chiark / gitweb /
resolved: don't resolve the hostname "gateway" with LLMNR, leave that to nss-myhostname
[elogind.git] / src / resolve / resolved-dns-scope.c
index 96a2ff7fae0fcdbae47af863dbd4775994423e31..a43359f8c59c0d28939135565c5dddb79df7e2cb 100644 (file)
 
 #include <netinet/tcp.h>
 
+#include "missing.h"
 #include "strv.h"
 #include "socket-util.h"
 #include "af-list.h"
 #include "resolved-dns-domain.h"
 #include "resolved-dns-scope.h"
 
-#define SEND_TIMEOUT_USEC (2*USEC_PER_SEC)
+#define MULTICAST_RATELIMIT_INTERVAL_USEC (1*USEC_PER_SEC)
+#define MULTICAST_RATELIMIT_BURST 1000
 
 int dns_scope_new(Manager *m, DnsScope **ret, Link *l, DnsProtocol protocol, int family) {
         DnsScope *s;
@@ -50,12 +52,16 @@ int dns_scope_new(Manager *m, DnsScope **ret, Link *l, DnsProtocol protocol, int
 
         log_debug("New scope on link %s, protocol %s, family %s", l ? l->name : "*", dns_protocol_to_string(protocol), family == AF_UNSPEC ? "*" : af_to_name(family));
 
+        /* Enforce ratelimiting for the multicast protocols */
+        RATELIMIT_INIT(s->ratelimit, MULTICAST_RATELIMIT_INTERVAL_USEC, MULTICAST_RATELIMIT_BURST);
+
         *ret = s;
         return 0;
 }
 
 DnsScope* dns_scope_free(DnsScope *s) {
-        DnsQueryTransaction *t;
+        DnsTransaction *t;
+        DnsResourceRecord *rr;
 
         if (!s)
                 return NULL;
@@ -70,13 +76,20 @@ DnsScope* dns_scope_free(DnsScope *s) {
                  * freed while we still look at it */
 
                 t->block_gc++;
-                dns_query_transaction_complete(t, DNS_QUERY_ABORTED);
+                dns_transaction_complete(t, DNS_TRANSACTION_ABORTED);
                 t->block_gc--;
 
-                dns_query_transaction_free(t);
+                dns_transaction_free(t);
         }
 
+        while ((rr = ordered_hashmap_steal_first(s->conflict_queue)))
+                dns_resource_record_unref(rr);
+
+        ordered_hashmap_free(s->conflict_queue);
+        sd_event_source_unref(s->conflict_event_source);
+
         dns_cache_flush(&s->cache);
+        dns_zone_flush(&s->zone);
 
         LIST_REMOVE(scopes, s->manager->dns_scopes, s);
         strv_free(s->domains);
@@ -85,7 +98,7 @@ DnsScope* dns_scope_free(DnsScope *s) {
         return NULL;
 }
 
-DnsServer *dns_scope_get_server(DnsScope *s) {
+DnsServer *dns_scope_get_dns_server(DnsScope *s) {
         assert(s);
 
         if (s->protocol != DNS_PROTOCOL_DNS)
@@ -109,7 +122,7 @@ void dns_scope_next_dns_server(DnsScope *s) {
                 manager_next_dns_server(s->manager);
 }
 
-int dns_scope_send(DnsScope *s, DnsPacket *p) {
+int dns_scope_emit(DnsScope *s, DnsPacket *p) {
         union in_addr_union addr;
         int ifindex = 0, r;
         int family;
@@ -130,7 +143,10 @@ int dns_scope_send(DnsScope *s, DnsPacket *p) {
         if (s->protocol == DNS_PROTOCOL_DNS) {
                 DnsServer *srv;
 
-                srv = dns_scope_get_server(s);
+                if (DNS_PACKET_QDCOUNT(p) > 1)
+                        return -ENOTSUP;
+
+                srv = dns_scope_get_dns_server(s);
                 if (!srv)
                         return -ESRCH;
 
@@ -158,17 +174,18 @@ int dns_scope_send(DnsScope *s, DnsPacket *p) {
                 if (DNS_PACKET_QDCOUNT(p) > 1)
                         return -ENOTSUP;
 
+                if (!ratelimit_test(&s->ratelimit))
+                        return -EBUSY;
+
                 family = s->family;
                 port = 5355;
 
                 if (family == AF_INET) {
                         addr.in = LLMNR_MULTICAST_IPV4_ADDRESS;
-                        /* fd = manager_dns_ipv4_fd(s->manager); */
                         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);
-                        /* fd = manager_dns_ipv6_fd(s->manager); */
                 } else
                         return -EAFNOSUPPORT;
                 if (fd < 0)
@@ -183,39 +200,86 @@ int dns_scope_send(DnsScope *s, DnsPacket *p) {
         return 1;
 }
 
-int dns_scope_tcp_socket(DnsScope *s) {
+int dns_scope_tcp_socket(DnsScope *s, int family, const union in_addr_union *address, uint16_t port) {
         _cleanup_close_ int fd = -1;
         union sockaddr_union sa = {};
         socklen_t salen;
-        int one, ret;
-        DnsServer *srv;
-        int r;
+        static const int one = 1;
+        int ret, r;
 
         assert(s);
+        assert((family == AF_UNSPEC) == !address);
 
-        srv = dns_scope_get_server(s);
-        if (!srv)
-                return -ESRCH;
-
-        sa.sa.sa_family = srv->family;
-        if (srv->family == AF_INET) {
-                sa.in.sin_port = htobe16(53);
-                sa.in.sin_addr = srv->address.in;
-                salen = sizeof(sa.in);
-        } else if (srv->family == AF_INET6) {
-                sa.in6.sin6_port = htobe16(53);
-                sa.in6.sin6_addr = srv->address.in6;
-                sa.in6.sin6_scope_id = s->link ? s->link->ifindex : 0;
-                salen = sizeof(sa.in6);
-        } else
-                return -EAFNOSUPPORT;
+        if (family == AF_UNSPEC) {
+                DnsServer *srv;
 
-        fd = socket(srv->family, SOCK_STREAM|SOCK_CLOEXEC|SOCK_NONBLOCK, 0);
+                srv = dns_scope_get_dns_server(s);
+                if (!srv)
+                        return -ESRCH;
+
+                sa.sa.sa_family = srv->family;
+                if (srv->family == AF_INET) {
+                        sa.in.sin_port = htobe16(port);
+                        sa.in.sin_addr = srv->address.in;
+                        salen = sizeof(sa.in);
+                } else if (srv->family == AF_INET6) {
+                        sa.in6.sin6_port = htobe16(port);
+                        sa.in6.sin6_addr = srv->address.in6;
+                        sa.in6.sin6_scope_id = s->link ? s->link->ifindex : 0;
+                        salen = sizeof(sa.in6);
+                } else
+                        return -EAFNOSUPPORT;
+        } else {
+                sa.sa.sa_family = family;
+
+                if (family == AF_INET) {
+                        sa.in.sin_port = htobe16(port);
+                        sa.in.sin_addr = address->in;
+                        salen = sizeof(sa.in);
+                } else if (family == AF_INET6) {
+                        sa.in6.sin6_port = htobe16(port);
+                        sa.in6.sin6_addr = address->in6;
+                        sa.in6.sin6_scope_id = s->link ? s->link->ifindex : 0;
+                        salen = sizeof(sa.in6);
+                } else
+                        return -EAFNOSUPPORT;
+        }
+
+        fd = socket(sa.sa.sa_family, SOCK_STREAM|SOCK_CLOEXEC|SOCK_NONBLOCK, 0);
         if (fd < 0)
                 return -errno;
 
-        one = 1;
-        setsockopt(fd, IPPROTO_TCP, TCP_NODELAY, &one, sizeof(one));
+        r = setsockopt(fd, IPPROTO_TCP, TCP_NODELAY, &one, sizeof(one));
+        if (r < 0)
+                return -errno;
+
+        if (s->link) {
+                uint32_t ifindex = htobe32(s->link->ifindex);
+
+                if (sa.sa.sa_family == AF_INET) {
+                        r = setsockopt(fd, IPPROTO_IP, IP_UNICAST_IF, &ifindex, sizeof(ifindex));
+                        if (r < 0)
+                                return -errno;
+                } else if (sa.sa.sa_family == AF_INET6) {
+                        r = setsockopt(fd, IPPROTO_IPV6, IPV6_UNICAST_IF, &ifindex, sizeof(ifindex));
+                        if (r < 0)
+                                return -errno;
+                }
+        }
+
+        if (s->protocol == DNS_PROTOCOL_LLMNR) {
+                /* RFC 4795, section 2.5 requires the TTL to be set to 1 */
+
+                if (sa.sa.sa_family == AF_INET) {
+                        r = setsockopt(fd, IPPROTO_IP, IP_TTL, &one, sizeof(one));
+                        if (r < 0)
+                                return -errno;
+                } else if (sa.sa.sa_family == AF_INET6) {
+                        r = setsockopt(fd, IPPROTO_IPV6, IPV6_UNICAST_HOPS, &one, sizeof(one));
+                        if (r < 0)
+                                return -errno;
+                }
+        }
 
         r = connect(fd, &sa.sa, salen);
         if (r < 0 && errno != EINPROGRESS)
@@ -223,15 +287,22 @@ int dns_scope_tcp_socket(DnsScope *s) {
 
         ret = fd;
         fd = -1;
+
         return ret;
 }
 
-DnsScopeMatch dns_scope_good_domain(DnsScope *s, const char *domain) {
+DnsScopeMatch dns_scope_good_domain(DnsScope *s, int ifindex, uint64_t flags, const char *domain) {
         char **i;
 
         assert(s);
         assert(domain);
 
+        if (ifindex != 0 && (!s->link || s->link->ifindex != ifindex))
+                return DNS_SCOPE_NO;
+
+        if ((SD_RESOLVED_FLAGS_MAKE(s->protocol, s->family) & flags) == 0)
+                return DNS_SCOPE_NO;
+
         STRV_FOREACH(i, s->domains)
                 if (dns_name_endswith(domain, *i) > 0)
                         return DNS_SCOPE_YES;
@@ -261,9 +332,10 @@ DnsScopeMatch dns_scope_good_domain(DnsScope *s, const char *domain) {
         }
 
         if (s->protocol == DNS_PROTOCOL_LLMNR) {
-                if (dns_name_endswith(domain, "254.169.in-addr.arpa") > 0 ||
-                    dns_name_endswith(domain, "0.8.e.f.ip6.arpa") > 0 ||
-                    dns_name_single_label(domain) > 0)
+                if (dns_name_endswith(domain, "in-addr.arpa") > 0 ||
+                    dns_name_endswith(domain, "ip6.arpa") > 0 ||
+                    (dns_name_single_label(domain) > 0 &&
+                     dns_name_equal(domain, "gateway") <= 0))  /* don't resolve "gateway" with LLMNR, let nss-myhostname handle this */
                         return DNS_SCOPE_MAYBE;
 
                 return DNS_SCOPE_NO;
@@ -294,6 +366,13 @@ int dns_scope_good_key(DnsScope *s, DnsResourceKey *key) {
 int dns_scope_llmnr_membership(DnsScope *s, bool b) {
         int fd;
 
+        assert(s);
+
+        if (s->protocol != DNS_PROTOCOL_LLMNR)
+                return 0;
+
+        assert(s->link);
+
         if (s->family == AF_INET) {
                 struct ip_mreqn mreqn = {
                         .imr_multiaddr = LLMNR_MULTICAST_IPV4_ADDRESS,
@@ -304,6 +383,12 @@ int dns_scope_llmnr_membership(DnsScope *s, bool b) {
                 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;
 
@@ -317,6 +402,9 @@ int dns_scope_llmnr_membership(DnsScope *s, bool b) {
                 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
@@ -324,3 +412,383 @@ int dns_scope_llmnr_membership(DnsScope *s, bool b) {
 
         return 0;
 }
+
+int dns_scope_good_dns_server(DnsScope *s, int family, const union in_addr_union *address) {
+        assert(s);
+        assert(address);
+
+        if (s->protocol != DNS_PROTOCOL_DNS)
+                return 1;
+
+        if (s->link)
+                return !!link_find_dns_server(s->link,  family, address);
+        else
+                return !!manager_find_dns_server(s->manager, family, address);
+}
+
+static 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 i;
+        int r;
+
+        assert(s);
+        assert(ret);
+
+        if ((!q || q->n_keys <= 0)
+            && (!answer || answer->n_rrs <= 0)
+            && (!soa || soa->n_rrs <= 0))
+                return -EINVAL;
+
+        r = dns_packet_new(&p, s->protocol, 0);
+        if (r < 0)
+                return r;
+
+        DNS_PACKET_HEADER(p)->id = id;
+        DNS_PACKET_HEADER(p)->flags = htobe16(DNS_PACKET_MAKE_FLAGS(
+                                                              1 /* qr */,
+                                                              0 /* opcode */,
+                                                              0 /* c */,
+                                                              0 /* tc */,
+                                                              tentative,
+                                                              0 /* (ra) */,
+                                                              0 /* (ad) */,
+                                                              0 /* (cd) */,
+                                                              rcode));
+
+        if (q) {
+                for (i = 0; i < q->n_keys; i++) {
+                        r = dns_packet_append_key(p, q->keys[i], NULL);
+                        if (r < 0)
+                                return r;
+                }
+
+                DNS_PACKET_HEADER(p)->qdcount = htobe16(q->n_keys);
+        }
+
+        if (answer) {
+                for (i = 0; i < answer->n_rrs; i++) {
+                        r = dns_packet_append_rr(p, answer->rrs[i], NULL);
+                        if (r < 0)
+                                return r;
+                }
+
+                DNS_PACKET_HEADER(p)->ancount = htobe16(answer->n_rrs);
+        }
+
+        if (soa) {
+                for (i = 0; i < soa->n_rrs; i++) {
+                        r = dns_packet_append_rr(p, soa->rrs[i], NULL);
+                        if (r < 0)
+                                return r;
+                }
+
+                DNS_PACKET_HEADER(p)->arcount = htobe16(soa->n_rrs);
+        }
+
+        *ret = p;
+        p = NULL;
+
+        return 0;
+}
+
+static void dns_scope_verify_conflicts(DnsScope *s, DnsPacket *p) {
+        unsigned n;
+
+        assert(s);
+        assert(p);
+
+        if (p->question)
+                for (n = 0; n < p->question->n_keys; n++)
+                        dns_zone_verify_conflicts(&s->zone, p->question->keys[n]);
+        if (p->answer)
+                for (n = 0; n < p->answer->n_rrs; n++)
+                        dns_zone_verify_conflicts(&s->zone, p->answer->rrs[n]->key);
+}
+
+void dns_scope_process_query(DnsScope *s, DnsStream *stream, DnsPacket *p) {
+        _cleanup_(dns_packet_unrefp) DnsPacket *reply = NULL;
+        _cleanup_(dns_answer_unrefp) DnsAnswer *answer = NULL, *soa = NULL;
+        bool tentative = false;
+        int r, fd;
+
+        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 && !in_addr_equal(AF_INET, &p->destination, (union in_addr_union*) &LLMNR_MULTICAST_IPV4_ADDRESS))
+                        return;
+
+                if (p->family == AF_INET6 && !in_addr_equal(AF_INET6, &p->destination, (union in_addr_union*) &LLMNR_MULTICAST_IPV6_ADDRESS))
+                        return;
+        }
+
+        r = dns_packet_extract(p);
+        if (r < 0) {
+                log_debug_errno(r, "Failed to extract resources from incoming packet: %m");
+                return;
+        }
+
+        if (DNS_PACKET_C(p)) {
+                /* Somebody notified us about a possible conflict */
+                dns_scope_verify_conflicts(s, p);
+                return;
+        }
+
+        r = dns_zone_lookup(&s->zone, p->question, &answer, &soa, &tentative);
+        if (r < 0) {
+                log_debug_errno(r, "Failed to lookup 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);
+        else {
+                if (!ratelimit_test(&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, reply);
+        }
+
+        if (r < 0) {
+                log_debug_errno(r, "Failed to send reply packet: %m");
+                return;
+        }
+}
+
+DnsTransaction *dns_scope_find_transaction(DnsScope *scope, DnsQuestion *question, bool cache_ok) {
+        DnsTransaction *t;
+
+        assert(scope);
+        assert(question);
+
+        /* Try to find an ongoing transaction that is a equal or a
+         * superset of the specified question */
+
+        LIST_FOREACH(transactions_by_scope, t, scope->transactions) {
+
+                /* Refuse reusing transactions that completed based on
+                 * cached data instead of a real packet, if that's
+                 * requested. */
+                if (!cache_ok &&
+                    IN_SET(t->state, DNS_TRANSACTION_SUCCESS, DNS_TRANSACTION_FAILURE) &&
+                    !t->received)
+                        continue;
+
+                if (dns_question_is_superset(t->question, question) > 0)
+                        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);
+        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));
+        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, NULL);
+        if (r < 0)
+                return r;
+
+        r = dns_packet_append_rr(p, rr, NULL);
+        if (r < 0)
+                return r;
+
+        *ret = p;
+        p = NULL;
+
+        return 0;
+}
+
+static int on_conflict_dispatch(sd_event_source *es, usec_t usec, void *userdata) {
+        DnsScope *scope = userdata;
+        int r;
+
+        assert(es);
+        assert(scope);
+
+        scope->conflict_event_source = sd_event_source_unref(scope->conflict_event_source);
+
+        for (;;) {
+                _cleanup_(dns_resource_record_unrefp) DnsResourceRecord *rr = NULL;
+                _cleanup_(dns_packet_unrefp) DnsPacket *p = NULL;
+
+                rr = ordered_hashmap_steal_first(scope->conflict_queue);
+                if (!rr)
+                        break;
+
+                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(scope, 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) {
+        usec_t jitter;
+        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 (r == -EEXIST || r == 0)
+                return 0;
+        if (r < 0)
+                return log_debug_errno(r, "Failed to queue conflicting RR: %m");
+
+        dns_resource_record_ref(rr);
+
+        if (scope->conflict_event_source)
+                return 0;
+
+        random_bytes(&jitter, sizeof(jitter));
+        jitter %= LLMNR_JITTER_INTERVAL_USEC;
+
+        r = sd_event_add_time(scope->manager->event,
+                              &scope->conflict_event_source,
+                              clock_boottime_or_monotonic(),
+                              now(clock_boottime_or_monotonic()) + jitter,
+                              LLMNR_JITTER_INTERVAL_USEC,
+                              on_conflict_dispatch, scope);
+        if (r < 0)
+                return log_debug_errno(r, "Failed to add conflict dispatch event: %m");
+
+        return 0;
+}
+
+void dns_scope_check_conflicts(DnsScope *scope, DnsPacket *p) {
+        unsigned i;
+        int r;
+
+        assert(scope);
+        assert(p);
+
+        if (p->protocol != DNS_PROTOCOL_LLMNR)
+                return;
+
+        if (DNS_PACKET_RRCOUNT(p) <= 0)
+                return;
+
+        if (DNS_PACKET_C(p) != 0)
+                return;
+
+        if (DNS_PACKET_T(p) != 0)
+                return;
+
+        if (manager_our_packet(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...");
+
+        for (i = 0; i < p->answer->n_rrs; i++) {
+
+                /* Check for conflicts against the local zone. If we
+                 * found one, we won't check any further */
+                r = dns_zone_check_conflicts(&scope->zone, p->answer->rrs[i]);
+                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, p->answer->rrs[i], p->family, &p->sender);
+                if (r <= 0)
+                        continue;
+
+                dns_scope_notify_conflict(scope, p->answer->rrs[i]);
+        }
+}