1 /*
2 * validator/val_neg.c - validator aggressive negative caching functions.
3 *
4 * Copyright (c) 2008, NLnet Labs. All rights reserved.
5 *
6 * This software is open source.
7 *
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
10 * are met:
11 *
12 * Redistributions of source code must retain the above copyright notice,
13 * this list of conditions and the following disclaimer.
14 *
15 * Redistributions in binary form must reproduce the above copyright notice,
16 * this list of conditions and the following disclaimer in the documentation
17 * and/or other materials provided with the distribution.
18 *
19 * Neither the name of the NLNET LABS nor the names of its contributors may
20 * be used to endorse or promote products derived from this software without
21 * specific prior written permission.
22 *
23 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
24 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
25 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
26 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
27 * HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
28 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED
29 * TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
30 * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
31 * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
32 * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
33 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
34 */
35
36 /**
37 * \file
38 *
39 * This file contains helper functions for the validator module.
40 * The functions help with aggressive negative caching.
41 * This creates new denials of existence, and proofs for absence of types
42 * from cached NSEC records.
43 */
44 #include "config.h"
45 #ifdef HAVE_OPENSSL_SSL_H
46 #include <openssl/ssl.h>
47 #define NSEC3_SHA_LEN SHA_DIGEST_LENGTH
48 #else
49 #define NSEC3_SHA_LEN 20
50 #endif
51 #include "validator/val_neg.h"
52 #include "validator/val_nsec.h"
53 #include "validator/val_nsec3.h"
54 #include "validator/val_utils.h"
55 #include "util/data/dname.h"
56 #include "util/data/msgreply.h"
57 #include "util/log.h"
58 #include "util/net_help.h"
59 #include "util/config_file.h"
60 #include "services/cache/rrset.h"
61 #include "services/cache/dns.h"
62 #include "sldns/rrdef.h"
63 #include "sldns/sbuffer.h"
64
65 /**
66 * The maximum salt length that the negative cache is willing to use.
67 * Larger salt increases the computation time, while recommendations are
68 * for zero salt length for zones.
69 */
70 #define MAX_SALT_LENGTH 64
71
val_neg_data_compare(const void * a,const void * b)72 int val_neg_data_compare(const void* a, const void* b)
73 {
74 struct val_neg_data* x = (struct val_neg_data*)a;
75 struct val_neg_data* y = (struct val_neg_data*)b;
76 int m;
77 return dname_canon_lab_cmp(x->name, x->labs, y->name, y->labs, &m);
78 }
79
val_neg_zone_compare(const void * a,const void * b)80 int val_neg_zone_compare(const void* a, const void* b)
81 {
82 struct val_neg_zone* x = (struct val_neg_zone*)a;
83 struct val_neg_zone* y = (struct val_neg_zone*)b;
84 int m;
85 if(x->dclass != y->dclass) {
86 if(x->dclass < y->dclass)
87 return -1;
88 return 1;
89 }
90 return dname_canon_lab_cmp(x->name, x->labs, y->name, y->labs, &m);
91 }
92
val_neg_create(struct config_file * cfg,size_t maxiter)93 struct val_neg_cache* val_neg_create(struct config_file* cfg, size_t maxiter)
94 {
95 struct val_neg_cache* neg = (struct val_neg_cache*)calloc(1,
96 sizeof(*neg));
97 if(!neg) {
98 log_err("Could not create neg cache: out of memory");
99 return NULL;
100 }
101 neg->nsec3_max_iter = maxiter;
102 neg->max = 1024*1024; /* 1 M is thousands of entries */
103 if(cfg) neg->max = cfg->neg_cache_size;
104 rbtree_init(&neg->tree, &val_neg_zone_compare);
105 lock_basic_init(&neg->lock);
106 lock_protect(&neg->lock, neg, sizeof(*neg));
107 return neg;
108 }
109
val_neg_get_mem(struct val_neg_cache * neg)110 size_t val_neg_get_mem(struct val_neg_cache* neg)
111 {
112 size_t result;
113 lock_basic_lock(&neg->lock);
114 result = sizeof(*neg) + neg->use;
115 lock_basic_unlock(&neg->lock);
116 return result;
117 }
118
119 /** clear datas on cache deletion */
120 static void
neg_clear_datas(rbnode_type * n,void * ATTR_UNUSED (arg))121 neg_clear_datas(rbnode_type* n, void* ATTR_UNUSED(arg))
122 {
123 struct val_neg_data* d = (struct val_neg_data*)n;
124 free(d->name);
125 free(d);
126 }
127
128 /** clear zones on cache deletion */
129 static void
neg_clear_zones(rbnode_type * n,void * ATTR_UNUSED (arg))130 neg_clear_zones(rbnode_type* n, void* ATTR_UNUSED(arg))
131 {
132 struct val_neg_zone* z = (struct val_neg_zone*)n;
133 /* delete all the rrset entries in the tree */
134 traverse_postorder(&z->tree, &neg_clear_datas, NULL);
135 free(z->nsec3_salt);
136 free(z->name);
137 free(z);
138 }
139
neg_cache_delete(struct val_neg_cache * neg)140 void neg_cache_delete(struct val_neg_cache* neg)
141 {
142 if(!neg) return;
143 lock_basic_destroy(&neg->lock);
144 /* delete all the zones in the tree */
145 traverse_postorder(&neg->tree, &neg_clear_zones, NULL);
146 free(neg);
147 }
148
149 /**
150 * Put data element at the front of the LRU list.
151 * @param neg: negative cache with LRU start and end.
152 * @param data: this data is fronted.
153 */
neg_lru_front(struct val_neg_cache * neg,struct val_neg_data * data)154 static void neg_lru_front(struct val_neg_cache* neg,
155 struct val_neg_data* data)
156 {
157 data->prev = NULL;
158 data->next = neg->first;
159 if(!neg->first)
160 neg->last = data;
161 else neg->first->prev = data;
162 neg->first = data;
163 }
164
165 /**
166 * Remove data element from LRU list.
167 * @param neg: negative cache with LRU start and end.
168 * @param data: this data is removed from the list.
169 */
neg_lru_remove(struct val_neg_cache * neg,struct val_neg_data * data)170 static void neg_lru_remove(struct val_neg_cache* neg,
171 struct val_neg_data* data)
172 {
173 if(data->prev)
174 data->prev->next = data->next;
175 else neg->first = data->next;
176 if(data->next)
177 data->next->prev = data->prev;
178 else neg->last = data->prev;
179 }
180
181 /**
182 * Touch LRU for data element, put it at the start of the LRU list.
183 * @param neg: negative cache with LRU start and end.
184 * @param data: this data is used.
185 */
neg_lru_touch(struct val_neg_cache * neg,struct val_neg_data * data)186 static void neg_lru_touch(struct val_neg_cache* neg,
187 struct val_neg_data* data)
188 {
189 if(data == neg->first)
190 return; /* nothing to do */
191 /* remove from current lru position */
192 neg_lru_remove(neg, data);
193 /* add at front */
194 neg_lru_front(neg, data);
195 }
196
197 /**
198 * Delete a zone element from the negative cache.
199 * May delete other zone elements to keep tree coherent, or
200 * only mark the element as 'not in use'.
201 * @param neg: negative cache.
202 * @param z: zone element to delete.
203 */
neg_delete_zone(struct val_neg_cache * neg,struct val_neg_zone * z)204 static void neg_delete_zone(struct val_neg_cache* neg, struct val_neg_zone* z)
205 {
206 struct val_neg_zone* p, *np;
207 if(!z) return;
208 log_assert(z->in_use);
209 log_assert(z->count > 0);
210 z->in_use = 0;
211
212 /* go up the tree and reduce counts */
213 p = z;
214 while(p) {
215 log_assert(p->count > 0);
216 p->count --;
217 p = p->parent;
218 }
219
220 /* remove zones with zero count */
221 p = z;
222 while(p && p->count == 0) {
223 np = p->parent;
224 (void)rbtree_delete(&neg->tree, &p->node);
225 neg->use -= p->len + sizeof(*p);
226 free(p->nsec3_salt);
227 free(p->name);
228 free(p);
229 p = np;
230 }
231 }
232
neg_delete_data(struct val_neg_cache * neg,struct val_neg_data * el)233 void neg_delete_data(struct val_neg_cache* neg, struct val_neg_data* el)
234 {
235 struct val_neg_zone* z;
236 struct val_neg_data* p, *np;
237 if(!el) return;
238 z = el->zone;
239 log_assert(el->in_use);
240 log_assert(el->count > 0);
241 el->in_use = 0;
242
243 /* remove it from the lru list */
244 neg_lru_remove(neg, el);
245 log_assert(neg->first != el && neg->last != el);
246
247 /* go up the tree and reduce counts */
248 p = el;
249 while(p) {
250 log_assert(p->count > 0);
251 p->count --;
252 p = p->parent;
253 }
254
255 /* delete 0 count items from tree */
256 p = el;
257 while(p && p->count == 0) {
258 np = p->parent;
259 (void)rbtree_delete(&z->tree, &p->node);
260 neg->use -= p->len + sizeof(*p);
261 free(p->name);
262 free(p);
263 p = np;
264 }
265
266 /* check if the zone is now unused */
267 if(z->tree.count == 0) {
268 neg_delete_zone(neg, z);
269 }
270 }
271
272 /**
273 * Create more space in negative cache
274 * The oldest elements are deleted until enough space is present.
275 * Empty zones are deleted.
276 * @param neg: negative cache.
277 * @param need: how many bytes are needed.
278 */
neg_make_space(struct val_neg_cache * neg,size_t need)279 static void neg_make_space(struct val_neg_cache* neg, size_t need)
280 {
281 /* delete elements until enough space or its empty */
282 while(neg->last && neg->max < neg->use + need) {
283 neg_delete_data(neg, neg->last);
284 }
285 }
286
neg_find_zone(struct val_neg_cache * neg,uint8_t * nm,size_t len,uint16_t dclass)287 struct val_neg_zone* neg_find_zone(struct val_neg_cache* neg,
288 uint8_t* nm, size_t len, uint16_t dclass)
289 {
290 struct val_neg_zone lookfor;
291 struct val_neg_zone* result;
292 lookfor.node.key = &lookfor;
293 lookfor.name = nm;
294 lookfor.len = len;
295 lookfor.labs = dname_count_labels(lookfor.name);
296 lookfor.dclass = dclass;
297
298 result = (struct val_neg_zone*)
299 rbtree_search(&neg->tree, lookfor.node.key);
300 return result;
301 }
302
303 /**
304 * Find the given data
305 * @param zone: negative zone
306 * @param nm: what to look for.
307 * @param len: length of nm
308 * @param labs: labels in nm
309 * @return data or NULL if not found.
310 */
neg_find_data(struct val_neg_zone * zone,uint8_t * nm,size_t len,int labs)311 static struct val_neg_data* neg_find_data(struct val_neg_zone* zone,
312 uint8_t* nm, size_t len, int labs)
313 {
314 struct val_neg_data lookfor;
315 struct val_neg_data* result;
316 lookfor.node.key = &lookfor;
317 lookfor.name = nm;
318 lookfor.len = len;
319 lookfor.labs = labs;
320
321 result = (struct val_neg_data*)
322 rbtree_search(&zone->tree, lookfor.node.key);
323 return result;
324 }
325
326 /**
327 * Calculate space needed for the data and all its parents
328 * @param rep: NSEC entries.
329 * @return size.
330 */
calc_data_need(struct reply_info * rep)331 static size_t calc_data_need(struct reply_info* rep)
332 {
333 uint8_t* d;
334 size_t i, len, res = 0;
335
336 for(i=rep->an_numrrsets; i<rep->an_numrrsets+rep->ns_numrrsets; i++) {
337 if(ntohs(rep->rrsets[i]->rk.type) == LDNS_RR_TYPE_NSEC) {
338 d = rep->rrsets[i]->rk.dname;
339 len = rep->rrsets[i]->rk.dname_len;
340 res = sizeof(struct val_neg_data) + len;
341 while(!dname_is_root(d)) {
342 log_assert(len > 1); /* not root label */
343 dname_remove_label(&d, &len);
344 res += sizeof(struct val_neg_data) + len;
345 }
346 }
347 }
348 return res;
349 }
350
351 /**
352 * Calculate space needed for zone and all its parents
353 * @param d: name of zone
354 * @param len: length of name
355 * @return size.
356 */
calc_zone_need(uint8_t * d,size_t len)357 static size_t calc_zone_need(uint8_t* d, size_t len)
358 {
359 size_t res = sizeof(struct val_neg_zone) + len;
360 while(!dname_is_root(d)) {
361 log_assert(len > 1); /* not root label */
362 dname_remove_label(&d, &len);
363 res += sizeof(struct val_neg_zone) + len;
364 }
365 return res;
366 }
367
368 /**
369 * Find closest existing parent zone of the given name.
370 * @param neg: negative cache.
371 * @param nm: name to look for
372 * @param nm_len: length of nm
373 * @param labs: labelcount of nm.
374 * @param qclass: class.
375 * @return the zone or NULL if none found.
376 */
neg_closest_zone_parent(struct val_neg_cache * neg,uint8_t * nm,size_t nm_len,int labs,uint16_t qclass)377 static struct val_neg_zone* neg_closest_zone_parent(struct val_neg_cache* neg,
378 uint8_t* nm, size_t nm_len, int labs, uint16_t qclass)
379 {
380 struct val_neg_zone key;
381 struct val_neg_zone* result;
382 rbnode_type* res = NULL;
383 key.node.key = &key;
384 key.name = nm;
385 key.len = nm_len;
386 key.labs = labs;
387 key.dclass = qclass;
388 if(rbtree_find_less_equal(&neg->tree, &key, &res)) {
389 /* exact match */
390 result = (struct val_neg_zone*)res;
391 } else {
392 /* smaller element (or no element) */
393 int m;
394 result = (struct val_neg_zone*)res;
395 if(!result || result->dclass != qclass)
396 return NULL;
397 /* count number of labels matched */
398 (void)dname_lab_cmp(result->name, result->labs, key.name,
399 key.labs, &m);
400 while(result) { /* go up until qname is subdomain of stub */
401 if(result->labs <= m)
402 break;
403 result = result->parent;
404 }
405 }
406 return result;
407 }
408
409 /**
410 * Find closest existing parent data for the given name.
411 * @param zone: to look in.
412 * @param nm: name to look for
413 * @param nm_len: length of nm
414 * @param labs: labelcount of nm.
415 * @return the data or NULL if none found.
416 */
neg_closest_data_parent(struct val_neg_zone * zone,uint8_t * nm,size_t nm_len,int labs)417 static struct val_neg_data* neg_closest_data_parent(
418 struct val_neg_zone* zone, uint8_t* nm, size_t nm_len, int labs)
419 {
420 struct val_neg_data key;
421 struct val_neg_data* result;
422 rbnode_type* res = NULL;
423 key.node.key = &key;
424 key.name = nm;
425 key.len = nm_len;
426 key.labs = labs;
427 if(rbtree_find_less_equal(&zone->tree, &key, &res)) {
428 /* exact match */
429 result = (struct val_neg_data*)res;
430 } else {
431 /* smaller element (or no element) */
432 int m;
433 result = (struct val_neg_data*)res;
434 if(!result)
435 return NULL;
436 /* count number of labels matched */
437 (void)dname_lab_cmp(result->name, result->labs, key.name,
438 key.labs, &m);
439 while(result) { /* go up until qname is subdomain of stub */
440 if(result->labs <= m)
441 break;
442 result = result->parent;
443 }
444 }
445 return result;
446 }
447
448 /**
449 * Create a single zone node
450 * @param nm: name for zone (copied)
451 * @param nm_len: length of name
452 * @param labs: labels in name.
453 * @param dclass: class of zone, host order.
454 * @return new zone or NULL on failure
455 */
neg_setup_zone_node(uint8_t * nm,size_t nm_len,int labs,uint16_t dclass)456 static struct val_neg_zone* neg_setup_zone_node(
457 uint8_t* nm, size_t nm_len, int labs, uint16_t dclass)
458 {
459 struct val_neg_zone* zone =
460 (struct val_neg_zone*)calloc(1, sizeof(*zone));
461 if(!zone) {
462 return NULL;
463 }
464 zone->node.key = zone;
465 zone->name = memdup(nm, nm_len);
466 if(!zone->name) {
467 free(zone);
468 return NULL;
469 }
470 zone->len = nm_len;
471 zone->labs = labs;
472 zone->dclass = dclass;
473
474 rbtree_init(&zone->tree, &val_neg_data_compare);
475 return zone;
476 }
477
478 /**
479 * Create a linked list of parent zones, starting at longname ending on
480 * the parent (can be NULL, creates to the root).
481 * @param nm: name for lowest in chain
482 * @param nm_len: length of name
483 * @param labs: labels in name.
484 * @param dclass: class of zone.
485 * @param parent: NULL for to root, else so it fits under here.
486 * @return zone; a chain of zones and their parents up to the parent.
487 * or NULL on malloc failure
488 */
neg_zone_chain(uint8_t * nm,size_t nm_len,int labs,uint16_t dclass,struct val_neg_zone * parent)489 static struct val_neg_zone* neg_zone_chain(
490 uint8_t* nm, size_t nm_len, int labs, uint16_t dclass,
491 struct val_neg_zone* parent)
492 {
493 int i;
494 int tolabs = parent?parent->labs:0;
495 struct val_neg_zone* zone, *prev = NULL, *first = NULL;
496
497 /* create the new subtree, i is labelcount of current creation */
498 /* this creates a 'first' to z->parent=NULL list of zones */
499 for(i=labs; i!=tolabs; i--) {
500 /* create new item */
501 zone = neg_setup_zone_node(nm, nm_len, i, dclass);
502 if(!zone) {
503 /* need to delete other allocations in this routine!*/
504 struct val_neg_zone* p=first, *np;
505 while(p) {
506 np = p->parent;
507 free(p->name);
508 free(p);
509 p = np;
510 }
511 return NULL;
512 }
513 if(i == labs) {
514 first = zone;
515 } else {
516 prev->parent = zone;
517 }
518 /* prepare for next name */
519 prev = zone;
520 dname_remove_label(&nm, &nm_len);
521 }
522 return first;
523 }
524
val_neg_zone_take_inuse(struct val_neg_zone * zone)525 void val_neg_zone_take_inuse(struct val_neg_zone* zone)
526 {
527 if(!zone->in_use) {
528 struct val_neg_zone* p;
529 zone->in_use = 1;
530 /* increase usage count of all parents */
531 for(p=zone; p; p = p->parent) {
532 p->count++;
533 }
534 }
535 }
536
neg_create_zone(struct val_neg_cache * neg,uint8_t * nm,size_t nm_len,uint16_t dclass)537 struct val_neg_zone* neg_create_zone(struct val_neg_cache* neg,
538 uint8_t* nm, size_t nm_len, uint16_t dclass)
539 {
540 struct val_neg_zone* zone;
541 struct val_neg_zone* parent;
542 struct val_neg_zone* p, *np;
543 int labs = dname_count_labels(nm);
544
545 /* find closest enclosing parent zone that (still) exists */
546 parent = neg_closest_zone_parent(neg, nm, nm_len, labs, dclass);
547 if(parent && query_dname_compare(parent->name, nm) == 0)
548 return parent; /* already exists, weird */
549 /* if parent exists, it is in use */
550 log_assert(!parent || parent->count > 0);
551 zone = neg_zone_chain(nm, nm_len, labs, dclass, parent);
552 if(!zone) {
553 return NULL;
554 }
555
556 /* insert the list of zones into the tree */
557 p = zone;
558 while(p) {
559 np = p->parent;
560 /* mem use */
561 neg->use += sizeof(struct val_neg_zone) + p->len;
562 /* insert in tree */
563 (void)rbtree_insert(&neg->tree, &p->node);
564 /* last one needs proper parent pointer */
565 if(np == NULL)
566 p->parent = parent;
567 p = np;
568 }
569 return zone;
570 }
571
572 /** find zone name of message, returns the SOA record */
reply_find_soa(struct reply_info * rep)573 static struct ub_packed_rrset_key* reply_find_soa(struct reply_info* rep)
574 {
575 size_t i;
576 for(i=rep->an_numrrsets; i< rep->an_numrrsets+rep->ns_numrrsets; i++){
577 if(ntohs(rep->rrsets[i]->rk.type) == LDNS_RR_TYPE_SOA)
578 return rep->rrsets[i];
579 }
580 return NULL;
581 }
582
583 /** see if the reply has NSEC records worthy of caching */
reply_has_nsec(struct reply_info * rep)584 static int reply_has_nsec(struct reply_info* rep)
585 {
586 size_t i;
587 struct packed_rrset_data* d;
588 if(rep->security != sec_status_secure)
589 return 0;
590 for(i=rep->an_numrrsets; i< rep->an_numrrsets+rep->ns_numrrsets; i++){
591 if(ntohs(rep->rrsets[i]->rk.type) == LDNS_RR_TYPE_NSEC) {
592 d = (struct packed_rrset_data*)rep->rrsets[i]->
593 entry.data;
594 if(d->security == sec_status_secure)
595 return 1;
596 }
597 }
598 return 0;
599 }
600
601
602 /**
603 * Create single node of data element.
604 * @param nm: name (copied)
605 * @param nm_len: length of name
606 * @param labs: labels in name.
607 * @return element with name nm, or NULL malloc failure.
608 */
neg_setup_data_node(uint8_t * nm,size_t nm_len,int labs)609 static struct val_neg_data* neg_setup_data_node(
610 uint8_t* nm, size_t nm_len, int labs)
611 {
612 struct val_neg_data* el;
613 el = (struct val_neg_data*)calloc(1, sizeof(*el));
614 if(!el) {
615 return NULL;
616 }
617 el->node.key = el;
618 el->name = memdup(nm, nm_len);
619 if(!el->name) {
620 free(el);
621 return NULL;
622 }
623 el->len = nm_len;
624 el->labs = labs;
625 return el;
626 }
627
628 /**
629 * Create chain of data element and parents
630 * @param nm: name
631 * @param nm_len: length of name
632 * @param labs: labels in name.
633 * @param parent: up to where to make, if NULL up to root label.
634 * @return lowest element with name nm, or NULL malloc failure.
635 */
neg_data_chain(uint8_t * nm,size_t nm_len,int labs,struct val_neg_data * parent)636 static struct val_neg_data* neg_data_chain(
637 uint8_t* nm, size_t nm_len, int labs, struct val_neg_data* parent)
638 {
639 int i;
640 int tolabs = parent?parent->labs:0;
641 struct val_neg_data* el, *first = NULL, *prev = NULL;
642
643 /* create the new subtree, i is labelcount of current creation */
644 /* this creates a 'first' to z->parent=NULL list of zones */
645 for(i=labs; i!=tolabs; i--) {
646 /* create new item */
647 el = neg_setup_data_node(nm, nm_len, i);
648 if(!el) {
649 /* need to delete other allocations in this routine!*/
650 struct val_neg_data* p = first, *np;
651 while(p) {
652 np = p->parent;
653 free(p->name);
654 free(p);
655 p = np;
656 }
657 return NULL;
658 }
659 if(i == labs) {
660 first = el;
661 } else {
662 prev->parent = el;
663 }
664
665 /* prepare for next name */
666 prev = el;
667 dname_remove_label(&nm, &nm_len);
668 }
669 return first;
670 }
671
672 /**
673 * Remove NSEC records between start and end points.
674 * By walking the tree, the tree is sorted canonically.
675 * @param neg: negative cache.
676 * @param zone: the zone
677 * @param el: element to start walking at.
678 * @param nsec: the nsec record with the end point
679 */
wipeout(struct val_neg_cache * neg,struct val_neg_zone * zone,struct val_neg_data * el,struct ub_packed_rrset_key * nsec)680 static void wipeout(struct val_neg_cache* neg, struct val_neg_zone* zone,
681 struct val_neg_data* el, struct ub_packed_rrset_key* nsec)
682 {
683 struct packed_rrset_data* d = (struct packed_rrset_data*)nsec->
684 entry.data;
685 uint8_t* end;
686 size_t end_len;
687 int end_labs, m;
688 rbnode_type* walk, *next;
689 struct val_neg_data* cur;
690 uint8_t buf[257];
691 /* get endpoint */
692 if(!d || d->count == 0 || d->rr_len[0] < 2+1)
693 return;
694 if(ntohs(nsec->rk.type) == LDNS_RR_TYPE_NSEC) {
695 end = d->rr_data[0]+2;
696 end_len = dname_valid(end, d->rr_len[0]-2);
697 end_labs = dname_count_labels(end);
698 } else {
699 /* NSEC3 */
700 if(!nsec3_get_nextowner_b32(nsec, 0, buf, sizeof(buf)))
701 return;
702 end = buf;
703 end_labs = dname_count_size_labels(end, &end_len);
704 }
705
706 /* sanity check, both owner and end must be below the zone apex */
707 if(!dname_subdomain_c(el->name, zone->name) ||
708 !dname_subdomain_c(end, zone->name))
709 return;
710
711 /* detect end of zone NSEC ; wipe until the end of zone */
712 if(query_dname_compare(end, zone->name) == 0) {
713 end = NULL;
714 }
715
716 walk = rbtree_next(&el->node);
717 while(walk && walk != RBTREE_NULL) {
718 cur = (struct val_neg_data*)walk;
719 /* sanity check: must be larger than start */
720 if(dname_canon_lab_cmp(cur->name, cur->labs,
721 el->name, el->labs, &m) <= 0) {
722 /* r == 0 skip original record. */
723 /* r < 0 too small! */
724 walk = rbtree_next(walk);
725 continue;
726 }
727 /* stop at endpoint, also data at empty nonterminals must be
728 * removed (no NSECs there) so everything between
729 * start and end */
730 if(end && dname_canon_lab_cmp(cur->name, cur->labs,
731 end, end_labs, &m) >= 0) {
732 break;
733 }
734 /* this element has to be deleted, but we cannot do it
735 * now, because we are walking the tree still ... */
736 /* get the next element: */
737 next = rbtree_next(walk);
738 /* now delete the original element, this may trigger
739 * rbtree rebalances, but really, the next element is
740 * the one we need.
741 * But it may trigger delete of other data and the
742 * entire zone. However, if that happens, this is done
743 * by deleting the *parents* of the element for deletion,
744 * and maybe also the entire zone if it is empty.
745 * But parents are smaller in canonical compare, thus,
746 * if a larger element exists, then it is not a parent,
747 * it cannot get deleted, the zone cannot get empty.
748 * If the next==NULL, then zone can be empty. */
749 if(cur->in_use)
750 neg_delete_data(neg, cur);
751 walk = next;
752 }
753 }
754
neg_insert_data(struct val_neg_cache * neg,struct val_neg_zone * zone,struct ub_packed_rrset_key * nsec)755 void neg_insert_data(struct val_neg_cache* neg,
756 struct val_neg_zone* zone, struct ub_packed_rrset_key* nsec)
757 {
758 struct packed_rrset_data* d;
759 struct val_neg_data* parent;
760 struct val_neg_data* el;
761 uint8_t* nm = nsec->rk.dname;
762 size_t nm_len = nsec->rk.dname_len;
763 int labs = dname_count_labels(nsec->rk.dname);
764
765 d = (struct packed_rrset_data*)nsec->entry.data;
766 if( !(d->security == sec_status_secure ||
767 (d->security == sec_status_unchecked && d->rrsig_count > 0)))
768 return;
769 log_nametypeclass(VERB_ALGO, "negcache rr",
770 nsec->rk.dname, ntohs(nsec->rk.type),
771 ntohs(nsec->rk.rrset_class));
772
773 /* find closest enclosing parent data that (still) exists */
774 parent = neg_closest_data_parent(zone, nm, nm_len, labs);
775 if(parent && query_dname_compare(parent->name, nm) == 0) {
776 /* perfect match already exists */
777 log_assert(parent->count > 0);
778 el = parent;
779 } else {
780 struct val_neg_data* p, *np;
781
782 /* create subtree for perfect match */
783 /* if parent exists, it is in use */
784 log_assert(!parent || parent->count > 0);
785
786 el = neg_data_chain(nm, nm_len, labs, parent);
787 if(!el) {
788 log_err("out of memory inserting NSEC negative cache");
789 return;
790 }
791 el->in_use = 0; /* set on below */
792
793 /* insert the list of zones into the tree */
794 p = el;
795 while(p) {
796 np = p->parent;
797 /* mem use */
798 neg->use += sizeof(struct val_neg_data) + p->len;
799 /* insert in tree */
800 p->zone = zone;
801 (void)rbtree_insert(&zone->tree, &p->node);
802 /* last one needs proper parent pointer */
803 if(np == NULL)
804 p->parent = parent;
805 p = np;
806 }
807 }
808
809 if(!el->in_use) {
810 struct val_neg_data* p;
811
812 el->in_use = 1;
813 /* increase usage count of all parents */
814 for(p=el; p; p = p->parent) {
815 p->count++;
816 }
817
818 neg_lru_front(neg, el);
819 } else {
820 /* in use, bring to front, lru */
821 neg_lru_touch(neg, el);
822 }
823
824 /* if nsec3 store last used parameters */
825 if(ntohs(nsec->rk.type) == LDNS_RR_TYPE_NSEC3) {
826 int h;
827 uint8_t* s;
828 size_t slen, it;
829 if(nsec3_get_params(nsec, 0, &h, &it, &s, &slen) &&
830 it <= neg->nsec3_max_iter &&
831 (h != zone->nsec3_hash || it != zone->nsec3_iter ||
832 slen != zone->nsec3_saltlen ||
833 (slen != 0 && zone->nsec3_salt && s
834 && memcmp(zone->nsec3_salt, s, slen) != 0))) {
835
836 if(slen > MAX_SALT_LENGTH) {
837 /* RFC 9276 s3.1: operators SHOULD NOT use a salt; large
838 * salts inflate per-hash block count. Decline to cache. */
839 return;
840 } else if(slen > 0) {
841 uint8_t* sa = memdup(s, slen);
842 if(sa) {
843 free(zone->nsec3_salt);
844 zone->nsec3_salt = sa;
845 zone->nsec3_saltlen = slen;
846 zone->nsec3_iter = it;
847 zone->nsec3_hash = h;
848 }
849 } else {
850 free(zone->nsec3_salt);
851 zone->nsec3_salt = NULL;
852 zone->nsec3_saltlen = 0;
853 zone->nsec3_iter = it;
854 zone->nsec3_hash = h;
855 }
856 }
857 }
858
859 /* wipe out the cache items between NSEC start and end */
860 wipeout(neg, zone, el, nsec);
861 }
862
863 /** see if the reply has signed NSEC records and return the signer */
reply_nsec_signer(struct reply_info * rep,size_t * signer_len,uint16_t * dclass)864 static uint8_t* reply_nsec_signer(struct reply_info* rep, size_t* signer_len,
865 uint16_t* dclass)
866 {
867 size_t i;
868 struct packed_rrset_data* d;
869 uint8_t* s;
870 for(i=rep->an_numrrsets; i< rep->an_numrrsets+rep->ns_numrrsets; i++){
871 if(ntohs(rep->rrsets[i]->rk.type) == LDNS_RR_TYPE_NSEC ||
872 ntohs(rep->rrsets[i]->rk.type) == LDNS_RR_TYPE_NSEC3) {
873 d = (struct packed_rrset_data*)rep->rrsets[i]->
874 entry.data;
875 /* return first signer name of first NSEC */
876 if(d->rrsig_count != 0) {
877 val_find_rrset_signer(rep->rrsets[i],
878 &s, signer_len);
879 if(s && *signer_len) {
880 *dclass = ntohs(rep->rrsets[i]->
881 rk.rrset_class);
882 return s;
883 }
884 }
885 }
886 }
887 return 0;
888 }
889
val_neg_addreply(struct val_neg_cache * neg,struct reply_info * rep)890 void val_neg_addreply(struct val_neg_cache* neg, struct reply_info* rep)
891 {
892 size_t i, need;
893 struct ub_packed_rrset_key* soa;
894 uint8_t* dname = NULL;
895 size_t dname_len;
896 uint16_t rrset_class;
897 struct val_neg_zone* zone;
898 /* see if secure nsecs inside */
899 if(!reply_has_nsec(rep))
900 return;
901 /* find the zone name in message */
902 if((soa = reply_find_soa(rep))) {
903 dname = soa->rk.dname;
904 dname_len = soa->rk.dname_len;
905 rrset_class = ntohs(soa->rk.rrset_class);
906 }
907 else {
908 /* No SOA in positive (wildcard) answer. Use signer from the
909 * validated answer RRsets' signature. */
910 if(!(dname = reply_nsec_signer(rep, &dname_len, &rrset_class)))
911 return;
912 }
913
914 log_nametypeclass(VERB_ALGO, "negcache insert for zone",
915 dname, LDNS_RR_TYPE_SOA, rrset_class);
916
917 /* ask for enough space to store all of it */
918 need = calc_data_need(rep) +
919 calc_zone_need(dname, dname_len);
920 lock_basic_lock(&neg->lock);
921 neg_make_space(neg, need);
922
923 /* find or create the zone entry */
924 zone = neg_find_zone(neg, dname, dname_len, rrset_class);
925 if(!zone) {
926 if(!(zone = neg_create_zone(neg, dname, dname_len,
927 rrset_class))) {
928 lock_basic_unlock(&neg->lock);
929 log_err("out of memory adding negative zone");
930 return;
931 }
932 }
933 val_neg_zone_take_inuse(zone);
934
935 /* insert the NSECs */
936 for(i=rep->an_numrrsets; i< rep->an_numrrsets+rep->ns_numrrsets; i++){
937 if(ntohs(rep->rrsets[i]->rk.type) != LDNS_RR_TYPE_NSEC)
938 continue;
939 if(!dname_subdomain_c(rep->rrsets[i]->rk.dname,
940 zone->name)) continue;
941 if(ntohs(rep->rrsets[i]->rk.type) == LDNS_RR_TYPE_NSEC &&
942 !nsec_nextowner_subdomain(rep->rrsets[i], zone->name)) {
943 continue; /* nextowner not in zone */
944 }
945 /* insert NSEC into this zone's tree */
946 neg_insert_data(neg, zone, rep->rrsets[i]);
947 }
948 if(zone->tree.count == 0) {
949 /* remove empty zone if inserts failed */
950 neg_delete_zone(neg, zone);
951 }
952 lock_basic_unlock(&neg->lock);
953 }
954
955 /**
956 * Lookup closest data record. For NSEC denial.
957 * @param zone: zone to look in
958 * @param qname: name to look for.
959 * @param len: length of name
960 * @param labs: labels in name
961 * @param data: data element, exact or smaller or NULL
962 * @return true if exact match.
963 */
neg_closest_data(struct val_neg_zone * zone,uint8_t * qname,size_t len,int labs,struct val_neg_data ** data)964 static int neg_closest_data(struct val_neg_zone* zone,
965 uint8_t* qname, size_t len, int labs, struct val_neg_data** data)
966 {
967 struct val_neg_data key;
968 rbnode_type* r;
969 key.node.key = &key;
970 key.name = qname;
971 key.len = len;
972 key.labs = labs;
973 if(rbtree_find_less_equal(&zone->tree, &key, &r)) {
974 /* exact match */
975 *data = (struct val_neg_data*)r;
976 return 1;
977 } else {
978 /* smaller match */
979 *data = (struct val_neg_data*)r;
980 return 0;
981 }
982 }
983
val_neg_addreferral(struct val_neg_cache * neg,struct reply_info * rep,uint8_t * zone_name)984 void val_neg_addreferral(struct val_neg_cache* neg, struct reply_info* rep,
985 uint8_t* zone_name)
986 {
987 size_t i, need;
988 uint8_t* signer;
989 size_t signer_len;
990 uint16_t dclass;
991 struct val_neg_zone* zone;
992 /* no SOA in this message, find RRSIG over NSEC's signer name.
993 * note the NSEC records are maybe not validated yet */
994 signer = reply_nsec_signer(rep, &signer_len, &dclass);
995 if(!signer)
996 return;
997 if(!dname_subdomain_c(signer, zone_name)) {
998 /* the signer is not in the bailiwick, throw it out */
999 return;
1000 }
1001
1002 log_nametypeclass(VERB_ALGO, "negcache insert referral ",
1003 signer, LDNS_RR_TYPE_NS, dclass);
1004
1005 /* ask for enough space to store all of it */
1006 need = calc_data_need(rep) + calc_zone_need(signer, signer_len);
1007 lock_basic_lock(&neg->lock);
1008 neg_make_space(neg, need);
1009
1010 /* find or create the zone entry */
1011 zone = neg_find_zone(neg, signer, signer_len, dclass);
1012 if(!zone) {
1013 if(!(zone = neg_create_zone(neg, signer, signer_len,
1014 dclass))) {
1015 lock_basic_unlock(&neg->lock);
1016 log_err("out of memory adding negative zone");
1017 return;
1018 }
1019 }
1020 val_neg_zone_take_inuse(zone);
1021
1022 /* insert the NSECs */
1023 for(i=rep->an_numrrsets; i< rep->an_numrrsets+rep->ns_numrrsets; i++){
1024 if(ntohs(rep->rrsets[i]->rk.type) != LDNS_RR_TYPE_NSEC &&
1025 ntohs(rep->rrsets[i]->rk.type) != LDNS_RR_TYPE_NSEC3)
1026 continue;
1027 if(!dname_subdomain_c(rep->rrsets[i]->rk.dname,
1028 zone->name)) continue;
1029 if(ntohs(rep->rrsets[i]->rk.type) == LDNS_RR_TYPE_NSEC &&
1030 !nsec_nextowner_subdomain(rep->rrsets[i], zone->name)) {
1031 continue; /* nextowner not in zone */
1032 }
1033 /* insert NSEC into this zone's tree */
1034 neg_insert_data(neg, zone, rep->rrsets[i]);
1035 }
1036 if(zone->tree.count == 0) {
1037 /* remove empty zone if inserts failed */
1038 neg_delete_zone(neg, zone);
1039 }
1040 lock_basic_unlock(&neg->lock);
1041 }
1042
1043 /**
1044 * Check that an NSEC3 rrset does not have a type set.
1045 * None of the nsec3s in a hash-collision are allowed to have the type.
1046 * (since we do not know which one is the nsec3 looked at, flags, ..., we
1047 * ignore the cached item and let it bypass negative caching).
1048 * @param k: the nsec3 rrset to check.
1049 * @param t: type to check
1050 * @return true if no RRs have the type.
1051 */
nsec3_no_type(struct ub_packed_rrset_key * k,uint16_t t)1052 static int nsec3_no_type(struct ub_packed_rrset_key* k, uint16_t t)
1053 {
1054 int count = (int)((struct packed_rrset_data*)k->entry.data)->count;
1055 int i;
1056 for(i=0; i<count; i++)
1057 if(nsec3_has_type(k, i, t))
1058 return 0;
1059 return 1;
1060 }
1061
1062 /**
1063 * See if rrset exists in rrset cache.
1064 * If it does, the bit is checked, and if not expired, it is returned
1065 * allocated in region.
1066 * @param rrset_cache: rrset cache
1067 * @param qname: to lookup rrset name
1068 * @param qname_len: length of qname.
1069 * @param qtype: type of rrset to lookup, host order
1070 * @param qclass: class of rrset to lookup, host order
1071 * @param flags: flags for rrset to lookup
1072 * @param region: where to alloc result
1073 * @param checkbit: if true, a bit in the nsec typemap is checked for absence.
1074 * @param checktype: which bit to check
1075 * @param now: to check ttl against
1076 * @return rrset or NULL
1077 */
1078 static struct ub_packed_rrset_key*
grab_nsec(struct rrset_cache * rrset_cache,uint8_t * qname,size_t qname_len,uint16_t qtype,uint16_t qclass,uint32_t flags,struct regional * region,int checkbit,uint16_t checktype,time_t now)1079 grab_nsec(struct rrset_cache* rrset_cache, uint8_t* qname, size_t qname_len,
1080 uint16_t qtype, uint16_t qclass, uint32_t flags,
1081 struct regional* region, int checkbit, uint16_t checktype,
1082 time_t now)
1083 {
1084 struct ub_packed_rrset_key* r, *k = rrset_cache_lookup(rrset_cache,
1085 qname, qname_len, qtype, qclass, flags, now, 0);
1086 struct packed_rrset_data* d;
1087 if(!k) return NULL;
1088 d = k->entry.data;
1089 /* only secure or unchecked records that have signatures. */
1090 if( ! ( d->security == sec_status_secure ||
1091 (d->security == sec_status_unchecked &&
1092 d->rrsig_count > 0) ) ) {
1093 lock_rw_unlock(&k->entry.lock);
1094 return NULL;
1095 }
1096 /* check if checktype is absent */
1097 if(checkbit && (
1098 (qtype == LDNS_RR_TYPE_NSEC && nsec_has_type(k, checktype)) ||
1099 (qtype == LDNS_RR_TYPE_NSEC3 && !nsec3_no_type(k, checktype))
1100 )) {
1101 lock_rw_unlock(&k->entry.lock);
1102 return NULL;
1103 }
1104 /* looks OK! copy to region and return it */
1105 r = packed_rrset_copy_region(k, region, now);
1106 /* if it failed, we return the NULL */
1107 lock_rw_unlock(&k->entry.lock);
1108 return r;
1109 }
1110
1111 /**
1112 * Get best NSEC record for qname. Might be matching, covering or totally
1113 * useless.
1114 * @param neg_cache: neg cache
1115 * @param qname: to lookup rrset name
1116 * @param qname_len: length of qname.
1117 * @param qclass: class of rrset to lookup, host order
1118 * @param rrset_cache: rrset cache
1119 * @param now: to check ttl against
1120 * @param region: where to alloc result
1121 * @param topname: do not look higher than this name, so that the
1122 * result cannot be taken from a zone above the current trust anchor.
1123 * @return rrset or NULL
1124 */
1125 static struct ub_packed_rrset_key*
neg_find_nsec(struct val_neg_cache * neg_cache,uint8_t * qname,size_t qname_len,uint16_t qclass,struct rrset_cache * rrset_cache,time_t now,struct regional * region,uint8_t * topname)1126 neg_find_nsec(struct val_neg_cache* neg_cache, uint8_t* qname, size_t qname_len,
1127 uint16_t qclass, struct rrset_cache* rrset_cache, time_t now,
1128 struct regional* region, uint8_t* topname)
1129 {
1130 int labs;
1131 uint32_t flags;
1132 struct val_neg_zone* zone;
1133 struct val_neg_data* data;
1134 struct ub_packed_rrset_key* nsec;
1135
1136 labs = dname_count_labels(qname);
1137 lock_basic_lock(&neg_cache->lock);
1138 zone = neg_closest_zone_parent(neg_cache, qname, qname_len, labs,
1139 qclass);
1140 while(zone && !zone->in_use)
1141 zone = zone->parent;
1142 if(!zone) {
1143 lock_basic_unlock(&neg_cache->lock);
1144 return NULL;
1145 }
1146 if(topname && !dname_subdomain_c(zone->name, topname)) {
1147 /* Reject NSEC not within trust anchor's bailiwick */
1148 lock_basic_unlock(&neg_cache->lock);
1149 return NULL;
1150 }
1151
1152 /* NSEC only for now */
1153 if(zone->nsec3_hash) {
1154 lock_basic_unlock(&neg_cache->lock);
1155 return NULL;
1156 }
1157
1158 /* ignore return value, don't care if it is an exact or smaller match */
1159 (void)neg_closest_data(zone, qname, qname_len, labs, &data);
1160 if(!data) {
1161 lock_basic_unlock(&neg_cache->lock);
1162 return NULL;
1163 }
1164
1165 /* ENT nodes are not in use, try the previous node. If the previous node
1166 * is not in use, we don't have an useful NSEC and give up. */
1167 if(!data->in_use) {
1168 data = (struct val_neg_data*)rbtree_previous((rbnode_type*)data);
1169 if((rbnode_type*)data == RBTREE_NULL || !data->in_use) {
1170 lock_basic_unlock(&neg_cache->lock);
1171 return NULL;
1172 }
1173 }
1174
1175 flags = 0;
1176 if(query_dname_compare(data->name, zone->name) == 0)
1177 flags = PACKED_RRSET_NSEC_AT_APEX;
1178
1179 nsec = grab_nsec(rrset_cache, data->name, data->len, LDNS_RR_TYPE_NSEC,
1180 zone->dclass, flags, region, 0, 0, now);
1181 lock_basic_unlock(&neg_cache->lock);
1182 return nsec;
1183 }
1184
1185 /** find nsec3 closest encloser in neg cache */
1186 static struct val_neg_data*
neg_find_nsec3_ce(struct val_neg_zone * zone,uint8_t * qname,size_t qname_len,int qlabs,sldns_buffer * buf,uint8_t * hashnc,size_t * nclen)1187 neg_find_nsec3_ce(struct val_neg_zone* zone, uint8_t* qname, size_t qname_len,
1188 int qlabs, sldns_buffer* buf, uint8_t* hashnc, size_t* nclen)
1189 {
1190 struct val_neg_data* data;
1191 uint8_t hashce[NSEC3_SHA_LEN];
1192 uint8_t b32[257];
1193 size_t celen, b32len;
1194 int hashmax = MAX_NSEC3_CALCULATIONS;
1195 if(qlabs > hashmax) {
1196 /* strip leading labels so the walk costs at most
1197 * MAX_NSEC3_CALCULATIONS hashes, mirroring val_nsec3.c */
1198 while(qlabs > hashmax) {
1199 dname_remove_label(&qname, &qname_len);
1200 qlabs--;
1201 }
1202 }
1203
1204 *nclen = 0;
1205 while(qlabs > 0) {
1206 /* hash */
1207 if(!(celen=nsec3_get_hashed(buf, qname, qname_len,
1208 zone->nsec3_hash, zone->nsec3_iter, zone->nsec3_salt,
1209 zone->nsec3_saltlen, hashce, sizeof(hashce))))
1210 return NULL;
1211 if(!(b32len=nsec3_hash_to_b32(hashce, celen, zone->name,
1212 zone->len, b32, sizeof(b32))))
1213 return NULL;
1214
1215 /* lookup (exact match only) */
1216 data = neg_find_data(zone, b32, b32len, zone->labs+1);
1217 if(data && data->in_use) {
1218 /* found ce match! */
1219 return data;
1220 }
1221
1222 *nclen = celen;
1223 memmove(hashnc, hashce, celen);
1224 dname_remove_label(&qname, &qname_len);
1225 qlabs --;
1226 }
1227 return NULL;
1228 }
1229
1230 /** check nsec3 parameters on nsec3 rrset with current zone values */
1231 static int
neg_params_ok(struct val_neg_zone * zone,struct ub_packed_rrset_key * rrset)1232 neg_params_ok(struct val_neg_zone* zone, struct ub_packed_rrset_key* rrset)
1233 {
1234 int h;
1235 uint8_t* s;
1236 size_t slen, it;
1237 if(!nsec3_get_params(rrset, 0, &h, &it, &s, &slen))
1238 return 0;
1239 return (h == zone->nsec3_hash && it == zone->nsec3_iter &&
1240 slen == zone->nsec3_saltlen &&
1241 (slen == 0 || (slen != 0 && zone->nsec3_salt && s
1242 && memcmp(zone->nsec3_salt, s, slen) == 0)));
1243 }
1244
1245 /** get next closer for nsec3 proof */
1246 static struct ub_packed_rrset_key*
neg_nsec3_getnc(struct val_neg_zone * zone,uint8_t * hashnc,size_t nclen,struct rrset_cache * rrset_cache,struct regional * region,time_t now,uint8_t * b32,size_t maxb32)1247 neg_nsec3_getnc(struct val_neg_zone* zone, uint8_t* hashnc, size_t nclen,
1248 struct rrset_cache* rrset_cache, struct regional* region,
1249 time_t now, uint8_t* b32, size_t maxb32)
1250 {
1251 struct ub_packed_rrset_key* nc_rrset;
1252 struct val_neg_data* data;
1253 size_t b32len;
1254
1255 if(!(b32len=nsec3_hash_to_b32(hashnc, nclen, zone->name,
1256 zone->len, b32, maxb32)))
1257 return NULL;
1258 (void)neg_closest_data(zone, b32, b32len, zone->labs+1, &data);
1259 if(!data && zone->tree.count != 0) {
1260 /* could be before the first entry ; return the last
1261 * entry (possibly the rollover nsec3 at end) */
1262 data = (struct val_neg_data*)rbtree_last(&zone->tree);
1263 }
1264 while(data && !data->in_use)
1265 data = data->parent;
1266 if(!data)
1267 return NULL;
1268 /* got a data element in tree, grab it */
1269 nc_rrset = grab_nsec(rrset_cache, data->name, data->len,
1270 LDNS_RR_TYPE_NSEC3, zone->dclass, 0, region, 0, 0, now);
1271 if(!nc_rrset)
1272 return NULL;
1273 if(!neg_params_ok(zone, nc_rrset))
1274 return NULL;
1275 return nc_rrset;
1276 }
1277
1278 /** neg cache nsec3 proof procedure*/
1279 static struct dns_msg*
neg_nsec3_proof_ds(struct val_neg_zone * zone,uint8_t * qname,size_t qname_len,int qlabs,sldns_buffer * buf,struct rrset_cache * rrset_cache,struct regional * region,time_t now,uint8_t * topname)1280 neg_nsec3_proof_ds(struct val_neg_zone* zone, uint8_t* qname, size_t qname_len,
1281 int qlabs, sldns_buffer* buf, struct rrset_cache* rrset_cache,
1282 struct regional* region, time_t now, uint8_t* topname)
1283 {
1284 struct dns_msg* msg;
1285 struct val_neg_data* data;
1286 uint8_t hashnc[NSEC3_SHA_LEN];
1287 size_t nclen;
1288 struct ub_packed_rrset_key* ce_rrset, *nc_rrset;
1289 struct nsec3_cached_hash c;
1290 uint8_t nc_b32[257];
1291
1292 /* for NSEC3 ; determine the closest encloser for which we
1293 * can find an exact match. Remember the hashed lower name,
1294 * since that is the one we need a closest match for.
1295 * If we find a match straight away, then it becomes NODATA.
1296 * Otherwise, NXDOMAIN or if OPTOUT, an insecure delegation.
1297 * Also check that parameters are the same on closest encloser
1298 * and on closest match.
1299 */
1300 if(!zone->nsec3_hash)
1301 return NULL; /* not nsec3 zone */
1302
1303 if(!topname && qlabs > zone->labs + 1)
1304 return NULL; /* iterator caller; opt-out proof would be discarded
1305 * at the !topname check below anyway.
1306 * The qlabs check allows the exact-match for
1307 * the one-label-below-zone case. */
1308
1309 if(!(data=neg_find_nsec3_ce(zone, qname, qname_len, qlabs, buf,
1310 hashnc, &nclen))) {
1311 return NULL;
1312 }
1313
1314 /* grab the ce rrset */
1315 ce_rrset = grab_nsec(rrset_cache, data->name, data->len,
1316 LDNS_RR_TYPE_NSEC3, zone->dclass, 0, region, 1,
1317 LDNS_RR_TYPE_DS, now);
1318 if(!ce_rrset)
1319 return NULL;
1320 if(!neg_params_ok(zone, ce_rrset))
1321 return NULL;
1322
1323 if(nclen == 0) {
1324 /* exact match, just check the type bits */
1325 /* need: -SOA, -DS, +NS */
1326 if(nsec3_has_type(ce_rrset, 0, LDNS_RR_TYPE_SOA) ||
1327 nsec3_has_type(ce_rrset, 0, LDNS_RR_TYPE_DS) ||
1328 !nsec3_has_type(ce_rrset, 0, LDNS_RR_TYPE_NS))
1329 return NULL;
1330 if(!(msg = dns_msg_create(qname, qname_len,
1331 LDNS_RR_TYPE_DS, zone->dclass, region, 2))) /* ce + soa */
1332 return NULL;
1333 /* The cache response means recursion is available. */
1334 msg->rep->flags |= BIT_RA;
1335 /* TTL reduced in grab_nsec */
1336 if(!dns_msg_authadd(msg, region, ce_rrset, 0))
1337 return NULL;
1338 return msg;
1339 }
1340
1341 /* optout is not allowed without knowing the trust-anchor in use,
1342 * otherwise the optout could spoof away that anchor */
1343 if(!topname)
1344 return NULL;
1345
1346 /* if there is no exact match, it must be in an optout span
1347 * (an existing DS implies an NSEC3 must exist) */
1348 nc_rrset = neg_nsec3_getnc(zone, hashnc, nclen, rrset_cache,
1349 region, now, nc_b32, sizeof(nc_b32));
1350 if(!nc_rrset)
1351 return NULL;
1352 if(!neg_params_ok(zone, nc_rrset))
1353 return NULL;
1354 if(!nsec3_has_optout(nc_rrset, 0))
1355 return NULL;
1356 c.hash = hashnc;
1357 c.hash_len = nclen;
1358 c.b32 = nc_b32+1;
1359 c.b32_len = (size_t)nc_b32[0];
1360 if(nsec3_covers(zone->name, &c, nc_rrset, 0, buf)) {
1361 /* nc_rrset covers the next closer name.
1362 * ce_rrset equals a closer encloser.
1363 * nc_rrset is optout.
1364 * No need to check wildcard for type DS */
1365 /* capacity=3: ce + nc + soa(if needed) */
1366 if(!(msg = dns_msg_create(qname, qname_len,
1367 LDNS_RR_TYPE_DS, zone->dclass, region, 3)))
1368 return NULL;
1369 /* The cache response means recursion is available. */
1370 msg->rep->flags |= BIT_RA;
1371 /* now=0 because TTL was reduced in grab_nsec */
1372 if(!dns_msg_authadd(msg, region, ce_rrset, 0))
1373 return NULL;
1374 if(!dns_msg_authadd(msg, region, nc_rrset, 0))
1375 return NULL;
1376 return msg;
1377 }
1378 return NULL;
1379 }
1380
1381 /**
1382 * Add SOA record for external responses.
1383 * @param rrset_cache: to look into.
1384 * @param now: current time.
1385 * @param region: where to perform the allocation
1386 * @param msg: current msg with NSEC.
1387 * @param zone: val_neg_zone if we have one.
1388 * @return false on lookup or alloc failure.
1389 */
add_soa(struct rrset_cache * rrset_cache,time_t now,struct regional * region,struct dns_msg * msg,struct val_neg_zone * zone)1390 static int add_soa(struct rrset_cache* rrset_cache, time_t now,
1391 struct regional* region, struct dns_msg* msg, struct val_neg_zone* zone)
1392 {
1393 struct ub_packed_rrset_key* soa;
1394 uint8_t* nm;
1395 size_t nmlen;
1396 uint16_t dclass;
1397 if(zone) {
1398 nm = zone->name;
1399 nmlen = zone->len;
1400 dclass = zone->dclass;
1401 } else {
1402 /* Assumes the signer is the zone SOA to add */
1403 nm = reply_nsec_signer(msg->rep, &nmlen, &dclass);
1404 if(!nm)
1405 return 0;
1406 }
1407 soa = rrset_cache_lookup(rrset_cache, nm, nmlen, LDNS_RR_TYPE_SOA,
1408 dclass, PACKED_RRSET_SOA_NEG, now, 0);
1409 if(!soa)
1410 return 0;
1411 if(!dns_msg_authadd(msg, region, soa, now)) {
1412 lock_rw_unlock(&soa->entry.lock);
1413 return 0;
1414 }
1415 lock_rw_unlock(&soa->entry.lock);
1416 return 1;
1417 }
1418
1419 struct dns_msg*
val_neg_getmsg(struct val_neg_cache * neg,struct query_info * qinfo,struct regional * region,struct rrset_cache * rrset_cache,sldns_buffer * buf,time_t now,int addsoa,uint8_t * topname,struct config_file * cfg)1420 val_neg_getmsg(struct val_neg_cache* neg, struct query_info* qinfo,
1421 struct regional* region, struct rrset_cache* rrset_cache,
1422 sldns_buffer* buf, time_t now, int addsoa, uint8_t* topname,
1423 struct config_file* cfg)
1424 {
1425 struct dns_msg* msg;
1426 struct ub_packed_rrset_key* nsec; /* qname matching/covering nsec */
1427 struct ub_packed_rrset_key* wcrr; /* wildcard record or nsec */
1428 uint8_t* nodata_wc = NULL;
1429 uint8_t* ce = NULL;
1430 size_t ce_len;
1431 uint8_t wc_ce[LDNS_MAX_DOMAINLEN+3];
1432 struct query_info wc_qinfo;
1433 struct ub_packed_rrset_key* cache_wc;
1434 struct packed_rrset_data* wcrr_data;
1435 int rcode = LDNS_RCODE_NOERROR;
1436 uint8_t* zname;
1437 size_t zname_len;
1438 int zname_labs;
1439 struct val_neg_zone* zone;
1440
1441 /* only for DS queries when aggressive use of NSEC is disabled */
1442 if(qinfo->qtype != LDNS_RR_TYPE_DS && !cfg->aggressive_nsec)
1443 return NULL;
1444 log_assert(!topname || dname_subdomain_c(qinfo->qname, topname));
1445
1446 /* Get best available NSEC for qname */
1447 nsec = neg_find_nsec(neg, qinfo->qname, qinfo->qname_len, qinfo->qclass,
1448 rrset_cache, now, region, topname);
1449
1450 /* Matching NSEC, use to generate No Data answer. Not creating answers
1451 * yet for No Data proven using wildcard. */
1452 if(nsec && nsec_proves_nodata(nsec, qinfo, &nodata_wc) && !nodata_wc) {
1453 /* do not create nodata answers for qtype ANY, it is a query
1454 * type, not an rrtype to disprove. Nameerrors are useful for
1455 * qtype ANY, in the else branch. */
1456 if(qinfo->qtype == LDNS_RR_TYPE_ANY)
1457 return NULL;
1458 if(!(msg = dns_msg_create(qinfo->qname, qinfo->qname_len,
1459 qinfo->qtype, qinfo->qclass, region, 2)))
1460 return NULL;
1461 /* The cache response means recursion is available. */
1462 msg->rep->flags |= BIT_RA;
1463 if(!dns_msg_authadd(msg, region, nsec, 0))
1464 return NULL;
1465 if(addsoa && !add_soa(rrset_cache, now, region, msg, NULL))
1466 return NULL;
1467
1468 lock_basic_lock(&neg->lock);
1469 neg->num_neg_cache_noerror++;
1470 lock_basic_unlock(&neg->lock);
1471 return msg;
1472 } else if(nsec && val_nsec_proves_name_error(nsec, qinfo->qname)) {
1473 if(!(msg = dns_msg_create(qinfo->qname, qinfo->qname_len,
1474 qinfo->qtype, qinfo->qclass, region, 3)))
1475 return NULL;
1476 /* The cache response means recursion is available. */
1477 msg->rep->flags |= BIT_RA;
1478 if(!(ce = nsec_closest_encloser(qinfo->qname, nsec)))
1479 return NULL;
1480 dname_count_size_labels(ce, &ce_len);
1481
1482 /* No extra extra NSEC required if both nameerror qname and
1483 * nodata *.ce. are proven already. */
1484 if(!nodata_wc || query_dname_compare(nodata_wc, ce) != 0) {
1485 /* Qname proven non existing, get wildcard record for
1486 * QTYPE or NSEC covering or matching wildcard. */
1487
1488 /* Num labels in ce is always smaller than in qname,
1489 * therefore adding the wildcard label cannot overflow
1490 * buffer. */
1491 wc_ce[0] = 1;
1492 wc_ce[1] = (uint8_t)'*';
1493 memmove(wc_ce+2, ce, ce_len);
1494 wc_qinfo.qname = wc_ce;
1495 wc_qinfo.qname_len = ce_len + 2;
1496 wc_qinfo.qtype = qinfo->qtype;
1497
1498
1499 if((cache_wc = rrset_cache_lookup(rrset_cache, wc_qinfo.qname,
1500 wc_qinfo.qname_len, wc_qinfo.qtype,
1501 qinfo->qclass, 0/*flags*/, now, 0/*read only*/))) {
1502 /* Synthesize wildcard answer */
1503 wcrr_data = (struct packed_rrset_data*)cache_wc->entry.data;
1504 if(!(wcrr_data->security == sec_status_secure ||
1505 (wcrr_data->security == sec_status_unchecked &&
1506 wcrr_data->rrsig_count > 0))) {
1507 lock_rw_unlock(&cache_wc->entry.lock);
1508 return NULL;
1509 }
1510 if(!(wcrr = packed_rrset_copy_region(cache_wc,
1511 region, now))) {
1512 lock_rw_unlock(&cache_wc->entry.lock);
1513 return NULL;
1514 };
1515 lock_rw_unlock(&cache_wc->entry.lock);
1516 wcrr->rk.dname = qinfo->qname;
1517 wcrr->rk.dname_len = qinfo->qname_len;
1518 if(!dns_msg_ansadd(msg, region, wcrr, 0))
1519 return NULL;
1520 /* No SOA needed for wildcard synthesised
1521 * answer. */
1522 addsoa = 0;
1523 } else {
1524 /* Get wildcard NSEC for possible non existence
1525 * proof */
1526 if(!(wcrr = neg_find_nsec(neg, wc_qinfo.qname,
1527 wc_qinfo.qname_len, qinfo->qclass,
1528 rrset_cache, now, region, topname)))
1529 return NULL;
1530
1531 nodata_wc = NULL;
1532 if(val_nsec_proves_name_error(wcrr, wc_ce))
1533 rcode = LDNS_RCODE_NXDOMAIN;
1534 else if(!nsec_proves_nodata(wcrr, &wc_qinfo,
1535 &nodata_wc) || nodata_wc)
1536 /* &nodata_wc shouldn't be set, wc_qinfo
1537 * already contains wildcard domain. */
1538 /* NSEC doesn't prove anything for
1539 * wildcard. */
1540 return NULL;
1541 if(query_dname_compare(wcrr->rk.dname,
1542 nsec->rk.dname) != 0)
1543 if(!dns_msg_authadd(msg, region, wcrr, 0))
1544 return NULL;
1545 }
1546 }
1547
1548 if(!dns_msg_authadd(msg, region, nsec, 0))
1549 return NULL;
1550 if(addsoa && !add_soa(rrset_cache, now, region, msg, NULL))
1551 return NULL;
1552
1553 /* Increment statistic counters */
1554 lock_basic_lock(&neg->lock);
1555 if(rcode == LDNS_RCODE_NOERROR)
1556 neg->num_neg_cache_noerror++;
1557 else if(rcode == LDNS_RCODE_NXDOMAIN)
1558 neg->num_neg_cache_nxdomain++;
1559 lock_basic_unlock(&neg->lock);
1560
1561 FLAGS_SET_RCODE(msg->rep->flags, rcode);
1562 return msg;
1563 }
1564
1565 /* No aggressive use of NSEC3 for now, only proceed for DS types. */
1566 if(qinfo->qtype != LDNS_RR_TYPE_DS){
1567 return NULL;
1568 }
1569 /* check NSEC3 neg cache for type DS */
1570 /* need to look one zone higher for DS type */
1571 zname = qinfo->qname;
1572 zname_len = qinfo->qname_len;
1573 dname_remove_label(&zname, &zname_len);
1574 zname_labs = dname_count_labels(zname);
1575
1576 /* lookup closest zone */
1577 lock_basic_lock(&neg->lock);
1578 zone = neg_closest_zone_parent(neg, zname, zname_len, zname_labs,
1579 qinfo->qclass);
1580 while(zone && !zone->in_use)
1581 zone = zone->parent;
1582 /* check that the zone is not too high up so that we do not pick data
1583 * out of a zone that is above the last-seen key (or trust-anchor). */
1584 if(zone && topname) {
1585 if(!dname_subdomain_c(zone->name, topname))
1586 zone = NULL;
1587 }
1588 if(!zone) {
1589 lock_basic_unlock(&neg->lock);
1590 return NULL;
1591 }
1592
1593 msg = neg_nsec3_proof_ds(zone, qinfo->qname, qinfo->qname_len,
1594 zname_labs+1, buf, rrset_cache, region, now, topname);
1595 if(msg && addsoa && !add_soa(rrset_cache, now, region, msg, zone)) {
1596 lock_basic_unlock(&neg->lock);
1597 return NULL;
1598 }
1599 lock_basic_unlock(&neg->lock);
1600 return msg;
1601 }
1602
1603 void
val_neg_adjust_size(struct val_neg_cache * neg,size_t max)1604 val_neg_adjust_size(struct val_neg_cache* neg, size_t max)
1605 {
1606 lock_basic_lock(&neg->lock);
1607 neg->max = max;
1608 neg_make_space(neg, 0);
1609 lock_basic_unlock(&neg->lock);
1610 }
1611