibs), ("fib4_lookup_nh_basic: bad fibnum")); rnh = rt_tables_get_rnh(fibnum, AF_INET); if (rnh == NULL) return (ENOENT); /* Prepare lookup key */ memset(&sin, 0, sizeof(sin)); sin.sin_len = sizeof(struct sockaddr_in); sin.sin_addr = dst; RADIX_NODE_HEAD_RLOCK(rnh); rn = rnh->rnh_matchaddr((void *)&sin, rnh); if (rn != NULL && ((rn->rn_flags & RNF_ROOT) == 0)) { rte = RNTORT(rn); /* Ensure route & ifp is UP */ if (RT_LINK_IS_UP(rte->rt_ifp)) { fib4_rte_to_nh_extended(rte, dst, pnh4); RADIX_NODE_HEAD_RUNLOCK(rnh); return (0); } } RADIX_NODE_HEAD_RUNLOCK(rnh); return (ENOENT); } void fib4_free_nh_ext(uint32_t fibnum, struct nhop4_extended *pnh4) { } #endif #ifdef INET6 void fib6_free_nh(uint32_t fibnum, struct nhop_data *nh) { fib_free_nh(fibnum, nh, AF_INET6); } void fib6_choose_prepend(uint32_t fibnum, struct nhop_data *nh_src, uint32_t flowid, struct nhop_data *nh, struct nhop6_extended *nh_ext) { fib_choose_prepend(fibnum, nh_src, flowid, nh, AF_INET6); if (nh_ext == NULL) return; nh_ext->nh_ifp = NH_LIFP(nh); nh_ext->nh_mtu = nh->nh_mtu; nh_ext->nh_flags = nh->nh_flags; /* nh_ext->nh_addr = ; nh_ext->nh_src= ; */ } static void fib6_rte_to_nh_basic(struct rtentry *rte, struct in6_addr dst, struct nhop6_basic *pnh6) { struct sockaddr_in6 *gw; pnh6->nh_ifp = rte->rt_ifa->ifa_ifp; pnh6->nh_mtu = min(rte->rt_mtu, rte->rt_ifp->if_mtu); if (rte->rt_flags & RTF_GATEWAY) { gw = (struct sockaddr_in6 *)rte->rt_gateway; pnh6->nh_addr = gw->sin6_addr; } else pnh6->nh_addr = dst; /* Set flags */ pnh6->nh_flags = rte->rt_flags & NHOP_FLAGS_MASK; gw = (struct sockaddr_in6 *)rt_key(rte); if (IN6_IS_ADDR_UNSPECIFIED(&gw->sin6_addr)) pnh6->nh_flags |= NHOP_DEFAULT; } int fib6_lookup_nh_basic(uint32_t fibnum, struct in6_addr dst, uint32_t flowid, struct nhop6_basic *pnh6) { struct radix_node_head *rnh; struct radix_node *rn; struct sockaddr_in6 sin6; struct rtentry *rte; KASSERT((fibnum < rt_numfibs), ("fib6_lookup_nh_basic: bad fibnum")); rnh = rt_tables_get_rnh(fibnum, AF_INET); if (rnh == NULL) return (ENOENT); /* Prepare lookup key */ memset(&sin6, 0, sizeof(sin6)); sin6.sin6_addr = dst; RADIX_NODE_HEAD_RLOCK(rnh); rn = rnh->rnh_matchaddr((void *)&sin6, rnh); if (rn != NULL && ((rn->rn_flags & RNF_ROOT) == 0)) { rte = RNTORT(rn); /* Ensure route & ifp is UP */ if (RT_LINK_IS_UP(rte->rt_ifp)) { fib6_rte_to_nh_basic(rte, dst, pnh6); RADIX_NODE_HEAD_RUNLOCK(rnh); return (0); } } RADIX_NODE_HEAD_RUNLOCK(rnh); return (ENOENT); } #endif #if 0 typedef void nhop_change_cb_t(void *state); struct nhop_tracker { TAILQ_ENTRY(nhop_tracker) next; nhop_change_cb_t *f; void *state; uint32_t fibnum; struct sockaddr_storage ss; }; struct nhop_tracker * nhop_alloc_tracked(uint32_t fibnum, struct sockaddr *sa, nhop_change_cb_t *f, void *state) { struct nhop_tracker *nt; nt = malloc(sizeof(struct nhop_tracker), M_RTFIB, M_WAITOK | M_ZERO); nt->f = f; nt-state = state; nt->fibnum = fibnum; memcpy(&nt->ss, sa, sa->sa_len); return (nt); } int nhop_bind(struct nhop_tracker *nt) { NHOP_LOCK(nnh); NHOP_UNLOCK(nnh); return (0); } #endif Index: projects/routing/sys/net/rt_nhops.h =================================================================== --- projects/routing/sys/net/rt_nhops.h (revision 287051) +++ projects/routing/sys/net/rt_nhops.h (revision 287052) @@ -1,267 +1,268 @@ /*- * Copyright (c) 2014 * Alexander V. Chernikov * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #ifndef _NET_RT_NHOPS_H_ #define _NET_RT_NHOPS_H_ #define NH_TYPE_DIRECT 1 /* Directly reachable, no data */ #define NH_TYPE_BLACKHOLE 2 /* Blackhole route */ #define NH_TYPE_REJECT 3 /* Send reject */ #define NH_TYPE_L2 4 /* Provides full prepend header */ #define NH_TYPE_MUTATOR 5 /* NH+callback function */ #define NH_TYPE_MULTIPATH 6 /* Multipath route */ struct nhop_ctl_info { uint64_t refcnt; /* Use references */ uint64_t flags; /* Options */ }; /* Multipath nhop info */ struct nhop_mpath_info { uint16_t nhop; /* Netxthop id */ }; /* mutator info */ struct nhop_mutator_info; struct nhop_data; typedef int nhop_mutate_t(struct mbuf **, struct nhop_data *nd, void *storage); struct nhop_mutator_info { nhop_mutate_t *func; char data[]; }; /* Structures used for forwarding purposes */ #define MAX_PREPEND_LEN 56 /* Max data that can be prepended */ /* Non-recursive nexthop */ struct nhop_data { uint8_t nh_flags; /* NH flags */ uint8_t nh_count; /* Number of nexthops or data length */ uint16_t nh_mtu; /* given nhop MTU */ uint16_t lifp_idx; /* Logical interface index */ union { uint16_t ifp_idx; /* Transmit interface index */ uint16_t nhop_idx; /* L2 multipath nhop index */ } i; union { char data[MAX_PREPEND_LEN]; /* data to prepend */ #ifdef INET struct in_addr gw4; /* IPv4 gw address */ #endif #ifdef INET6 struct in6_addr gw6; /* IPv4 gw address */ #endif } d; }; /* Internal flags */ #define NH_FLAGS_RECURSE 0x01 /* Nexthop structure is recursive */ #define NH_FLAGS_L2_NHOP 0x02 /* L2 interface has to be selected */ #define NH_FLAGS_L2_ME 0x04 /* dst L2 address is our address */ #define NH_FLAGS_L2_INCOMPLETE 0x08 /* L2 header not prepended */ #define NH_LIFP(nh) ifnet_byindex_locked((nh)->lifp_idx) #define NH_TIFP(nh) ifnet_byindex_locked((nh)->i.ifp_idx) /* L2/L3 recursive nexthop */ struct nhop_multi { uint8_t nh_flags; /* NH flags */ uint8_t nh_count; /* Number of nexthops or data length */ uint8_t spare[2]; uint16_t nh_nhops[30]; /* Nexthop indexes */ }; /* Control plane nexthop data */ struct nhop_info { }; /* Per-AF per-fib nhop table */ struct nhops_descr { uint32_t nhop_size; /* Nehthop data size */ uint32_t nhops_max; /* Max number of nhops */ void *nhops_data; /* Pointer to nhop data table */ void *nhops_info; /* Pointer to nhop info table */ }; #if 0 typedef int nhop_resolve_t(struct sockaddr *dst, u_int fib, struct nhop_data *nd, struct nhop_info *nf); int lla_create_notify(struct sockaddr *dst, u_int fib, lla_notify_t *func, void *state, int flags); #endif /* Basic nexthop info used for uRPF/mtu checks */ struct nhop4_basic { struct ifnet *nh_ifp; /* Logical egress interface */ uint16_t nh_mtu; /* nexthop mtu */ uint16_t nh_flags; /* nhop flags */ struct in_addr nh_addr; /* GW/DST IPv4 address */ }; struct nhop6_basic { struct ifnet *nh_ifp; /* Logical egress interface */ uint16_t nh_mtu; /* nexthop mtu */ uint16_t nh_flags; /* nhop flags */ uint8_t spare[4]; struct in6_addr nh_addr; /* GW/DST IPv4 address */ }; struct nhop64_basic { union { struct nhop4_basic nh4; struct nhop6_basic nh6; } u; }; /* Extended nexthop info used for control protocols */ struct nhop4_extended { struct ifnet *nh_ifp; /* Logical egress interface */ uint16_t nh_mtu; /* nexthop mtu */ uint16_t nh_flags; /* nhop flags */ uint8_t spare[4]; struct in_addr nh_addr; /* GW/DST IPv4 address */ struct in_addr nh_src; /* default source IPv4 address */ uint64_t spare2[2]; }; struct nhop6_extended { struct ifnet *nh_ifp; /* Logical egress interface */ uint16_t nh_mtu; /* nexthop mtu */ uint16_t nh_flags; /* nhop flags */ uint8_t spare[4]; struct in6_addr nh_addr; /* GW/DST IPv6 address */ struct in6_addr nh_src; /* default source IPv6 address */ uint64_t spare2[2]; }; struct nhop64_extended { union { struct nhop4_extended nh4; struct nhop6_extended nh6; } u; }; struct route_info { struct nhop_data *ri_nh; /* Desired nexthop to use */ struct nhop64_basic *ri_nh_info; /* Get selected route info */ uint16_t ri_mtu; uint16_t spare[3]; }; struct route_compat { struct nhop_data *ro_nh; void *spare0; void *spare1; int ro_flags; }; int fib4_lookup_nh_basic(uint32_t fibnum, struct in_addr dst, uint32_t flowid, struct nhop4_basic *pnh4); int fib6_lookup_nh_basic(uint32_t fibnum, struct in6_addr dst, uint32_t flowid, struct nhop6_basic *pnh6); int fib4_lookup_nh_extended(uint32_t fibnum, struct in_addr dst, uint32_t flowid, struct nhop4_extended *pnh4); void fib4_free_nh_ext(uint32_t fibnum, struct nhop4_extended *pnh4); void fib4_free_nh(uint32_t fibnum, struct nhop_data *nh); void fib4_choose_prepend(uint32_t fibnum, struct nhop_data *nh_src, uint32_t flowid, struct nhop_data *nh, struct nhop4_extended *nh_ext); int fib4_lookup_prepend(uint32_t fibnum, struct in_addr dst, struct mbuf *m, struct nhop_data *nh, struct nhop4_extended *nh_ext); int fib4_sendmbuf(struct ifnet *ifp, struct mbuf *m, struct nhop_data *nh, struct in_addr dst); void fib6_free_nh(uint32_t fibnum, struct nhop_data *nh); void fib6_choose_prepend(uint32_t fibnum, struct nhop_data *nh_src, uint32_t flowid, struct nhop_data *nh, struct nhop6_extended *nh_ext); -#define NHOP_REJECT RTF_REJECT -#define NHOP_BLACKHOLE RTF_BLACKHOLE +#define NHOP_REJECT 0x08 /* RTF_REJECT */ +#define NHOP_BLACKHOLE 0x1000 /* RTF_BLACKHOLE */ +#define NHOP_REDIRECT 0x10 /* RTF_DYNAMIC|RTF_MODIFIED */ #define NHOP_DEFAULT 0x80 /* Default route */ #define FWD_INET 0 #define FWD_INET6 1 #define FWD_SIZE 2 #define FWD_NAME_MAX 15 #define FWD_MULTIPATH 0x0001 /* has multipath support */ #define FWD_OLDMASKS 0x0002 /* has support for non-contig masks */ #define FWD_DEFAULT 0x0004 /* installs as default fib mechanism */ #define FWD_MANAGELOCK 0x0004 /* manage its own locking */ typedef void *fib_init_t(u_int fibnum); typedef void fib_destroy_t(void *state); typedef int fib_dump_t(void *state, struct radix_node_head *rnh); typedef int fib_change_t(void *state, int req, struct rtentry *rte, struct rt_addrinfo *info); typedef int fib_lookup_t(void *state, void *key, uint64_t *attr, u_int flowid, void *nhop); /* Structure used by external module */ struct fwd_module_info { uint8_t fwd_family; /* family we're registering to */ char name[FWD_NAME_MAX]; /* fwd module name */ uint32_t capabilities; fib_init_t *fib_init; fib_destroy_t *fib_destroy; fib_dump_t *fib_dump; fib_change_t *fib_change; fib_lookup_t *fib_lookup; }; /* Internal version of previous structure */ struct fwd_module { TAILQ_ENTRY(fwd_module) list; uint8_t fwd_family; char name[FWD_NAME_MAX]; uint32_t capabilities; fib_init_t *fib_init; fib_destroy_t *fib_destroy; fib_dump_t *fib_dump; fib_change_t *fib_change; fib_lookup_t *fib_lookup; }; int fwd_attach_module(struct fwd_module_info *m, void **); int fwd_destroy_module(void *state); int fwd_change_fib(struct radix_node_head *rnh, int req, struct rtentry *rte, struct rt_addrinfo *info); #endif Index: projects/routing/sys/netinet/ip_input.c =================================================================== --- projects/routing/sys/netinet/ip_input.c (revision 287051) +++ projects/routing/sys/netinet/ip_input.c (revision 287052) @@ -1,1273 +1,1258 @@ /*- * Copyright (c) 1982, 1986, 1988, 1993 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * @(#)ip_input.c 8.2 (Berkeley) 1/4/94 */ #include __FBSDID("$FreeBSD$"); #include "opt_bootp.h" #include "opt_ipfw.h" #include "opt_ipstealth.h" #include "opt_ipsec.h" #include "opt_route.h" #include "opt_rss.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef IPSEC #include #endif /* IPSEC */ #include #include #include #include #ifdef CTASSERT CTASSERT(sizeof(struct ip) == 20); #endif /* IP reassembly functions are defined in ip_reass.c. */ extern void ipreass_init(void); extern void ipreass_drain(void); extern void ipreass_slowtimo(void); #ifdef VIMAGE extern void ipreass_destroy(void); #endif struct rmlock in_ifaddr_lock; RM_SYSINIT(in_ifaddr_lock, &in_ifaddr_lock, "in_ifaddr_lock"); VNET_DEFINE(int, rsvp_on); VNET_DEFINE(int, ipforwarding); SYSCTL_INT(_net_inet_ip, IPCTL_FORWARDING, forwarding, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(ipforwarding), 0, "Enable IP forwarding between interfaces"); static VNET_DEFINE(int, ipsendredirects) = 1; /* XXX */ #define V_ipsendredirects VNET(ipsendredirects) SYSCTL_INT(_net_inet_ip, IPCTL_SENDREDIRECTS, redirect, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(ipsendredirects), 0, "Enable sending IP redirects"); /* * XXX - Setting ip_checkinterface mostly implements the receive side of * the Strong ES model described in RFC 1122, but since the routing table * and transmit implementation do not implement the Strong ES model, * setting this to 1 results in an odd hybrid. * * XXX - ip_checkinterface currently must be disabled if you use ipnat * to translate the destination address to another local interface. * * XXX - ip_checkinterface must be disabled if you add IP aliases * to the loopback interface instead of the interface where the * packets for those addresses are received. */ static VNET_DEFINE(int, ip_checkinterface); #define V_ip_checkinterface VNET(ip_checkinterface) SYSCTL_INT(_net_inet_ip, OID_AUTO, check_interface, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(ip_checkinterface), 0, "Verify packet arrives on correct interface"); VNET_DEFINE(struct pfil_head, inet_pfil_hook); /* Packet filter hooks */ static struct netisr_handler ip_nh = { .nh_name = "ip", .nh_handler = ip_input, .nh_proto = NETISR_IP, #ifdef RSS .nh_m2cpuid = rss_soft_m2cpuid, .nh_policy = NETISR_POLICY_CPU, .nh_dispatch = NETISR_DISPATCH_HYBRID, #else .nh_policy = NETISR_POLICY_FLOW, #endif }; #ifdef RSS /* * Directly dispatched frames are currently assumed * to have a flowid already calculated. * * It should likely have something that assert it * actually has valid flow details. */ static struct netisr_handler ip_direct_nh = { .nh_name = "ip_direct", .nh_handler = ip_direct_input, .nh_proto = NETISR_IP_DIRECT, .nh_m2cpuid = rss_m2cpuid, .nh_policy = NETISR_POLICY_CPU, .nh_dispatch = NETISR_DISPATCH_HYBRID, }; #endif extern struct domain inetdomain; extern struct protosw inetsw[]; u_char ip_protox[IPPROTO_MAX]; VNET_DEFINE(struct in_ifaddrhead, in_ifaddrhead); /* first inet address */ VNET_DEFINE(struct in_ifaddrhashhead *, in_ifaddrhashtbl); /* inet addr hash table */ VNET_DEFINE(u_long, in_ifaddrhmask); /* mask for hash table */ #ifdef IPCTL_DEFMTU SYSCTL_INT(_net_inet_ip, IPCTL_DEFMTU, mtu, CTLFLAG_RW, &ip_mtu, 0, "Default MTU"); #endif #ifdef IPSTEALTH VNET_DEFINE(int, ipstealth); SYSCTL_INT(_net_inet_ip, OID_AUTO, stealth, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(ipstealth), 0, "IP stealth mode, no TTL decrementation on forwarding"); #endif /* * IP statistics are stored in the "array" of counter(9)s. */ VNET_PCPUSTAT_DEFINE(struct ipstat, ipstat); VNET_PCPUSTAT_SYSINIT(ipstat); SYSCTL_VNET_PCPUSTAT(_net_inet_ip, IPCTL_STATS, stats, struct ipstat, ipstat, "IP statistics (struct ipstat, netinet/ip_var.h)"); #ifdef VIMAGE VNET_PCPUSTAT_SYSUNINIT(ipstat); #endif /* VIMAGE */ /* * Kernel module interface for updating ipstat. The argument is an index * into ipstat treated as an array. */ void kmod_ipstat_inc(int statnum) { counter_u64_add(VNET(ipstat)[statnum], 1); } void kmod_ipstat_dec(int statnum) { counter_u64_add(VNET(ipstat)[statnum], -1); } static int sysctl_netinet_intr_queue_maxlen(SYSCTL_HANDLER_ARGS) { int error, qlimit; netisr_getqlimit(&ip_nh, &qlimit); error = sysctl_handle_int(oidp, &qlimit, 0, req); if (error || !req->newptr) return (error); if (qlimit < 1) return (EINVAL); return (netisr_setqlimit(&ip_nh, qlimit)); } SYSCTL_PROC(_net_inet_ip, IPCTL_INTRQMAXLEN, intr_queue_maxlen, CTLTYPE_INT|CTLFLAG_RW, 0, 0, sysctl_netinet_intr_queue_maxlen, "I", "Maximum size of the IP input queue"); static int sysctl_netinet_intr_queue_drops(SYSCTL_HANDLER_ARGS) { u_int64_t qdrops_long; int error, qdrops; netisr_getqdrops(&ip_nh, &qdrops_long); qdrops = qdrops_long; error = sysctl_handle_int(oidp, &qdrops, 0, req); if (error || !req->newptr) return (error); if (qdrops != 0) return (EINVAL); netisr_clearqdrops(&ip_nh); return (0); } SYSCTL_PROC(_net_inet_ip, IPCTL_INTRQDROPS, intr_queue_drops, CTLTYPE_INT|CTLFLAG_RD, 0, 0, sysctl_netinet_intr_queue_drops, "I", "Number of packets dropped from the IP input queue"); #ifdef RSS static int sysctl_netinet_intr_direct_queue_maxlen(SYSCTL_HANDLER_ARGS) { int error, qlimit; netisr_getqlimit(&ip_direct_nh, &qlimit); error = sysctl_handle_int(oidp, &qlimit, 0, req); if (error || !req->newptr) return (error); if (qlimit < 1) return (EINVAL); return (netisr_setqlimit(&ip_direct_nh, qlimit)); } SYSCTL_PROC(_net_inet_ip, IPCTL_INTRQMAXLEN, intr_direct_queue_maxlen, CTLTYPE_INT|CTLFLAG_RW, 0, 0, sysctl_netinet_intr_direct_queue_maxlen, "I", "Maximum size of the IP direct input queue"); static int sysctl_netinet_intr_direct_queue_drops(SYSCTL_HANDLER_ARGS) { u_int64_t qdrops_long; int error, qdrops; netisr_getqdrops(&ip_direct_nh, &qdrops_long); qdrops = qdrops_long; error = sysctl_handle_int(oidp, &qdrops, 0, req); if (error || !req->newptr) return (error); if (qdrops != 0) return (EINVAL); netisr_clearqdrops(&ip_direct_nh); return (0); } SYSCTL_PROC(_net_inet_ip, IPCTL_INTRQDROPS, intr_direct_queue_drops, CTLTYPE_INT|CTLFLAG_RD, 0, 0, sysctl_netinet_intr_direct_queue_drops, "I", "Number of packets dropped from the IP direct input queue"); #endif /* RSS */ /* * IP initialization: fill in IP protocol switch table. * All protocols not implemented in kernel go to raw IP protocol handler. */ void ip_init(void) { struct protosw *pr; int i; TAILQ_INIT(&V_in_ifaddrhead); V_in_ifaddrhashtbl = hashinit(INADDR_NHASH, M_IFADDR, &V_in_ifaddrhmask); /* Initialize IP reassembly queue. */ ipreass_init(); /* Initialize packet filter hooks. */ V_inet_pfil_hook.ph_type = PFIL_TYPE_AF; V_inet_pfil_hook.ph_af = AF_INET; if ((i = pfil_head_register(&V_inet_pfil_hook)) != 0) printf("%s: WARNING: unable to register pfil hook, " "error %d\n", __func__, i); /* Skip initialization of globals for non-default instances. */ if (!IS_DEFAULT_VNET(curvnet)) return; pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); if (pr == NULL) panic("ip_init: PF_INET not found"); /* Initialize the entire ip_protox[] array to IPPROTO_RAW. */ for (i = 0; i < IPPROTO_MAX; i++) ip_protox[i] = pr - inetsw; /* * Cycle through IP protocols and put them into the appropriate place * in ip_protox[]. */ for (pr = inetdomain.dom_protosw; pr < inetdomain.dom_protoswNPROTOSW; pr++) if (pr->pr_domain->dom_family == PF_INET && pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) { /* Be careful to only index valid IP protocols. */ if (pr->pr_protocol < IPPROTO_MAX) ip_protox[pr->pr_protocol] = pr - inetsw; } netisr_register(&ip_nh); #ifdef RSS netisr_register(&ip_direct_nh); #endif } #ifdef VIMAGE void ip_destroy(void) { int i; if ((i = pfil_head_unregister(&V_inet_pfil_hook)) != 0) printf("%s: WARNING: unable to unregister pfil hook, " "error %d\n", __func__, i); /* Cleanup in_ifaddr hash table; should be empty. */ hashdestroy(V_in_ifaddrhashtbl, M_IFADDR, V_in_ifaddrhmask); /* Destroy IP reassembly queue. */ ipreass_destroy(); } #endif #ifdef RSS /* * IP direct input routine. * * This is called when reinjecting completed fragments where * all of the previous checking and book-keeping has been done. */ void ip_direct_input(struct mbuf *m) { struct ip *ip; int hlen; ip = mtod(m, struct ip *); hlen = ip->ip_hl << 2; IPSTAT_INC(ips_delivered); (*inetsw[ip_protox[ip->ip_p]].pr_input)(&m, &hlen, ip->ip_p); return; } #endif /* * Ip input routine. Checksum and byte swap header. If fragmented * try to reassemble. Process options. Pass to next level. */ void ip_input(struct mbuf *m) { struct ip *ip = NULL; struct in_ifaddr *ia = NULL; struct ifaddr *ifa; struct ifnet *ifp; int checkif, hlen = 0; uint16_t sum, ip_len; int dchg = 0; /* dest changed after fw */ struct in_addr odst; /* original dst address */ M_ASSERTPKTHDR(m); if (m->m_flags & M_FASTFWD_OURS) { m->m_flags &= ~M_FASTFWD_OURS; /* Set up some basics that will be used later. */ ip = mtod(m, struct ip *); hlen = ip->ip_hl << 2; ip_len = ntohs(ip->ip_len); goto ours; } IPSTAT_INC(ips_total); if (m->m_pkthdr.len < sizeof(struct ip)) goto tooshort; if (m->m_len < sizeof (struct ip) && (m = m_pullup(m, sizeof (struct ip))) == NULL) { IPSTAT_INC(ips_toosmall); return; } ip = mtod(m, struct ip *); if (ip->ip_v != IPVERSION) { IPSTAT_INC(ips_badvers); goto bad; } hlen = ip->ip_hl << 2; if (hlen < sizeof(struct ip)) { /* minimum header length */ IPSTAT_INC(ips_badhlen); goto bad; } if (hlen > m->m_len) { if ((m = m_pullup(m, hlen)) == NULL) { IPSTAT_INC(ips_badhlen); return; } ip = mtod(m, struct ip *); } IP_PROBE(receive, NULL, NULL, ip, m->m_pkthdr.rcvif, ip, NULL); /* 127/8 must not appear on wire - RFC1122 */ ifp = m->m_pkthdr.rcvif; if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET || (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) { if ((ifp->if_flags & IFF_LOOPBACK) == 0) { IPSTAT_INC(ips_badaddr); goto bad; } } if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) { sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID); } else { if (hlen == sizeof(struct ip)) { sum = in_cksum_hdr(ip); } else { sum = in_cksum(m, hlen); } } if (sum) { IPSTAT_INC(ips_badsum); goto bad; } #ifdef ALTQ if (altq_input != NULL && (*altq_input)(m, AF_INET) == 0) /* packet is dropped by traffic conditioner */ return; #endif ip_len = ntohs(ip->ip_len); if (ip_len < hlen) { IPSTAT_INC(ips_badlen); goto bad; } /* * Check that the amount of data in the buffers * is as at least much as the IP header would have us expect. * Trim mbufs if longer than we expect. * Drop packet if shorter than we expect. */ if (m->m_pkthdr.len < ip_len) { tooshort: IPSTAT_INC(ips_tooshort); goto bad; } if (m->m_pkthdr.len > ip_len) { if (m->m_len == m->m_pkthdr.len) { m->m_len = ip_len; m->m_pkthdr.len = ip_len; } else m_adj(m, ip_len - m->m_pkthdr.len); } #ifdef IPSEC /* * Bypass packet filtering for packets previously handled by IPsec. */ if (ip_ipsec_filtertunnel(m)) goto passin; #endif /* IPSEC */ /* * Run through list of hooks for input packets. * * NB: Beware of the destination address changing (e.g. * by NAT rewriting). When this happens, tell * ip_forward to do the right thing. */ /* Jump over all PFIL processing if hooks are not active. */ if (!PFIL_HOOKED(&V_inet_pfil_hook)) goto passin; odst = ip->ip_dst; if (pfil_run_hooks(&V_inet_pfil_hook, &m, ifp, PFIL_IN, NULL) != 0) return; if (m == NULL) /* consumed by filter */ return; ip = mtod(m, struct ip *); dchg = (odst.s_addr != ip->ip_dst.s_addr); ifp = m->m_pkthdr.rcvif; if (m->m_flags & M_FASTFWD_OURS) { m->m_flags &= ~M_FASTFWD_OURS; goto ours; } if (m->m_flags & M_IP_NEXTHOP) { dchg = (m_tag_find(m, PACKET_TAG_IPFORWARD, NULL) != NULL); if (dchg != 0) { /* * Directly ship the packet on. This allows * forwarding packets originally destined to us * to some other directly connected host. */ ip_forward(m, 1); return; } } passin: /* * Process options and, if not destined for us, * ship it on. ip_dooptions returns 1 when an * error was detected (causing an icmp message * to be sent and the original packet to be freed). */ if (hlen > sizeof (struct ip) && ip_dooptions(m, 0)) return; /* greedy RSVP, snatches any PATH packet of the RSVP protocol and no * matter if it is destined to another node, or whether it is * a multicast one, RSVP wants it! and prevents it from being forwarded * anywhere else. Also checks if the rsvp daemon is running before * grabbing the packet. */ if (V_rsvp_on && ip->ip_p==IPPROTO_RSVP) goto ours; /* * Check our list of addresses, to see if the packet is for us. * If we don't have any addresses, assume any unicast packet * we receive might be for us (and let the upper layers deal * with it). */ if (TAILQ_EMPTY(&V_in_ifaddrhead) && (m->m_flags & (M_MCAST|M_BCAST)) == 0) goto ours; /* * Enable a consistency check between the destination address * and the arrival interface for a unicast packet (the RFC 1122 * strong ES model) if IP forwarding is disabled and the packet * is not locally generated and the packet is not subject to * 'ipfw fwd'. * * XXX - Checking also should be disabled if the destination * address is ipnat'ed to a different interface. * * XXX - Checking is incompatible with IP aliases added * to the loopback interface instead of the interface where * the packets are received. * * XXX - This is the case for carp vhost IPs as well so we * insert a workaround. If the packet got here, we already * checked with carp_iamatch() and carp_forus(). */ checkif = V_ip_checkinterface && (V_ipforwarding == 0) && ifp != NULL && ((ifp->if_flags & IFF_LOOPBACK) == 0) && ifp->if_carp == NULL && (dchg == 0); /* * Check for exact addresses in the hash bucket. */ /* IN_IFADDR_RLOCK(); */ LIST_FOREACH(ia, INADDR_HASH(ip->ip_dst.s_addr), ia_hash) { /* * If the address matches, verify that the packet * arrived via the correct interface if checking is * enabled. */ if (IA_SIN(ia)->sin_addr.s_addr == ip->ip_dst.s_addr && (!checkif || ia->ia_ifp == ifp)) { counter_u64_add(ia->ia_ifa.ifa_ipackets, 1); counter_u64_add(ia->ia_ifa.ifa_ibytes, m->m_pkthdr.len); /* IN_IFADDR_RUNLOCK(); */ goto ours; } } /* IN_IFADDR_RUNLOCK(); */ /* * Check for broadcast addresses. * * Only accept broadcast packets that arrive via the matching * interface. Reception of forwarded directed broadcasts would * be handled via ip_forward() and ether_output() with the loopback * into the stack for SIMPLEX interfaces handled by ether_output(). */ if (ifp != NULL && ifp->if_flags & IFF_BROADCAST) { IF_ADDR_RLOCK(ifp); TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { if (ifa->ifa_addr->sa_family != AF_INET) continue; ia = ifatoia(ifa); if (satosin(&ia->ia_broadaddr)->sin_addr.s_addr == ip->ip_dst.s_addr) { counter_u64_add(ia->ia_ifa.ifa_ipackets, 1); counter_u64_add(ia->ia_ifa.ifa_ibytes, m->m_pkthdr.len); IF_ADDR_RUNLOCK(ifp); goto ours; } #ifdef BOOTP_COMPAT if (IA_SIN(ia)->sin_addr.s_addr == INADDR_ANY) { counter_u64_add(ia->ia_ifa.ifa_ipackets, 1); counter_u64_add(ia->ia_ifa.ifa_ibytes, m->m_pkthdr.len); IF_ADDR_RUNLOCK(ifp); goto ours; } #endif } IF_ADDR_RUNLOCK(ifp); ia = NULL; } /* RFC 3927 2.7: Do not forward datagrams for 169.254.0.0/16. */ if (IN_LINKLOCAL(ntohl(ip->ip_dst.s_addr))) { IPSTAT_INC(ips_cantforward); m_freem(m); return; } if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) { if (V_ip_mrouter) { /* * If we are acting as a multicast router, all * incoming multicast packets are passed to the * kernel-level multicast forwarding function. * The packet is returned (relatively) intact; if * ip_mforward() returns a non-zero value, the packet * must be discarded, else it may be accepted below. */ if (ip_mforward && ip_mforward(ip, ifp, m, 0) != 0) { IPSTAT_INC(ips_cantforward); m_freem(m); return; } /* * The process-level routing daemon needs to receive * all multicast IGMP packets, whether or not this * host belongs to their destination groups. */ if (ip->ip_p == IPPROTO_IGMP) goto ours; IPSTAT_INC(ips_forward); } /* * Assume the packet is for us, to avoid prematurely taking * a lock on the in_multi hash. Protocols must perform * their own filtering and update statistics accordingly. */ goto ours; } if (ip->ip_dst.s_addr == (u_long)INADDR_BROADCAST) goto ours; if (ip->ip_dst.s_addr == INADDR_ANY) goto ours; /* * Not for us; forward if possible and desirable. */ if (V_ipforwarding == 0) { IPSTAT_INC(ips_cantforward); m_freem(m); } else { ip_forward(m, dchg); } return; ours: #ifdef IPSTEALTH /* * IPSTEALTH: Process non-routing options only * if the packet is destined for us. */ if (V_ipstealth && hlen > sizeof (struct ip) && ip_dooptions(m, 1)) return; #endif /* IPSTEALTH */ /* * Attempt reassembly; if it succeeds, proceed. * ip_reass() will return a different mbuf. */ if (ip->ip_off & htons(IP_MF | IP_OFFMASK)) { /* XXXGL: shouldn't we save & set m_flags? */ m = ip_reass(m); if (m == NULL) return; ip = mtod(m, struct ip *); /* Get the header length of the reassembled packet */ hlen = ip->ip_hl << 2; } #ifdef 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 (ip_ipsec_input(m, ip->ip_p) != 0) goto bad; #endif /* IPSEC */ /* * Switch out to protocol's input routine. */ IPSTAT_INC(ips_delivered); (*inetsw[ip_protox[ip->ip_p]].pr_input)(&m, &hlen, ip->ip_p); return; bad: m_freem(m); } /* * IP timer processing; * if a timer expires on a reassembly * queue, discard it. */ void ip_slowtimo(void) { VNET_ITERATOR_DECL(vnet_iter); VNET_LIST_RLOCK_NOSLEEP(); VNET_FOREACH(vnet_iter) { CURVNET_SET(vnet_iter); ipreass_slowtimo(); CURVNET_RESTORE(); } VNET_LIST_RUNLOCK_NOSLEEP(); } void ip_drain(void) { VNET_ITERATOR_DECL(vnet_iter); VNET_LIST_RLOCK_NOSLEEP(); VNET_FOREACH(vnet_iter) { CURVNET_SET(vnet_iter); ipreass_drain(); CURVNET_RESTORE(); } VNET_LIST_RUNLOCK_NOSLEEP(); } /* * The protocol to be inserted into ip_protox[] must be already registered * in inetsw[], either statically or through pf_proto_register(). */ int ipproto_register(short ipproto) { struct protosw *pr; /* Sanity checks. */ if (ipproto <= 0 || ipproto >= IPPROTO_MAX) return (EPROTONOSUPPORT); /* * The protocol slot must not be occupied by another protocol * already. An index pointing to IPPROTO_RAW is unused. */ pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); if (pr == NULL) return (EPFNOSUPPORT); if (ip_protox[ipproto] != pr - inetsw) /* IPPROTO_RAW */ return (EEXIST); /* Find the protocol position in inetsw[] and set the index. */ for (pr = inetdomain.dom_protosw; pr < inetdomain.dom_protoswNPROTOSW; pr++) { if (pr->pr_domain->dom_family == PF_INET && pr->pr_protocol && pr->pr_protocol == ipproto) { ip_protox[pr->pr_protocol] = pr - inetsw; return (0); } } return (EPROTONOSUPPORT); } int ipproto_unregister(short ipproto) { struct protosw *pr; /* Sanity checks. */ if (ipproto <= 0 || ipproto >= IPPROTO_MAX) return (EPROTONOSUPPORT); /* Check if the protocol was indeed registered. */ pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); if (pr == NULL) return (EPFNOSUPPORT); if (ip_protox[ipproto] == pr - inetsw) /* IPPROTO_RAW */ return (ENOENT); /* Reset the protocol slot to IPPROTO_RAW. */ ip_protox[ipproto] = pr - inetsw; return (0); } u_char inetctlerrmap[PRC_NCMDS] = { 0, 0, 0, 0, 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, EMSGSIZE, EHOSTUNREACH, 0, 0, 0, 0, EHOSTUNREACH, 0, ENOPROTOOPT, ECONNREFUSED }; /* * Forward a packet. If some error occurs return the sender * an icmp packet. Note we can't always generate a meaningful * icmp message because icmp doesn't have a large enough repertoire * of codes and types. * * If not forwarding, just drop the packet. This could be confusing * if ipforwarding was zero but some routing protocol was advancing * us as a gateway to somewhere. However, we must let the routing * protocol deal with that. * * The srcrt parameter indicates whether the packet is being forwarded * via a source route. */ void ip_forward(struct mbuf *m, int srcrt) { struct ip *ip = mtod(m, struct ip *); struct mbuf *mcopy; struct in_addr dest; struct nhop4_basic nh4, *pnh4; - struct route ro; struct route_info ri; int error, type = 0, code = 0, mtu = 0; if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) { IPSTAT_INC(ips_cantforward); m_freem(m); return; } #ifdef IPSEC if (ip_ipsec_fwd(m) != 0) { IPSTAT_INC(ips_cantforward); m_freem(m); return; } #endif /* IPSEC */ #ifdef IPSTEALTH if (!V_ipstealth) { #endif if (ip->ip_ttl <= IPTTLDEC) { icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, 0, 0); return; } #ifdef IPSTEALTH } #endif pnh4 = &nh4; if (fib4_lookup_nh_basic(M_GETFIB(m), ip->ip_dst, 0, &nh4) != 0) pnh4 = NULL; #ifndef IPSEC /* * 'ia' may be NULL if there is no route for this destination. * In case of IPsec, Don't discard it just yet, but pass it to * ip_output in case of outgoing IPsec policy. */ if (!srcrt && pnh4 == NULL) { icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0); RO_RTFREE(&ro); return; } #endif /* * Save the IP header and at most 8 bytes of the payload, * in case we need to generate an ICMP message to the src. * * XXX this can be optimized a lot by saving the data in a local * buffer on the stack (72 bytes at most), and only allocating the * mbuf if really necessary. The vast majority of the packets * are forwarded without having to send an ICMP back (either * because unnecessary, or because rate limited), so we are * really we are wasting a lot of work here. * * We don't use m_copy() because it might return a reference * to a shared cluster. Both this function and ip_output() * assume exclusive access to the IP header in `m', so any * data in a cluster may change before we reach icmp_error(). */ mcopy = m_gethdr(M_NOWAIT, m->m_type); if (mcopy != NULL && !m_dup_pkthdr(mcopy, m, M_NOWAIT)) { /* * It's probably ok if the pkthdr dup fails (because * the deep copy of the tag chain failed), but for now * be conservative and just discard the copy since * code below may some day want the tags. */ m_free(mcopy); mcopy = NULL; } if (mcopy != NULL) { mcopy->m_len = min(ntohs(ip->ip_len), M_TRAILINGSPACE(mcopy)); mcopy->m_pkthdr.len = mcopy->m_len; m_copydata(m, 0, mcopy->m_len, mtod(mcopy, caddr_t)); } #ifdef IPSTEALTH if (!V_ipstealth) { #endif ip->ip_ttl -= IPTTLDEC; #ifdef IPSTEALTH } #endif /* * If forwarding packet using same interface that it came in on, * perhaps should send a redirect to sender to shortcut a hop. * Only send redirect if source is sending directly to us, * and if packet was not source routed (or has any options). * Also, don't send redirect if forwarding using a default route * or a route modified by a redirect. */ dest.s_addr = 0; if (!srcrt && V_ipsendredirects && pnh4 != NULL && nh4.nh_ifp == m->m_pkthdr.rcvif) { - struct rtentry *rt; - - rt = ro.ro_rt; - - if (rt && (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && - satosin(rt_key(rt))->sin_addr.s_addr != 0) { -#define RTA(rt) ((struct in_ifaddr *)(rt->rt_ifa)) - u_long src = ntohl(ip->ip_src.s_addr); - - if (RTA(rt) && - (src & RTA(rt)->ia_subnetmask) == RTA(rt)->ia_subnet) { - if (rt->rt_flags & RTF_GATEWAY) - dest.s_addr = satosin(rt->rt_gateway)->sin_addr.s_addr; - else - dest.s_addr = ip->ip_dst.s_addr; - /* Router requirements says to only send host redirects */ - type = ICMP_REDIRECT; - code = ICMP_REDIRECT_HOST; - } + if ((nh4.nh_flags & (NHOP_REDIRECT|NHOP_DEFAULT)) == 0) { + dest = nh4.nh_addr; + /*Router requirements says to only send host redirects*/ + type = ICMP_REDIRECT; + code = ICMP_REDIRECT_HOST; } } bzero(&ri, sizeof(ri)); error = ip_output(m, NULL, &ri, IP_FORWARDING, NULL, NULL); if (error == EMSGSIZE) mtu = ri.ri_mtu; if (error) IPSTAT_INC(ips_cantforward); else { IPSTAT_INC(ips_forward); if (type) IPSTAT_INC(ips_redirectsent); else { if (mcopy) m_freem(mcopy); return; } } if (mcopy == NULL) { return; } switch (error) { case 0: /* forwarded, but need redirect */ /* type, code set above */ break; case ENETUNREACH: case EHOSTUNREACH: case ENETDOWN: case EHOSTDOWN: default: type = ICMP_UNREACH; code = ICMP_UNREACH_HOST; break; case EMSGSIZE: type = ICMP_UNREACH; code = ICMP_UNREACH_NEEDFRAG; #ifdef IPSEC /* * If IPsec is configured for this path, * override any possibly mtu value set by ip_output. */ mtu = min(ri.ri_mtu, ip_ipsec_mtu(mcopy, mtu)); #endif /* IPSEC */ IPSTAT_INC(ips_cantfrag); break; case ENOBUFS: case EACCES: /* ipfw denied packet */ m_freem(mcopy); return; } icmp_error(mcopy, type, code, dest.s_addr, mtu); } void ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip, struct mbuf *m) { if (inp->inp_socket->so_options & (SO_BINTIME | SO_TIMESTAMP)) { struct bintime bt; bintime(&bt); if (inp->inp_socket->so_options & SO_BINTIME) { *mp = sbcreatecontrol((caddr_t)&bt, sizeof(bt), SCM_BINTIME, SOL_SOCKET); if (*mp) mp = &(*mp)->m_next; } if (inp->inp_socket->so_options & SO_TIMESTAMP) { struct timeval tv; bintime2timeval(&bt, &tv); *mp = sbcreatecontrol((caddr_t)&tv, sizeof(tv), SCM_TIMESTAMP, SOL_SOCKET); if (*mp) mp = &(*mp)->m_next; } } if (inp->inp_flags & INP_RECVDSTADDR) { *mp = sbcreatecontrol((caddr_t)&ip->ip_dst, sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); if (*mp) mp = &(*mp)->m_next; } if (inp->inp_flags & INP_RECVTTL) { *mp = sbcreatecontrol((caddr_t)&ip->ip_ttl, sizeof(u_char), IP_RECVTTL, IPPROTO_IP); if (*mp) mp = &(*mp)->m_next; } #ifdef notyet /* XXX * Moving these out of udp_input() made them even more broken * than they already were. */ /* options were tossed already */ if (inp->inp_flags & INP_RECVOPTS) { *mp = sbcreatecontrol((caddr_t)opts_deleted_above, sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP); if (*mp) mp = &(*mp)->m_next; } /* ip_srcroute doesn't do what we want here, need to fix */ if (inp->inp_flags & INP_RECVRETOPTS) { *mp = sbcreatecontrol((caddr_t)ip_srcroute(m), sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP); if (*mp) mp = &(*mp)->m_next; } #endif if (inp->inp_flags & INP_RECVIF) { struct ifnet *ifp; struct sdlbuf { struct sockaddr_dl sdl; u_char pad[32]; } sdlbuf; struct sockaddr_dl *sdp; struct sockaddr_dl *sdl2 = &sdlbuf.sdl; if ((ifp = m->m_pkthdr.rcvif) && ifp->if_index && ifp->if_index <= V_if_index) { sdp = (struct sockaddr_dl *)ifp->if_addr->ifa_addr; /* * Change our mind and don't try copy. */ if (sdp->sdl_family != AF_LINK || sdp->sdl_len > sizeof(sdlbuf)) { goto makedummy; } bcopy(sdp, sdl2, sdp->sdl_len); } else { makedummy: sdl2->sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]); sdl2->sdl_family = AF_LINK; sdl2->sdl_index = 0; sdl2->sdl_nlen = sdl2->sdl_alen = sdl2->sdl_slen = 0; } *mp = sbcreatecontrol((caddr_t)sdl2, sdl2->sdl_len, IP_RECVIF, IPPROTO_IP); if (*mp) mp = &(*mp)->m_next; } if (inp->inp_flags & INP_RECVTOS) { *mp = sbcreatecontrol((caddr_t)&ip->ip_tos, sizeof(u_char), IP_RECVTOS, IPPROTO_IP); if (*mp) mp = &(*mp)->m_next; } if (inp->inp_flags2 & INP_RECVFLOWID) { uint32_t flowid, flow_type; flowid = m->m_pkthdr.flowid; flow_type = M_HASHTYPE_GET(m); /* * XXX should handle the failure of one or the * other - don't populate both? */ *mp = sbcreatecontrol((caddr_t) &flowid, sizeof(uint32_t), IP_FLOWID, IPPROTO_IP); if (*mp) mp = &(*mp)->m_next; *mp = sbcreatecontrol((caddr_t) &flow_type, sizeof(uint32_t), IP_FLOWTYPE, IPPROTO_IP); if (*mp) mp = &(*mp)->m_next; } #ifdef RSS if (inp->inp_flags2 & INP_RECVRSSBUCKETID) { uint32_t flowid, flow_type; uint32_t rss_bucketid; flowid = m->m_pkthdr.flowid; flow_type = M_HASHTYPE_GET(m); if (rss_hash2bucket(flowid, flow_type, &rss_bucketid) == 0) { *mp = sbcreatecontrol((caddr_t) &rss_bucketid, sizeof(uint32_t), IP_RSSBUCKETID, IPPROTO_IP); if (*mp) mp = &(*mp)->m_next; } } #endif } /* * XXXRW: Multicast routing code in ip_mroute.c is generally MPSAFE, but the * ip_rsvp and ip_rsvp_on variables need to be interlocked with rsvp_on * locking. This code remains in ip_input.c as ip_mroute.c is optionally * compiled. */ static VNET_DEFINE(int, ip_rsvp_on); VNET_DEFINE(struct socket *, ip_rsvpd); #define V_ip_rsvp_on VNET(ip_rsvp_on) int ip_rsvp_init(struct socket *so) { if (so->so_type != SOCK_RAW || so->so_proto->pr_protocol != IPPROTO_RSVP) return EOPNOTSUPP; if (V_ip_rsvpd != NULL) return EADDRINUSE; V_ip_rsvpd = so; /* * This may seem silly, but we need to be sure we don't over-increment * the RSVP counter, in case something slips up. */ if (!V_ip_rsvp_on) { V_ip_rsvp_on = 1; V_rsvp_on++; } return 0; } int ip_rsvp_done(void) { V_ip_rsvpd = NULL; /* * This may seem silly, but we need to be sure we don't over-decrement * the RSVP counter, in case something slips up. */ if (V_ip_rsvp_on) { V_ip_rsvp_on = 0; V_rsvp_on--; } return 0; } int rsvp_input(struct mbuf **mp, int *offp, int proto) { struct mbuf *m; m = *mp; *mp = NULL; if (rsvp_input_p) { /* call the real one if loaded */ *mp = m; rsvp_input_p(mp, offp, proto); return (IPPROTO_DONE); } /* Can still get packets with rsvp_on = 0 if there is a local member * of the group to which the RSVP packet is addressed. But in this * case we want to throw the packet away. */ if (!V_rsvp_on) { m_freem(m); return (IPPROTO_DONE); } if (V_ip_rsvpd != NULL) { *mp = m; rip_input(mp, offp, proto); return (IPPROTO_DONE); } /* Drop the packet */ m_freem(m); return (IPPROTO_DONE); }