Adding upstream version 2:4.20.0+dfsg.upstream/2%4.20.0+dfsg

Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>

1 files changed, 1176 insertions, 0 deletions

diff --git a/source4/dns_server/dns_query.c b/source4/dns_server/dns_query.c
new file mode 100644
index 0000000..181beda
--- /dev/null
+++ b/source4/dns_server/dns_query.c
@@ -0,0 +1,1176 @@
+/*
+   Unix SMB/CIFS implementation.
+
+   DNS server handler for queries
+
+   Copyright (C) 2010 Kai Blin  <kai@samba.org>
+
+   This program is free software; you can redistribute it and/or modify
+   it under the terms of the GNU General Public License as published by
+   the Free Software Foundation; either version 3 of the License, or
+   (at your option) any later version.
+
+   This program is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+   GNU General Public License for more details.
+
+   You should have received a copy of the GNU General Public License
+   along with this program.  If not, see <http://www.gnu.org/licenses/>.
+*/
+
+#include "includes.h"
+#include "samba/service_task.h"
+#include "libcli/util/werror.h"
+#include "librpc/ndr/libndr.h"
+#include "librpc/gen_ndr/ndr_dns.h"
+#include "librpc/gen_ndr/ndr_dnsp.h"
+#include <ldb.h>
+#include "param/param.h"
+#include "dsdb/samdb/samdb.h"
+#include "dsdb/common/util.h"
+#include "dns_server/dns_server.h"
+#include "libcli/dns/libdns.h"
+#include "lib/util/dlinklist.h"
+#include "lib/util/util_net.h"
+#include "lib/util/tevent_werror.h"
+#include "auth/auth.h"
+#include "auth/credentials/credentials.h"
+#include "auth/gensec/gensec.h"
+
+#undef DBGC_CLASS
+#define DBGC_CLASS DBGC_DNS
+#define MAX_Q_RECURSION_DEPTH 20
+
+struct forwarder_string {
+	const char *forwarder;
+	struct forwarder_string *prev, *next;
+};
+
+static WERROR add_response_rr(const char *name,
+			      const struct dnsp_DnssrvRpcRecord *rec,
+			      struct dns_res_rec **answers)
+{
+	struct dns_res_rec *ans = *answers;
+	uint16_t ai = talloc_array_length(ans);
+	enum ndr_err_code ndr_err;
+
+	if (ai == UINT16_MAX) {
+		return WERR_BUFFER_OVERFLOW;
+	}
+
+	/*
+	 * "ans" is always non-NULL and thus its own talloc context
+	 */
+	ans = talloc_realloc(ans, ans, struct dns_res_rec, ai+1);
+	if (ans == NULL) {
+		return WERR_NOT_ENOUGH_MEMORY;
+	}
+
+	ZERO_STRUCT(ans[ai]);
+
+	switch (rec->wType) {
+	case DNS_QTYPE_CNAME:
+		ans[ai].rdata.cname_record = talloc_strdup(ans, rec->data.cname);
+		W_ERROR_HAVE_NO_MEMORY(ans[ai].rdata.cname_record);
+		break;
+	case DNS_QTYPE_A:
+		ans[ai].rdata.ipv4_record = talloc_strdup(ans, rec->data.ipv4);
+		W_ERROR_HAVE_NO_MEMORY(ans[ai].rdata.ipv4_record);
+		break;
+	case DNS_QTYPE_AAAA:
+		ans[ai].rdata.ipv6_record = talloc_strdup(ans, rec->data.ipv6);
+		W_ERROR_HAVE_NO_MEMORY(ans[ai].rdata.ipv6_record);
+		break;
+	case DNS_TYPE_NS:
+		ans[ai].rdata.ns_record = talloc_strdup(ans, rec->data.ns);
+		W_ERROR_HAVE_NO_MEMORY(ans[ai].rdata.ns_record);
+		break;
+	case DNS_QTYPE_SRV:
+		ans[ai].rdata.srv_record.priority = rec->data.srv.wPriority;
+		ans[ai].rdata.srv_record.weight   = rec->data.srv.wWeight;
+		ans[ai].rdata.srv_record.port     = rec->data.srv.wPort;
+		ans[ai].rdata.srv_record.target   = talloc_strdup(
+			ans, rec->data.srv.nameTarget);
+		W_ERROR_HAVE_NO_MEMORY(ans[ai].rdata.srv_record.target);
+		break;
+	case DNS_QTYPE_SOA:
+		ans[ai].rdata.soa_record.mname	 = talloc_strdup(
+			ans, rec->data.soa.mname);
+		W_ERROR_HAVE_NO_MEMORY(ans[ai].rdata.soa_record.mname);
+		ans[ai].rdata.soa_record.rname	 = talloc_strdup(
+			ans, rec->data.soa.rname);
+		W_ERROR_HAVE_NO_MEMORY(ans[ai].rdata.soa_record.rname);
+		ans[ai].rdata.soa_record.serial	 = rec->data.soa.serial;
+		ans[ai].rdata.soa_record.refresh = rec->data.soa.refresh;
+		ans[ai].rdata.soa_record.retry	 = rec->data.soa.retry;
+		ans[ai].rdata.soa_record.expire	 = rec->data.soa.expire;
+		ans[ai].rdata.soa_record.minimum = rec->data.soa.minimum;
+		break;
+	case DNS_QTYPE_PTR:
+		ans[ai].rdata.ptr_record = talloc_strdup(ans, rec->data.ptr);
+		W_ERROR_HAVE_NO_MEMORY(ans[ai].rdata.ptr_record);
+		break;
+	case DNS_QTYPE_MX:
+		ans[ai].rdata.mx_record.preference = rec->data.mx.wPriority;
+		ans[ai].rdata.mx_record.exchange = talloc_strdup(
+			ans, rec->data.mx.nameTarget);
+		if (ans[ai].rdata.mx_record.exchange == NULL) {
+			return WERR_NOT_ENOUGH_MEMORY;
+		}
+		break;
+	case DNS_QTYPE_TXT:
+		ndr_err = ndr_dnsp_string_list_copy(ans,
+						    &rec->data.txt,
+						    &ans[ai].rdata.txt_record.txt);
+		if (!NDR_ERR_CODE_IS_SUCCESS(ndr_err)) {
+			return WERR_NOT_ENOUGH_MEMORY;
+		}
+		break;
+	default:
+		DEBUG(0, ("Got unhandled type %u query.\n", rec->wType));
+		return DNS_ERR(NOT_IMPLEMENTED);
+	}
+
+	ans[ai].name = talloc_strdup(ans, name);
+	W_ERROR_HAVE_NO_MEMORY(ans[ai].name);
+	ans[ai].rr_type = (enum dns_qtype)rec->wType;
+	ans[ai].rr_class = DNS_QCLASS_IN;
+	ans[ai].ttl = rec->dwTtlSeconds;
+	ans[ai].length = UINT16_MAX;
+
+	*answers = ans;
+
+	return WERR_OK;
+}
+
+static WERROR add_dns_res_rec(struct dns_res_rec **pdst,
+			      const struct dns_res_rec *src)
+{
+	struct dns_res_rec *dst = *pdst;
+	uint16_t di = talloc_array_length(dst);
+	enum ndr_err_code ndr_err;
+
+	if (di == UINT16_MAX) {
+		return WERR_BUFFER_OVERFLOW;
+	}
+
+	dst = talloc_realloc(dst, dst, struct dns_res_rec, di+1);
+	if (dst == NULL) {
+		return WERR_NOT_ENOUGH_MEMORY;
+	}
+
+	ZERO_STRUCT(dst[di]);
+
+	dst[di] = (struct dns_res_rec) {
+		.name = talloc_strdup(dst, src->name),
+		.rr_type = src->rr_type,
+		.rr_class = src->rr_class,
+		.ttl = src->ttl,
+		.length = src->length
+	};
+
+	if (dst[di].name == NULL) {
+		return WERR_NOT_ENOUGH_MEMORY;
+	}
+
+	switch (src->rr_type) {
+	case DNS_QTYPE_CNAME:
+		dst[di].rdata.cname_record = talloc_strdup(
+			dst, src->rdata.cname_record);
+		if (dst[di].rdata.cname_record == NULL) {
+			return WERR_NOT_ENOUGH_MEMORY;
+		}
+		break;
+	case DNS_QTYPE_A:
+		dst[di].rdata.ipv4_record = talloc_strdup(
+			dst, src->rdata.ipv4_record);
+		if (dst[di].rdata.ipv4_record == NULL) {
+			return WERR_NOT_ENOUGH_MEMORY;
+		}
+		break;
+	case DNS_QTYPE_AAAA:
+		dst[di].rdata.ipv6_record = talloc_strdup(
+			dst, src->rdata.ipv6_record);
+		if (dst[di].rdata.ipv6_record == NULL) {
+			return WERR_NOT_ENOUGH_MEMORY;
+		}
+		break;
+	case DNS_TYPE_NS:
+		dst[di].rdata.ns_record = talloc_strdup(
+			dst, src->rdata.ns_record);
+		if (dst[di].rdata.ns_record == NULL) {
+			return WERR_NOT_ENOUGH_MEMORY;
+		}
+		break;
+	case DNS_QTYPE_SRV:
+		dst[di].rdata.srv_record = (struct dns_srv_record) {
+			.priority = src->rdata.srv_record.priority,
+			.weight   = src->rdata.srv_record.weight,
+			.port     = src->rdata.srv_record.port,
+			.target   = talloc_strdup(
+				dst, src->rdata.srv_record.target)
+		};
+		if (dst[di].rdata.srv_record.target == NULL) {
+			return WERR_NOT_ENOUGH_MEMORY;
+		}
+		break;
+	case DNS_QTYPE_SOA:
+		dst[di].rdata.soa_record = (struct dns_soa_record) {
+			.mname	 = talloc_strdup(
+				dst, src->rdata.soa_record.mname),
+			.rname	 = talloc_strdup(
+				dst, src->rdata.soa_record.rname),
+			.serial	 = src->rdata.soa_record.serial,
+			.refresh = src->rdata.soa_record.refresh,
+			.retry   = src->rdata.soa_record.retry,
+			.expire  = src->rdata.soa_record.expire,
+			.minimum = src->rdata.soa_record.minimum
+		};
+
+		if ((dst[di].rdata.soa_record.mname == NULL) ||
+		    (dst[di].rdata.soa_record.rname == NULL)) {
+			return WERR_NOT_ENOUGH_MEMORY;
+		}
+
+		break;
+	case DNS_QTYPE_PTR:
+		dst[di].rdata.ptr_record = talloc_strdup(
+			dst, src->rdata.ptr_record);
+		if (dst[di].rdata.ptr_record == NULL) {
+			return WERR_NOT_ENOUGH_MEMORY;
+		}
+		break;
+	case DNS_QTYPE_MX:
+		dst[di].rdata.mx_record = (struct dns_mx_record) {
+			.preference = src->rdata.mx_record.preference,
+			.exchange   = talloc_strdup(
+				src, src->rdata.mx_record.exchange)
+		};
+
+		if (dst[di].rdata.mx_record.exchange == NULL) {
+			return WERR_NOT_ENOUGH_MEMORY;
+		}
+		break;
+	case DNS_QTYPE_TXT:
+		ndr_err = ndr_dnsp_string_list_copy(dst,
+						    &src->rdata.txt_record.txt,
+						    &dst[di].rdata.txt_record.txt);
+		if (!NDR_ERR_CODE_IS_SUCCESS(ndr_err)) {
+			return WERR_NOT_ENOUGH_MEMORY;
+		}
+		break;
+	default:
+		DBG_WARNING("Got unhandled type %u query.\n", src->rr_type);
+		return DNS_ERR(NOT_IMPLEMENTED);
+	}
+
+	*pdst = dst;
+
+	return WERR_OK;
+}
+
+struct ask_forwarder_state {
+	struct dns_name_packet *reply;
+};
+
+static void ask_forwarder_done(struct tevent_req *subreq);
+
+static struct tevent_req *ask_forwarder_send(
+	TALLOC_CTX *mem_ctx, struct tevent_context *ev,
+	const char *forwarder, struct dns_name_question *question)
+{
+	struct tevent_req *req, *subreq;
+	struct ask_forwarder_state *state;
+
+	req = tevent_req_create(mem_ctx, &state, struct ask_forwarder_state);
+	if (req == NULL) {
+		return NULL;
+	}
+
+	subreq = dns_cli_request_send(state, ev, forwarder,
+				      question->name, question->question_class,
+				      question->question_type);
+	if (tevent_req_nomem(subreq, req)) {
+		return tevent_req_post(req, ev);
+	}
+	tevent_req_set_callback(subreq, ask_forwarder_done, req);
+	return req;
+}
+
+static void ask_forwarder_done(struct tevent_req *subreq)
+{
+	struct tevent_req *req = tevent_req_callback_data(
+		subreq, struct tevent_req);
+	struct ask_forwarder_state *state = tevent_req_data(
+		req, struct ask_forwarder_state);
+	int ret;
+
+	ret = dns_cli_request_recv(subreq, state, &state->reply);
+	TALLOC_FREE(subreq);
+
+	if (ret != 0) {
+		tevent_req_werror(req, unix_to_werror(ret));
+		return;
+	}
+
+	tevent_req_done(req);
+}
+
+static WERROR ask_forwarder_recv(
+	struct tevent_req *req, TALLOC_CTX *mem_ctx,
+	struct dns_res_rec **answers, uint16_t *ancount,
+	struct dns_res_rec **nsrecs, uint16_t *nscount,
+	struct dns_res_rec **additional, uint16_t *arcount)
+{
+	struct ask_forwarder_state *state = tevent_req_data(
+		req, struct ask_forwarder_state);
+	struct dns_name_packet *in_packet = state->reply;
+	WERROR err;
+
+	if (tevent_req_is_werror(req, &err)) {
+		return err;
+	}
+
+	*ancount = in_packet->ancount;
+	*answers = talloc_move(mem_ctx, &in_packet->answers);
+
+	*nscount = in_packet->nscount;
+	*nsrecs = talloc_move(mem_ctx, &in_packet->nsrecs);
+
+	*arcount = in_packet->arcount;
+	*additional = talloc_move(mem_ctx, &in_packet->additional);
+
+	return WERR_OK;
+}
+
+static WERROR add_zone_authority_record(struct dns_server *dns,
+					TALLOC_CTX *mem_ctx,
+					const struct dns_name_question *question,
+					struct dns_res_rec **nsrecs)
+{
+	const char *zone = NULL;
+	struct dnsp_DnssrvRpcRecord *recs;
+	struct dns_res_rec *ns = *nsrecs;
+	uint16_t rec_count;
+	struct ldb_dn *dn = NULL;
+	unsigned int ri;
+	WERROR werror;
+
+	zone = dns_get_authoritative_zone(dns, question->name);
+	DEBUG(10, ("Creating zone authority record for '%s'\n", zone));
+
+	werror = dns_name2dn(dns, mem_ctx, zone, &dn);
+	if (!W_ERROR_IS_OK(werror)) {
+		return werror;
+	}
+
+	werror = dns_lookup_records(dns, mem_ctx, dn, &recs, &rec_count);
+	if (!W_ERROR_IS_OK(werror)) {
+		return werror;
+	}
+
+	for (ri = 0; ri < rec_count; ri++) {
+		if (recs[ri].wType == DNS_TYPE_SOA) {
+			werror = add_response_rr(zone, &recs[ri], &ns);
+			if (!W_ERROR_IS_OK(werror)) {
+				return werror;
+			}
+		}
+	}
+
+	*nsrecs = ns;
+
+	return WERR_OK;
+}
+
+static struct tevent_req *handle_authoritative_send(
+	TALLOC_CTX *mem_ctx, struct tevent_context *ev,
+	struct dns_server *dns, const char *forwarder,
+	struct dns_name_question *question,
+	struct dns_res_rec **answers, struct dns_res_rec **nsrecs,
+	size_t cname_depth);
+static WERROR handle_authoritative_recv(struct tevent_req *req);
+
+struct handle_dnsrpcrec_state {
+	struct dns_res_rec **answers;
+	struct dns_res_rec **nsrecs;
+};
+
+static void handle_dnsrpcrec_gotauth(struct tevent_req *subreq);
+static void handle_dnsrpcrec_gotforwarded(struct tevent_req *subreq);
+
+static struct tevent_req *handle_dnsrpcrec_send(
+	TALLOC_CTX *mem_ctx, struct tevent_context *ev,
+	struct dns_server *dns, const char *forwarder,
+	const struct dns_name_question *question,
+	struct dnsp_DnssrvRpcRecord *rec,
+	struct dns_res_rec **answers, struct dns_res_rec **nsrecs,
+	size_t cname_depth)
+{
+	struct tevent_req *req, *subreq;
+	struct handle_dnsrpcrec_state *state;
+	struct dns_name_question *new_q;
+	bool resolve_cname;
+	WERROR werr;
+
+	req = tevent_req_create(mem_ctx, &state,
+				struct handle_dnsrpcrec_state);
+	if (req == NULL) {
+		return NULL;
+	}
+	state->answers = answers;
+	state->nsrecs = nsrecs;
+
+	if (cname_depth >= MAX_Q_RECURSION_DEPTH) {
+		tevent_req_done(req);
+		return tevent_req_post(req, ev);
+	}
+
+	resolve_cname = ((rec->wType == DNS_TYPE_CNAME) &&
+			 ((question->question_type == DNS_QTYPE_A) ||
+			  (question->question_type == DNS_QTYPE_AAAA)));
+
+	if (!resolve_cname) {
+		if ((question->question_type != DNS_QTYPE_ALL) &&
+		    (rec->wType !=
+		     (enum dns_record_type) question->question_type)) {
+			tevent_req_done(req);
+			return tevent_req_post(req, ev);
+		}
+
+		werr = add_response_rr(question->name, rec, state->answers);
+		if (tevent_req_werror(req, werr)) {
+			return tevent_req_post(req, ev);
+		}
+
+		tevent_req_done(req);
+		return tevent_req_post(req, ev);
+	}
+
+	werr = add_response_rr(question->name, rec, state->answers);
+	if (tevent_req_werror(req, werr)) {
+		return tevent_req_post(req, ev);
+	}
+
+	new_q = talloc(state, struct dns_name_question);
+	if (tevent_req_nomem(new_q, req)) {
+		return tevent_req_post(req, ev);
+	}
+
+	*new_q = (struct dns_name_question) {
+		.question_type = question->question_type,
+		.question_class = question->question_class,
+		.name = rec->data.cname
+	};
+
+	if (dns_authoritative_for_zone(dns, new_q->name)) {
+		subreq = handle_authoritative_send(
+			state, ev, dns, forwarder, new_q,
+			state->answers, state->nsrecs,
+			cname_depth + 1);
+		if (tevent_req_nomem(subreq, req)) {
+			return tevent_req_post(req, ev);
+		}
+		tevent_req_set_callback(subreq, handle_dnsrpcrec_gotauth, req);
+		return req;
+	}
+
+	subreq = ask_forwarder_send(state, ev, forwarder, new_q);
+	if (tevent_req_nomem(subreq, req)) {
+		return tevent_req_post(req, ev);
+	}
+	tevent_req_set_callback(subreq, handle_dnsrpcrec_gotforwarded, req);
+
+	return req;
+}
+
+static void handle_dnsrpcrec_gotauth(struct tevent_req *subreq)
+{
+	struct tevent_req *req = tevent_req_callback_data(
+		subreq, struct tevent_req);
+	WERROR werr;
+
+	werr = handle_authoritative_recv(subreq);
+	TALLOC_FREE(subreq);
+	if (tevent_req_werror(req, werr)) {
+		return;
+	}
+	tevent_req_done(req);
+}
+
+static void handle_dnsrpcrec_gotforwarded(struct tevent_req *subreq)
+{
+	struct tevent_req *req = tevent_req_callback_data(
+		subreq, struct tevent_req);
+	struct handle_dnsrpcrec_state *state = tevent_req_data(
+		req, struct handle_dnsrpcrec_state);
+	struct dns_res_rec *answers, *nsrecs, *additional;
+	uint16_t ancount = 0;
+	uint16_t nscount = 0;
+	uint16_t arcount = 0;
+	uint16_t i;
+	WERROR werr;
+
+	werr = ask_forwarder_recv(subreq, state, &answers, &ancount,
+				  &nsrecs, &nscount, &additional, &arcount);
+	if (tevent_req_werror(req, werr)) {
+		return;
+	}
+
+	for (i=0; i<ancount; i++) {
+		werr = add_dns_res_rec(state->answers, &answers[i]);
+		if (tevent_req_werror(req, werr)) {
+			return;
+		}
+	}
+
+	for (i=0; i<nscount; i++) {
+		werr = add_dns_res_rec(state->nsrecs, &nsrecs[i]);
+		if (tevent_req_werror(req, werr)) {
+			return;
+		}
+	}
+
+	tevent_req_done(req);
+}
+
+static WERROR handle_dnsrpcrec_recv(struct tevent_req *req)
+{
+	return tevent_req_simple_recv_werror(req);
+}
+
+struct handle_authoritative_state {
+	struct tevent_context *ev;
+	struct dns_server *dns;
+	struct dns_name_question *question;
+	const char *forwarder;
+
+	struct dnsp_DnssrvRpcRecord *recs;
+	uint16_t rec_count;
+	uint16_t recs_done;
+
+	struct dns_res_rec **answers;
+	struct dns_res_rec **nsrecs;
+
+	size_t cname_depth;
+};
+
+static void handle_authoritative_done(struct tevent_req *subreq);
+
+static struct tevent_req *handle_authoritative_send(
+	TALLOC_CTX *mem_ctx, struct tevent_context *ev,
+	struct dns_server *dns, const char *forwarder,
+	struct dns_name_question *question,
+	struct dns_res_rec **answers, struct dns_res_rec **nsrecs,
+	size_t cname_depth)
+{
+	struct tevent_req *req, *subreq;
+	struct handle_authoritative_state *state;
+	struct ldb_dn *dn = NULL;
+	WERROR werr;
+
+	req = tevent_req_create(mem_ctx, &state,
+				struct handle_authoritative_state);
+	if (req == NULL) {
+		return NULL;
+	}
+	state->ev = ev;
+	state->dns = dns;
+	state->question = question;
+	state->forwarder = forwarder;
+	state->answers = answers;
+	state->nsrecs = nsrecs;
+	state->cname_depth = cname_depth;
+
+	werr = dns_name2dn(dns, state, question->name, &dn);
+	if (tevent_req_werror(req, werr)) {
+		return tevent_req_post(req, ev);
+	}
+	werr = dns_lookup_records_wildcard(dns, state, dn, &state->recs,
+				           &state->rec_count);
+	TALLOC_FREE(dn);
+	if (tevent_req_werror(req, werr)) {
+		return tevent_req_post(req, ev);
+	}
+
+	if (state->rec_count == 0) {
+		tevent_req_werror(req, DNS_ERR(NAME_ERROR));
+		return tevent_req_post(req, ev);
+	}
+
+	subreq = handle_dnsrpcrec_send(
+		state, state->ev, state->dns, state->forwarder,
+		state->question, &state->recs[state->recs_done],
+		state->answers, state->nsrecs,
+		state->cname_depth);
+	if (tevent_req_nomem(subreq, req)) {
+		return tevent_req_post(req, ev);
+	}
+	tevent_req_set_callback(subreq, handle_authoritative_done, req);
+	return req;
+}
+
+static void handle_authoritative_done(struct tevent_req *subreq)
+{
+	struct tevent_req *req = tevent_req_callback_data(
+		subreq, struct tevent_req);
+	struct handle_authoritative_state *state = tevent_req_data(
+		req, struct handle_authoritative_state);
+	WERROR werr;
+
+	werr = handle_dnsrpcrec_recv(subreq);
+	TALLOC_FREE(subreq);
+	if (tevent_req_werror(req, werr)) {
+		return;
+	}
+
+	state->recs_done += 1;
+
+	if (state->recs_done == state->rec_count) {
+		tevent_req_done(req);
+		return;
+	}
+
+	subreq = handle_dnsrpcrec_send(
+		state, state->ev, state->dns, state->forwarder,
+		state->question, &state->recs[state->recs_done],
+		state->answers, state->nsrecs,
+		state->cname_depth);
+	if (tevent_req_nomem(subreq, req)) {
+		return;
+	}
+	tevent_req_set_callback(subreq, handle_authoritative_done, req);
+}
+
+static WERROR handle_authoritative_recv(struct tevent_req *req)
+{
+	WERROR werr;
+
+	if (tevent_req_is_werror(req, &werr)) {
+		return werr;
+	}
+
+	return WERR_OK;
+}
+
+static NTSTATUS create_tkey(struct dns_server *dns,
+			    const char* name,
+			    const char* algorithm,
+			    const struct tsocket_address *remote_address,
+			    const struct tsocket_address *local_address,
+			    struct dns_server_tkey **tkey)
+{
+	NTSTATUS status;
+	struct dns_server_tkey_store *store = dns->tkeys;
+	struct dns_server_tkey *k = talloc_zero(store, struct dns_server_tkey);
+
+	if (k == NULL) {
+		return NT_STATUS_NO_MEMORY;
+	}
+
+	k->name = talloc_strdup(k, name);
+
+	if (k->name  == NULL) {
+		return NT_STATUS_NO_MEMORY;
+	}
+
+	k->algorithm = talloc_strdup(k, algorithm);
+	if (k->algorithm == NULL) {
+		return NT_STATUS_NO_MEMORY;
+	}
+
+	/*
+	 * We only allow SPNEGO/KRB5 currently
+	 * and rely on the backend to be RPC/IPC free.
+	 *
+	 * It allows gensec_update() not to block.
+	 */
+	status = samba_server_gensec_krb5_start(k,
+						dns->task->event_ctx,
+						dns->task->msg_ctx,
+						dns->task->lp_ctx,
+						dns->server_credentials,
+						"dns",
+						&k->gensec);
+	if (!NT_STATUS_IS_OK(status)) {
+		DEBUG(1, ("Failed to start GENSEC server code: %s\n", nt_errstr(status)));
+		*tkey = NULL;
+		return status;
+	}
+
+	gensec_want_feature(k->gensec, GENSEC_FEATURE_SIGN);
+
+	status = gensec_set_remote_address(k->gensec,
+					   remote_address);
+	if (!NT_STATUS_IS_OK(status)) {
+		DEBUG(1, ("Failed to set remote address into GENSEC: %s\n",
+			  nt_errstr(status)));
+		*tkey = NULL;
+		return status;
+	}
+
+	status = gensec_set_local_address(k->gensec,
+					  local_address);
+	if (!NT_STATUS_IS_OK(status)) {
+		DEBUG(1, ("Failed to set local address into GENSEC: %s\n",
+			  nt_errstr(status)));
+		*tkey = NULL;
+		return status;
+	}
+
+	status = gensec_start_mech_by_oid(k->gensec, GENSEC_OID_SPNEGO);
+
+	if (!NT_STATUS_IS_OK(status)) {
+		DEBUG(1, ("Failed to start GENSEC server code: %s\n",
+			  nt_errstr(status)));
+		*tkey = NULL;
+		return status;
+	}
+
+	TALLOC_FREE(store->tkeys[store->next_idx]);
+
+	store->tkeys[store->next_idx] = k;
+	(store->next_idx)++;
+	store->next_idx %= store->size;
+
+	*tkey = k;
+	return NT_STATUS_OK;
+}
+
+static NTSTATUS accept_gss_ticket(TALLOC_CTX *mem_ctx,
+				  struct dns_server *dns,
+				  struct dns_server_tkey *tkey,
+				  const DATA_BLOB *key,
+				  DATA_BLOB *reply,
+				  uint16_t *dns_auth_error)
+{
+	NTSTATUS status;
+
+	/*
+	 * We use samba_server_gensec_krb5_start(),
+	 * which only allows SPNEGO/KRB5 currently
+	 * and makes sure the backend to be RPC/IPC free.
+	 *
+	 * See gensec_gssapi_update_internal() as
+	 * GENSEC_SERVER.
+	 *
+	 * It allows gensec_update() not to block.
+	 *
+	 * If that changes in future we need to use
+	 * gensec_update_send/recv here!
+	 */
+	status = gensec_update(tkey->gensec, mem_ctx,
+			       *key, reply);
+
+	if (NT_STATUS_EQUAL(NT_STATUS_MORE_PROCESSING_REQUIRED, status)) {
+		*dns_auth_error = DNS_RCODE_OK;
+		return status;
+	}
+
+	if (NT_STATUS_IS_OK(status)) {
+
+		status = gensec_session_info(tkey->gensec, tkey, &tkey->session_info);
+		if (!NT_STATUS_IS_OK(status)) {
+			*dns_auth_error = DNS_RCODE_BADKEY;
+			return status;
+		}
+		*dns_auth_error = DNS_RCODE_OK;
+	}
+
+	return status;
+}
+
+static WERROR handle_tkey(struct dns_server *dns,
+                          TALLOC_CTX *mem_ctx,
+                          const struct dns_name_packet *in,
+			  struct dns_request_state *state,
+                          struct dns_res_rec **answers,
+                          uint16_t *ancount)
+{
+	struct dns_res_rec *in_tkey = NULL;
+	struct dns_res_rec *ret_tkey;
+	uint16_t i;
+
+	for (i = 0; i < in->arcount; i++) {
+		if (in->additional[i].rr_type == DNS_QTYPE_TKEY) {
+			in_tkey = &in->additional[i];
+			break;
+		}
+	}
+
+	/* If this is a TKEY query, it should have a TKEY RR.
+	 * Behaviour is not really specified in RFC 2930 or RFC 3645, but
+	 * FORMAT_ERROR seems to be what BIND uses .*/
+	if (in_tkey == NULL) {
+		return DNS_ERR(FORMAT_ERROR);
+	}
+
+	ret_tkey = talloc_zero(mem_ctx, struct dns_res_rec);
+	if (ret_tkey == NULL) {
+		return WERR_NOT_ENOUGH_MEMORY;
+	}
+
+	ret_tkey->name = talloc_strdup(ret_tkey, in_tkey->name);
+	if (ret_tkey->name == NULL) {
+		return WERR_NOT_ENOUGH_MEMORY;
+	}
+
+	ret_tkey->rr_type = DNS_QTYPE_TKEY;
+	ret_tkey->rr_class = DNS_QCLASS_ANY;
+	ret_tkey->length = UINT16_MAX;
+
+	ret_tkey->rdata.tkey_record.algorithm = talloc_strdup(ret_tkey,
+			in_tkey->rdata.tkey_record.algorithm);
+	if (ret_tkey->rdata.tkey_record.algorithm  == NULL) {
+		return WERR_NOT_ENOUGH_MEMORY;
+	}
+
+	ret_tkey->rdata.tkey_record.inception = in_tkey->rdata.tkey_record.inception;
+	ret_tkey->rdata.tkey_record.expiration = in_tkey->rdata.tkey_record.expiration;
+	ret_tkey->rdata.tkey_record.mode = in_tkey->rdata.tkey_record.mode;
+
+	switch (in_tkey->rdata.tkey_record.mode) {
+	case DNS_TKEY_MODE_DH:
+		/* FIXME: According to RFC 2930, we MUST support this, but we don't.
+		 * Still, claim it's a bad key instead of a bad mode */
+		ret_tkey->rdata.tkey_record.error = DNS_RCODE_BADKEY;
+		break;
+	case DNS_TKEY_MODE_GSSAPI: {
+		NTSTATUS status;
+		struct dns_server_tkey *tkey;
+		DATA_BLOB key;
+		DATA_BLOB reply;
+
+		tkey = dns_find_tkey(dns->tkeys, in->questions[0].name);
+		if (tkey != NULL && tkey->complete) {
+			/* TODO: check if the key is still valid */
+			DEBUG(1, ("Rejecting tkey negotiation for already established key\n"));
+			ret_tkey->rdata.tkey_record.error = DNS_RCODE_BADNAME;
+			break;
+		}
+
+		if (tkey == NULL) {
+			status  = create_tkey(dns, in->questions[0].name,
+					      in_tkey->rdata.tkey_record.algorithm,
+					      state->remote_address,
+					      state->local_address,
+					      &tkey);
+			if (!NT_STATUS_IS_OK(status)) {
+				ret_tkey->rdata.tkey_record.error = DNS_RCODE_BADKEY;
+				return ntstatus_to_werror(status);
+			}
+		}
+
+		key.data = in_tkey->rdata.tkey_record.key_data;
+		key.length = in_tkey->rdata.tkey_record.key_size;
+
+		status = accept_gss_ticket(ret_tkey, dns, tkey, &key, &reply,
+					   &ret_tkey->rdata.tkey_record.error);
+		if (NT_STATUS_EQUAL(status, NT_STATUS_MORE_PROCESSING_REQUIRED)) {
+			DEBUG(1, ("More processing required\n"));
+			ret_tkey->rdata.tkey_record.error = DNS_RCODE_BADKEY;
+		} else if (NT_STATUS_IS_OK(status)) {
+			DBG_DEBUG("Tkey handshake completed\n");
+			ret_tkey->rdata.tkey_record.key_size = reply.length;
+			ret_tkey->rdata.tkey_record.key_data = talloc_memdup(ret_tkey,
+								reply.data,
+								reply.length);
+			if (ret_tkey->rdata.tkey_record.key_data == NULL) {
+				return WERR_NOT_ENOUGH_MEMORY;
+			}
+			state->sign = true;
+			state->key_name = talloc_strdup(state->mem_ctx, tkey->name);
+			if (state->key_name == NULL) {
+				return WERR_NOT_ENOUGH_MEMORY;
+			}
+		} else {
+			DEBUG(1, ("GSS key negotiation returned %s\n", nt_errstr(status)));
+			ret_tkey->rdata.tkey_record.error = DNS_RCODE_BADKEY;
+		}
+
+		break;
+		}
+	case DNS_TKEY_MODE_DELETE:
+		/* TODO: implement me */
+		DEBUG(1, ("Should delete tkey here\n"));
+		ret_tkey->rdata.tkey_record.error = DNS_RCODE_OK;
+		break;
+	case DNS_TKEY_MODE_NULL:
+	case DNS_TKEY_MODE_SERVER:
+	case DNS_TKEY_MODE_CLIENT:
+	case DNS_TKEY_MODE_LAST:
+		/* We don't have to implement these, return a mode error */
+		ret_tkey->rdata.tkey_record.error = DNS_RCODE_BADMODE;
+		break;
+	default:
+		DEBUG(1, ("Unsupported TKEY mode %d\n",
+		      in_tkey->rdata.tkey_record.mode));
+	}
+
+	*answers = ret_tkey;
+	*ancount = 1;
+
+	return WERR_OK;
+}
+
+struct dns_server_process_query_state {
+	struct tevent_context *ev;
+	struct dns_server *dns;
+	struct dns_name_question *question;
+
+	struct dns_res_rec *answers;
+	uint16_t ancount;
+	struct dns_res_rec *nsrecs;
+	uint16_t nscount;
+	struct dns_res_rec *additional;
+	uint16_t arcount;
+	struct forwarder_string *forwarders;
+};
+
+static void dns_server_process_query_got_auth(struct tevent_req *subreq);
+static void dns_server_process_query_got_response(struct tevent_req *subreq);
+
+struct tevent_req *dns_server_process_query_send(
+	TALLOC_CTX *mem_ctx, struct tevent_context *ev,
+	struct dns_server *dns,	struct dns_request_state *req_state,
+	const struct dns_name_packet *in)
+{
+	struct tevent_req *req, *subreq;
+	struct dns_server_process_query_state *state;
+	const char **forwarders = NULL;
+	unsigned int i;
+
+	req = tevent_req_create(mem_ctx, &state,
+				struct dns_server_process_query_state);
+	if (req == NULL) {
+		return NULL;
+	}
+	if (in->qdcount != 1) {
+		tevent_req_werror(req, DNS_ERR(FORMAT_ERROR));
+		return tevent_req_post(req, ev);
+	}
+
+	/* Windows returns NOT_IMPLEMENTED on this as well */
+	if (in->questions[0].question_class == DNS_QCLASS_NONE) {
+		tevent_req_werror(req, DNS_ERR(NOT_IMPLEMENTED));
+		return tevent_req_post(req, ev);
+	}
+
+	if (in->questions[0].question_type == DNS_QTYPE_TKEY) {
+                WERROR err;
+
+		err = handle_tkey(dns, state, in, req_state,
+				  &state->answers, &state->ancount);
+		if (tevent_req_werror(req, err)) {
+			return tevent_req_post(req, ev);
+		}
+		tevent_req_done(req);
+		return tevent_req_post(req, ev);
+	}
+
+	state->dns = dns;
+	state->ev = ev;
+	state->question = &in->questions[0];
+
+	forwarders = lpcfg_dns_forwarder(dns->task->lp_ctx);
+	for (i = 0; forwarders != NULL && forwarders[i] != NULL; i++) {
+		struct forwarder_string *f = talloc_zero(state,
+							 struct forwarder_string);
+		f->forwarder = forwarders[i];
+		DLIST_ADD_END(state->forwarders, f);
+	}
+
+	if (dns_authoritative_for_zone(dns, in->questions[0].name)) {
+
+		req_state->flags |= DNS_FLAG_AUTHORITATIVE;
+
+		/*
+		 * Initialize the response arrays, so that we can use
+		 * them as their own talloc contexts when doing the
+		 * realloc
+		 */
+		state->answers = talloc_array(state, struct dns_res_rec, 0);
+		if (tevent_req_nomem(state->answers, req)) {
+			return tevent_req_post(req, ev);
+		}
+		state->nsrecs = talloc_array(state, struct dns_res_rec, 0);
+		if (tevent_req_nomem(state->nsrecs, req)) {
+			return tevent_req_post(req, ev);
+		}
+
+		subreq = handle_authoritative_send(
+			state, ev, dns, (forwarders == NULL ? NULL : forwarders[0]),
+			&in->questions[0], &state->answers, &state->nsrecs,
+			0); /* cname_depth */
+		if (tevent_req_nomem(subreq, req)) {
+			return tevent_req_post(req, ev);
+		}
+		tevent_req_set_callback(
+			subreq, dns_server_process_query_got_auth, req);
+		return req;
+	}
+
+	if ((req_state->flags & DNS_FLAG_RECURSION_DESIRED) &&
+	    (req_state->flags & DNS_FLAG_RECURSION_AVAIL)) {
+		DEBUG(5, ("Not authoritative for '%s', forwarding\n",
+			  in->questions[0].name));
+
+		subreq = ask_forwarder_send(state, ev,
+					    (forwarders == NULL ? NULL : forwarders[0]),
+					    &in->questions[0]);
+		if (tevent_req_nomem(subreq, req)) {
+			return tevent_req_post(req, ev);
+		}
+		tevent_req_set_callback(
+			subreq, dns_server_process_query_got_response, req);
+		return req;
+	}
+
+	tevent_req_werror(req, DNS_ERR(NAME_ERROR));
+	return tevent_req_post(req, ev);
+}
+
+static void dns_server_process_query_got_response(struct tevent_req *subreq)
+{
+	struct tevent_req *req = tevent_req_callback_data(
+		subreq, struct tevent_req);
+	struct dns_server_process_query_state *state = tevent_req_data(
+		req, struct dns_server_process_query_state);
+	WERROR werr;
+
+	werr = ask_forwarder_recv(subreq, state,
+				  &state->answers, &state->ancount,
+				  &state->nsrecs, &state->nscount,
+				  &state->additional, &state->arcount);
+	TALLOC_FREE(subreq);
+
+	/* If you get an error, attempt a different forwarder */
+	if (!W_ERROR_IS_OK(werr)) {
+		if (state->forwarders != NULL) {
+			DLIST_REMOVE(state->forwarders, state->forwarders);
+		}
+
+		/* If you have run out of forwarders, simply finish */
+		if (state->forwarders == NULL) {
+			tevent_req_werror(req, werr);
+			return;
+		}
+
+		DEBUG(5, ("DNS query returned %s, trying another forwarder.\n",
+			  win_errstr(werr)));
+		subreq = ask_forwarder_send(state, state->ev,
+					    state->forwarders->forwarder,
+					    state->question);
+
+		if (tevent_req_nomem(subreq, req)) {
+			return;
+		}
+
+		tevent_req_set_callback(subreq,
+					dns_server_process_query_got_response,
+					req);
+		return;
+	}
+
+	tevent_req_done(req);
+}
+
+static void dns_server_process_query_got_auth(struct tevent_req *subreq)
+{
+	struct tevent_req *req = tevent_req_callback_data(
+		subreq, struct tevent_req);
+	struct dns_server_process_query_state *state = tevent_req_data(
+		req, struct dns_server_process_query_state);
+	WERROR werr;
+	WERROR werr2;
+
+	werr = handle_authoritative_recv(subreq);
+	TALLOC_FREE(subreq);
+
+	/* If you get an error, attempt a different forwarder */
+	if (!W_ERROR_IS_OK(werr)) {
+		if (state->forwarders != NULL) {
+			DLIST_REMOVE(state->forwarders, state->forwarders);
+		}
+
+		/* If you have run out of forwarders, simply finish */
+		if (state->forwarders == NULL) {
+			werr2 = add_zone_authority_record(state->dns,
+							  state,
+							  state->question,
+							  &state->nsrecs);
+			if (tevent_req_werror(req, werr2)) {
+				DBG_WARNING("Failed to add SOA record: %s\n",
+					    win_errstr(werr2));
+				return;
+			}
+
+			state->ancount = talloc_array_length(state->answers);
+			state->nscount = talloc_array_length(state->nsrecs);
+			state->arcount = talloc_array_length(state->additional);
+
+			tevent_req_werror(req, werr);
+			return;
+		}
+
+		DEBUG(5, ("Error: %s, trying a different forwarder.\n",
+			  win_errstr(werr)));
+		subreq = handle_authoritative_send(state, state->ev, state->dns,
+						   state->forwarders->forwarder,
+						   state->question, &state->answers,
+						   &state->nsrecs,
+						   0); /* cname_depth */
+
+		if (tevent_req_nomem(subreq, req)) {
+			return;
+		}
+
+		tevent_req_set_callback(subreq,
+					dns_server_process_query_got_auth,
+					req);
+		return;
+	}
+
+	werr2 = add_zone_authority_record(state->dns,
+					  state,
+					  state->question,
+					  &state->nsrecs);
+	if (tevent_req_werror(req, werr2)) {
+		DBG_WARNING("Failed to add SOA record: %s\n",
+				win_errstr(werr2));
+		return;
+	}
+
+	state->ancount = talloc_array_length(state->answers);
+	state->nscount = talloc_array_length(state->nsrecs);
+	state->arcount = talloc_array_length(state->additional);
+
+	tevent_req_done(req);
+}
+
+WERROR dns_server_process_query_recv(
+	struct tevent_req *req, TALLOC_CTX *mem_ctx,
+	struct dns_res_rec **answers,    uint16_t *ancount,
+	struct dns_res_rec **nsrecs,     uint16_t *nscount,
+	struct dns_res_rec **additional, uint16_t *arcount)
+{
+	struct dns_server_process_query_state *state = tevent_req_data(
+		req, struct dns_server_process_query_state);
+	WERROR err = WERR_OK;
+
+	if (tevent_req_is_werror(req, &err)) {
+
+		if ((!W_ERROR_EQUAL(err, DNS_ERR(NAME_ERROR))) &&
+		    (!W_ERROR_EQUAL(err, WERR_DNS_ERROR_NAME_DOES_NOT_EXIST))) {
+			return err;
+		}
+	}
+	*answers = talloc_move(mem_ctx, &state->answers);
+	*ancount = state->ancount;
+	*nsrecs = talloc_move(mem_ctx, &state->nsrecs);
+	*nscount = state->nscount;
+	*additional = talloc_move(mem_ctx, &state->additional);
+	*arcount = state->arcount;
+	return err;
+}