xref: /freebsd/contrib/unbound/validator/val_sigcrypt.c (revision b2efd602aea8b3cbc3fb215b9611946d04fceb10)

/*
 * validator/val_sigcrypt.c - validator signature crypto functions.
 *
 * Copyright (c) 2007, NLnet Labs. All rights reserved.
 *
 * This software is open source.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 *
 * Redistributions of source code must retain the above copyright notice,
 * this list of conditions and the following disclaimer.
 *
 * Redistributions in binary form must reproduce the above copyright notice,
 * this list of conditions and the following disclaimer in the documentation
 * and/or other materials provided with the distribution.
 *
 * Neither the name of the NLNET LABS nor the names of its contributors may
 * be used to endorse or promote products derived from this software without
 * specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
 * HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED
 * TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
 * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
 * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
 * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 */

/**
 * \file
 *
 * This file contains helper functions for the validator module.
 * The functions help with signature verification and checking, the
 * bridging between RR wireformat data and crypto calls.
 */
#include "config.h"
#include "validator/val_sigcrypt.h"
#include "validator/val_secalgo.h"
#include "validator/validator.h"
#include "util/data/msgreply.h"
#include "util/data/msgparse.h"
#include "util/data/dname.h"
#include "util/rbtree.h"
#include "util/rfc_1982.h"
#include "util/module.h"
#include "util/net_help.h"
#include "util/regional.h"
#include "util/config_file.h"
#include "sldns/keyraw.h"
#include "sldns/sbuffer.h"
#include "sldns/parseutil.h"
#include "sldns/wire2str.h"
#include "services/mesh.h"

#include <ctype.h>
#if !defined(HAVE_SSL) && !defined(HAVE_NSS) && !defined(HAVE_NETTLE)
#error "Need crypto library to do digital signature cryptography"
#endif

#ifdef HAVE_OPENSSL_ERR_H
#include <openssl/err.h>
#endif

#ifdef HAVE_OPENSSL_RAND_H
#include <openssl/rand.h>
#endif

#ifdef HAVE_OPENSSL_CONF_H
#include <openssl/conf.h>
#endif

#ifdef HAVE_OPENSSL_ENGINE_H
#include <openssl/engine.h>
#endif

/** Maximum number of RRSIG validations for an RRset. */
#define MAX_VALIDATE_RRSIGS 8

/** return number of rrs in an rrset */
static size_t
rrset_get_count(struct ub_packed_rrset_key* rrset)
{
	struct packed_rrset_data* d = (struct packed_rrset_data*)
	rrset->entry.data;
	if(!d) return 0;
	return d->count;
}

/**
 * Get RR signature count
 */
static size_t
rrset_get_sigcount(struct ub_packed_rrset_key* k)
{
	struct packed_rrset_data* d = (struct packed_rrset_data*)k->entry.data;
	return d->rrsig_count;
}

/**
 * Get signature keytag value
 * @param k: rrset (with signatures)
 * @param sig_idx: signature index.
 * @return keytag or 0 if malformed rrsig.
 */
static uint16_t
rrset_get_sig_keytag(struct ub_packed_rrset_key* k, size_t sig_idx)
{
	uint16_t t;
	struct packed_rrset_data* d = (struct packed_rrset_data*)k->entry.data;
	log_assert(sig_idx < d->rrsig_count);
	if(d->rr_len[d->count + sig_idx] < 2+18)
		return 0;
	memmove(&t, d->rr_data[d->count + sig_idx]+2+16, 2);
	return ntohs(t);
}

/**
 * Get signature signing algorithm value
 * @param k: rrset (with signatures)
 * @param sig_idx: signature index.
 * @return algo or 0 if malformed rrsig.
 */
static int
rrset_get_sig_algo(struct ub_packed_rrset_key* k, size_t sig_idx)
{
	struct packed_rrset_data* d = (struct packed_rrset_data*)k->entry.data;
	log_assert(sig_idx < d->rrsig_count);
	if(d->rr_len[d->count + sig_idx] < 2+3)
		return 0;
	return (int)d->rr_data[d->count + sig_idx][2+2];
}

/** get rdata pointer and size */
static void
rrset_get_rdata(struct ub_packed_rrset_key* k, size_t idx, uint8_t** rdata,
	size_t* len)
{
	struct packed_rrset_data* d = (struct packed_rrset_data*)k->entry.data;
	log_assert(d && idx < (d->count + d->rrsig_count));
	*rdata = d->rr_data[idx];
	*len = d->rr_len[idx];
}

uint16_t
dnskey_get_flags(struct ub_packed_rrset_key* k, size_t idx)
{
	uint8_t* rdata;
	size_t len;
	uint16_t f;
	rrset_get_rdata(k, idx, &rdata, &len);
	if(len < 2+2)
		return 0;
	memmove(&f, rdata+2, 2);
	f = ntohs(f);
	return f;
}

/**
 * Get DNSKEY protocol value from rdata
 * @param k: DNSKEY rrset.
 * @param idx: which key.
 * @return protocol octet value
 */
static int
dnskey_get_protocol(struct ub_packed_rrset_key* k, size_t idx)
{
	uint8_t* rdata;
	size_t len;
	rrset_get_rdata(k, idx, &rdata, &len);
	if(len < 2+4)
		return 0;
	return (int)rdata[2+2];
}

int
dnskey_get_algo(struct ub_packed_rrset_key* k, size_t idx)
{
	uint8_t* rdata;
	size_t len;
	rrset_get_rdata(k, idx, &rdata, &len);
	if(len < 2+4)
		return 0;
	return (int)rdata[2+3];
}

/** get public key rdata field from a dnskey RR and do some checks */
static void
dnskey_get_pubkey(struct ub_packed_rrset_key* k, size_t idx,
	unsigned char** pk, unsigned int* pklen)
{
	uint8_t* rdata;
	size_t len;
	rrset_get_rdata(k, idx, &rdata, &len);
	if(len < 2+5) {
		*pk = NULL;
		*pklen = 0;
		return;
	}
	*pk = (unsigned char*)rdata+2+4;
	*pklen = (unsigned)len-2-4;
}

int
ds_get_key_algo(struct ub_packed_rrset_key* k, size_t idx)
{
	uint8_t* rdata;
	size_t len;
	rrset_get_rdata(k, idx, &rdata, &len);
	if(len < 2+3)
		return 0;
	return (int)rdata[2+2];
}

int
ds_get_digest_algo(struct ub_packed_rrset_key* k, size_t idx)
{
	uint8_t* rdata;
	size_t len;
	rrset_get_rdata(k, idx, &rdata, &len);
	if(len < 2+4)
		return 0;
	return (int)rdata[2+3];
}

uint16_t
ds_get_keytag(struct ub_packed_rrset_key* ds_rrset, size_t ds_idx)
{
	uint16_t t;
	uint8_t* rdata;
	size_t len;
	rrset_get_rdata(ds_rrset, ds_idx, &rdata, &len);
	if(len < 2+2)
		return 0;
	memmove(&t, rdata+2, 2);
	return ntohs(t);
}

/**
 * Return pointer to the digest in a DS RR.
 * @param k: DS rrset.
 * @param idx: which DS.
 * @param digest: digest data is returned.
 *	on error, this is NULL.
 * @param len: length of digest is returned.
 *	on error, the length is 0.
 */
static void
ds_get_sigdata(struct ub_packed_rrset_key* k, size_t idx, uint8_t** digest,
        size_t* len)
{
	uint8_t* rdata;
	size_t rdlen;
	rrset_get_rdata(k, idx, &rdata, &rdlen);
	if(rdlen < 2+5) {
		*digest = NULL;
		*len = 0;
		return;
	}
	*digest = rdata + 2 + 4;
	*len = rdlen - 2 - 4;
}

/**
 * Return size of DS digest according to its hash algorithm.
 * @param k: DS rrset.
 * @param idx: which DS.
 * @return size in bytes of digest, or 0 if not supported.
 */
static size_t
ds_digest_size_algo(struct ub_packed_rrset_key* k, size_t idx)
{
	return ds_digest_size_supported(ds_get_digest_algo(k, idx));
}

/**
 * Create a DS digest for a DNSKEY entry.
 *
 * @param env: module environment. Uses scratch space.
 * @param dnskey_rrset: DNSKEY rrset.
 * @param dnskey_idx: index of RR in rrset.
 * @param ds_rrset: DS rrset
 * @param ds_idx: index of RR in DS rrset.
 * @param digest: digest is returned in here (must be correctly sized).
 * @return false on error.
 */
static int
ds_create_dnskey_digest(struct module_env* env,
	struct ub_packed_rrset_key* dnskey_rrset, size_t dnskey_idx,
	struct ub_packed_rrset_key* ds_rrset, size_t ds_idx,
	uint8_t* digest)
{
	sldns_buffer* b = env->scratch_buffer;
	uint8_t* dnskey_rdata;
	size_t dnskey_len;
	rrset_get_rdata(dnskey_rrset, dnskey_idx, &dnskey_rdata, &dnskey_len);

	/* create digest source material in buffer
	 * digest = digest_algorithm( DNSKEY owner name | DNSKEY RDATA);
	 *	DNSKEY RDATA = Flags | Protocol | Algorithm | Public Key. */
	sldns_buffer_clear(b);
	sldns_buffer_write(b, dnskey_rrset->rk.dname,
		dnskey_rrset->rk.dname_len);
	query_dname_tolower(sldns_buffer_begin(b));
	sldns_buffer_write(b, dnskey_rdata+2, dnskey_len-2); /* skip rdatalen*/
	sldns_buffer_flip(b);

	return secalgo_ds_digest(ds_get_digest_algo(ds_rrset, ds_idx),
		(unsigned char*)sldns_buffer_begin(b), sldns_buffer_limit(b),
		(unsigned char*)digest);
}

int ds_digest_match_dnskey(struct module_env* env,
	struct ub_packed_rrset_key* dnskey_rrset, size_t dnskey_idx,
	struct ub_packed_rrset_key* ds_rrset, size_t ds_idx)
{
	uint8_t* ds;	/* DS digest */
	size_t dslen;
	uint8_t* digest; /* generated digest */
	size_t digestlen = ds_digest_size_algo(ds_rrset, ds_idx);

	if(digestlen == 0) {
		verbose(VERB_QUERY, "DS fail: not supported, or DS RR "
			"format error");
		return 0; /* not supported, or DS RR format error */
	}
#ifndef USE_SHA1
	if(fake_sha1 && ds_get_digest_algo(ds_rrset, ds_idx)==LDNS_SHA1)
		return 1;
#endif

	/* check digest length in DS with length from hash function */
	ds_get_sigdata(ds_rrset, ds_idx, &ds, &dslen);
	if(!ds || dslen != digestlen) {
		verbose(VERB_QUERY, "DS fail: DS RR algo and digest do not "
			"match each other");
		return 0; /* DS algorithm and digest do not match */
	}

	digest = regional_alloc(env->scratch, digestlen);
	if(!digest) {
		verbose(VERB_QUERY, "DS fail: out of memory");
		return 0; /* mem error */
	}
	if(!ds_create_dnskey_digest(env, dnskey_rrset, dnskey_idx, ds_rrset,
		ds_idx, digest)) {
		verbose(VERB_QUERY, "DS fail: could not calc key digest");
		return 0; /* digest algo failed */
	}
	if(memcmp(digest, ds, dslen) != 0) {
		verbose(VERB_QUERY, "DS fail: digest is different");
		return 0; /* digest different */
	}
	return 1;
}

int
ds_digest_algo_is_supported(struct ub_packed_rrset_key* ds_rrset,
	size_t ds_idx)
{
	return (ds_digest_size_algo(ds_rrset, ds_idx) != 0);
}

int
ds_key_algo_is_supported(struct ub_packed_rrset_key* ds_rrset,
	size_t ds_idx)
{
	return dnskey_algo_id_is_supported(ds_get_key_algo(ds_rrset, ds_idx));
}

uint16_t
dnskey_calc_keytag(struct ub_packed_rrset_key* dnskey_rrset, size_t dnskey_idx)
{
	uint8_t* data;
	size_t len;
	rrset_get_rdata(dnskey_rrset, dnskey_idx, &data, &len);
	/* do not pass rdatalen to ldns */
	return sldns_calc_keytag_raw(data+2, len-2);
}

int dnskey_algo_is_supported(struct ub_packed_rrset_key* dnskey_rrset,
        size_t dnskey_idx)
{
	return dnskey_algo_id_is_supported(dnskey_get_algo(dnskey_rrset,
		dnskey_idx));
}

int dnskey_size_is_supported(struct ub_packed_rrset_key* dnskey_rrset,
	size_t dnskey_idx)
{
#ifdef DEPRECATE_RSA_1024
	uint8_t* rdata;
	size_t len;
	int alg = dnskey_get_algo(dnskey_rrset, dnskey_idx);
	size_t keysize;

	rrset_get_rdata(dnskey_rrset, dnskey_idx, &rdata, &len);
	if(len < 2+4)
		return 0;
	keysize = sldns_rr_dnskey_key_size_raw(rdata+2+4, len-2-4, alg);

	switch((sldns_algorithm)alg) {
	case LDNS_RSAMD5:
	case LDNS_RSASHA1:
	case LDNS_RSASHA1_NSEC3:
	case LDNS_RSASHA256:
	case LDNS_RSASHA512:
		/* reject RSA keys of 1024 bits and shorter */
		if(keysize <= 1024)
			return 0;
		break;
	default:
		break;
	}
#else
	(void)dnskey_rrset; (void)dnskey_idx;
#endif /* DEPRECATE_RSA_1024 */
	return 1;
}

int dnskeyset_size_is_supported(struct ub_packed_rrset_key* dnskey_rrset)
{
	size_t i, num = rrset_get_count(dnskey_rrset);
	for(i=0; i<num; i++) {
		if(!dnskey_size_is_supported(dnskey_rrset, i))
			return 0;
	}
	return 1;
}

void algo_needs_init_dnskey_add(struct algo_needs* n,
        struct ub_packed_rrset_key* dnskey, uint8_t* sigalg)
{
	uint8_t algo;
	size_t i, total = n->num;
	size_t num = rrset_get_count(dnskey);

	for(i=0; i<num; i++) {
		algo = (uint8_t)dnskey_get_algo(dnskey, i);
		if(!dnskey_algo_id_is_supported((int)algo))
			continue;
		if(n->needs[algo] == 0) {
			n->needs[algo] = 1;
			sigalg[total] = algo;
			total++;
		}
	}
	sigalg[total] = 0;
	n->num = total;
}

void algo_needs_init_list(struct algo_needs* n, uint8_t* sigalg)
{
	uint8_t algo;
	size_t total = 0;

	memset(n->needs, 0, sizeof(uint8_t)*ALGO_NEEDS_MAX);
	while( (algo=*sigalg++) != 0) {
		log_assert(dnskey_algo_id_is_supported((int)algo));
		log_assert(n->needs[algo] == 0);
		n->needs[algo] = 1;
		total++;
	}
	n->num = total;
}

void algo_needs_init_ds(struct algo_needs* n, struct ub_packed_rrset_key* ds,
	int fav_ds_algo, uint8_t* sigalg)
{
	uint8_t algo;
	size_t i, total = 0;
	size_t num = rrset_get_count(ds);

	memset(n->needs, 0, sizeof(uint8_t)*ALGO_NEEDS_MAX);
	for(i=0; i<num; i++) {
		if(ds_get_digest_algo(ds, i) != fav_ds_algo)
			continue;
		algo = (uint8_t)ds_get_key_algo(ds, i);
		if(!dnskey_algo_id_is_supported((int)algo))
			continue;
		log_assert(algo != 0); /* we do not support 0 and is EOS */
		if(n->needs[algo] == 0) {
			n->needs[algo] = 1;
			sigalg[total] = algo;
			total++;
		}
	}
	sigalg[total] = 0;
	n->num = total;
}

int algo_needs_set_secure(struct algo_needs* n, uint8_t algo)
{
	if(n->needs[algo]) {
		n->needs[algo] = 0;
		n->num --;
		if(n->num == 0) /* done! */
			return 1;
	}
	return 0;
}

void algo_needs_set_bogus(struct algo_needs* n, uint8_t algo)
{
	if(n->needs[algo]) n->needs[algo] = 2; /* need it, but bogus */
}

size_t algo_needs_num_missing(struct algo_needs* n)
{
	return n->num;
}

int algo_needs_missing(struct algo_needs* n)
{
	int i, miss = -1;
	/* check if a needed algo was bogus - report that;
	 * check the first missing algo - report that;
	 * or return 0 */
	for(i=0; i<ALGO_NEEDS_MAX; i++) {
		if(n->needs[i] == 2)
			return 0;
		if(n->needs[i] == 1 && miss == -1)
			miss = i;
	}
	if(miss != -1) return miss;
	return 0;
}

/**
 * verify rrset, with dnskey rrset, for a specific rrsig in rrset
 * @param env: module environment, scratch space is used.
 * @param ve: validator environment, date settings.
 * @param now: current time for validation (can be overridden).
 * @param rrset: to be validated.
 * @param dnskey: DNSKEY rrset, keyset to try.
 * @param sig_idx: which signature to try to validate.
 * @param sortree: reused sorted order. Stored in region. Pass NULL at start,
 * 	and for a new rrset.
 * @param reason: if bogus, a string returned, fixed or alloced in scratch.
 * @param reason_bogus: EDE (RFC8914) code paired with the reason of failure.
 * @param section: section of packet where this rrset comes from.
 * @param qstate: qstate with region.
 * @param numverified: incremented when the number of RRSIG validations
 * 	increases.
 * @return secure if any key signs *this* signature. bogus if no key signs it,
 *	unchecked on error, or indeterminate if all keys are not supported by
 *	the crypto library (openssl3+ only).
 */
static enum sec_status
dnskeyset_verify_rrset_sig(struct module_env* env, struct val_env* ve,
	time_t now, struct ub_packed_rrset_key* rrset,
	struct ub_packed_rrset_key* dnskey, size_t sig_idx,
	struct rbtree_type** sortree,
	char** reason, sldns_ede_code *reason_bogus,
	sldns_pkt_section section, struct module_qstate* qstate,
	int* numverified)
{
	/* find matching keys and check them */
	enum sec_status sec = sec_status_bogus;
	uint16_t tag = rrset_get_sig_keytag(rrset, sig_idx);
	int algo = rrset_get_sig_algo(rrset, sig_idx);
	size_t i, num = rrset_get_count(dnskey);
	size_t numchecked = 0;
	size_t numindeterminate = 0;
	int buf_canon = 0;
	verbose(VERB_ALGO, "verify sig %d %d", (int)tag, algo);
	if(!dnskey_algo_id_is_supported(algo)) {
		if(reason_bogus)
			*reason_bogus = LDNS_EDE_UNSUPPORTED_DNSKEY_ALG;
		verbose(VERB_QUERY, "verify sig: unknown algorithm");
		return sec_status_insecure;
	}

	for(i=0; i<num; i++) {
		/* see if key matches keytag and algo */
		if(algo != dnskey_get_algo(dnskey, i) ||
			tag != dnskey_calc_keytag(dnskey, i))
			continue;
		numchecked ++;
		(*numverified)++;

		/* see if key verifies */
		sec = dnskey_verify_rrset_sig(env->scratch,
			env->scratch_buffer, ve, now, rrset, dnskey, i,
			sig_idx, sortree, &buf_canon, reason, reason_bogus,
			section, qstate);
		if(sec == sec_status_secure)
			return sec;
		else if(sec == sec_status_indeterminate)
			numindeterminate ++;
		if(*numverified > MAX_VALIDATE_RRSIGS) {
			*reason = "too many RRSIG validations";
			if(reason_bogus)
				*reason_bogus = LDNS_EDE_DNSSEC_BOGUS;
			verbose(VERB_ALGO, "verify sig: too many RRSIG validations");
			return sec_status_bogus;
		}
	}
	if(numchecked == 0) {
		*reason = "signatures from unknown keys";
		if(reason_bogus)
			*reason_bogus = LDNS_EDE_DNSKEY_MISSING;
		verbose(VERB_QUERY, "verify: could not find appropriate key");
		return sec_status_bogus;
	}
	if(numindeterminate == numchecked) {
		*reason = "unsupported algorithm by crypto library";
		if(reason_bogus)
			*reason_bogus = LDNS_EDE_UNSUPPORTED_DNSKEY_ALG;
		verbose(VERB_ALGO, "verify sig: unsupported algorithm by "
			"crypto library");
		return sec_status_indeterminate;
	}
	return sec_status_bogus;
}

enum sec_status
dnskeyset_verify_rrset(struct module_env* env, struct val_env* ve,
	struct ub_packed_rrset_key* rrset, struct ub_packed_rrset_key* dnskey,
	uint8_t* sigalg, char** reason, sldns_ede_code *reason_bogus,
	sldns_pkt_section section, struct module_qstate* qstate, int* verified,
	char* reasonbuf, size_t reasonlen)
{
	enum sec_status sec;
	size_t i, num;
	rbtree_type* sortree = NULL;
	/* make sure that for all DNSKEY algorithms there are valid sigs */
	struct algo_needs needs;
	int alg;
	*verified = 0;

	num = rrset_get_sigcount(rrset);
	if(num == 0) {
		verbose(VERB_QUERY, "rrset failed to verify due to a lack of "
			"signatures");
		*reason = "no signatures";
		if(reason_bogus)
			*reason_bogus = LDNS_EDE_RRSIGS_MISSING;
		return sec_status_bogus;
	}

	if(sigalg) {
		algo_needs_init_list(&needs, sigalg);
		if(algo_needs_num_missing(&needs) == 0) {
			verbose(VERB_QUERY, "zone has no known algorithms");
			*reason = "zone has no known algorithms";
			if(reason_bogus)
				*reason_bogus = LDNS_EDE_UNSUPPORTED_DNSKEY_ALG;
			return sec_status_insecure;
		}
	}
	for(i=0; i<num; i++) {
		sec = dnskeyset_verify_rrset_sig(env, ve, *env->now, rrset,
			dnskey, i, &sortree, reason, reason_bogus,
			section, qstate, verified);
		/* see which algorithm has been fixed up */
		if(sec == sec_status_secure) {
			if(!sigalg)
				return sec; /* done! */
			else if(algo_needs_set_secure(&needs,
				(uint8_t)rrset_get_sig_algo(rrset, i)))
				return sec; /* done! */
		} else if(sigalg && sec == sec_status_bogus) {
			algo_needs_set_bogus(&needs,
				(uint8_t)rrset_get_sig_algo(rrset, i));
		}
		if(*verified > MAX_VALIDATE_RRSIGS) {
			verbose(VERB_QUERY, "rrset failed to verify, too many RRSIG validations");
			*reason = "too many RRSIG validations";
			if(reason_bogus)
				*reason_bogus = LDNS_EDE_DNSSEC_BOGUS;
			return sec_status_bogus;
		}
	}
	if(sigalg && (alg=algo_needs_missing(&needs)) != 0) {
		verbose(VERB_ALGO, "rrset failed to verify: "
			"no valid signatures for %d algorithms",
			(int)algo_needs_num_missing(&needs));
		algo_needs_reason(alg, reason, "no signatures", reasonbuf,
			reasonlen);
	} else {
		verbose(VERB_ALGO, "rrset failed to verify: "
			"no valid signatures");
	}
	return sec_status_bogus;
}

void algo_needs_reason(int alg, char** reason, char* s, char* reasonbuf,
	size_t reasonlen)
{
	sldns_lookup_table *t = sldns_lookup_by_id(sldns_algorithms, alg);
	if(t&&t->name)
		snprintf(reasonbuf, reasonlen, "%s with algorithm %s", s,
			t->name);
	else	snprintf(reasonbuf, reasonlen, "%s with algorithm ALG%u", s,
			(unsigned)alg);
	*reason = reasonbuf;
}

enum sec_status
dnskey_verify_rrset(struct module_env* env, struct val_env* ve,
        struct ub_packed_rrset_key* rrset, struct ub_packed_rrset_key* dnskey,
	size_t dnskey_idx, char** reason, sldns_ede_code *reason_bogus,
	sldns_pkt_section section, struct module_qstate* qstate)
{
	enum sec_status sec;
	size_t i, num, numchecked = 0, numindeterminate = 0;
	rbtree_type* sortree = NULL;
	int buf_canon = 0;
	uint16_t tag = dnskey_calc_keytag(dnskey, dnskey_idx);
	int algo = dnskey_get_algo(dnskey, dnskey_idx);
	int numverified = 0;

	num = rrset_get_sigcount(rrset);
	if(num == 0) {
		verbose(VERB_QUERY, "rrset failed to verify due to a lack of "
			"signatures");
		*reason = "no signatures";
		if(reason_bogus)
			*reason_bogus = LDNS_EDE_RRSIGS_MISSING;
		return sec_status_bogus;
	}
	for(i=0; i<num; i++) {
		/* see if sig matches keytag and algo */
		if(algo != rrset_get_sig_algo(rrset, i) ||
			tag != rrset_get_sig_keytag(rrset, i))
			continue;
		buf_canon = 0;
		sec = dnskey_verify_rrset_sig(env->scratch,
			env->scratch_buffer, ve, *env->now, rrset,
			dnskey, dnskey_idx, i, &sortree, &buf_canon, reason,
			reason_bogus, section, qstate);
		if(sec == sec_status_secure)
			return sec;
		numchecked ++;
		numverified ++;
		if(sec == sec_status_indeterminate)
			numindeterminate ++;
		if(numverified > MAX_VALIDATE_RRSIGS) {
			verbose(VERB_QUERY, "rrset failed to verify, too many RRSIG validations");
			*reason = "too many RRSIG validations";
			if(reason_bogus)
				*reason_bogus = LDNS_EDE_DNSSEC_BOGUS;
			return sec_status_bogus;
		}
	}
	if(!numchecked) {
		*reason = "signature for expected key and algorithm missing";
		if(reason_bogus)
			*reason_bogus = LDNS_EDE_DNSSEC_BOGUS;
	} else if(numchecked == numindeterminate) {
		verbose(VERB_ALGO, "rrset failed to verify due to algorithm "
			"refusal by cryptolib");
		if(reason_bogus)
			*reason_bogus = LDNS_EDE_UNSUPPORTED_DNSKEY_ALG;
		*reason = "algorithm refused by cryptolib";
		return sec_status_indeterminate;
	}
	verbose(VERB_ALGO, "rrset failed to verify: all signatures are bogus");
	return sec_status_bogus;
}

/**
 * RR entries in a canonical sorted tree of RRs
 */
struct canon_rr {
	/** rbtree node, key is this structure */
	rbnode_type node;
	/** rrset the RR is in */
	struct ub_packed_rrset_key* rrset;
	/** which RR in the rrset */
	size_t rr_idx;
};

/**
 * Compare two RR for canonical order, in a field-style sweep.
 * @param d: rrset data
 * @param desc: ldns wireformat descriptor.
 * @param i: first RR to compare
 * @param j: first RR to compare
 * @return comparison code.
 */
static int
canonical_compare_byfield(struct packed_rrset_data* d,
	const sldns_rr_descriptor* desc, size_t i, size_t j)
{
	/* sweep across rdata, keep track of some state:
	 * 	which rr field, and bytes left in field.
	 * 	current position in rdata, length left.
	 * 	are we in a dname, length left in a label.
	 */
	int wfi = -1;	/* current wireformat rdata field (rdf) */
	int wfj = -1;
	uint8_t* di = d->rr_data[i]+2; /* ptr to current rdata byte */
	uint8_t* dj = d->rr_data[j]+2;
	size_t ilen = d->rr_len[i]-2; /* length left in rdata */
	size_t jlen = d->rr_len[j]-2;
	int dname_i = 0;  /* true if these bytes are part of a name */
	int dname_j = 0;
	size_t lablen_i = 0; /* 0 for label length byte,for first byte of rdf*/
	size_t lablen_j = 0; /* otherwise remaining length of rdf or label */
	int dname_num_i = (int)desc->_dname_count; /* decreased at root label */
	int dname_num_j = (int)desc->_dname_count;

	/* loop while there are rdata bytes available for both rrs,
	 * and still some lowercasing needs to be done; either the dnames
	 * have not been reached yet, or they are currently being processed */
	while(ilen > 0 && jlen > 0 && (dname_num_i > 0 || dname_num_j > 0)) {
		/* compare these two bytes */
		/* lowercase if in a dname and not a label length byte */
		if( ((dname_i && lablen_i)?(uint8_t)tolower((int)*di):*di)
		 != ((dname_j && lablen_j)?(uint8_t)tolower((int)*dj):*dj)
		 ) {
		  if(((dname_i && lablen_i)?(uint8_t)tolower((int)*di):*di)
		  < ((dname_j && lablen_j)?(uint8_t)tolower((int)*dj):*dj))
		 	return -1;
		    return 1;
		}
		ilen--;
		jlen--;
		/* bytes are equal */

		/* advance field i */
		/* lablen 0 means that this byte is the first byte of the
		 * next rdata field; inspect this rdata field and setup
		 * to process the rest of this rdata field.
		 * The reason to first read the byte, then setup the rdf,
		 * is that we are then sure the byte is available and short
		 * rdata is handled gracefully (even if it is a formerr). */
		if(lablen_i == 0) {
			if(dname_i) {
				/* scan this dname label */
				/* capture length to lowercase */
				lablen_i = (size_t)*di;
				if(lablen_i == 0) {
					/* end root label */
					dname_i = 0;
					dname_num_i--;
					/* if dname num is 0, then the
					 * remainder is binary only */
					if(dname_num_i == 0)
						lablen_i = ilen;
				}
			} else {
				/* scan this rdata field */
				wfi++;
				if(desc->_wireformat[wfi]
					== LDNS_RDF_TYPE_DNAME) {
					dname_i = 1;
					lablen_i = (size_t)*di;
					if(lablen_i == 0) {
						dname_i = 0;
						dname_num_i--;
						if(dname_num_i == 0)
							lablen_i = ilen;
					}
				} else if(desc->_wireformat[wfi]
					== LDNS_RDF_TYPE_STR)
					lablen_i = (size_t)*di;
				else	lablen_i = get_rdf_size(
					desc->_wireformat[wfi]) - 1;
			}
		} else	lablen_i--;

		/* advance field j; same as for i */
		if(lablen_j == 0) {
			if(dname_j) {
				lablen_j = (size_t)*dj;
				if(lablen_j == 0) {
					dname_j = 0;
					dname_num_j--;
					if(dname_num_j == 0)
						lablen_j = jlen;
				}
			} else {
				wfj++;
				if(desc->_wireformat[wfj]
					== LDNS_RDF_TYPE_DNAME) {
					dname_j = 1;
					lablen_j = (size_t)*dj;
					if(lablen_j == 0) {
						dname_j = 0;
						dname_num_j--;
						if(dname_num_j == 0)
							lablen_j = jlen;
					}
				} else if(desc->_wireformat[wfj]
					== LDNS_RDF_TYPE_STR)
					lablen_j = (size_t)*dj;
				else	lablen_j = get_rdf_size(
					desc->_wireformat[wfj]) - 1;
			}
		} else	lablen_j--;
		di++;
		dj++;
	}
	/* end of the loop; because we advanced byte by byte; now we have
	 * that the rdata has ended, or that there is a binary remainder */
	/* shortest first */
	if(ilen == 0 && jlen == 0)
		return 0;
	if(ilen == 0)
		return -1;
	if(jlen == 0)
		return 1;
	/* binary remainder, capture comparison in wfi variable */
	if((wfi = memcmp(di, dj, (ilen<jlen)?ilen:jlen)) != 0)
		return wfi;
	if(ilen < jlen)
		return -1;
	if(jlen < ilen)
		return 1;
	return 0;
}

/**
 * Compare two RRs in the same RRset and determine their relative
 * canonical order.
 * @param rrset: the rrset in which to perform compares.
 * @param i: first RR to compare
 * @param j: first RR to compare
 * @return 0 if RR i== RR j, -1 if <, +1 if >.
 */
static int
canonical_compare(struct ub_packed_rrset_key* rrset, size_t i, size_t j)
{
	struct packed_rrset_data* d = (struct packed_rrset_data*)
		rrset->entry.data;
	const sldns_rr_descriptor* desc;
	uint16_t type = ntohs(rrset->rk.type);
	size_t minlen;
	int c;

	if(i==j)
		return 0;

	switch(type) {
		/* These RR types have only a name as RDATA.
		 * This name has to be canonicalized.*/
		case LDNS_RR_TYPE_NS:
		case LDNS_RR_TYPE_MD:
		case LDNS_RR_TYPE_MF:
		case LDNS_RR_TYPE_CNAME:
		case LDNS_RR_TYPE_MB:
		case LDNS_RR_TYPE_MG:
		case LDNS_RR_TYPE_MR:
		case LDNS_RR_TYPE_PTR:
		case LDNS_RR_TYPE_DNAME:
			/* the wireread function has already checked these
			 * dname's for correctness, and this double checks */
			if(!dname_valid(d->rr_data[i]+2, d->rr_len[i]-2) ||
				!dname_valid(d->rr_data[j]+2, d->rr_len[j]-2))
				return 0;
			return query_dname_compare(d->rr_data[i]+2,
				d->rr_data[j]+2);

		/* These RR types have STR and fixed size rdata fields
		 * before one or more name fields that need canonicalizing,
		 * and after that a byte-for byte remainder can be compared.
		 */
		/* type starts with the name; remainder is binary compared */
		case LDNS_RR_TYPE_NXT:
		/* use rdata field formats */
		case LDNS_RR_TYPE_MINFO:
		case LDNS_RR_TYPE_RP:
		case LDNS_RR_TYPE_SOA:
		case LDNS_RR_TYPE_RT:
		case LDNS_RR_TYPE_AFSDB:
		case LDNS_RR_TYPE_KX:
		case LDNS_RR_TYPE_MX:
		case LDNS_RR_TYPE_SIG:
		/* RRSIG signer name has to be downcased */
		case LDNS_RR_TYPE_RRSIG:
		case LDNS_RR_TYPE_PX:
		case LDNS_RR_TYPE_NAPTR:
		case LDNS_RR_TYPE_SRV:
			desc = sldns_rr_descript(type);
			log_assert(desc);
			/* this holds for the types that need canonicalizing */
			log_assert(desc->_minimum == desc->_maximum);
			return canonical_compare_byfield(d, desc, i, j);

		case LDNS_RR_TYPE_HINFO: /* no longer downcased */
		case LDNS_RR_TYPE_NSEC:
	default:
		/* For unknown RR types, or types not listed above,
		 * no canonicalization is needed, do binary compare */
		/* byte for byte compare, equal means shortest first*/
		minlen = d->rr_len[i]-2;
		if(minlen > d->rr_len[j]-2)
			minlen = d->rr_len[j]-2;
		c = memcmp(d->rr_data[i]+2, d->rr_data[j]+2, minlen);
		if(c!=0)
			return c;
		/* rdata equal, shortest is first */
		if(d->rr_len[i] < d->rr_len[j])
			return -1;
		if(d->rr_len[i] > d->rr_len[j])
			return 1;
		/* rdata equal, length equal */
		break;
	}
	return 0;
}

int
canonical_tree_compare(const void* k1, const void* k2)
{
	struct canon_rr* r1 = (struct canon_rr*)k1;
	struct canon_rr* r2 = (struct canon_rr*)k2;
	log_assert(r1->rrset == r2->rrset);
	return canonical_compare(r1->rrset, r1->rr_idx, r2->rr_idx);
}

/**
 * Sort RRs for rrset in canonical order.
 * Does not actually canonicalize the RR rdatas.
 * Does not touch rrsigs.
 * @param rrset: to sort.
 * @param d: rrset data.
 * @param sortree: tree to sort into.
 * @param rrs: rr storage.
 */
static void
canonical_sort(struct ub_packed_rrset_key* rrset, struct packed_rrset_data* d,
	rbtree_type* sortree, struct canon_rr* rrs)
{
	size_t i;
	/* insert into rbtree to sort and detect duplicates */
	for(i=0; i<d->count; i++) {
		rrs[i].node.key = &rrs[i];
		rrs[i].rrset = rrset;
		rrs[i].rr_idx = i;
		if(!rbtree_insert(sortree, &rrs[i].node)) {
			/* this was a duplicate */
		}
	}
}

/**
 * Insert canonical owner name into buffer.
 * @param buf: buffer to insert into at current position.
 * @param k: rrset with its owner name.
 * @param sig: signature with signer name and label count.
 * 	must be length checked, at least 18 bytes long.
 * @param can_owner: position in buffer returned for future use.
 * @param can_owner_len: length of canonical owner name.
 */
static void
insert_can_owner(sldns_buffer* buf, struct ub_packed_rrset_key* k,
	uint8_t* sig, uint8_t** can_owner, size_t* can_owner_len)
{
	int rrsig_labels = (int)sig[3];
	int fqdn_labels = dname_signame_label_count(k->rk.dname);
	*can_owner = sldns_buffer_current(buf);
	if(rrsig_labels == fqdn_labels) {
		/* no change */
		sldns_buffer_write(buf, k->rk.dname, k->rk.dname_len);
		query_dname_tolower(*can_owner);
		*can_owner_len = k->rk.dname_len;
		return;
	}
	log_assert(rrsig_labels < fqdn_labels);
	/* *. | fqdn(rightmost rrsig_labels) */
	if(rrsig_labels < fqdn_labels) {
		int i;
		uint8_t* nm = k->rk.dname;
		size_t len = k->rk.dname_len;
		/* so skip fqdn_labels-rrsig_labels */
		for(i=0; i<fqdn_labels-rrsig_labels; i++) {
			dname_remove_label(&nm, &len);
		}
		*can_owner_len = len+2;
		sldns_buffer_write(buf, (uint8_t*)"\001*", 2);
		sldns_buffer_write(buf, nm, len);
		query_dname_tolower(*can_owner);
	}
}

/**
 * Canonicalize Rdata in buffer.
 * @param buf: buffer at position just after the rdata.
 * @param rrset: rrset with type.
 * @param len: length of the rdata (including rdatalen uint16).
 */
static void
canonicalize_rdata(sldns_buffer* buf, struct ub_packed_rrset_key* rrset,
	size_t len)
{
	uint8_t* datstart = sldns_buffer_current(buf)-len+2;
	switch(ntohs(rrset->rk.type)) {
		case LDNS_RR_TYPE_NXT:
		case LDNS_RR_TYPE_NS:
		case LDNS_RR_TYPE_MD:
		case LDNS_RR_TYPE_MF:
		case LDNS_RR_TYPE_CNAME:
		case LDNS_RR_TYPE_MB:
		case LDNS_RR_TYPE_MG:
		case LDNS_RR_TYPE_MR:
		case LDNS_RR_TYPE_PTR:
		case LDNS_RR_TYPE_DNAME:
			/* type only has a single argument, the name */
			query_dname_tolower(datstart);
			return;
		case LDNS_RR_TYPE_MINFO:
		case LDNS_RR_TYPE_RP:
		case LDNS_RR_TYPE_SOA:
			/* two names after another */
			query_dname_tolower(datstart);
			query_dname_tolower(datstart +
				dname_valid(datstart, len-2));
			return;
		case LDNS_RR_TYPE_RT:
		case LDNS_RR_TYPE_AFSDB:
		case LDNS_RR_TYPE_KX:
		case LDNS_RR_TYPE_MX:
			/* skip fixed part */
			if(len < 2+2+1) /* rdlen, skiplen, 1byteroot */
				return;
			datstart += 2;
			query_dname_tolower(datstart);
			return;
		case LDNS_RR_TYPE_SIG:
		/* downcase the RRSIG, compat with BIND (kept it from SIG) */
		case LDNS_RR_TYPE_RRSIG:
			/* skip fixed part */
			if(len < 2+18+1)
				return;
			datstart += 18;
			query_dname_tolower(datstart);
			return;
		case LDNS_RR_TYPE_PX:
			/* skip, then two names after another */
			if(len < 2+2+1)
				return;
			datstart += 2;
			query_dname_tolower(datstart);
			query_dname_tolower(datstart +
				dname_valid(datstart, len-2-2));
			return;
		case LDNS_RR_TYPE_NAPTR:
			if(len < 2+4)
				return;
			len -= 2+4;
			datstart += 4;
			if(len < (size_t)datstart[0]+1) /* skip text field */
				return;
			len -= (size_t)datstart[0]+1;
			datstart += (size_t)datstart[0]+1;
			if(len < (size_t)datstart[0]+1) /* skip text field */
				return;
			len -= (size_t)datstart[0]+1;
			datstart += (size_t)datstart[0]+1;
			if(len < (size_t)datstart[0]+1) /* skip text field */
				return;
			len -= (size_t)datstart[0]+1;
			datstart += (size_t)datstart[0]+1;
			if(len < 1)	/* check name is at least 1 byte*/
				return;
			query_dname_tolower(datstart);
			return;
		case LDNS_RR_TYPE_SRV:
			/* skip fixed part */
			if(len < 2+6+1)
				return;
			datstart += 6;
			query_dname_tolower(datstart);
			return;

		/* do not canonicalize NSEC rdata name, compat with
		 * from bind 9.4 signer, where it does not do so */
		case LDNS_RR_TYPE_NSEC: /* type starts with the name */
		case LDNS_RR_TYPE_HINFO: /* not downcased */
		/* A6 not supported */
		default:
			/* nothing to do for unknown types */
			return;
	}
}

int rrset_canonical_equal(struct regional* region,
	struct ub_packed_rrset_key* k1, struct ub_packed_rrset_key* k2)
{
	struct rbtree_type sortree1, sortree2;
	struct canon_rr *rrs1, *rrs2, *p1, *p2;
	struct packed_rrset_data* d1=(struct packed_rrset_data*)k1->entry.data;
	struct packed_rrset_data* d2=(struct packed_rrset_data*)k2->entry.data;
	struct ub_packed_rrset_key fk;
	struct packed_rrset_data fd;
	size_t flen[2];
	uint8_t* fdata[2];

	/* basic compare */
	if(k1->rk.dname_len != k2->rk.dname_len ||
		k1->rk.flags != k2->rk.flags ||
		k1->rk.type != k2->rk.type ||
		k1->rk.rrset_class != k2->rk.rrset_class ||
		query_dname_compare(k1->rk.dname, k2->rk.dname) != 0)
		return 0;
	if(d1->ttl != d2->ttl ||
		d1->count != d2->count ||
		d1->rrsig_count != d2->rrsig_count ||
		d1->trust != d2->trust ||
		d1->security != d2->security)
		return 0;

	/* init */
	memset(&fk, 0, sizeof(fk));
	memset(&fd, 0, sizeof(fd));
	fk.entry.data = &fd;
	fd.count = 2;
	fd.rr_len = flen;
	fd.rr_data = fdata;
	rbtree_init(&sortree1, &canonical_tree_compare);
	rbtree_init(&sortree2, &canonical_tree_compare);
	if(d1->count > RR_COUNT_MAX || d2->count > RR_COUNT_MAX)
		return 1; /* protection against integer overflow */
	rrs1 = regional_alloc(region, sizeof(struct canon_rr)*d1->count);
	rrs2 = regional_alloc(region, sizeof(struct canon_rr)*d2->count);
	if(!rrs1 || !rrs2) return 1; /* alloc failure */

	/* sort */
	canonical_sort(k1, d1, &sortree1, rrs1);
	canonical_sort(k2, d2, &sortree2, rrs2);

	/* compare canonical-sorted RRs for canonical-equality */
	if(sortree1.count != sortree2.count)
		return 0;
	p1 = (struct canon_rr*)rbtree_first(&sortree1);
	p2 = (struct canon_rr*)rbtree_first(&sortree2);
	while(p1 != (struct canon_rr*)RBTREE_NULL &&
		p2 != (struct canon_rr*)RBTREE_NULL) {
		flen[0] = d1->rr_len[p1->rr_idx];
		flen[1] = d2->rr_len[p2->rr_idx];
		fdata[0] = d1->rr_data[p1->rr_idx];
		fdata[1] = d2->rr_data[p2->rr_idx];

		if(canonical_compare(&fk, 0, 1) != 0)
			return 0;
		p1 = (struct canon_rr*)rbtree_next(&p1->node);
		p2 = (struct canon_rr*)rbtree_next(&p2->node);
	}
	return 1;
}

/**
 * Create canonical form of rrset in the scratch buffer.
 * @param region: temporary region.
 * @param buf: the buffer to use.
 * @param k: the rrset to insert.
 * @param sig: RRSIG rdata to include.
 * @param siglen: RRSIG rdata len excluding signature field, but inclusive
 * 	signer name length.
 * @param sortree: if NULL is passed a new sorted rrset tree is built.
 * 	Otherwise it is reused.
 * @param section: section of packet where this rrset comes from.
 * @param qstate: qstate with region.
 * @return false on alloc error.
 */
static int
rrset_canonical(struct regional* region, sldns_buffer* buf,
	struct ub_packed_rrset_key* k, uint8_t* sig, size_t siglen,
	struct rbtree_type** sortree, sldns_pkt_section section,
	struct module_qstate* qstate)
{
	struct packed_rrset_data* d = (struct packed_rrset_data*)k->entry.data;
	uint8_t* can_owner = NULL;
	size_t can_owner_len = 0;
	struct canon_rr* walk;
	struct canon_rr* rrs;

	if(!*sortree) {
		*sortree = (struct rbtree_type*)regional_alloc(region,
			sizeof(rbtree_type));
		if(!*sortree)
			return 0;
		if(d->count > RR_COUNT_MAX)
			return 0; /* integer overflow protection */
		rrs = regional_alloc(region, sizeof(struct canon_rr)*d->count);
		if(!rrs) {
			*sortree = NULL;
			return 0;
		}
		rbtree_init(*sortree, &canonical_tree_compare);
		canonical_sort(k, d, *sortree, rrs);
	}

	sldns_buffer_clear(buf);
	sldns_buffer_write(buf, sig, siglen);
	/* canonicalize signer name */
	query_dname_tolower(sldns_buffer_begin(buf)+18);
	RBTREE_FOR(walk, struct canon_rr*, (*sortree)) {
		/* see if there is enough space left in the buffer */
		if(sldns_buffer_remaining(buf) < can_owner_len + 2 + 2 + 4
			+ d->rr_len[walk->rr_idx]) {
			log_err("verify: failed to canonicalize, "
				"rrset too big");
			return 0;
		}
		/* determine canonical owner name */
		if(can_owner)
			sldns_buffer_write(buf, can_owner, can_owner_len);
		else	insert_can_owner(buf, k, sig, &can_owner,
				&can_owner_len);
		sldns_buffer_write(buf, &k->rk.type, 2);
		sldns_buffer_write(buf, &k->rk.rrset_class, 2);
		sldns_buffer_write(buf, sig+4, 4);
		sldns_buffer_write(buf, d->rr_data[walk->rr_idx],
			d->rr_len[walk->rr_idx]);
		canonicalize_rdata(buf, k, d->rr_len[walk->rr_idx]);
	}
	sldns_buffer_flip(buf);

	/* Replace RR owner with canonical owner for NSEC records in authority
	 * section, to prevent that a wildcard synthesized NSEC can be used in
	 * the non-existence proves. */
	if(ntohs(k->rk.type) == LDNS_RR_TYPE_NSEC &&
		section == LDNS_SECTION_AUTHORITY && qstate) {
		k->rk.dname = regional_alloc_init(qstate->region, can_owner,
			can_owner_len);
		if(!k->rk.dname)
			return 0;
		k->rk.dname_len = can_owner_len;
	}


	return 1;
}

int
rrset_canonicalize_to_buffer(struct regional* region, sldns_buffer* buf,
	struct ub_packed_rrset_key* k)
{
	struct rbtree_type* sortree = NULL;
	struct packed_rrset_data* d = (struct packed_rrset_data*)k->entry.data;
	uint8_t* can_owner = NULL;
	size_t can_owner_len = 0;
	struct canon_rr* walk;
	struct canon_rr* rrs;

	sortree = (struct rbtree_type*)regional_alloc(region,
		sizeof(rbtree_type));
	if(!sortree)
		return 0;
	if(d->count > RR_COUNT_MAX)
		return 0; /* integer overflow protection */
	rrs = regional_alloc(region, sizeof(struct canon_rr)*d->count);
	if(!rrs) {
		return 0;
	}
	rbtree_init(sortree, &canonical_tree_compare);
	canonical_sort(k, d, sortree, rrs);

	sldns_buffer_clear(buf);
	RBTREE_FOR(walk, struct canon_rr*, sortree) {
		/* see if there is enough space left in the buffer */
		if(sldns_buffer_remaining(buf) < can_owner_len + 2 + 2 + 4
			+ d->rr_len[walk->rr_idx]) {
			log_err("verify: failed to canonicalize, "
				"rrset too big");
			return 0;
		}
		/* determine canonical owner name */
		if(can_owner)
			sldns_buffer_write(buf, can_owner, can_owner_len);
		else	{
			can_owner = sldns_buffer_current(buf);
			sldns_buffer_write(buf, k->rk.dname, k->rk.dname_len);
			query_dname_tolower(can_owner);
			can_owner_len = k->rk.dname_len;
		}
		sldns_buffer_write(buf, &k->rk.type, 2);
		sldns_buffer_write(buf, &k->rk.rrset_class, 2);
		sldns_buffer_write_u32(buf, d->rr_ttl[walk->rr_idx]);
		sldns_buffer_write(buf, d->rr_data[walk->rr_idx],
			d->rr_len[walk->rr_idx]);
		canonicalize_rdata(buf, k, d->rr_len[walk->rr_idx]);
	}
	sldns_buffer_flip(buf);
	return 1;
}

/** pretty print rrsig error with dates */
static void
sigdate_error(const char* str, int32_t expi, int32_t incep, int32_t now)
{
	struct tm tm;
	char expi_buf[16];
	char incep_buf[16];
	char now_buf[16];
	time_t te, ti, tn;

	if(verbosity < VERB_QUERY)
		return;
	te = (time_t)expi;
	ti = (time_t)incep;
	tn = (time_t)now;
	memset(&tm, 0, sizeof(tm));
	if(gmtime_r(&te, &tm) && strftime(expi_buf, 15, "%Y%m%d%H%M%S", &tm)
	 &&gmtime_r(&ti, &tm) && strftime(incep_buf, 15, "%Y%m%d%H%M%S", &tm)
	 &&gmtime_r(&tn, &tm) && strftime(now_buf, 15, "%Y%m%d%H%M%S", &tm)) {
		log_info("%s expi=%s incep=%s now=%s", str, expi_buf,
			incep_buf, now_buf);
	} else
		log_info("%s expi=%u incep=%u now=%u", str, (unsigned)expi,
			(unsigned)incep, (unsigned)now);
}

/** check rrsig dates */
static int
check_dates(struct val_env* ve, uint32_t unow, uint8_t* expi_p,
	uint8_t* incep_p, char** reason, sldns_ede_code *reason_bogus)
{
	/* read out the dates */
	uint32_t expi, incep, now;
	memmove(&expi, expi_p, sizeof(expi));
	memmove(&incep, incep_p, sizeof(incep));
	expi = ntohl(expi);
	incep = ntohl(incep);

	/* get current date */
	if(ve->date_override) {
		if(ve->date_override == -1) {
			verbose(VERB_ALGO, "date override: ignore date");
			return 1;
		}
		now = ve->date_override;
		verbose(VERB_ALGO, "date override option %d", (int)now);
	} else	now = unow;

	/* check them */
	if(compare_1982(incep, expi) > 0) {
		sigdate_error("verify: inception after expiration, "
			"signature bad", expi, incep, now);
		*reason = "signature inception after expiration";
		if(reason_bogus){
			/* from RFC8914 on Signature Not Yet Valid: The resolver
			 * attempted to perform DNSSEC validation, but no
			 * signatures are presently valid and at least some are
			 * not yet valid. */
			*reason_bogus = LDNS_EDE_SIGNATURE_NOT_YET_VALID;
		}

		return 0;
	}
	if(compare_1982(incep, now) > 0) {
		/* within skew ? (calc here to avoid calculation normally) */
		uint32_t skew = subtract_1982(incep, expi)/10;
		if(skew < (uint32_t)ve->skew_min) skew = ve->skew_min;
		if(skew > (uint32_t)ve->skew_max) skew = ve->skew_max;
		if(subtract_1982(now, incep) > skew) {
			sigdate_error("verify: signature bad, current time is"
				" before inception date", expi, incep, now);
			*reason = "signature before inception date";
			if(reason_bogus)
				*reason_bogus = LDNS_EDE_SIGNATURE_NOT_YET_VALID;
			return 0;
		}
		sigdate_error("verify warning suspicious signature inception "
			" or bad local clock", expi, incep, now);
	}
	if(compare_1982(now, expi) > 0) {
		uint32_t skew = subtract_1982(incep, expi)/10;
		if(skew < (uint32_t)ve->skew_min) skew = ve->skew_min;
		if(skew > (uint32_t)ve->skew_max) skew = ve->skew_max;
		if(subtract_1982(expi, now) > skew) {
			sigdate_error("verify: signature expired", expi,
				incep, now);
			*reason = "signature expired";
			if(reason_bogus)
				*reason_bogus = LDNS_EDE_SIGNATURE_EXPIRED;
			return 0;
		}
		sigdate_error("verify warning suspicious signature expiration "
			" or bad local clock", expi, incep, now);
	}
	return 1;
}

/** adjust rrset TTL for verified rrset, compare to original TTL and expi */
static void
adjust_ttl(struct val_env* ve, uint32_t unow,
	struct ub_packed_rrset_key* rrset, uint8_t* orig_p,
	uint8_t* expi_p, uint8_t* incep_p)
{
	struct packed_rrset_data* d =
		(struct packed_rrset_data*)rrset->entry.data;
	/* read out the dates */
	int32_t origttl, expittl, expi, incep, now;
	memmove(&origttl, orig_p, sizeof(origttl));
	memmove(&expi, expi_p, sizeof(expi));
	memmove(&incep, incep_p, sizeof(incep));
	expi = ntohl(expi);
	incep = ntohl(incep);
	origttl = ntohl(origttl);

	/* get current date */
	if(ve->date_override) {
		now = ve->date_override;
	} else	now = (int32_t)unow;
	expittl = (int32_t)((uint32_t)expi - (uint32_t)now);

	/* so now:
	 * d->ttl: rrset ttl read from message or cache. May be reduced
	 * origttl: original TTL from signature, authoritative TTL max.
	 * MIN_TTL: minimum TTL from config.
	 * expittl: TTL until the signature expires.
	 *
	 * Use the smallest of these, but don't let origttl set the TTL
	 * below the minimum.
	 */
	if(MIN_TTL > (time_t)origttl && d->ttl > MIN_TTL) {
		verbose(VERB_QUERY, "rrset TTL larger than original and minimum"
			" TTL, adjusting TTL downwards to minimum ttl");
		d->ttl = MIN_TTL;
	}
	else if(MIN_TTL <= origttl && d->ttl > (time_t)origttl) {
		verbose(VERB_QUERY, "rrset TTL larger than original TTL, "
		"adjusting TTL downwards to original ttl");
		d->ttl = origttl;
	}

	if(expittl > 0 && d->ttl > (time_t)expittl) {
		verbose(VERB_ALGO, "rrset TTL larger than sig expiration ttl,"
			" adjusting TTL downwards");
		d->ttl = expittl;
	}
}

enum sec_status
dnskey_verify_rrset_sig(struct regional* region, sldns_buffer* buf,
	struct val_env* ve, time_t now,
        struct ub_packed_rrset_key* rrset, struct ub_packed_rrset_key* dnskey,
        size_t dnskey_idx, size_t sig_idx,
	struct rbtree_type** sortree, int* buf_canon,
	char** reason, sldns_ede_code *reason_bogus,
	sldns_pkt_section section, struct module_qstate* qstate)
{
	enum sec_status sec;
	uint8_t* sig;		/* RRSIG rdata */
	size_t siglen;
	size_t rrnum = rrset_get_count(rrset);
	uint8_t* signer;	/* rrsig signer name */
	size_t signer_len;
	unsigned char* sigblock; /* signature rdata field */
	unsigned int sigblock_len;
	uint16_t ktag;		/* DNSKEY key tag */
	unsigned char* key;	/* public key rdata field */
	unsigned int keylen;
	rrset_get_rdata(rrset, rrnum + sig_idx, &sig, &siglen);
	/* min length of rdatalen, fixed rrsig, root signer, 1 byte sig */
	if(siglen < 2+20) {
		verbose(VERB_QUERY, "verify: signature too short");
		*reason = "signature too short";
		if(reason_bogus)
			*reason_bogus = LDNS_EDE_DNSSEC_BOGUS;
		return sec_status_bogus;
	}

	if(!(dnskey_get_flags(dnskey, dnskey_idx) & DNSKEY_BIT_ZSK)) {
		verbose(VERB_QUERY, "verify: dnskey without ZSK flag");
		*reason = "dnskey without ZSK flag";
		if(reason_bogus)
			*reason_bogus = LDNS_EDE_NO_ZONE_KEY_BIT_SET;
		return sec_status_bogus;
	}

	if(dnskey_get_protocol(dnskey, dnskey_idx) != LDNS_DNSSEC_KEYPROTO) {
		/* RFC 4034 says DNSKEY PROTOCOL MUST be 3 */
		verbose(VERB_QUERY, "verify: dnskey has wrong key protocol");
		*reason = "dnskey has wrong protocolnumber";
		if(reason_bogus)
			*reason_bogus = LDNS_EDE_DNSSEC_BOGUS;
		return sec_status_bogus;
	}

	/* verify as many fields in rrsig as possible */
	signer = sig+2+18;
	signer_len = dname_valid(signer, siglen-2-18);
	if(!signer_len) {
		verbose(VERB_QUERY, "verify: malformed signer name");
		*reason = "signer name malformed";
		if(reason_bogus)
			*reason_bogus = LDNS_EDE_DNSSEC_BOGUS;
		return sec_status_bogus; /* signer name invalid */
	}
	if(!dname_subdomain_c(rrset->rk.dname, signer)) {
		verbose(VERB_QUERY, "verify: signer name is off-tree");
		*reason = "signer name off-tree";
		if(reason_bogus)
			*reason_bogus = LDNS_EDE_DNSSEC_BOGUS;
		return sec_status_bogus; /* signer name offtree */
	}
	sigblock = (unsigned char*)signer+signer_len;
	if(siglen < 2+18+signer_len+1) {
		verbose(VERB_QUERY, "verify: too short, no signature data");
		*reason = "signature too short, no signature data";
		if(reason_bogus)
			*reason_bogus = LDNS_EDE_DNSSEC_BOGUS;
		return sec_status_bogus; /* sig rdf is < 1 byte */
	}
	sigblock_len = (unsigned int)(siglen - 2 - 18 - signer_len);

	/* verify key dname == sig signer name */
	if(query_dname_compare(signer, dnskey->rk.dname) != 0) {
		verbose(VERB_QUERY, "verify: wrong key for rrsig");
		log_nametypeclass(VERB_QUERY, "RRSIG signername is",
			signer, 0, 0);
		log_nametypeclass(VERB_QUERY, "the key name is",
			dnskey->rk.dname, 0, 0);
		*reason = "signer name mismatches key name";
		if(reason_bogus)
			*reason_bogus = LDNS_EDE_DNSSEC_BOGUS;
		return sec_status_bogus;
	}

	/* verify covered type */
	/* memcmp works because type is in network format for rrset */
	if(memcmp(sig+2, &rrset->rk.type, 2) != 0) {
		verbose(VERB_QUERY, "verify: wrong type covered");
		*reason = "signature covers wrong type";
		if(reason_bogus)
			*reason_bogus = LDNS_EDE_DNSSEC_BOGUS;
		return sec_status_bogus;
	}
	/* verify keytag and sig algo (possibly again) */
	if((int)sig[2+2] != dnskey_get_algo(dnskey, dnskey_idx)) {
		verbose(VERB_QUERY, "verify: wrong algorithm");
		*reason = "signature has wrong algorithm";
		if(reason_bogus)
			*reason_bogus = LDNS_EDE_DNSSEC_BOGUS;
		return sec_status_bogus;
	}
	ktag = htons(dnskey_calc_keytag(dnskey, dnskey_idx));
	if(memcmp(sig+2+16, &ktag, 2) != 0) {
		verbose(VERB_QUERY, "verify: wrong keytag");
		*reason = "signature has wrong keytag";
		if(reason_bogus)
			*reason_bogus = LDNS_EDE_DNSSEC_BOGUS;
		return sec_status_bogus;
	}

	/* verify labels is in a valid range */
	if((int)sig[2+3] > dname_signame_label_count(rrset->rk.dname)) {
		verbose(VERB_QUERY, "verify: labelcount out of range");
		*reason = "signature labelcount out of range";
		if(reason_bogus)
			*reason_bogus = LDNS_EDE_DNSSEC_BOGUS;
		return sec_status_bogus;
	}

	/* original ttl, always ok */

	if(!*buf_canon) {
		/* create rrset canonical format in buffer, ready for
		 * signature */
		if(!rrset_canonical(region, buf, rrset, sig+2,
			18 + signer_len, sortree, section, qstate)) {
			log_err("verify: failed due to alloc error");
			return sec_status_unchecked;
		}
		*buf_canon = 1;
	}

	/* check that dnskey is available */
	dnskey_get_pubkey(dnskey, dnskey_idx, &key, &keylen);
	if(!key) {
		verbose(VERB_QUERY, "verify: short DNSKEY RR");
		return sec_status_unchecked;
	}

	/* verify */
	sec = verify_canonrrset(buf, (int)sig[2+2],
		sigblock, sigblock_len, key, keylen, reason);

	/* count validation operation */
	if(qstate && qstate->env && qstate->env->mesh)
		qstate->env->mesh->val_ops++;

	if(sec == sec_status_secure) {
		/* check if TTL is too high - reduce if so */
		adjust_ttl(ve, now, rrset, sig+2+4, sig+2+8, sig+2+12);

		/* verify inception, expiration dates
		 * Do this last so that if you ignore expired-sigs the
		 * rest is sure to be OK. */
		if(!check_dates(ve, now, sig+2+8, sig+2+12,
			reason, reason_bogus)) {
			return sec_status_bogus;
		}
	}

	return sec;
}