/* SPDX-License-Identifier: LGPL-2.1-or-later */ #include #include "af-list.h" #include "alloc-util.h" #include "dns-domain.h" #include "errno-util.h" #include "fd-util.h" #include "hostname-util.h" #include "missing_network.h" #include "random-util.h" #include "resolved-dnssd.h" #include "resolved-dns-scope.h" #include "resolved-dns-zone.h" #include "resolved-llmnr.h" #include "resolved-mdns.h" #include "socket-util.h" #include "strv.h" #define MULTICAST_RATELIMIT_INTERVAL_USEC (1*USEC_PER_SEC) #define MULTICAST_RATELIMIT_BURST 1000 /* After how much time to repeat LLMNR requests, see RFC 4795 Section 7 */ #define MULTICAST_RESEND_TIMEOUT_MIN_USEC (100 * USEC_PER_MSEC) #define MULTICAST_RESEND_TIMEOUT_MAX_USEC (1 * USEC_PER_SEC) int dns_scope_new(Manager *m, DnsScope **ret, Link *l, DnsProtocol protocol, int family) { DnsScope *s; assert(m); assert(ret); s = new(DnsScope, 1); if (!s) return -ENOMEM; *s = (DnsScope) { .manager = m, .link = l, .protocol = protocol, .family = family, .resend_timeout = MULTICAST_RESEND_TIMEOUT_MIN_USEC, }; if (protocol == DNS_PROTOCOL_DNS) { /* Copy DNSSEC mode from the link if it is set there, * otherwise take the manager's DNSSEC mode. Note that * we copy this only at scope creation time, and do * not update it from the on, even if the setting * changes. */ if (l) { s->dnssec_mode = link_get_dnssec_mode(l); s->dns_over_tls_mode = link_get_dns_over_tls_mode(l); } else { s->dnssec_mode = manager_get_dnssec_mode(m); s->dns_over_tls_mode = manager_get_dns_over_tls_mode(m); } } else { s->dnssec_mode = DNSSEC_NO; s->dns_over_tls_mode = DNS_OVER_TLS_NO; } LIST_PREPEND(scopes, m->dns_scopes, s); dns_scope_llmnr_membership(s, true); dns_scope_mdns_membership(s, true); log_debug("New scope on link %s, protocol %s, family %s", l ? l->ifname : "*", dns_protocol_to_string(protocol), family == AF_UNSPEC ? "*" : af_to_name(family)); /* Enforce ratelimiting for the multicast protocols */ s->ratelimit = (RateLimit) { MULTICAST_RATELIMIT_INTERVAL_USEC, MULTICAST_RATELIMIT_BURST }; *ret = s; return 0; } static void dns_scope_abort_transactions(DnsScope *s) { assert(s); while (s->transactions) { DnsTransaction *t = s->transactions; /* Abort the transaction, but make sure it is not * freed while we still look at it */ t->block_gc++; if (DNS_TRANSACTION_IS_LIVE(t->state)) dns_transaction_complete(t, DNS_TRANSACTION_ABORTED); t->block_gc--; dns_transaction_free(t); } } DnsScope* dns_scope_free(DnsScope *s) { if (!s) return NULL; log_debug("Removing scope on link %s, protocol %s, family %s", s->link ? s->link->ifname : "*", dns_protocol_to_string(s->protocol), s->family == AF_UNSPEC ? "*" : af_to_name(s->family)); dns_scope_llmnr_membership(s, false); dns_scope_mdns_membership(s, false); dns_scope_abort_transactions(s); while (s->query_candidates) dns_query_candidate_unref(s->query_candidates); hashmap_free(s->transactions_by_key); ordered_hashmap_free_with_destructor(s->conflict_queue, dns_resource_record_unref); sd_event_source_disable_unref(s->conflict_event_source); sd_event_source_disable_unref(s->announce_event_source); dns_cache_flush(&s->cache); dns_zone_flush(&s->zone); LIST_REMOVE(scopes, s->manager->dns_scopes, s); return mfree(s); } DnsServer *dns_scope_get_dns_server(DnsScope *s) { assert(s); if (s->protocol != DNS_PROTOCOL_DNS) return NULL; if (s->link) return link_get_dns_server(s->link); else return manager_get_dns_server(s->manager); } unsigned dns_scope_get_n_dns_servers(DnsScope *s) { unsigned n = 0; DnsServer *i; assert(s); if (s->protocol != DNS_PROTOCOL_DNS) return 0; if (s->link) i = s->link->dns_servers; else i = s->manager->dns_servers; for (; i; i = i->servers_next) n++; return n; } void dns_scope_next_dns_server(DnsScope *s, DnsServer *if_current) { assert(s); if (s->protocol != DNS_PROTOCOL_DNS) return; /* Changes to the next DNS server in the list. If 'if_current' is passed will do so only if the * current DNS server still matches it. */ if (s->link) link_next_dns_server(s->link, if_current); else manager_next_dns_server(s->manager, if_current); } void dns_scope_packet_received(DnsScope *s, usec_t rtt) { assert(s); if (rtt <= s->max_rtt) return; s->max_rtt = rtt; s->resend_timeout = MIN(MAX(MULTICAST_RESEND_TIMEOUT_MIN_USEC, s->max_rtt * 2), MULTICAST_RESEND_TIMEOUT_MAX_USEC); } void dns_scope_packet_lost(DnsScope *s, usec_t usec) { assert(s); if (s->resend_timeout <= usec) s->resend_timeout = MIN(s->resend_timeout * 2, MULTICAST_RESEND_TIMEOUT_MAX_USEC); } static int dns_scope_emit_one(DnsScope *s, int fd, int family, DnsPacket *p) { int r; assert(s); assert(p); assert(p->protocol == s->protocol); if (family == AF_UNSPEC) { if (s->family == AF_UNSPEC) return -EAFNOSUPPORT; family = s->family; } switch (s->protocol) { case DNS_PROTOCOL_DNS: { size_t mtu, udp_size, min_mtu, socket_mtu = 0; assert(fd >= 0); if (DNS_PACKET_QDCOUNT(p) > 1) /* Classic DNS only allows one question per packet */ return -EOPNOTSUPP; if (p->size > DNS_PACKET_UNICAST_SIZE_MAX) return -EMSGSIZE; /* Determine the local most accurate MTU */ if (s->link) mtu = s->link->mtu; else mtu = manager_find_mtu(s->manager); /* Acquire the socket's PMDU MTU */ r = socket_get_mtu(fd, family, &socket_mtu); if (r < 0 && !ERRNO_IS_DISCONNECT(r)) /* Will return ENOTCONN if no information is available yet */ return log_debug_errno(r, "Failed to read socket MTU: %m"); /* Determine the appropriate UDP header size */ udp_size = udp_header_size(family); min_mtu = udp_size + DNS_PACKET_HEADER_SIZE; log_debug("Emitting UDP, link MTU is %zu, socket MTU is %zu, minimal MTU is %zu", mtu, socket_mtu, min_mtu); /* Clamp by the kernel's idea of the (path) MTU */ if (socket_mtu != 0 && socket_mtu < mtu) mtu = socket_mtu; /* Put a lower limit, in case all MTU data we acquired was rubbish */ if (mtu < min_mtu) mtu = min_mtu; /* Now check our packet size against the MTU we determined */ if (udp_size + p->size > mtu) return -EMSGSIZE; /* This means: try TCP instead */ r = manager_write(s->manager, fd, p); if (r < 0) return r; break; } case DNS_PROTOCOL_LLMNR: { union in_addr_union addr; assert(fd < 0); if (DNS_PACKET_QDCOUNT(p) > 1) return -EOPNOTSUPP; if (!ratelimit_below(&s->ratelimit)) return -EBUSY; if (family == AF_INET) { addr.in = LLMNR_MULTICAST_IPV4_ADDRESS; fd = manager_llmnr_ipv4_udp_fd(s->manager); } else if (family == AF_INET6) { addr.in6 = LLMNR_MULTICAST_IPV6_ADDRESS; fd = manager_llmnr_ipv6_udp_fd(s->manager); } else return -EAFNOSUPPORT; if (fd < 0) return fd; r = manager_send(s->manager, fd, s->link->ifindex, family, &addr, LLMNR_PORT, NULL, p); if (r < 0) return r; break; } case DNS_PROTOCOL_MDNS: { union in_addr_union addr; assert(fd < 0); if (!ratelimit_below(&s->ratelimit)) return -EBUSY; if (family == AF_INET) { if (in4_addr_is_null(&p->destination.in)) addr.in = MDNS_MULTICAST_IPV4_ADDRESS; else addr = p->destination; fd = manager_mdns_ipv4_fd(s->manager); } else if (family == AF_INET6) { if (in6_addr_is_null(&p->destination.in6)) addr.in6 = MDNS_MULTICAST_IPV6_ADDRESS; else addr = p->destination; fd = manager_mdns_ipv6_fd(s->manager); } else return -EAFNOSUPPORT; if (fd < 0) return fd; r = manager_send(s->manager, fd, s->link->ifindex, family, &addr, p->destination_port ?: MDNS_PORT, NULL, p); if (r < 0) return r; break; } default: return -EAFNOSUPPORT; } return 1; } int dns_scope_emit_udp(DnsScope *s, int fd, int af, DnsPacket *p) { int r; assert(s); assert(p); assert(p->protocol == s->protocol); assert((s->protocol == DNS_PROTOCOL_DNS) == (fd >= 0)); do { /* If there are multiple linked packets, set the TC bit in all but the last of them */ if (p->more) { assert(p->protocol == DNS_PROTOCOL_MDNS); dns_packet_set_flags(p, true, true); } r = dns_scope_emit_one(s, fd, af, p); if (r < 0) return r; p = p->more; } while (p); return 0; } static int dns_scope_socket( DnsScope *s, int type, int family, const union in_addr_union *address, DnsServer *server, uint16_t port, union sockaddr_union *ret_socket_address) { _cleanup_close_ int fd = -1; union sockaddr_union sa; socklen_t salen; int r, ifindex; assert(s); if (server) { assert(family == AF_UNSPEC); assert(!address); ifindex = dns_server_ifindex(server); switch (server->family) { case AF_INET: sa = (union sockaddr_union) { .in.sin_family = server->family, .in.sin_port = htobe16(port), .in.sin_addr = server->address.in, }; salen = sizeof(sa.in); break; case AF_INET6: sa = (union sockaddr_union) { .in6.sin6_family = server->family, .in6.sin6_port = htobe16(port), .in6.sin6_addr = server->address.in6, .in6.sin6_scope_id = ifindex, }; salen = sizeof(sa.in6); break; default: return -EAFNOSUPPORT; } } else { assert(family != AF_UNSPEC); assert(address); ifindex = s->link ? s->link->ifindex : 0; switch (family) { case AF_INET: sa = (union sockaddr_union) { .in.sin_family = family, .in.sin_port = htobe16(port), .in.sin_addr = address->in, }; salen = sizeof(sa.in); break; case AF_INET6: sa = (union sockaddr_union) { .in6.sin6_family = family, .in6.sin6_port = htobe16(port), .in6.sin6_addr = address->in6, .in6.sin6_scope_id = ifindex, }; salen = sizeof(sa.in6); break; default: return -EAFNOSUPPORT; } } fd = socket(sa.sa.sa_family, type|SOCK_CLOEXEC|SOCK_NONBLOCK, 0); if (fd < 0) return -errno; if (type == SOCK_STREAM) { r = setsockopt_int(fd, IPPROTO_TCP, TCP_NODELAY, true); if (r < 0) return r; } if (ifindex != 0) { r = socket_set_unicast_if(fd, sa.sa.sa_family, ifindex); if (r < 0) return r; } if (s->protocol == DNS_PROTOCOL_LLMNR) { /* RFC 4795, section 2.5 requires the TTL to be set to 1 */ r = socket_set_ttl(fd, sa.sa.sa_family, 1); if (r < 0) return r; } if (type == SOCK_DGRAM) { /* Set IP_RECVERR or IPV6_RECVERR to get ICMP error feedback. See discussion in #10345. */ r = socket_set_recverr(fd, sa.sa.sa_family, true); if (r < 0) return r; r = socket_set_recvpktinfo(fd, sa.sa.sa_family, true); if (r < 0) return r; /* Turn of path MTU discovery for security reasons */ r = socket_disable_pmtud(fd, sa.sa.sa_family); if (r < 0) log_debug_errno(r, "Failed to disable UDP PMTUD, ignoring: %m"); /* Learn about fragmentation taking place */ r = socket_set_recvfragsize(fd, sa.sa.sa_family, true); if (r < 0) log_debug_errno(r, "Failed to enable fragment size reception, ignoring: %m"); } if (ret_socket_address) *ret_socket_address = sa; else { bool bound = false; /* Let's temporarily bind the socket to the specified ifindex. The kernel currently takes * only the SO_BINDTODEVICE/SO_BINDTOINDEX ifindex into account when making routing decisions * in connect() — and not IP_UNICAST_IF. We don't really want any of the other semantics of * SO_BINDTODEVICE/SO_BINDTOINDEX, hence we immediately unbind the socket after the fact * again. * * As a special exception we don't do this if we notice that the specified IP address is on * the local host. SO_BINDTODEVICE in combination with destination addresses on the local * host result in EHOSTUNREACH, since Linux won't send the packets out of the specified * interface, but delivers them directly to the local socket. */ if (s->link && !manager_find_link_address(s->manager, sa.sa.sa_family, sockaddr_in_addr(&sa.sa)) && in_addr_is_localhost(sa.sa.sa_family, sockaddr_in_addr(&sa.sa)) == 0) { r = socket_bind_to_ifindex(fd, ifindex); if (r < 0) return r; bound = true; } r = connect(fd, &sa.sa, salen); if (r < 0 && errno != EINPROGRESS) return -errno; if (bound) { r = socket_bind_to_ifindex(fd, 0); if (r < 0) return r; } } return TAKE_FD(fd); } int dns_scope_socket_udp(DnsScope *s, DnsServer *server) { return dns_scope_socket(s, SOCK_DGRAM, AF_UNSPEC, NULL, server, dns_server_port(server), NULL); } int dns_scope_socket_tcp(DnsScope *s, int family, const union in_addr_union *address, DnsServer *server, uint16_t port, union sockaddr_union *ret_socket_address) { /* If ret_socket_address is not NULL, the caller is responsible * for calling connect() or sendmsg(). This is required by TCP * Fast Open, to be able to send the initial SYN packet along * with the first data packet. */ return dns_scope_socket(s, SOCK_STREAM, family, address, server, port, ret_socket_address); } static DnsScopeMatch match_link_local_reverse_lookups(const char *domain) { assert(domain); if (dns_name_endswith(domain, "254.169.in-addr.arpa") > 0) return DNS_SCOPE_YES_BASE + 4; /* 4 labels match */ if (dns_name_endswith(domain, "8.e.f.ip6.arpa") > 0 || dns_name_endswith(domain, "9.e.f.ip6.arpa") > 0 || dns_name_endswith(domain, "a.e.f.ip6.arpa") > 0 || dns_name_endswith(domain, "b.e.f.ip6.arpa") > 0) return DNS_SCOPE_YES_BASE + 5; /* 5 labels match */ return _DNS_SCOPE_MATCH_INVALID; } static DnsScopeMatch match_subnet_reverse_lookups( DnsScope *s, const char *domain, bool exclude_own) { union in_addr_union ia; int f, r; assert(s); assert(domain); /* Checks whether the specified domain is a reverse address domain (i.e. in the .in-addr.arpa or * .ip6.arpa area), and if so, whether the address matches any of the local subnets of the link the * scope is associated with. If so, our scope should consider itself relevant for any lookup in the * domain, since it apparently refers to hosts on this link's subnet. * * If 'exclude_own' is true this will return DNS_SCOPE_NO for any IP addresses assigned locally. This * is useful for LLMNR/mDNS as we never want to look up our own hostname on LLMNR/mDNS but always use * the locally synthesized one. */ if (!s->link) return _DNS_SCOPE_MATCH_INVALID; /* No link, hence no local addresses to check */ r = dns_name_address(domain, &f, &ia); if (r < 0) log_debug_errno(r, "Failed to determine whether '%s' is an address domain: %m", domain); if (r <= 0) return _DNS_SCOPE_MATCH_INVALID; if (s->family != AF_UNSPEC && f != s->family) return _DNS_SCOPE_MATCH_INVALID; /* Don't look for IPv4 addresses on LLMNR/mDNS over IPv6 and vice versa */ LIST_FOREACH(addresses, a, s->link->addresses) { if (a->family != f) continue; /* Equals our own address? nah, let's not use this scope. The local synthesizer will pick it up for us. */ if (exclude_own && in_addr_equal(f, &a->in_addr, &ia) > 0) return DNS_SCOPE_NO; if (a->prefixlen == UCHAR_MAX) /* don't know subnet mask */ continue; /* Check if the address is in the local subnet */ r = in_addr_prefix_covers(f, &a->in_addr, a->prefixlen, &ia); if (r < 0) log_debug_errno(r, "Failed to determine whether link address covers lookup address '%s': %m", domain); if (r > 0) /* Note that we only claim zero labels match. This is so that this is at the same * priority a DNS scope with "." as routing domain is. */ return DNS_SCOPE_YES_BASE + 0; } return _DNS_SCOPE_MATCH_INVALID; } DnsScopeMatch dns_scope_good_domain( DnsScope *s, DnsQuery *q) { DnsQuestion *question; const char *domain; uint64_t flags; int ifindex; /* This returns the following return values: * * DNS_SCOPE_NO → This scope is not suitable for lookups of this domain, at all * DNS_SCOPE_MAYBE → This scope is suitable, but only if nothing else wants it * DNS_SCOPE_YES_BASE+n → This scope is suitable, and 'n' suffix labels match * * (The idea is that the caller will only use the scopes with the longest 'n' returned. If no scopes return * DNS_SCOPE_YES_BASE+n, then it should use those which returned DNS_SCOPE_MAYBE. It should never use those * which returned DNS_SCOPE_NO.) */ assert(s); assert(q); question = dns_query_question_for_protocol(q, s->protocol); if (!question) return DNS_SCOPE_NO; domain = dns_question_first_name(question); if (!domain) return DNS_SCOPE_NO; ifindex = q->ifindex; flags = q->flags; /* Checks if the specified domain is something to look up on this scope. Note that this accepts * non-qualified hostnames, i.e. those without any search path suffixed. */ if (ifindex != 0 && (!s->link || s->link->ifindex != ifindex)) return DNS_SCOPE_NO; if ((SD_RESOLVED_FLAGS_MAKE(s->protocol, s->family, false, false) & flags) == 0) return DNS_SCOPE_NO; /* Never resolve any loopback hostname or IP address via DNS, LLMNR or mDNS. Instead, always rely on * synthesized RRs for these. */ if (is_localhost(domain) || dns_name_endswith(domain, "127.in-addr.arpa") > 0 || dns_name_equal(domain, "1.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.ip6.arpa") > 0) return DNS_SCOPE_NO; /* Never respond to some of the domains listed in RFC6303 + RFC6761 */ if (dns_name_dont_resolve(domain)) return DNS_SCOPE_NO; /* Never go to network for the _gateway or _outbound domain — they're something special, synthesized locally. */ if (is_gateway_hostname(domain) || is_outbound_hostname(domain)) return DNS_SCOPE_NO; switch (s->protocol) { case DNS_PROTOCOL_DNS: { bool has_search_domains = false; DnsScopeMatch m; int n_best = -1; if (dns_name_is_empty(domain)) { DnsResourceKey *t; bool found = false; /* Refuse empty name if only A and/or AAAA records are requested. */ DNS_QUESTION_FOREACH(t, question) if (!IN_SET(t->type, DNS_TYPE_A, DNS_TYPE_AAAA)) { found = true; break; } if (!found) return DNS_SCOPE_NO; } /* Never route things to scopes that lack DNS servers */ if (!dns_scope_get_dns_server(s)) return DNS_SCOPE_NO; /* Always honour search domains for routing queries, except if this scope lacks DNS servers. Note that * we return DNS_SCOPE_YES here, rather than just DNS_SCOPE_MAYBE, which means other wildcard scopes * won't be considered anymore. */ LIST_FOREACH(domains, d, dns_scope_get_search_domains(s)) { if (!d->route_only && !dns_name_is_root(d->name)) has_search_domains = true; if (dns_name_endswith(domain, d->name) > 0) { int c; c = dns_name_count_labels(d->name); if (c < 0) continue; if (c > n_best) n_best = c; } } /* If there's a true search domain defined for this scope, and the query is single-label, * then let's resolve things here, prefereably. Note that LLMNR considers itself * authoritative for single-label names too, at the same preference, see below. */ if (has_search_domains && dns_name_is_single_label(domain)) return DNS_SCOPE_YES_BASE + 1; /* If ResolveUnicastSingleLabel=yes and the query is single-label, then bump match result to prevent LLMNR monopoly among candidates. */ if (s->manager->resolve_unicast_single_label && dns_name_is_single_label(domain)) return DNS_SCOPE_YES_BASE + 1; /* Let's return the number of labels in the best matching result */ if (n_best >= 0) { assert(n_best <= DNS_SCOPE_YES_END - DNS_SCOPE_YES_BASE); return DNS_SCOPE_YES_BASE + n_best; } /* Exclude link-local IP ranges */ if (match_link_local_reverse_lookups(domain) >= DNS_SCOPE_YES_BASE || /* If networks use .local in their private setups, they are supposed to also add .local * to their search domains, which we already checked above. Otherwise, we consider .local * specific to mDNS and won't send such queries ordinary DNS servers. */ dns_name_endswith(domain, "local") > 0) return DNS_SCOPE_NO; /* If the IP address to look up matches the local subnet, then implicitly synthesizes * DNS_SCOPE_YES_BASE + 0 on this interface, i.e. preferably resolve IP addresses via the DNS * server belonging to this interface. */ m = match_subnet_reverse_lookups(s, domain, false); if (m >= 0) return m; /* If there was no match at all, then see if this scope is suitable as default route. */ if (!dns_scope_is_default_route(s)) return DNS_SCOPE_NO; return DNS_SCOPE_MAYBE; } case DNS_PROTOCOL_MDNS: { DnsScopeMatch m; m = match_link_local_reverse_lookups(domain); if (m >= 0) return m; m = match_subnet_reverse_lookups(s, domain, true); if (m >= 0) return m; if ((s->family == AF_INET && dns_name_endswith(domain, "in-addr.arpa") > 0) || (s->family == AF_INET6 && dns_name_endswith(domain, "ip6.arpa") > 0)) return DNS_SCOPE_MAYBE; if ((dns_name_endswith(domain, "local") > 0 && /* only resolve names ending in .local via mDNS */ dns_name_equal(domain, "local") == 0 && /* but not the single-label "local" name itself */ manager_is_own_hostname(s->manager, domain) <= 0)) /* never resolve the local hostname via mDNS */ return DNS_SCOPE_YES_BASE + 1; /* Return +1, as the top-level .local domain matches, i.e. one label */ return DNS_SCOPE_NO; } case DNS_PROTOCOL_LLMNR: { DnsScopeMatch m; m = match_link_local_reverse_lookups(domain); if (m >= 0) return m; m = match_subnet_reverse_lookups(s, domain, true); if (m >= 0) return m; if ((s->family == AF_INET && dns_name_endswith(domain, "in-addr.arpa") > 0) || (s->family == AF_INET6 && dns_name_endswith(domain, "ip6.arpa") > 0)) return DNS_SCOPE_MAYBE; if ((dns_name_is_single_label(domain) && /* only resolve single label names via LLMNR */ !is_gateway_hostname(domain) && /* don't resolve "_gateway" with LLMNR, let local synthesizing logic handle that */ !is_outbound_hostname(domain) && /* similar for "_outbound" */ dns_name_equal(domain, "local") == 0 && /* don't resolve "local" with LLMNR, it's the top-level domain of mDNS after all, see above */ manager_is_own_hostname(s->manager, domain) <= 0)) /* never resolve the local hostname via LLMNR */ return DNS_SCOPE_YES_BASE + 1; /* Return +1, as we consider ourselves authoritative * for single-label names, i.e. one label. This is * particularly relevant as it means a "." route on some * other scope won't pull all traffic away from * us. (If people actually want to pull traffic away * from us they should turn off LLMNR on the * link). Note that unicast DNS scopes with search * domains also consider themselves authoritative for * single-label domains, at the same preference (see * above). */ return DNS_SCOPE_NO; } default: assert_not_reached(); } } bool dns_scope_good_key(DnsScope *s, const DnsResourceKey *key) { int key_family; assert(s); assert(key); /* Check if it makes sense to resolve the specified key on this scope. Note that this call assumes a * fully qualified name, i.e. the search suffixes already appended. */ if (!IN_SET(key->class, DNS_CLASS_IN, DNS_CLASS_ANY)) return false; if (s->protocol == DNS_PROTOCOL_DNS) { /* On classic DNS, looking up non-address RRs is always fine. (Specifically, we want to * permit looking up DNSKEY and DS records on the root and top-level domains.) */ if (!dns_resource_key_is_address(key)) return true; /* Unless explicitly overridden, we refuse to look up A and AAAA RRs on the root and * single-label domains, under the assumption that those should be resolved via LLMNR or * search path only, and should not be leaked onto the internet. */ const char* name = dns_resource_key_name(key); if (!s->manager->resolve_unicast_single_label && dns_name_is_single_label(name)) return false; return !dns_name_is_root(name); } /* Never route DNSSEC RR queries to LLMNR/mDNS scopes */ if (dns_type_is_dnssec(key->type)) return false; /* On mDNS and LLMNR, send A and AAAA queries only on the respective scopes */ key_family = dns_type_to_af(key->type); if (key_family < 0) return true; return key_family == s->family; } static int dns_scope_multicast_membership(DnsScope *s, bool b, struct in_addr in, struct in6_addr in6) { int fd; assert(s); assert(s->link); if (s->family == AF_INET) { struct ip_mreqn mreqn = { .imr_multiaddr = in, .imr_ifindex = s->link->ifindex, }; if (s->protocol == DNS_PROTOCOL_LLMNR) fd = manager_llmnr_ipv4_udp_fd(s->manager); else fd = manager_mdns_ipv4_fd(s->manager); if (fd < 0) return fd; /* Always first try to drop membership before we add * one. This is necessary on some devices, such as * veth. */ if (b) (void) setsockopt(fd, IPPROTO_IP, IP_DROP_MEMBERSHIP, &mreqn, sizeof(mreqn)); if (setsockopt(fd, IPPROTO_IP, b ? IP_ADD_MEMBERSHIP : IP_DROP_MEMBERSHIP, &mreqn, sizeof(mreqn)) < 0) return -errno; } else if (s->family == AF_INET6) { struct ipv6_mreq mreq = { .ipv6mr_multiaddr = in6, .ipv6mr_interface = s->link->ifindex, }; if (s->protocol == DNS_PROTOCOL_LLMNR) fd = manager_llmnr_ipv6_udp_fd(s->manager); else fd = manager_mdns_ipv6_fd(s->manager); if (fd < 0) return fd; if (b) (void) setsockopt(fd, IPPROTO_IPV6, IPV6_DROP_MEMBERSHIP, &mreq, sizeof(mreq)); if (setsockopt(fd, IPPROTO_IPV6, b ? IPV6_ADD_MEMBERSHIP : IPV6_DROP_MEMBERSHIP, &mreq, sizeof(mreq)) < 0) return -errno; } else return -EAFNOSUPPORT; return 0; } int dns_scope_llmnr_membership(DnsScope *s, bool b) { assert(s); if (s->protocol != DNS_PROTOCOL_LLMNR) return 0; return dns_scope_multicast_membership(s, b, LLMNR_MULTICAST_IPV4_ADDRESS, LLMNR_MULTICAST_IPV6_ADDRESS); } int dns_scope_mdns_membership(DnsScope *s, bool b) { assert(s); if (s->protocol != DNS_PROTOCOL_MDNS) return 0; return dns_scope_multicast_membership(s, b, MDNS_MULTICAST_IPV4_ADDRESS, MDNS_MULTICAST_IPV6_ADDRESS); } int dns_scope_make_reply_packet( DnsScope *s, uint16_t id, int rcode, DnsQuestion *q, DnsAnswer *answer, DnsAnswer *soa, bool tentative, DnsPacket **ret) { _cleanup_(dns_packet_unrefp) DnsPacket *p = NULL; unsigned n_answer = 0, n_soa = 0; int r; bool c_or_aa; assert(s); assert(ret); if (dns_question_isempty(q) && dns_answer_isempty(answer) && dns_answer_isempty(soa)) return -EINVAL; r = dns_packet_new(&p, s->protocol, 0, DNS_PACKET_SIZE_MAX); if (r < 0) return r; /* mDNS answers must have the Authoritative Answer bit set, see RFC 6762, section 18.4. */ c_or_aa = s->protocol == DNS_PROTOCOL_MDNS; DNS_PACKET_HEADER(p)->id = id; DNS_PACKET_HEADER(p)->flags = htobe16(DNS_PACKET_MAKE_FLAGS( 1 /* qr */, 0 /* opcode */, c_or_aa, 0 /* tc */, tentative, 0 /* (ra) */, 0 /* (ad) */, 0 /* (cd) */, rcode)); r = dns_packet_append_question(p, q); if (r < 0) return r; DNS_PACKET_HEADER(p)->qdcount = htobe16(dns_question_size(q)); r = dns_packet_append_answer(p, answer, &n_answer); if (r < 0) return r; DNS_PACKET_HEADER(p)->ancount = htobe16(n_answer); r = dns_packet_append_answer(p, soa, &n_soa); if (r < 0) return r; DNS_PACKET_HEADER(p)->arcount = htobe16(n_soa); *ret = TAKE_PTR(p); return 0; } static void dns_scope_verify_conflicts(DnsScope *s, DnsPacket *p) { DnsResourceRecord *rr; DnsResourceKey *key; assert(s); assert(p); DNS_QUESTION_FOREACH(key, p->question) dns_zone_verify_conflicts(&s->zone, key); DNS_ANSWER_FOREACH(rr, p->answer) dns_zone_verify_conflicts(&s->zone, rr->key); } void dns_scope_process_query(DnsScope *s, DnsStream *stream, DnsPacket *p) { _cleanup_(dns_answer_unrefp) DnsAnswer *answer = NULL, *soa = NULL; _cleanup_(dns_packet_unrefp) DnsPacket *reply = NULL; DnsResourceKey *key = NULL; bool tentative = false; int r; assert(s); assert(p); if (p->protocol != DNS_PROTOCOL_LLMNR) return; if (p->ipproto == IPPROTO_UDP) { /* Don't accept UDP queries directed to anything but * the LLMNR multicast addresses. See RFC 4795, * section 2.5. */ if (p->family == AF_INET && !in4_addr_equal(&p->destination.in, &LLMNR_MULTICAST_IPV4_ADDRESS)) return; if (p->family == AF_INET6 && !in6_addr_equal(&p->destination.in6, &LLMNR_MULTICAST_IPV6_ADDRESS)) return; } r = dns_packet_extract(p); if (r < 0) { log_debug_errno(r, "Failed to extract resource records from incoming packet: %m"); return; } if (DNS_PACKET_LLMNR_C(p)) { /* Somebody notified us about a possible conflict */ dns_scope_verify_conflicts(s, p); return; } if (dns_question_size(p->question) != 1) return (void) log_debug("Received LLMNR query without question or multiple questions, ignoring."); key = dns_question_first_key(p->question); r = dns_zone_lookup(&s->zone, key, 0, &answer, &soa, &tentative); if (r < 0) { log_debug_errno(r, "Failed to look up key: %m"); return; } if (r == 0) return; if (answer) dns_answer_order_by_scope(answer, in_addr_is_link_local(p->family, &p->sender) > 0); r = dns_scope_make_reply_packet(s, DNS_PACKET_ID(p), DNS_RCODE_SUCCESS, p->question, answer, soa, tentative, &reply); if (r < 0) { log_debug_errno(r, "Failed to build reply packet: %m"); return; } if (stream) { r = dns_stream_write_packet(stream, reply); if (r < 0) { log_debug_errno(r, "Failed to enqueue reply packet: %m"); return; } /* Let's take an extra reference on this stream, so that it stays around after returning. The reference * will be dangling until the stream is disconnected, and the default completion handler of the stream * will then unref the stream and destroy it */ if (DNS_STREAM_QUEUED(stream)) dns_stream_ref(stream); } else { int fd; if (!ratelimit_below(&s->ratelimit)) return; if (p->family == AF_INET) fd = manager_llmnr_ipv4_udp_fd(s->manager); else if (p->family == AF_INET6) fd = manager_llmnr_ipv6_udp_fd(s->manager); else { log_debug("Unknown protocol"); return; } if (fd < 0) { log_debug_errno(fd, "Failed to get reply socket: %m"); return; } /* Note that we always immediately reply to all LLMNR * requests, and do not wait any time, since we * verified uniqueness for all records. Also see RFC * 4795, Section 2.7 */ r = manager_send(s->manager, fd, p->ifindex, p->family, &p->sender, p->sender_port, NULL, reply); if (r < 0) { log_debug_errno(r, "Failed to send reply packet: %m"); return; } } } DnsTransaction *dns_scope_find_transaction( DnsScope *scope, DnsResourceKey *key, uint64_t query_flags) { DnsTransaction *first; assert(scope); assert(key); /* Iterate through the list of transactions with a matching key */ first = hashmap_get(scope->transactions_by_key, key); LIST_FOREACH(transactions_by_key, t, first) { /* These four flags must match exactly: we cannot use a validated response for a * non-validating client, and we cannot use a non-validated response for a validating * client. Similar, if the sources don't match things aren't usable either. */ if (((query_flags ^ t->query_flags) & (SD_RESOLVED_NO_VALIDATE| SD_RESOLVED_NO_ZONE| SD_RESOLVED_NO_TRUST_ANCHOR| SD_RESOLVED_NO_NETWORK)) != 0) continue; /* We can reuse a primary query if a regular one is requested, but not vice versa */ if ((query_flags & SD_RESOLVED_REQUIRE_PRIMARY) && !(t->query_flags & SD_RESOLVED_REQUIRE_PRIMARY)) continue; /* Don't reuse a transaction that allowed caching when we got told not to use it */ if ((query_flags & SD_RESOLVED_NO_CACHE) && !(t->query_flags & SD_RESOLVED_NO_CACHE)) continue; /* If we are asked to clamp ttls an the existing transaction doesn't do it, we can't * reuse */ if ((query_flags & SD_RESOLVED_CLAMP_TTL) && !(t->query_flags & SD_RESOLVED_CLAMP_TTL)) continue; return t; } return NULL; } static int dns_scope_make_conflict_packet( DnsScope *s, DnsResourceRecord *rr, DnsPacket **ret) { _cleanup_(dns_packet_unrefp) DnsPacket *p = NULL; int r; assert(s); assert(rr); assert(ret); r = dns_packet_new(&p, s->protocol, 0, DNS_PACKET_SIZE_MAX); if (r < 0) return r; DNS_PACKET_HEADER(p)->flags = htobe16(DNS_PACKET_MAKE_FLAGS( 0 /* qr */, 0 /* opcode */, 1 /* conflict */, 0 /* tc */, 0 /* t */, 0 /* (ra) */, 0 /* (ad) */, 0 /* (cd) */, 0)); /* For mDNS, the transaction ID should always be 0 */ if (s->protocol != DNS_PROTOCOL_MDNS) random_bytes(&DNS_PACKET_HEADER(p)->id, sizeof(uint16_t)); DNS_PACKET_HEADER(p)->qdcount = htobe16(1); DNS_PACKET_HEADER(p)->arcount = htobe16(1); r = dns_packet_append_key(p, rr->key, 0, NULL); if (r < 0) return r; r = dns_packet_append_rr(p, rr, 0, NULL, NULL); if (r < 0) return r; *ret = TAKE_PTR(p); return 0; } static int on_conflict_dispatch(sd_event_source *es, usec_t usec, void *userdata) { DnsScope *scope = ASSERT_PTR(userdata); int r; assert(es); scope->conflict_event_source = sd_event_source_disable_unref(scope->conflict_event_source); for (;;) { _cleanup_(dns_resource_key_unrefp) DnsResourceKey *key = NULL; _cleanup_(dns_resource_record_unrefp) DnsResourceRecord *rr = NULL; _cleanup_(dns_packet_unrefp) DnsPacket *p = NULL; key = ordered_hashmap_first_key(scope->conflict_queue); if (!key) break; rr = ordered_hashmap_remove(scope->conflict_queue, key); assert(rr); r = dns_scope_make_conflict_packet(scope, rr, &p); if (r < 0) { log_error_errno(r, "Failed to make conflict packet: %m"); return 0; } r = dns_scope_emit_udp(scope, -1, AF_UNSPEC, p); if (r < 0) log_debug_errno(r, "Failed to send conflict packet: %m"); } return 0; } int dns_scope_notify_conflict(DnsScope *scope, DnsResourceRecord *rr) { int r; assert(scope); assert(rr); /* We don't send these queries immediately. Instead, we queue * them, and send them after some jitter delay. */ r = ordered_hashmap_ensure_allocated(&scope->conflict_queue, &dns_resource_key_hash_ops); if (r < 0) { log_oom(); return r; } /* We only place one RR per key in the conflict * messages, not all of them. That should be enough to * indicate where there might be a conflict */ r = ordered_hashmap_put(scope->conflict_queue, rr->key, rr); if (IN_SET(r, 0, -EEXIST)) return 0; if (r < 0) return log_debug_errno(r, "Failed to queue conflicting RR: %m"); dns_resource_key_ref(rr->key); dns_resource_record_ref(rr); if (scope->conflict_event_source) return 0; r = sd_event_add_time_relative( scope->manager->event, &scope->conflict_event_source, CLOCK_BOOTTIME, random_u64_range(LLMNR_JITTER_INTERVAL_USEC), 0, on_conflict_dispatch, scope); if (r < 0) return log_debug_errno(r, "Failed to add conflict dispatch event: %m"); (void) sd_event_source_set_description(scope->conflict_event_source, "scope-conflict"); return 0; } void dns_scope_check_conflicts(DnsScope *scope, DnsPacket *p) { DnsResourceRecord *rr; int r; assert(scope); assert(p); if (!IN_SET(p->protocol, DNS_PROTOCOL_LLMNR, DNS_PROTOCOL_MDNS)) return; if (DNS_PACKET_RRCOUNT(p) <= 0) return; if (p->protocol == DNS_PROTOCOL_LLMNR) { if (DNS_PACKET_LLMNR_C(p) != 0) return; if (DNS_PACKET_LLMNR_T(p) != 0) return; } if (manager_packet_from_local_address(scope->manager, p)) return; r = dns_packet_extract(p); if (r < 0) { log_debug_errno(r, "Failed to extract packet: %m"); return; } log_debug("Checking for conflicts..."); DNS_ANSWER_FOREACH(rr, p->answer) { /* No conflict if it is DNS-SD RR used for service enumeration. */ if (dns_resource_key_is_dnssd_ptr(rr->key)) continue; /* Check for conflicts against the local zone. If we * found one, we won't check any further */ r = dns_zone_check_conflicts(&scope->zone, rr); if (r != 0) continue; /* Check for conflicts against the local cache. If so, * send out an advisory query, to inform everybody */ r = dns_cache_check_conflicts(&scope->cache, rr, p->family, &p->sender); if (r <= 0) continue; dns_scope_notify_conflict(scope, rr); } } void dns_scope_dump(DnsScope *s, FILE *f) { assert(s); if (!f) f = stdout; fputs("[Scope protocol=", f); fputs(dns_protocol_to_string(s->protocol), f); if (s->link) { fputs(" interface=", f); fputs(s->link->ifname, f); } if (s->family != AF_UNSPEC) { fputs(" family=", f); fputs(af_to_name(s->family), f); } fputs("]\n", f); if (!dns_zone_is_empty(&s->zone)) { fputs("ZONE:\n", f); dns_zone_dump(&s->zone, f); } if (!dns_cache_is_empty(&s->cache)) { fputs("CACHE:\n", f); dns_cache_dump(&s->cache, f); } } DnsSearchDomain *dns_scope_get_search_domains(DnsScope *s) { assert(s); if (s->protocol != DNS_PROTOCOL_DNS) return NULL; if (s->link) return s->link->search_domains; return s->manager->search_domains; } bool dns_scope_name_wants_search_domain(DnsScope *s, const char *name) { assert(s); if (s->protocol != DNS_PROTOCOL_DNS) return false; return dns_name_is_single_label(name); } bool dns_scope_network_good(DnsScope *s) { /* Checks whether the network is in good state for lookups on this scope. For mDNS/LLMNR/Classic DNS scopes * bound to links this is easy, as they don't even exist if the link isn't in a suitable state. For the global * DNS scope we check whether there are any links that are up and have an address. * * Note that Linux routing is complex and even systems that superficially have no IPv4 address might * be able to route IPv4 (and similar for IPv6), hence let's make a check here independent of address * family. */ if (s->link) return true; return manager_routable(s->manager); } int dns_scope_ifindex(DnsScope *s) { assert(s); if (s->link) return s->link->ifindex; return 0; } static int on_announcement_timeout(sd_event_source *s, usec_t usec, void *userdata) { DnsScope *scope = userdata; assert(s); scope->announce_event_source = sd_event_source_disable_unref(scope->announce_event_source); (void) dns_scope_announce(scope, false); return 0; } int dns_scope_announce(DnsScope *scope, bool goodbye) { _cleanup_(dns_answer_unrefp) DnsAnswer *answer = NULL; _cleanup_(dns_packet_unrefp) DnsPacket *p = NULL; _cleanup_set_free_ Set *types = NULL; DnsZoneItem *z; unsigned size = 0; char *service_type; int r; if (!scope) return 0; if (scope->protocol != DNS_PROTOCOL_MDNS) return 0; r = sd_event_get_state(scope->manager->event); if (r < 0) return log_debug_errno(r, "Failed to get event loop state: %m"); /* If this is called on exit, through manager_free() -> link_free(), then we cannot announce. */ if (r == SD_EVENT_FINISHED) return 0; /* Check if we're done with probing. */ LIST_FOREACH(transactions_by_scope, t, scope->transactions) if (DNS_TRANSACTION_IS_LIVE(t->state)) return 0; /* Check if there're services pending conflict resolution. */ if (manager_next_dnssd_names(scope->manager)) return 0; /* we reach this point only if changing hostname didn't help */ /* Calculate answer's size. */ HASHMAP_FOREACH(z, scope->zone.by_key) { if (z->state != DNS_ZONE_ITEM_ESTABLISHED) continue; if (z->rr->key->type == DNS_TYPE_PTR && !dns_zone_contains_name(&scope->zone, z->rr->ptr.name)) { char key_str[DNS_RESOURCE_KEY_STRING_MAX]; log_debug("Skip PTR RR <%s> since its counterparts seem to be withdrawn", dns_resource_key_to_string(z->rr->key, key_str, sizeof key_str)); z->state = DNS_ZONE_ITEM_WITHDRAWN; continue; } /* Collect service types for _services._dns-sd._udp.local RRs in a set */ if (!scope->announced && dns_resource_key_is_dnssd_ptr(z->rr->key)) { if (!set_contains(types, dns_resource_key_name(z->rr->key))) { r = set_ensure_put(&types, &dns_name_hash_ops, dns_resource_key_name(z->rr->key)); if (r < 0) return log_debug_errno(r, "Failed to add item to set: %m"); } } LIST_FOREACH(by_key, i, z) size++; } answer = dns_answer_new(size + set_size(types)); if (!answer) return log_oom(); /* Second iteration, actually add RRs to the answer. */ HASHMAP_FOREACH(z, scope->zone.by_key) LIST_FOREACH (by_key, i, z) { DnsAnswerFlags flags; if (i->state != DNS_ZONE_ITEM_ESTABLISHED) continue; if (dns_resource_key_is_dnssd_ptr(i->rr->key)) flags = goodbye ? DNS_ANSWER_GOODBYE : 0; else flags = goodbye ? (DNS_ANSWER_GOODBYE|DNS_ANSWER_CACHE_FLUSH) : DNS_ANSWER_CACHE_FLUSH; r = dns_answer_add(answer, i->rr, 0, flags, NULL); if (r < 0) return log_debug_errno(r, "Failed to add RR to announce: %m"); } /* Since all the active services are in the zone make them discoverable now. */ SET_FOREACH(service_type, types) { _cleanup_(dns_resource_record_unrefp) DnsResourceRecord *rr = NULL; rr = dns_resource_record_new_full(DNS_CLASS_IN, DNS_TYPE_PTR, "_services._dns-sd._udp.local"); if (!rr) return log_oom(); rr->ptr.name = strdup(service_type); if (!rr->ptr.name) return log_oom(); rr->ttl = MDNS_DEFAULT_TTL; r = dns_zone_put(&scope->zone, scope, rr, false); if (r < 0) log_warning_errno(r, "Failed to add DNS-SD PTR record to MDNS zone, ignoring: %m"); r = dns_answer_add(answer, rr, 0, 0, NULL); if (r < 0) return log_debug_errno(r, "Failed to add RR to announce: %m"); } if (dns_answer_isempty(answer)) return 0; r = dns_scope_make_reply_packet(scope, 0, DNS_RCODE_SUCCESS, NULL, answer, NULL, false, &p); if (r < 0) return log_debug_errno(r, "Failed to build reply packet: %m"); r = dns_scope_emit_udp(scope, -1, AF_UNSPEC, p); if (r < 0) return log_debug_errno(r, "Failed to send reply packet: %m"); /* In section 8.3 of RFC6762: "The Multicast DNS responder MUST send at least two unsolicited * responses, one second apart." */ if (!scope->announced) { scope->announced = true; r = sd_event_add_time_relative( scope->manager->event, &scope->announce_event_source, CLOCK_BOOTTIME, MDNS_ANNOUNCE_DELAY, 0, on_announcement_timeout, scope); if (r < 0) return log_debug_errno(r, "Failed to schedule second announcement: %m"); (void) sd_event_source_set_description(scope->announce_event_source, "mdns-announce"); } return 0; } int dns_scope_add_dnssd_services(DnsScope *scope) { DnssdService *service; int r; assert(scope); if (hashmap_size(scope->manager->dnssd_services) == 0) return 0; scope->announced = false; HASHMAP_FOREACH(service, scope->manager->dnssd_services) { service->withdrawn = false; r = dns_zone_put(&scope->zone, scope, service->ptr_rr, false); if (r < 0) log_warning_errno(r, "Failed to add PTR record to MDNS zone: %m"); r = dns_zone_put(&scope->zone, scope, service->srv_rr, true); if (r < 0) log_warning_errno(r, "Failed to add SRV record to MDNS zone: %m"); LIST_FOREACH(items, txt_data, service->txt_data_items) { r = dns_zone_put(&scope->zone, scope, txt_data->rr, true); if (r < 0) log_warning_errno(r, "Failed to add TXT record to MDNS zone: %m"); } } return 0; } int dns_scope_remove_dnssd_services(DnsScope *scope) { _cleanup_(dns_resource_key_unrefp) DnsResourceKey *key = NULL; DnssdService *service; int r; assert(scope); key = dns_resource_key_new(DNS_CLASS_IN, DNS_TYPE_PTR, "_services._dns-sd._udp.local"); if (!key) return log_oom(); r = dns_zone_remove_rrs_by_key(&scope->zone, key); if (r < 0) return r; HASHMAP_FOREACH(service, scope->manager->dnssd_services) { dns_zone_remove_rr(&scope->zone, service->ptr_rr); dns_zone_remove_rr(&scope->zone, service->srv_rr); LIST_FOREACH(items, txt_data, service->txt_data_items) dns_zone_remove_rr(&scope->zone, txt_data->rr); } return 0; } static bool dns_scope_has_route_only_domains(DnsScope *scope) { DnsSearchDomain *first; bool route_only = false; assert(scope); assert(scope->protocol == DNS_PROTOCOL_DNS); /* Returns 'true' if this scope is suitable for queries to specific domains only. For that we check * if there are any route-only domains on this interface, as a heuristic to discern VPN-style links * from non-VPN-style links. Returns 'false' for all other cases, i.e. if the scope is intended to * take queries to arbitrary domains, i.e. has no routing domains set. */ if (scope->link) first = scope->link->search_domains; else first = scope->manager->search_domains; LIST_FOREACH(domains, domain, first) { /* "." means "any domain", thus the interface takes any kind of traffic. Thus, we exit early * here, as it doesn't really matter whether this link has any route-only domains or not, * "~." really trumps everything and clearly indicates that this interface shall receive all * traffic it can get. */ if (dns_name_is_root(DNS_SEARCH_DOMAIN_NAME(domain))) return false; if (domain->route_only) route_only = true; } return route_only; } bool dns_scope_is_default_route(DnsScope *scope) { assert(scope); /* Only use DNS scopes as default routes */ if (scope->protocol != DNS_PROTOCOL_DNS) return false; /* The global DNS scope is always suitable as default route */ if (!scope->link) return true; /* Honour whatever is explicitly configured. This is really the best approach, and trumps any * automatic logic. */ if (scope->link->default_route >= 0) return scope->link->default_route; /* Otherwise check if we have any route-only domains, as a sensible heuristic: if so, let's not * volunteer as default route. */ return !dns_scope_has_route_only_domains(scope); }