-
Notifications
You must be signed in to change notification settings - Fork 0
/
CHANGES
2103 lines (1478 loc) · 70.3 KB
/
CHANGES
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
####################### V 1.7.4.2:
Corrections:
The per address parameters for OpenSSL overlapped in memory with socket
parameters. Magically this did not seem to cause problems except on
MacOS Catalina that reported errors like:
socat[3458] E Select(7, &0x80, NULL, NULL, {140392884396544.000000}):
Invalid argument
Test: OPENSSL_PARA_OVERLAP
Thanks to Ryo Ota for reporting this bug.
Fixed a few minor coding issues
A VSOCK warning message was generated with all listening addresses
instead of only with VSOCK-LISTEN
When an OPENSSL-CONNECT client presented a certificate with IPv6
subject alternate name and the OPENSSL-LISTEN server had no commonname
option, the server crashed with SIGSEGV in xioip6_pton().
Test: OPENSSL_CLIENT_IP6_CN
Red Hat bug 1981308
Thanks to Vlad Slepukhin for reporting this issue and providing a patch
Corrected a typo in configure.ac that broke option --enable-openssl-base
Thanks to john1doe for reporting this issue.
Socat looped endlessly, not responding to SIGTERM, when a service name
(for port) could not be resolved.
Test: BAD_SERVICE
Using options of NAMED group, e.g.chown, with abstract UNIX domain
sockets, produced errors because the function was applied with a normal
file system related call, e.g.chown(), using file "" (empty name). Instead of
chown(), Socat now uses fchown() on the file descriptor. However, such
a call usually has no real effect.
Test: ABSTRACT_USER
Thanks to Andreas Fink for reporting this issue.
Option -R did not only dump ("sniff") right-to-left, but also
left-to-right traffic to the given file.
Test: SNIFF_RIGHT_TO_LEFT
Thanks to 1314 gsf for reporting this bug and sending a patch.
Options -r and -R, when opening a named pipe that has no actual reader,
failed with "No such device or address". To solve this problem, Socat
now opens the pipe in rw-Mode.
Thanks to Cody J.Soultz for sending a patch.
The call "socat -r - PIPE" traced to file ./- instead of issuing a
syntax error.
Print a message when readbytes option causes EOF
The ip-recverr option had no effect. Corrected and improved its
handling of ancilliary messages, so it is able to analyze ICMP error
packets (Linux only?)
Setgui(), Setuid() calls in xio-progcall.c were useless.
Testing:
Prevent the TIMESTAMP tests from sporadically failing due do seconds
overflow
Fixed in test.sh a few issues reported by shellcheck
Documentation:
Added missing docu of OpenSSL options min-proto-version,
max-proto-version.
Added missing closing parenthesis in socat.yo.
Thanks to Emanuele Torre for reporting this issue.
Corrected more typos and added missing bug info to CHANGES, performed
some non functional corrections.
Porting:
Corrected building when clock_gettime() not available, with or without
gettimeofday().
####################### V 1.7.4.1:
Corrections:
Socat 1.7.4.0 failed to compile especially on 32 bit systems.
Thanks to Wang Mingyu and others for sending a patch or reporting this
issue.
Under certain conditions OpenSSL stream connections, in particular bulk
data transfer in unidirectional mode, failed during transfer or near
its with Connection reset by peer on receiver side.
This happened with Socat versions 1.7.3.3 to 1.7.4.0. Reasons were
lazy SSL shutdown handling on the sender side in combination with
SSL_MODE_AUTO_RETRY turned off.
Fix: After SSH_shutdown but before socket shutdown call SSL_read()
Test: OPENSSL_STREAM_TO_SERVER
Fixes Red Hat issue 1870279.
####################### V 1.7.4.0:
Security:
Buffer size option (-b) is internally doubled for CR-CRLF conversion,
but not checked for integer overflow. This could lead to heap based
buffer overflow, assuming the attacker could provide this parameter.
Test: BLKSIZE_INT_OVERFL
Thanks to Lê Hiếu Bùi for reporting this issue and sending an
example exploit.
Corrections:
Socats address parser read over end of string when there were unbalanced
quotes
Test: UNBALANCED_QUOTE
Removed unused usleep() call from sycls.c
Unsetenv() was conditional in sysutils.c but not in xio-openssl.c thus
building failed on Solaris 9.
Thanks to Greg Earle for reporting this issue and providing a patch.
Mitigated race condition of quickly terminating SYSTEM or EXEC child
processes.
Option o-direct might require alignment of read/write buffer to, e.g.,
512 bytes, Socat now takes care of this when allocating the buffer.
With this fix read() succeeds, however, write() still might fail when
not writing complete pages.
Test: O_DIRECT
There was a race condition in the way Socat UDP-RECVFROM and similar
addresses with option fork prevents one packet from triggering
multiple processes. The symptom was that Socat master process seemed to
hang and did not process further packets. The fix makes use of
pselect() system call.
Thanks to Fulvio Scapin for reporting this issue.
UNIX domain client addresses applied file system entry options (group
NAMED) to the server socket instead of the client (bind) socket entry.
Tests: UNIX_SENDTO_UNLINK UNIX_CONNECT_UNLINK
Thanks to Nico Williams for reporting this major issue.
Length of single address options was limited to 511 bytes. This value
is now increased to 2047 bytes.
Change suggested by Mario Camou.
Addresses of type RECVFROM with option fork looped with an error
message in case that the second address failed before consuming the
packet. The fix makes RECVFROM drop the packet when the second address
failed before reading it. Use retry or forever option with the second
address if you want to avoid data loss.
Fixes Red Hat bug 1907718
Thanks to Chunmei Xu for reporting this issue and proving the patch.
Socats DTLS implementation has been reworked and appears to work now
reasonably over UDP.
New addresses: OPENSSL-DTLS-SERVER (DTLS-L),
OPENSSL-DTLS-CLIENT (DTLS)
Tests: OPENSSL_DTLS_CLIENT OPENSSL_DTLS_SERVER
OPENSSL_METHOD_DTLS1 OPENSSL_METHOD_DTLS1.2
Thanks to Brandon Carpenter, Qing Wan, and Pavel Nakonechnyi for
sending patches.
filan did not output the socket protocol.
filan -s assumed each stream socket to be TCP and each datagram socket
to be UDP. Now it uses SO_PROTOCOL and getprotoent() for correct output.
Help text showed two parameters for UDP4-RECVFROM address, but only
<port> is allowed.
Thanks to John the Scott for reporting this issue.
Error messages from SSL_read() and SSL_write() sometimes stated
SSL_connect instead of originating function name.
Fixed some more non functional minor issues.
Porting:
In gcc version 10 the default changed from -fcommon to -fno-common.
Consequently, linking filan and procan failed with error
"multiple definition of `deny_severity'" and `allow_severity'
Fixed by removing definitions in filan.c and procan.c
Debian issue 957823
Thanks to László Böszörményi and others for reporting this issue.
Solaris 9 does not provide strndup(); added substitute code.
Thanks to Greg Earle for providing a patch.
Added configure option --enable-openssl-base to specify the location of
a non-OS OpenSSL installation
There are systems whose kernel understands SCTP but getaddrinfo does
not. As workaround after EIA_SOCKTYPE on name and service resolution
fall back to ai_socktype=0; if it fails with EAI_SERVICE, set
ai_protocol=0 and try again
Test: SCTP_SERVICENAME
Per file filesystem options were still named ext2-* and depended on
<linux/ext2_fs.h>. Now they are called fs-* and depend on <linux/fs.h>.
These fs-* options are also available on old systems with ext2_fs.h
New options openssl-min-proto-version (min-version) and
openssl-max-proto-version (max-version) give access to the related
OpenSSL set-macros and substitute deprecated version-specific methods.
Test: OPENSSL_MIN_VERSION
With OpenSSL use OPENSSL_init_SSL when available, instead of deprecated
SSL_library_init.
With OPENSSL_API_COMPAT=0x10000000L the files openssl/dh.h, openssl/bn.h
must explicitely be included.
Thanks to Rosen Penev for reporting and sending a patch.
Testing:
test.sh now produces a list of tests that could not be performed for
any reason. This helps to analyse these cases.
OpenSSL s_server appearently started to neglect TCPs half close feature.
Test OPENSSL_TCP4 has been changed to tolerate this.
OpenSSL changed its behaviour when connection is rejected. Tests
OPENSSLCERTSERVER, OPENSSL_CN_CLIENT_SECURITY, and
OPENSSL_CN_SERVER_SECURITY now tolerate this.
OpenSSL no longer allows explicit renegotiation with TLSv1.3, thus the
appropriate tests failed.
Fix: use TLSv1.2 for renegotiation tests
Tests: OPENSSLRENEG1 OPENSSLRENEG2
Ubuntu 20.04 requires 2048 bit certificates with OpenSSL
Archlinux 2020 has not which command; its ip,ss commands have modified
version strings
More testing issues solved:
* ss to pipe might omit column separator
* UDP6MULTICAST_UNIDIR fails on newer Linux kernels
* do not use sort -V
* renamed testaddrs() to testfeats(), and introduced new testaddrs()
New features:
GOPEN and UNIX-CLIENT addresses now support sockets of type SEQPACKET.
Test: GOPENUNIXSEQPACKET
Feature suggested by vi0oss.
The generic setsockopt-int and related options are, in case of
listening/accepting addresses, applied to the connected socket(s). To enable
setting options on the listening socket, a new option setsockopt-listen
has been implemented. See the documentation for info on data types.
Tests: SETSOCKOPT SETSOCKOPT_LISTEN
Thanks to Steven Danna and Korian Edeline for reporting this issue.
Filan option -S gives short description like -s but with improved
format
Socat OpenSSL client, when server was specified using IP address, did
not verify connection on certificates SubjectAltName IP entries.
Tests: OPENSSL_SERVERALTAUTH OPENSSL_SERVERALTIP4AUTH OPENSSL_SERVERALTIP6AUTH
Fixes Red Hat bug 1805132
Added options -r and -R for raw dump of transferred data to files.
Test: OPTION_RAW_DUMP
Added option ip-transparent (socket option IP_TRANSPARENT)
Thanks to Wang Shanker for sending a patch.
OPENSSL-CONNECT now automatically uses the SNI feature, option
openssl-no-sni turns it off. Option openssl-snihost overrides the value
of option openssl-commonname or the server name.
Tests: OPENSSL_SNI OPENSSL_NO_SNI
Thanks to Travis Burtrum for providing the initial patch
New option accept-timeout (listen-timeout)
Test: ACCEPTTIMEOUT
Proposed by Roland
New option ip-add-source-membership
Feature inspired by Brian (b f31415)
INCOMPATIBLE CHANGE: Address UDP-DATAGRAM now does not check peerport
of replies, as it did up to version 1.7.3.4. Use option sourceport when
you need the old behaviour.
Test: UDP_DATAGRAM_SOURCEPORT
Feature inspired by Hans Bueckler for SSDP inquiry (for UPnP)
New option proxy-authorization-file reads PROXY-CONNECT credentials
from file and makes it possible to hide this data from the process
table.
Test: PROXYAUTHFILE
Thanks to Charles Stephens for sending an initial patch.
Added AF_VSOCK support with VSOCK-CONNECT and VSOCK-LISTEN addresses.
Developed by Stefano Garzarella.
Coding:
Added printf formats for uint16_t etc.
Documentation:
Address UDP-RECV does not support option fork.
Thanks to Fulvio Scapin for reporting that mistake in docu.
TUN address documentation showed TCP for backend which may merge
consecutive packets which causes data loss.
Thanks to Tomasz Lakota for reporting this issue.
####################### V 1.7.3.4:
Corrections:
Header of xiotermios_speed() declared parameter unsigned int instead of
speed_t, thus compiling failed on MacOS
Thanks to Joe Strout and others for reporting this bug.
Thanks to Andrew Childs and others for sending a patch.
Under certain circumstances, termios options of the first address were
applied to the second address, resulting in error
"Inappropriate ioctl for device"
This affected version 1.7.3.3 only.
Test: TERMIOS_PH_ALL
Thanks to Ivan J. for reporting this issue.
Socat failed to compile when no poll() system call was found by
configure.
Thanks to Jason White for sending a patch.
Due to use of SSL_CTX_clear_mode() Socat failed to compile on old
systems with, e.g., OpenSSL-0.9.8. Thanks to Simon Matter and Moritz B.
for reporting this problem and sending initial patches.
getaddrinfo() in IP4-SENDTO and IP6-SENDTO addresses failed with
"ai_socktype not supported" when protocol 6 was addressed.
The fix removes the possibility to use service names with SCTP.
Test: IP_SENDTO_6
Thanks to Sören for sending an initial patch.
Under certain circumstances, Socat printed the "socket ... is at EOF"
multiple times.
Test: MULTIPLE_EOF
Newer parts of test.sh used substitutions ${x,,*} or ${x^^*} that are
not implemented in older bash versions.
####################### V 1.7.3.3:
Corrections:
Makefile.in did not specify dependencies of filan on vsnprintf_r.o
and snprinterr.o
Added definition of FILAN_OBJS
Thanks to Craig Leres, Clayton Shotwell, and Chris Packham for
providing patches.
configure option --enable-msglevel did not work with numbers
The autoconf mechanism for determining SHIFT_OFFSET did not work when
cross compiling.
Thanks to Max Freisinger from Gentoo for sending a patch.
Socat still depended on obsolete gethostbyname() function, thus
compiling with MUSL libc failed.
Problem reported by Kennedy33.
The async signal safe diagnostic system used FDs 3 and 4 internally, so
use of appropriate fdin or fdout led to failures.
Test: DIAG_FDIN
Problem reported by Onur Sentürk.
The socket based mechanism for passing messages and signal information
from signal handler to process could reach and kill the wrong process.
Introduces functions diag_sock_pair(), diag_fork()
Thanks to Darren Zhao for analysing and reporting this problem.
Option ipv6-join-group did not work because it was applied in the wrong
phase
Test: UDP6MULTICAST_UNIDIR
Thanks to Angus Gratton for sending a patch.
Setting ispeed and ospeed failed for some serial devices because the
two settings were applied with two different get/set cycles, Thanks to
Alexandre Fenyo for providing an initial patch.
However, the actual fix is part of a conceptual change of the termios
module that aims for applying all changes in a single tcsetaddr call.
Fixes FreeBSD Bug 198441
Termios options TAB0,TAB1,TAB2,TAB3, and XTABS did not have an effect.
Thanks to Alan Walters for reporting this bug.
Substituted cumbersom ISPEED_OFFSET mechanism for cfsetispeed() calls
With TCP6-LISTEN and the other passive IPv6 addresses the range option
just failed: due to a bug in the syntax parser and two more bugs in
the xiocheckrange_ip6() function.
The syntax has now been changed from "[::1/128]" to "[::1]/128"!
Thanks Leah Neukirchen for sending an initial fix.
For name resolution Socat only checked the first character of the host
name to decide if it is an IPv4 address. This was not RFC conform. This
fix removes the possibility for use of IPv4 addresses with IPv6, e.g.
TCP6:127.0.0.1:80
Debian issue 695885
Thanks to Nicolas Fournil for reporting this issue.
Print a useful error message when single character options appear to be
merged in Socat invocation
Test: SOCAT_OPT_HINT
Fixed some docu typos.
Thanks to Travis Wellman, Thomas <tjps636>, Dan Kenigsberg,
Julian Zinn, and Simon Matter
Porting:
OpenSSL functions TLS1_client_method() and similar are
deprecated. Socat now uses recommended TLS_client_method(). The old
functions and dependend option openssl-method can still be
used when configuring socat with --enable-openssl-method
Shell scripts in socat distribution are now headed with:
#! /usr/bin/env bash
to make them better portable to systems without /bin/bash
Thanks to Maya Rashish for sending a patch
RES_AAONLY, RES_PRIMARY are deprecated. You can still enable them with
configure option --enable-res-deprecated.
New versions of OpenSSL preset SSL_MODE_AUTO_RETRY which may hang socat.
Solution: clear SSL_MODE_AUTO_RETRY when it is set.
Renamed configure.in to configure.ac and set an appropriate symlink for
older environments.
Related Gentoo bug 426262: Warning on configure.in
Thanks to Francesco Turco for reporting that warning.
Fixed new IPv6 range code for platforms without s6_addr32 component.
Testing:
test.sh: Show a warning when phase-1 (insecure phase) of a security
test fails
OpenSSL tests failed on actual Linux distributions. Measures:
Increased key lengths from 768 to 1024 bits
Added test.sh option -C to delete temp certs from prevsious runs
Provide DH-parameter in certificate in PEM
OpenSSL s_server option -verify 0 must be omitted
OpenSSL authentication method aNULL no longer works
Failure of cipher aNULL is not a failure
Failure of methods SSL3 and SSL23 is desired
test.sh depended on ifconfig and netstat utilities which are no longer
availabie in some distributions. test.sh now checks for and prefers
ip and ss.
Thanks to Ruediger Meier for reporting this problem.
More corrections to test.sh:
Language settings could still influence test results
netstat was still required
Suppress usleep deprecated messag
Force use of IPv4 with some certificates
Set timeout for UDPxMAXCHILDREN tests
Git:
Added missing Config/Makefile.DragonFly-2-8-2,
Config/config.DragonFly-2-8-2.h
Removed testcert.conf (to be generated by test.sh)
Cosmetics:
Simplified handling of missing termios defines.
New features:
Permit combined -d options as -dd etc.
porting:
ext2 options are now fs options.
####################### V 1.7.3.2:
corrections:
SIGSEGV and other signals could lead to a 100% CPU loop
Failing name resolution could lead to SIGSEGV
Thanks to Max for reporting this issue.
Include <stddef.h> for ptrdiff_t
Thanks to Jeroen Roovers for reporting this issue.
Building with --disable-sycls failed due to missing sslcls.h defines
Socat hung when configured with --disable-sycls.
Some minor corrections with includes etc.
Option so-reuseport did not work. Thanks to Some Raghavendra Prabhu
for sending a patch.
Programs invoked with EXEC, nofork, and -u or -U had stdin and stdout
incorrectly assigned
Test: EXEC_NOFORK_UNIDIR
Thanks to David Reiss for reporting this problem.
Socat exited with status 0 even when a program invoked with SYSTEM or
EXEC failed.
Tests: SYSTEM_RC EXEC_RC
Issue reported by Felix Winkelmann.
AddressSanitizer reported a few buffer overflows (false positives).
Nevertheless fixed Socat source.
Issue reported by Hanno Böck.
Socat did not use option ipv6-join-group.
Test: USE_IPV6_JOIN_GROUP
Thanks to Linus Lüssing for sending a patch.
UDP-LISTEN did not honor the max-children option.
Test: UDP4MAXCHILDREN UDP6MAXCHILDREN
Thanks to Leander Berwers for reporting this issue.
Options so-rcvtimeo and so-sndtimeo do not work with poll()/select()
and therefore were useless.
Thanks to Steve Borenstein for reporting this issue.
Option dhparam was documented as dhparams. Added the alias name
dhparams to fix this.
Thanks to Alexander Neumann for sending a patch.
Options shut-down and shut-close did not work.
Thanks to Stefan Schimanski for providing a patch.
There was a bug in printing readline log message caused by a misleading
indentation.
Thanks to Paul Wouters for reporting.
The internal vsnprintf_r function looped or crashed on size parameter
with hexadecimal output.
Ignore exit code of child process when it was killed by master due to
EOF
Corrected byte order on read of IPV6_TCLASS value from ancillary
message
Fixed type of the bool element in options. This had bug caused failures
e.g. of ignoreeof on big-endian systems when bool was not based on int.
On systems with predefined bool type whose size differs from int some
IPv6 and TCP options (per setsockopt()) failed.
Length of integral data in ancillary messages varies (TOS: 1 byte,
TTL: 4 bytes), the old implementation failed for TTL on big-endian
hosts.
Fixed an issue in options processing: TUN and DNS flags had failed on
big-endian systems and the NO- forms had probable never worked.
porting:
Type conflict between int and sig_atomic_t between declaration and
definition of diag_immediate_type and diag_immediate_exit broke
compilation on FreeBSD 10.1 with clang. Thanks to Emanuel Haupt for
reporting this bug.
Socat failed to compile on platforms with OpenSSL without
DTLSv1_client_method or DTLSv1_server_method.
Thanks to Simon Matter for sending a patch.
NuttX OS headers do not provide struct ip, thus socat did not compile.
Made struct ip subject to configure.
Thanks to SP for reporting this issue.
Socat failed to compile with OpenSSL version 1.0.2d where
SSLv3_server_method and SSLv3_client_method are no longer defined.
Thanks to Mischa ter Smitten for reporting this issue and providing
a patch.
configure checked for OpenSSL EC_KEY assuming it is a define but it
is a type, thus OpenSSL ECDHE ciphers failed even on Linux.
Thanks to Andrey Arapov for reporting this bug.
Changes to make socat compile with OpenSSL 1.1.
Thanks to Sebastian Andrzej Siewior e.a. from the Debian team for
providing the base patch.
Debian Bug#828550
Make Socat compatible with BoringSSL.
Thanks to Matt Braithwaite for providing a patch.
OpenSSL: Use RAND_status to determine PRNG state
Thanks to Adam Langley for providing a patch
AIX-7 uses an extended O_ACCMODE that does not fit socat's internal
requirements. Thanks to Garrick Trowsdale for providing a patch
LibreSSL support: check for OPENSSL_NO_COMP
Thanks to Bernard Spil for providing a patch
testing:
socks4echo.sh and socks4a-echo.sh hung with new bash with read -n
test.sh: stderr; option -v (verbose); FDOUT_ERROR description
improved proxy.sh - it now also takes hostnames
A few corrections in test.sh
DTLS1 test hangs on some distributions. Test is now only performed
with OpenSSL 1.0.2 or higher.
More corrections to test.sh that reveal a mistake with IPV6_TCLASS
docu:
Corrected source of socat man page to correctly show man references
like socket(2); removed obseolete entries from See Also
Docu and some comments mentioned addresses SSL-LISTEN and SSL-CONNECT
that do not exist (OPENSSL-LISTEN, SSL-L; and OPENNSSL-CONNECT, SSL
are correct).
Thanks to Zhigang Wang for reporting this issue.
Fixed a couple of English spelling and grammar mistakes.
Thanks to Jakub Wild for sending the patches.
NOEXPAND() was not resolved 2 times.
More minor docu corrections
legal:
Added contributors to copyright notices. Suggested by Matt Braithwaite.
####################### V 1.7.3.1:
security:
Socat security advisory 8
A stack overflow in vulnerability was found that can be triggered when
command line arguments (complete address specifications, host names,
file names) are longer than 512 bytes.
Successful exploitation might allow an attacker to execute arbitrary
code with the privileges of the socat process.
This vulnerability can only be exploited when an attacker is able to
inject data into socat's command line.
A vulnerable scenario would be a CGI script that reads data from clients
and uses (parts of) this data as hostname for a Socat invocation.
Test: NESTEDOVFL
Credits to Takumi Akiyama for finding and reporting this issue.
Socat security advisory 7
MSVR-1499
In the OpenSSL address implementation the hard coded 1024 bit DH p
parameter was not prime. The effective cryptographic strength of a key
exchange using these parameters was weaker than the one one could get by
using a prime p. Moreover, since there is no indication of how these
parameters were chosen, the existence of a trapdoor that makes possible
for an eavesdropper to recover the shared secret from a key exchange
that uses them cannot be ruled out.
Futhermore, 1024bit is not considered sufficiently secure.
Fix: generated a new 2048bit prime.
Thanks to Santiago Zanella-Beguelin and Microsoft Vulnerability
Research (MSVR) for finding and reporting this issue.
####################### V 1.7.3.0:
security:
Socat security advisory 6
CVE-2015-1379: Possible DoS with fork
Fixed problems with signal handling caused by use of not async signal
safe functions in signal handlers that could freeze socat, allowing
denial of service attacks.
Many changes in signal handling and the diagnostic messages system were
applied to make the code async signal safe but still provide detailled
logging from signal handlers:
Coded function vsnprintf_r() as async signal safe incomplete substitute
of libc vsnprintf()
Coded function snprinterr() to replace %m in strings with a system error
message
Instead of gettimeofday() use clock_gettime() when available
Pass Diagnostic messages from signal handler per unix socket to the main
program flow
Use sigaction() instead of signal() for better control
Turn off nested signal handler invocations
Thanks to Peter Lobsinger for reporting and explaining this issue.
Red Hat issue 1019975: add TLS host name checks
OpenSSL client checks if the server certificates names in
extensions/subjectAltName/DNS or in subject/commonName match the name
used to connect or the value of the openssl-commonname option.
Test: OPENSSL_CN_CLIENT_SECURITY
OpenSSL server checks if the client certificates names in
extensions/subjectAltNames/DNS or subject/commonName match the value of
the openssl-commonname option when it is used.
Test: OPENSSL_CN_SERVER_SECURITY
Red Hat issue 1019964: socat now uses the system certificate store with
OPENSSL when neither options cafile nor capath are used
Red Hat issue 1019972: needs to specify OpenSSL cipher suites
Default cipherlist is now "HIGH:-NULL:-PSK:-aNULL" instead of empty to
prevent downgrade attacks
new features:
OpenSSL addresses set couple of environment variables from values in
peer certificate, e.g.:
SOCAT_OPENSSL_X509_SUBJECT, SOCAT_OPENSSL_X509_ISSUER,
SOCAT_OPENSSL_X509_COMMONNAME,
SOCAT_OPENSSL_X509V3_SUBJECTALTNAME_DNS
Tests: ENV_OPENSSL_{CLIENT,SERVER}_X509_*
Added support for methods TLSv1, TLSv1.1, TLSv1.2, and DTLS1
Tests: OPENSSL_METHOD_*
Enabled OpenSSL server side use of ECDHE ciphers. Feature suggested
by Andrey Arapov.
Added a new option termios-rawer for ptys.
Thanks to Christian Vogelgsang for pointing me to this requirement
corrections:
Bind with ABSTRACT commands used non-abstract namespace (Linux).
Test: ABSTRACT_BIND
Thanks to Denis Shatov for reporting this bug.
Fixed return value of nestlex()
Option ignoreeof on the right address hung.
Test: IGNOREEOF_REV
Thanks to Franz Fasching for reporting this bug.
Address SYSTEM, when terminating, shut down its parent addresses,
e.g. an SSL connection which the parent assumed to still be active.
Test: SYSTEM_SHUTDOWN
Passive (listening or receiving) addresses with empty port field bound
to a random port instead of terminating with error.
Test: TCP4_NOPORT
configure with some combination of disable options produced config
files that failed to compile due to missing IPPROTO_TCP.
Thanks to Thierry Fournier for report and patch.
fixed a few minor bugs with OpenSSL in configure and with messages
Socat did not work in FIPS mode because 1024 instead of 512 bit DH prime
is required. Thanks to Zhigang Wang for reporting and sending a patch.
Christophe Leroy provided a patch that fixes memory leaks reported by
valgrind
Help for filan -L was bad, is now corrected to:
"follow symbolic links instead of showing their properties"
Address options fdin and fdout were silently ignored when not applicable
due to -u or -U option. Now these combinations are caught as errors.
Test: FDOUT_ERROR
Issue reported by Hendrik.
Added option termios-cfmakeraw that calls cfmakeraw() and is preferred
over option raw which is now obsolote. On SysV systems this call is
simulated by appropriate setting.
Thanks to Youfu Zhang for reporting issue with option raw.
porting:
Socat included <sys/poll.h> instead of POSIX <poll.h>
Thanks to John Spencer for reporting this issue.
Version 1.7.2.4 changed the check for gcc in configure.ac; this
broke cross compiling. The particular check gets reverted.
Thanks to Ross Burton and Danomi Manchego for reporting this issue.
Debian Bug#764251: Set the build timestamp to a deterministic time:
support external BUILD_DATE env var to allow to build reproducable
binaries
Joachim Fenkes provided an new adapted spec file.
Type bool and macros Min and Max are defined by socat which led to
compile errors when they were already provided by build framework.
Thanks to Liyu Liu for providing a patch.
David Arnstein contributed a patch for NetBSD 5.1 including stdbool.h
support and appropriate files in Config/
Lauri Tirkkonen contributed a patch regarding netinet/if_ether.h
on Illumos
Changes for Openindiana: define _XPG4_2, __EXTENSIONS__,
_POSIX_PTHREAD_SEMANTICS; and minor changes
Red Hat issue 1182005: socat 1.7.2.4 build failure missing
linux/errqueue.h
Socat failed to compile on on PPC due to new requirements for
including <linux/errqueue.h> and a weakness in the conditional code.
Thanks to Michel Normand for reporting this issue.
doc:
In the man page the PTY example was badly formatted. Thanks to
J.F.Sebastian for sending a patch.
Added missing CVE ids to security issues in CHANGES
testing:
Do not distribute testcert.conf with socat source but generate it
(and new testcert6.conf) during test.sh run.
####################### V 1.7.2.4:
corrections:
LISTEN based addresses applied some address options, e.g. so-keepalive,
to the listening file descriptor instead of the connected file
descriptor
Thanks to Ulises Alonso for reporting this bug
make failed after configure with non gcc compiler due to missing
include. Thanks to Horacio Mijail for reporting this problem
configure checked for --disable-rawsocket but printed
--disable-genericsocket in the help text. Thanks to Ben Gardiner for
reporting and patching this bug
In xioshutdown() a wrong branch was chosen after RECVFROM type addresses.
Probably no impact.
Thanks to David Binderman for reporting this issue.
procan could not cleanly format ulimit values longer than 16 decimal
digits. Thanks to Frank Dana for providing a patch that increases field
width to 24 digits.
OPENSSL-CONNECT with bind option failed on some systems, eg.FreeBSD, with
"Invalid argument"
Thanks to Emile den Tex for reporting this bug.
Changed some variable definitions to make gcc -O2 aliasing checker happy
Thanks to Ilya Gordeev for reporting these warnings
On big endian platforms with type long >32bit the range option applied a
bad base address. Thanks to hejia hejia for reporting and fixing this bug.
Red Hat issue 1022070: missing length check in xiolog_ancillary_socket()
Red Hat issue 1022063: out-of-range shifts on net mask bits
Red Hat issue 1022062: strcpy misuse in xiosetsockaddrenv_ip4()
Red Hat issue 1022048: strncpy hardening: corrected suspicious strncpy()
uses
Red Hat issue 1021958: fixed a bug with faulty buffer/data length
calculation in xio-ascii.c:_xiodump()
Red Hat issue 1021972: fixed a missing NUL termination in return string
of sysutils.c:sockaddr_info() for the AF_UNIX case
fixed some typos and minor issues, including:
Red Hat issue 1021967: formatting error in manual page
UNIX-LISTEN with fork option did not remove the socket file system entry
when exiting. Other file system based passive address types had similar
issues or failed to apply options umask, user e.a.
Thanks to Lorenzo Monti for pointing me to this issue
porting:
Red Hat issue 1020203: configure checks fail with some compilers.
Use case: clang
Performed changes for Fedora release 19
Adapted, improved test.sh script
Red Hat issue 1021429: getgroupent fails with large number of groups;
use getgrouplist() when available instead of sequence of calls to
getgrent()
Red Hat issue 1021948: snprintf API change;
Implemented xio_snprintf() function as wrapper that tries to emulate C99
behaviour on old glibc systems, and adapted all affected calls
appropriately
Mike Frysinger provided a patch that supports long long for time_t,
socklen_t and a few other libc types.
Artem Mygaiev extended Cedril Priscals Android build script with pty code
The check for fips.h required stddef.h
Thanks to Matt Hilt for reporting this issue and sending a patch
Check for linux/errqueue.h failed on some systems due to lack of
linux/types.h inclusion. Thanks to Michael Vastola for sending a patch.
autoconf now prefers configure.ac over configure.in
Thanks to Michael Vastola for sending a patch.
type of struct cmsghdr.cmsg is system dependend, determine it with
configure; some more print format corrections
docu:
libwrap always logs to syslog
added actual text version of GPLv2
####################### V 1.7.2.3:
security:
Socat security advisory 5
CVE-2014-0019: socats PROXY-CONNECT address was vulnerable to a buffer
overflow with data from command line (see socat-secadv5.txt)
Credits to Florian Weimer of the Red Hat Product Security Team
####################### V 1.7.2.2:
security:
Socat security advisory 4
CVE-2013-3571:
after refusing a client connection due to bad source address or source
port socat shutdown() the socket but did not close() it, resulting in
a file descriptor leak in the listening process, visible with lsof and
possibly resulting in EMFILE Too many open files. This issue could be
misused for a denial of service attack.
Full credits to Catalin Mitrofan for finding and reporting this issue.
####################### V 1.7.2.1:
security:
Socat security advisory 3
CVE-2012-0219:
fixed a possible heap buffer overflow in the readline address. This bug
could be exploited when all of the following conditions were met:
1) one of the addresses is READLINE without the noprompt and without the
prompt options.
2) the other (almost arbitrary address) reads malicious data (which is
then transferred by socat to READLINE).
Workaround: when using the READLINE address apply option prompt or
noprompt.
Full credits to Johan Thillemann for finding and reporting this issue.
####################### V 1.7.2.0:
corrections:
when UNIX-LISTEN was applied to an existing file it failed as expected
but removed the file. Thanks to Bjoern Bosselmann for reporting this
problem
fixed a bug where socat might crash when connecting to a unix domain
socket using address GOPEN. Thanks to Martin Forssen for bug report and
patch.
UDP-LISTEN would alway set SO_REUSEADDR even without fork option and
when user set it to 0. Thanks to Michal Svoboda for reporting this bug.
UNIX-CONNECT did not support half-close. Thanks to Greg Hughes who
pointed me to that bug
TCP-CONNECT with option nonblock reported successful connect even when
it was still pending
address option ioctl-intp failed with "unimplemented type 26". Thanks
to Jeremy W. Sherman for reporting and fixing that bug
socat option -x did not print packet direction, timestamp etc; thanks
to Anthony Sharobaiko for sending a patch
address PTY does not take any parameters but did not report an error
when some were given
Marcus Meissner provided a patch that fixes invalid output and possible
process crash when socat prints info about an unnamed unix domain
socket
Michal Soltys reported the following problem and provided an initial
patch: when socat was interrupted, e.g. by SIGSTOP, and resumed during
data transfer only parts of the data might have been written.
Option o-nonblock in combination with large transfer block sizes
may result in partial writes and/or EAGAIN errors that were not handled
properly but resulted in data loss or process termination.
Fixed a bug that could freeze socat when during assembly of a log
message a signal was handled that also printed a log message. socat
development had been aware that localtime() is not thread safe but had
only expected broken messages, not corrupted stack (glibc 2.11.1,
Ubuntu 10.4)
an internal store for child pids was susceptible to pid reuse which
could lead to sporadic data loss when both fork option and exec address
were used. Thanks to Tetsuya Sodo for reporting this problem and
sending a patch
OpenSSL server failed with "no shared cipher" when using cipher aNULL.
Fixed by providing temporary DH parameters. Thanks to Philip Rowlands
for drawing my attention to this issue.
UDP-LISTEN slept 1s after accepting a connection. This is not required.
Thanks to Peter Valdemar Morch for reporting this issue
fixed a bug that could lead to error or socat crash after a client
connection with option retry had been established
fixed configure.in bug on net/if.h check that caused IF_NAMESIZE to be
undefined
improved dev_t print format definition
porting:
Cedril Priscal ported socat to Android (using Googles cross compiler).
The port includes the socat_buildscript_for_android.sh script
added check for component ipi_spec_dst in struct in_pktinfo so
compilation does not fail on Cygwin (thanks to Peter Wagemans for
reporting this problem)
build failed on RHEL6 due to presence of fips.h; configure now checks
for fipsld too. Thanks to Andreas Gruenbacher for reporting this