-
Notifications
You must be signed in to change notification settings - Fork 5
/
Copy pathCHANGELOG
3123 lines (2445 loc) · 123 KB
/
CHANGELOG
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
version 2.91
Fix spurious "resource limit exceeded messages". Thanks to
Dominik Derigs for the bug report.
Fix out-of-bounds heap read in order_qsort().
We only need to order two server records on the ->serial field.
Literal address records are smaller and don't have
this field and don't need to be ordered on it.
To actually provoke this bug seems to need the same server-literal
to be repeated twice, e.g., --address=/a/1.1.1.1 --address-/a/1.1.1.1
which is clearly rare in the wild, but if it did exist it could
provoke a SIGSEGV. Thanks to Daniel Rhea for fuzzing this one.
Fix buffer overflow when configured lease-change script name
is too long.
Thanks to Daniel Rhea for finding this one.
Improve behaviour in the face of non-responsive upstream TCP DNS
servers. Without shorter timeouts, clients are blocked for too long
and fail with their own timeouts.
Set --fast-dns-retries by default when doing DNSSEC. A single
downstream query can trigger many upstream queries. On an
unreliable network, there may not be enough downstream retries
to ensure that all these queries complete.
Improve behaviour in the face of truncated answers to queries
for DNSSEC records. Getting these answers by TCP doesn't now
involve a faked truncated answer to the downstream client to
force it to move to TCP. This improves performance and robustness
in the face of broken clients which can't fall back to TCP.
No longer remove data from truncated upstream answers. If an
upstream replies with a truncated answer, but the answer has some
RRs included, return those RRs, rather than returning and
empty answer.
Fix handling of EDNS0 UDP packet sizes.
When talking upstream we always add a pseudo header, and set the
UDP packet size to --edns-packet-max. Answering queries from
downstream, we get the answer (either from upstream or local
data) If local data won't fit the advertised size (or 512 if
there's not an EDNS0 header) return truncated. If upstream
returns truncated, do likewise. If upstream is OK, but the
answer is too big for downstream, truncate the answer.
Modify the behaviour of --synth-domain for IPv6.
When deriving a domain name from an IPv6 address, an address
such as 1234:: would become 1234--.example.com, which is
not legal in IDNA2008. Stop using the :: compression method,
so 1234:: becomes
1234-0000-0000-0000-0000-0000-0000-0000.example.com
Fix broken dhcp-relay on *BSD. Thanks to Harold for finding
this problem.
Add --dhcp-option-pxe config. This acts almost exactly like
--dhcp-option except that the defined option is only sent when
replying to PXE clients. More importantly, these options are sent
in reply PXE clients when dnsmasq in acting in PXE proxy mode. In
PXE proxy mode, the set of options sent is defined by the PXE standard
and the normal set of options is not sent. This config allows arbitrary
options in PXE-proxy replies. A typical use-case is to send option
175 to iPXE. Thanks to Jason Berry for finding the requirement for
this.
Support PXE proxy-DHCP and DHCP-relay at the same time.
When using PXE proxy-DHCP, dnsmasq supplies PXE information to
the client, which also talks to another "normal" DHCP server
for address allocation and similar. The normal DHCP server may
be on the local network, but it may also be remote, and accessed via
a DHCP relay. This change allows dnsmasq to act as both a
PXE proxy-DHCP server AND a DHCP relay for the same network.
Fix erroneous "DNSSEC validated" state with non-DNSSEC
upstream servers. Thanks to Dominik Derigs for the bug report.
Handle queries with EDNS client subnet fields better. If dnsmasq
is configured to add an EDNS client subnet to a query, it is careful
to suppress use of the cache, since a cached answer may not be valid
for a query with a different client subnet. Extend this behaviour
to queries which arrive a dnsmasq already carrying an EDNS client
subnet.
Handle DS queries to auth zones. When dnsmasq is configured to
act as an authoritative server and has an authoritative zone
configured, and receives a query for that zone _as_forwarder_
it answers the query directly rather than forwarding it. This
doesn't affect the answer, but it saves dnsmasq forwarding the
query to the recursor upstream, which then bounces it back to dnsmasq
in auth mode. The exception should be when the query is for the root
of zone, for a DS RR. The answer to that has to come from the parent,
via the recursor, and will typically be a proof-of-non-existence
since dnsmasq doesn't support signed zones. This patch suppresses
local answers and forces forwarding to the upstream recursor for such
queries. It stops breakage when a DNSSEC validating client makes
queries to dnsmasq acting as forwarder for a zone for which it is
authoritative.
Implement "DNS-0x20 encoding", for extra protection against
reply-spoof attacks. Since DNS queries are case-insensitive,
it's possible to randomly flip the case of letters in a query
and still get the correct answer back.
This adds an extra dimension for a cache-poisoning attacker
to guess when sending replies in-the-blind since it's expected
that the legitimate answer will have the same pattern of upper
and lower case as the query, so any replies which don't can be
ignored as malicious. The amount of extra entropy clearly depends
on the number of a-z and A-Z characters in the query, and this
implementation puts a hard limit of 32 bits to make resource
allocation easy. This about doubles entropy over the standard
random ID and random port combination. This technique can interact
badly with rare broken DNS servers which don't preserve the case
of the query in their reply. The first time a reply is returned
which matches the query in all respects except case, a warning
will be logged. In this release, 0x020-encoding is default-off
and must be explicitly enabled with --do-0x20-encoding. In future
releases it may default on. You can avoid a future release
changing the behaviour of an installation with --no-x20-encode.
Fix a long-standing problem when two queries which are identical
in every repect _except_ case, get combined by dnsmasq. If
dnsmasq gets eg, two queries for example.com and Example.com
in quick succession it will get the answer for example.com from
upstream and send that answer to both requestors. This means that
the query for Example.com will get an answer for example.com, and
in the modern DNS, that answer may not be accepted.
version 2.90
Fix reversion in --rev-server introduced in 2.88 which
caused breakage if the prefix length is not exactly divisible
by 8 (IPv4) or 4 (IPv6).
Fix possible SEGV when there server(s) for a particular
domain are configured, but no server which is not qualified
for a particular domain. Thanks to Daniel Danzberger for
spotting this bug.
Set the default maximum DNS UDP packet sice to 1232. This
has been the recommended value since 2020 because it's the
largest value that avoid fragmentation, and fragmentation
is just not reliable on the modern internet, especially
for IPv6. It's still possible to override this with
--edns-packet-max for special circumstances.
Add --no-dhcpv4-interface and --no-dhcpv6-interface for
better control over which inetrfaces are providing DHCP service.
Fix issue with stale caching: After replying with stale data,
dnsmasq sends the query upstream to refresh the cache asynchronously
and sometimes sends the wrong packet: packet length can be wrong,
and if an EDE marking stale data is added to the answer that can
end up in the query also. This bug only seems to cause problems
when the usptream server is a DOH/DOT proxy. Thanks to Justin He
for the bug report.
Add configurable caching for arbitrary RR-types.
Add --filter-rr option, to filter arbitrary RR-types.
--filter-rr=ANY has a special meaning: it filters the
answers to queries for the ANY RR-type.
Add limits on the resources used to do DNSSEC validation.
DNSSEC introduces a potential CPU DoS, because a crafted domain
can force a validator to a large number of cryptographic
operations whilst attempting to do validation. When using TCP
transport a DNSKEY RRset contain thousands of members and any
RRset can have thousands of signatures. The potential number
of signature validations to follow the RFC for validation
for one RRset is the cross product of the keys and signatures,
so millions. In practice, the actual numbers are much lower,
so attacks can be mitigated by limiting the amount of
cryptographic "work" to a much lower amount. The actual
limits are number a signature validation fails per RRset(20),
number of signature validations and hash computations
per query(200), number of sub-queries to fetch DS and DNSKEY
RRsets per query(40), and the number of iterations in a
NSEC3 record(150). These values are sensible, but there is, as yet,
no standardisation on the values for a "conforming" domain, so a
new option --dnssec-limit is provided should they need to be altered.
The algorithm to validate DS records has also been altered to reduce
the maximum work from cross product of the number of DS records and
number of DNSKEYs to the cross product of the number of DS records
and supported DS digest types. As the number of DS digest types
is in single figures, this reduces the exposure.
Credit is due to Elias Heftrig, Haya Schulmann, Niklas Vogel,
and Michael Waidner from the German National Research Center for
Applied Cybersecurity ATHENE for finding this vulnerability.
CVE 2023-50387 and CVE 2023-50868 apply.
Note that the is a security vulnerablity only when DNSSEC validation
is enabled.
Fix memory-leak when attempting to cache SRV records with zero TTL.
Thanks to Damian Sawicki for the bug report.
Add --max-tcp-connections option to make limit on TCP handling
processes configurable. Also keep stats on how near the limit
we're getting, to help with tuning. Patch from Damian Sawicki.
version 2.89
Fix bug introduced in 2.88 (commit fe91134b) which can result
in corruption of the DNS cache internal data structures and
logging of "cache internal error". This has only been seen
in one place in the wild, and it took considerable effort
to even generate a test case to reproduce it, but there's
no way to be sure it won't strike, and the effect is to break
the cache badly. Installations with DNSSEC enabled are more
likely to see the problem, but not running DNSSEC does not
guarantee that it won't happen. Thanks to Timo van Roermund
for reporting the bug and for his great efforts in chasing
it down.
version 2.88
Fix bug in --dynamic-host when an interface has /16 IPv4
address. Thanks to Mark Dietzer for spotting this.
Add --fast-dns-retry option. This gives dnsmasq the ability
to originate retries for upstream DNS queries itself, rather
than relying on the downstream client. This is most useful
when doing DNSSEC over unreliable upstream networks. It comes
with some cost in memory usage and network bandwidth.
Add --use-stale-cache option. When set, if a DNS name exists
in the cache, but its time-to-live has expired, dnsmasq will
return the data anyway. (It attempts to refresh the
data with an upstream query after returning the stale data.)
This can improve speed and reliability. It comes
at the expense of sometimes returning out-of-date data and
less efficient cache utilisation, since old data cannot be
flushed when its TTL expires, so the cache becomes
strictly least-recently-used.
Add --port-limit option which allows tuning for robustness in
the face of some upstream network errors. Thanks to
Prashant Kumar Singh, Ravi Nagayach and Mike Danilov,
all of Amazon Web Services, for their efforts in developing this
and the stale-cache and fast-retry options.
Make --hostsdir (but NOT --dhcp-hostsdir and --dhcp-optsdir)
handle removal of whole files or entries within files.
Thanks to Dominik Derigs for the initial patches for this.
Fix bug, introduced in 2.87, which could result in DNS
servers being removed from the configuration when reloading
server configuration from DBus, or re-reading /etc/resolv.conf
Only servers from the same source should be replaced, but some
servers from other sources (i.e., hard coded or another dynamic source)
could mysteriously disappear. Thanks to all reporting this,
but especially Christopher J. Madsen who reduced the problem
to an easily reproducible case which saved much labour in
finding it.
Add --no-round-robin option.
Allow domain names as well as IP addresses when specifying
upstream DNS servers. There are some gotchas associated with this
(it will mysteriously fail to work if the dnsmasq instance
being started is in the path from the system resolver to the DNS),
and a seemingly sensible configuration like
[email protected] is unactionable if domain.name
only resolves to an IPv6 address). There are, however,
cases where is can be useful. Thanks to Dominik Derigs for
the patch.
Handle DS records for unsupported crypto algorithms correctly.
Such a DS, as long as it is validated, should allow answers
in the domain it attests to be returned as unvalidated, and not
as a validation error.
Optimise reading large numbers of --server options. When re-reading
upstream servers from /etc/resolv.conf or other sources that
can change dnsmasq tries to avoid memory fragmentation by re-using
existing records that are being re-read unchanged. This involves
seaching all the server records for each new one installed.
During startup this search is pointless, and can cause long
start times with thousands of --server options because the work
needed is O(n^2). Handle this case more intelligently.
Thanks to Ye Zhou for spotting the problem and an initial patch.
If we detect that a DNS reply from upstream is malformed don't
return it to the requestor; send a SEVFAIL rcode instead.
version 2.87
Allow arbitrary prefix lengths in --rev-server and
--domain=....,local
Replace --address=/#/..... functionality which got
missed in the 2.86 domain search rewrite.
Add --nftset option, like --ipset but for the newer nftables.
Thanks to Chen Zhenge for the patch.
Add --filter-A and --filter-AAAA options, to remove IPv4 or IPv6
addresses from DNS answers.
Fix crash doing netbooting when --port is set to zero
to disable the DNS server. Thanks to Drexl Johannes
for the bug report.
Generalise --dhcp-relay. Sending via broadcast/multicast is
now supported for both IPv4 and IPv6 and the configuration
syntax made easier (but backwards compatible).
Add snooping of IPv6 prefix-delegations to the DHCP-relay system.
Finesse parsing of --dhcp-remoteid and --dhcp-subscrid. To be treated
as hex, the pattern must consist of only hex digits AND contain
at least one ':'. Thanks to Bengt-Erik Sandstrom who tripped
over a pattern consisting of a decimal number which was interpreted
surprisingly.
Include client address in TFTP file-not-found error reports.
Thanks to Stefan Rink for the initial patch, which has been
re-worked by me (srk). All bugs mine.
Note in manpage the change in behaviour of -address. This behaviour
actually changed in v2.86, but was undocumented there. From 2.86 on,
(eg) --address=/example.com/1.2.3.4 ONLY applies to A queries. All other
types of query will be sent upstream. Pre 2.86, that would catch the
whole example.com domain and queries for other types would get
a local NODATA answer. The pre-2.86 behaviour is still available,
by configuring --address=/example.com/1.2.3.4 --local=/example.com/
Fix problem with binding DHCP sockets to an individual interface.
Despite the fact that the system call tales the interface _name_ as
a parameter, it actually, binds the socket to interface _index_.
Deleting the interface and creating a new one with the same name
leaves the socket bound to the old index. (Creating new sockets
always allocates a fresh index, they are not reused). We now
take this behaviour into account and keep up with changing indexes.
Add --conf-script configuration option.
Enhance --domain to accept, for instance,
--domain=net2.thekelleys.org.uk,eth2 so that hosts get a domain
which relects the interface they are attached to in a way which
doesn't require hard-coding addresses. Thanks to Sten Spans for
the idea.
Fix write-after-free error in DHCPv6 server code.
CVE-2022-0934 refers.
Add the ability to specify destination port in
DHCP-relay mode. This change also removes a previous bug
where --dhcp-alternate-port would affect the port used
to relay _to_ as well as the port being listened on.
The new feature allows configuration to provide bug-for-bug
compatibility, if required. Thanks to Damian Kaczkowski
for the feature suggestion.
Bound the value of UDP packet size in the EDNS0 header of
forwarded queries to the configured or default value of
edns-packet-max. There's no point letting a client set a larger
value if we're unable to return the answer. Thanks to Bertie
Taylor for pointing out the problem and supplying the patch.
Fix problem with the configuration
--server=/some.domain/# --address=/#/<ip> --server=<server_ip>
This would return <ip> for queries in some.domain, rather than
forwarding the query via the default server.
Tweak DHCPv6 relay code so that packets relayed towards a server
have source address on the server-facing network, not the
client facing network. Thanks to Luis Thomas for spotting this
and initial patch.
version 2.86
Handle DHCPREBIND requests in the DHCPv6 server code.
Thanks to Aichun Li for spotting this omission, and the initial
patch.
Fix bug which caused dnsmasq to lose track of processes forked
to handle TCP DNS connections under heavy load. The code
checked that at least one free process table slot was
available before listening on TCP sockets, but didn't take
into account that more than one TCP connection could
arrive, so that check was not sufficient to ensure that
there would be slots for all new processes. It compounded
this error by silently failing to store the process when
it did run out of slots. Even when this bug is triggered,
all the right things happen, and answers are still returned.
Only under very exceptional circumstances, does the bug
manifest itself: see
https://lists.thekelleys.org.uk/pipermail/dnsmasq-discuss/2021q2/014976.html
Thanks to Tijs Van Buggenhout for finding the conditions under
which the bug manifests itself, and then working out
exactly what was going on.
Major rewrite of the DNS server and domain handling code.
This should be largely transparent, but it drastically
improves performance and reduces memory foot-print when
configuring large numbers domains of the form
local=/adserver.com/
or
local=/adserver.com/#
Lookup times now grow as log-to-base-2 of the number of domains,
rather than greater than linearly, as before.
The change makes multiple addresses associated with a domain work
address=/example.com/1.2.3.4
address=/example.com/5.6.7.8
It also handles multiple upstream servers for a domain better; using
the same try/retry algorithms as non domain-specific servers. This
also applies to DNSSEC-generated queries.
Finally, some of the oldest and gnarliest code in dnsmasq has had
a significant clean-up. It's far from perfect, but it _is_ better.
Revise resource handling for number of concurrent DNS queries. This
used to have a global limit, but that has a problem when using
different servers for different upstream domains. Queries which are
routed by domain to an upstream server which is not responding will
build up and trigger the limit, which breaks DNS service for
all other domains which could be handled by other servers. The
change is to make the limit per server-group, where a server group
is the set of servers configured for a particular domain. In the
common case, where only default servers are declared, there is
no effective change.
Improve efficiency of DNSSEC. The sharing point for DNSSEC RR data
used to be when it entered the cache, having been validated. After
that queries requiring the KEY or DS records would share the cached
values. There is a common case in dual-stack hosts that queries for
A and AAAA records for the same domain are made simultaneously.
If required keys were not in the cache, this would result in two
requests being sent upstream for the same key data (and all the
subsequent chain-of-trust queries.) Now we combine these requests
and elide the duplicates, resulting in fewer queries upstream
and better performance. To keep a better handle on what's
going on, the "extra" logging mode has been modified to associate
queries and answers for DNSSEC queries in the same way as ordinary
queries. The requesting address and port have been removed from
DNSSEC logging lines, since this is no longer strictly defined.
Connection track mark based DNS query filtering. Thanks to
Etan Kissling for implementing this It extends query filtering
support beyond what is currently possible
with the `--ipset` configuration option, by adding support for:
1) Specifying allowlists on a per-client basis, based on their
associated Linux connection track mark.
2) Dynamic configuration of allowlists via Ubus.
3) Reporting when a DNS query resolves or is rejected via Ubus.
4) DNS name patterns containing wildcards.
Disallowed queries are not forwarded; they are rejected
with a REFUSED error code.
Allow smaller than 64 prefix lengths in synth-domain, with caveats.
--synth-domain=1234:4567::/56,example.com is now valid.
Make domains generated by --synth-domain appear in replies
when in authoritative mode.
Ensure CAP_NET_ADMIN capability is available when
conntrack is configured. Thanks to Yick Xie for spotting
the lack of this.
When --dhcp-hostsfile --dhcp-optsfile and --addn-hosts are
given a directory as argument, define the order in which
files within that directory are read (alphabetical order
of filename). Thanks to Ed Wildgoose for the initial patch
and motivation for this.
Allow adding IP address to nftables set in addition to
ipset.
version 2.85
Fix problem with DNS retries in 2.83/2.84.
The new logic in 2.83/2.84 which merges distinct requests
for the same domain causes problems with clients which do
retries as distinct requests (differing IDs and/or source ports.)
The retries just get piggy-backed on the first, failed, request.
The logic is now changed so that distinct requests for repeated
queries still get merged into a single ID/source port, but
they now always trigger a re-try upstream.
Thanks to Nicholas Mu for his analysis.
Tweak sort order of tags in get-version. v2.84 sorts
before v2.83, but v2.83 sorts before v2.83rc1 and 2.83rc1
sorts before v2.83test1. This fixes the problem which lead
to 2.84 announcing itself as 2.84rc2.
Avoid treating a --dhcp-host which has an IPv6 address
as eligible for use with DHCPv4 on the grounds that it has
no address, and vice-versa. Thanks to Viktor Papp for
spotting the problem. (This bug was fixed was back in 2.67, and
then regressed in 2.81).
Add --dynamic-host option: A and AAAA records which take their
network part from the network of a local interface. Useful
for routers with dynamically prefixes. Thanks
to Fred F for the suggestion.
Teach --bogus-nxdomain and --ignore-address to take an IPv4 subnet.
Use random source ports where possible if source
addresses/interfaces in use.
CVE-2021-3448 applies. Thanks to Petr Menšík for spotting this.
It's possible to specify the source address or interface to be
used when contacting upstream name servers: [email protected]
or [email protected]#66 or server=8.8.8.8@eth0, and all of
these have, until now, used a single socket, bound to a fixed
port. This was originally done to allow an error (non-existent
interface, or non-local address) to be detected at start-up. This
means that any upstream servers specified in such a way don't use
random source ports, and are more susceptible to cache-poisoning
attacks.
We now use random ports where possible, even when the
source is specified, so [email protected] or
server=8.8.8.8@eth0 will use random source
ports. [email protected]#66 or any use of --query-port will
use the explicitly configured port, and should only be done with
understanding of the security implications.
Note that this change changes non-existing interface, or non-local
source address errors from fatal to run-time. The error will be
logged and communication with the server not possible.
Change the method of allocation of random source ports for DNS.
Previously, without min-port or max-port configured, dnsmasq would
default to the compiled in defaults for those, which are 1024 and
65535. Now, when neither are configured, it defaults instead to
the kernel's ephemeral port range, which is typically
32768 to 60999 on Linux systems. This change eliminates the
possibility that dnsmasq may be using a registered port > 1024
when a long-running daemon starts up and wishes to claim it.
This change does likely slightly reduce the number of random ports
and therefore the protection from reply spoofing. The older
behaviour can be restored using the min-port and max-port config
switches should that be a concern.
Scale the size of the DNS random-port pool based on the
value of the --dns-forward-max configuration.
Tweak TFTP code to check sender of all received packets, as
specified in RFC 1350 para 4.
Support some wildcard matching of input tags to --tag-if.
Thanks to Geoff Back for the idea and the patch.
version 2.84
Fix a problem, introduced in 2.83, which could see DNS replies
being sent via the wrong socket. On machines running both
IPv4 and IPv6 this could result in sporadic messages of
the form "failed to send packet: Network is unreachable" and
the lost of the query. Since the error is sporadic and of
low probability, the client retry would normally succeed.
Change HAVE_NETTLEHASH compile-time to HAVE_CRYPTOHASH.
version 2.83
Use the values of --min-port and --max-port in outgoing
TCP connections to upstream DNS servers.
Fix a remote buffer overflow problem in the DNSSEC code. Any
dnsmasq with DNSSEC compiled in and enabled is vulnerable to this,
referenced by CVE-2020-25681, CVE-2020-25682, CVE-2020-25683
CVE-2020-25687.
Be sure to only accept UDP DNS query replies at the address
from which the query was originated. This keeps as much entropy
in the {query-ID, random-port} tuple as possible, to help defeat
cache poisoning attacks. Refer: CVE-2020-25684.
Use the SHA-256 hash function to verify that DNS answers
received are for the questions originally asked. This replaces
the slightly insecure SHA-1 (when compiled with DNSSEC) or
the very insecure CRC32 (otherwise). Refer: CVE-2020-25685.
Handle multiple identical near simultaneous DNS queries better.
Previously, such queries would all be forwarded
independently. This is, in theory, inefficient but in practise
not a problem, _except_ that is means that an answer for any
of the forwarded queries will be accepted and cached.
An attacker can send a query multiple times, and for each repeat,
another {port, ID} becomes capable of accepting the answer he is
sending in the blind, to random IDs and ports. The chance of a
successful attack is therefore multiplied by the number of repeats
of the query. The new behaviour detects repeated queries and
merely stores the clients sending repeats so that when the
first query completes, the answer can be sent to all the
clients who asked. Refer: CVE-2020-25686.
version 2.82
Improve behaviour in the face of network interfaces which come
and go and change index. Thanks to Petr Mensik for the patch.
Convert hard startup failure on NETLINK_NO_ENOBUFS under qemu-user
to a warning.
Allow IPv6 addresses ofthe form [::ffff:1.2.3.4] in --dhcp-option.
Fix crash under heavy TCP connection load introduced in 2.81.
Thanks to Frank for good work chasing this down.
Change default lease time for DHCPv6 to one day.
Alter calculation of preferred and valid times in router
advertisements, so that these do not have a floor applied
of the lease time in the dhcp-range if this is not explicitly
specified and is merely the default.
Thanks to Martin-Éric Racine for suggestions on this.
version 2.81
Improve cache behaviour for TCP connections. For ease of
implementation, dnsmasq has always forked a new process to handle
each incoming TCP connection. A side-effect of this is that
any DNS queries answered from TCP connections are not cached:
when TCP connections were rare, this was not a problem.
With the coming of DNSSEC, it is now the case that some
DNSSEC queries have answers which spill to TCP, and if,
for instance, this applies to the keys for the root, then
those never get cached, and performance is very bad.
This fix passes cache entries back from the TCP child process to
the main server process, and fixes the problem.
Remove the NO_FORK compile-time option, and support for uclinux.
In an era where everything has an MMU, this looks like
an anachronism, and it adds to (Ok, multiplies!) the
combinatorial explosion of compile-time options. Thanks to
Kevin Darbyshire-Bryant for the patch.
Fix line-counting when reading /etc/hosts and friends; for
correct error messages. Thanks to Christian Rosentreter
for reporting this.
Fix bug in DNS non-terminal code, added in 2.80, which could
sometimes cause a NODATA rather than an NXDOMAIN reply.
Thanks to Norman Rasmussen, Sven Mueller and Maciej Żenczykowski
for spotting and diagnosing the bug and providing patches.
Support TCP-fastopen (RFC-7413) on both incoming and
outgoing TCP connections, if supported and enabled in the OS.
Improve kernel-capability manipulation code under Linux. Dnsmasq
now fails early if a required capability is not available, and
tries not to request capabilities not required by its
configuration.
Add --shared-network config. This enables allocation of addresses
by the DHCP server in subnets where the server (or relay) does not
have an interface on the network in that subnet. Many thanks to
kamp.de for sponsoring this feature.
Fix broken contrib/lease_tools/dhcp_lease_time.c. A packet
validation check got borked in commit 2b38e382 and release 2.80.
Thanks to Tomasz Szajner for spotting this.
Fix compilation against nettle version 3.5 and later.
Fix spurious DNSSEC validation failures when the auth section
of a reply contains unsigned RRs from a signed zone,
with the exception that NSEC and NSEC3 RRs must always be signed.
Thanks to Tore Anderson for spotting and diagnosing the bug.
Add --dhcp-ignore-clid. This disables reading of DHCP client
identifier option (option 61), so clients are only identified by
MAC addresses.
Fix a bug which stopped --dhcp-name-match from working when a hostname
is supplied in --dhcp-host. Thanks to James Feeney for spotting this.
Fix bug which caused very rarely caused zero-length DHCPv6 packets.
Thanks to Dereck Higgins for spotting this.
Add --tftp-single-port option.
Enhance --conf-dir to load files in a deterministic order. Thanks to
Evgenii Seliavka for the suggestion and initial patch.
In the router advert code, handle case where we have two
different interfaces on the same IPv6 net, and we are doing
RA/DHCP service on only one of them. Thanks to NIIBE Yutaka
for spotting this case and making the initial patch.
Support prefixed ranges of ipv6 addresses in dhcp-host.
This eases problems chain-netbooting, where each link in the
chain requests an address using a different UID. With a single
address, only one gets the "static" address, but with this
fix, enough addresses can be reserved for all the stages of the
boot. Many thanks to Harald Jensås for his work on this idea and
earlier patches.
Add filtering by tag of --dhcp-host directives. Based on a patch
by Harald Jensås.
Allow empty server spec in --rev-server, to match --server.
Remove DSA signature verification from DNSSEC, as specified in
RFC 8624. Thanks to Loganaden Velvindron for the original patch.
Add --script-on-renewal option.
version 2.80
Add support for RFC 4039 DHCP rapid commit. Thanks to Ashram Method
for the initial patch and motivation.
Alter the default for dnssec-check-unsigned. Versions of
dnsmasq prior to 2.80 defaulted to not checking unsigned
replies, and used --dnssec-check-unsigned to switch
this on. Such configurations will continue to work as before,
but those which used the default of no checking will need to be
altered to explicitly select no checking. The new default is
because switching off checking for unsigned replies is
inherently dangerous. Not only does it open the possiblity of forged
replies, but it allows everything to appear to be working even
when the upstream namesevers do not support DNSSEC, and in this
case no DNSSEC validation at all is occuring.
Fix DHCP broken-ness when --no-ping AND --dhcp-sequential-ip
are set. Thanks to Daniel Miess for help with this.
Add a facilty to store DNS packets sent/recieved in a
pcap-format file for later debugging. The file location
is given by the --dumpfile option, and a bitmap controlling
which packets should be dumped is given by the --dumpmask
option.
Handle the case of both standard and constructed dhcp-ranges on the
same interface better. We don't now contruct a dhcp-range if there's
already one specified. This allows the specified interface to
have different parameters and avoids advertising the same
prefix twice. Thanks to Luis Marsano for spotting this case.
Allow zone transfer in authoritative mode if auth-peer is specified,
even if auth-sec-servers is not. Thanks to Raphaël Halimi for
the suggestion.
Fix bug which sometimes caused dnsmasq to wrongly return answers
without DNSSEC RRs to queries with the do-bit set, but only when
DNSSEC validation was not enabled.
Thanks to Petr Menšík for spotting this.
Fix missing fatal errors with some malformed options
(server, local, address, rebind-domain-ok, ipset, alias).
Thanks to Eugene Lozovoy for spotting the problem.
Fix crash on startup with a --synth-domain which has no prefix.
Introduced in 2.79. Thanks to Andreas Engel for the bug report.
Fix missing EDNS0 section in some replies generated by local
DNS configuration which confused systemd-resolvd. Thanks to
Steve Dodd for characterising the problem.
Add --dhcp-name-match config option.
Add --caa-record config option.
Implement --address=/example.com/# as (more efficient) syntactic
sugar for --address=/example.com/0.0.0.0 and
--address=/example.com/::
Returning null addresses is a useful technique for ad-blocking.
Thanks to Peter Russell for the suggestion.
Change anti cache-snooping behaviour with queries with the
recursion-desired bit unset. Instead to returning SERVFAIL, we
now always forward, and never answer from the cache. This
allows "dig +trace" command to work.
Include in the example config file a formulation which
stops DHCP clients from claiming the DNS name "wpad".
This is a fix for the CERT Vulnerability VU#598349.
version 2.79
Fix parsing of CNAME arguments, which are confused by extra spaces.
Thanks to Diego Aguirre for spotting the bug.
Where available, use IP_UNICAST_IF or IPV6_UNICAST_IF to bind
upstream servers to an interface, rather than SO_BINDTODEVICE.
Thanks to Beniamino Galvani for the patch.
Always return a SERVFAIL answer to DNS queries without the
recursion desired bit set, UNLESS acting as an authoritative
DNS server. This avoids a potential route to cache snooping.
Add support for Ed25519 signatures in DNSSEC validation.
No longer support RSA/MD5 signatures in DNSSEC validation,
since these are not secure. This behaviour is mandated in
RFC-6944.
Fix incorrect error exit code from dhcp_release6 utility.
Thanks Gaudenz Steinlin for the bug report.
Use SIGINT (instead of overloading SIGHUP) to turn on DNSSEC
time validation when --dnssec-no-timecheck is in use.
Note that this is an incompatible change from earlier releases.
Allow more than one --bridge-interface option to refer to an
interface, so that we can use
--bridge-interface=int1,alias1
--bridge-interface=int1,alias2
as an alternative to
--bridge-interface=int1,alias1,alias2
Thanks to Neil Jerram for work on this.
Fix for DNSSEC with wildcard-derived NSEC records.
It's OK for NSEC records to be expanded from wildcards,
but in that case, the proof of non-existence is only valid
starting at the wildcard name, *.<domain> NOT the name expanded
from the wildcard. Without this check it's possible for an
attacker to craft an NSEC which wrongly proves non-existence.
Thanks to Ralph Dolmans for finding this, and co-ordinating
the vulnerability tracking and fix release.
CVE-2017-15107 applies.
Remove special handling of A-for-A DNS queries. These
are no longer a significant problem in the global DNS.
http://cs.northwestern.edu/~ychen/Papers/DNS_ToN15.pdf
Thanks to Mattias Hellström for the initial patch.
Fix failure to delete dynamically created dhcp options
from files in -dhcp-optsdir directories. Thanks to
Lindgren Fredrik for the bug report.
Add to --synth-domain the ability to create names using
sequential numbers, as well as encodings of IP addresses.
For instance,
--synth-domain=thekelleys.org.uk,192.168.0.50,192.168.0.70,internal-*
creates 21 domain names of the form
internal-4.thekelleys.org.uk over the address range given, with
internal-0.thekelleys.org.uk being 192.168.0.50 and
internal-20.thekelleys.org.uk being 192.168.0.70
Thanks to Andy Hawkins for the suggestion.
Tidy up Crypto code, removing workarounds for ancient
versions of libnettle. We now require libnettle 3.
version 2.78
Fix logic of appending ".<layer>" to PXE basename. Thanks to Chris
Novakovic for the patch.
Revert ping-check of address in DHCPDISCOVER if there
already exists a lease for the address. Under some
circumstances, and netbooted windows installation can reply
to pings before if has a DHCP lease and block allocation
of the address it already used during netboot. Thanks to
Jan Psota for spotting this.
Fix DHCP relaying, broken in 2.76 and 2.77 by commit
ff325644c7afae2588583f935f4ea9b9694eb52e. Thanks to
John Fitzgibbon for the diagnosis and patch.
Try other servers if first returns REFUSED when
--strict-order active. Thanks to Hans Dedecker
for the patch
Fix regression in 2.77, ironically added as a security
improvement, which resulted in a crash when a DNS
query exceeded 512 bytes (or the EDNS0 packet size,
if different.) Thanks to Christian Kujau, Arne Woerner
Juan Manuel Fernandez and Kevin Darbyshire-Bryant for
chasing this one down. CVE-2017-13704 applies.
Fix heap overflow in DNS code. This is a potentially serious
security hole. It allows an attacker who can make DNS
requests to dnsmasq, and who controls the contents of
a domain, which is thereby queried, to overflow
(by 2 bytes) a heap buffer and either crash, or
even take control of, dnsmasq.
CVE-2017-14491 applies.
Credit to Felix Wilhelm, Fermin J. Serna, Gabriel Campana
Kevin Hamacher and Ron Bowes of the Google Security Team for
finding this.
Fix heap overflow in IPv6 router advertisement code.
This is a potentially serious security hole, as a
crafted RA request can overflow a buffer and crash or
control dnsmasq. Attacker must be on the local network.
CVE-2017-14492 applies.
Credit to Felix Wilhelm, Fermin J. Serna, Gabriel Campana
and Kevin Hamacher of the Google Security Team for
finding this.
Fix stack overflow in DHCPv6 code. An attacker who can send
a DHCPv6 request to dnsmasq can overflow the stack frame and
crash or control dnsmasq.
CVE-2017-14493 applies.
Credit to Felix Wilhelm, Fermin J. Serna, Gabriel Campana
Kevin Hamacher and Ron Bowes of the Google Security Team for
finding this.
Fix information leak in DHCPv6. A crafted DHCPv6 packet can
cause dnsmasq to forward memory from outside the packet
buffer to a DHCPv6 server when acting as a relay.
CVE-2017-14494 applies.
Credit to Felix Wilhelm, Fermin J. Serna, Gabriel Campana
Kevin Hamacher and Ron Bowes of the Google Security Team for
finding this.
Fix DoS in DNS. Invalid boundary checks in the
add_pseudoheader function allows a memcpy call with negative
size An attacker which can send malicious DNS queries
to dnsmasq can trigger a DoS remotely.
dnsmasq is vulnerable only if one of the following option is
specified: --add-mac, --add-cpe-id or --add-subnet.
CVE-2017-14496 applies.
Credit to Felix Wilhelm, Fermin J. Serna, Gabriel Campana
Kevin Hamacher and Ron Bowes of the Google Security Team for
finding this.
Fix out-of-memory Dos vulnerability. An attacker which can
send malicious DNS queries to dnsmasq can trigger memory
allocations in the add_pseudoheader function
The allocated memory is never freed which leads to a DoS
through memory exhaustion. dnsmasq is vulnerable only
if one of the following option is specified:
--add-mac, --add-cpe-id or --add-subnet.
CVE-2017-14495 applies.
Credit to Felix Wilhelm, Fermin J. Serna, Gabriel Campana
Kevin Hamacher and Ron Bowes of the Google Security Team for
finding this.
version 2.77
Generate an error when configured with a CNAME loop,
rather than a crash. Thanks to George Metz for
spotting this problem.
Calculate the length of TFTP error reply packet
correctly. This fixes a problem when the error
message in a TFTP packet exceeds the arbitrary
limit of 500 characters. The message was correctly
truncated, but not the packet length, so
extra data was appended. This is a possible
security risk, since the extra data comes from
a buffer which is also used for DNS, so that
previous DNS queries or replies may be leaked.
Thanks to Mozilla for funding the security audit
which spotted this bug.
Fix logic error in Linux netlink code. This could
cause dnsmasq to enter a tight loop on systems
with a very large number of network interfaces.
Thanks to Ivan Kokshaysky for the diagnosis and
patch.
Fix problem with --dnssec-timestamp whereby receipt
of SIGHUP would erroneously engage timestamp checking.
Thanks to Kevin Darbyshire-Bryant for this work.
Bump zone serial on reloading /etc/hosts and friends
when providing authoritative DNS. Thanks to Harrald
Dunkel for spotting this.
Handle v4-mapped IPv6 addresses sanely in --synth-domain.
These have standard representation like ::ffff:1.2.3.4
and are now converted to names like
<prefix>--ffff-1-2-3-4.<domain>
Handle binding upstream servers to an interface
(--server=1.2.3.4@eth0) when the named interface
is destroyed and recreated in the kernel. Thanks to
Beniamino Galvani for the patch.
Allow wildcard CNAME records in authoritative zones.
For example --cname=*.example.com,default.example.com
Thanks to Pro Backup for sponsoring this development.
Bump the allowed backlog of TCP connections from 5 to 32,
and make this a compile-time configurable option. Thanks
to Donatas Abraitis for diagnosing this as a potential
problem.
Add DNSMASQ_REQUESTED_OPTIONS environment variable to the
lease-change script. Thanks to ZHAO Yu for the patch.
Fix foobar in rrfilter code, that could cause malformed
replies, especially when DNSSEC validation on, and
the upstream server returns answer with the RRs in a
particular order. The only DNS server known to tickle
this is Nominum's. Thanks to Dave Täht for spotting the
bug and assisting in the fix.
Fix the manpage which lied that only the primary address
of an interface is used by --interface-name.
Make --localise-queries apply to names from --interface-name.
Thanks to Kevin Darbyshire-Bryant and Eric Luehrsen
for pushing this.
Improve connection handling when talking to TCP upstream
servers. Specifically, be prepared to open a new TCP
connection when we want to make multiple queries
but the upstream server accepts fewer queries per connection.
Improve logging of upstream servers when there are a lot