5 This is a high-level summary of the most important changes.
6 For a full list of changes, see the git commit log; for example,
7 https://github.com/openssl/openssl/commits/ and pick the appropriate
10 Changes between 1.1.0f and 1.1.1 [xx XXX xxxx]
12 *) The last traces of Netware support, first removed in 1.1.0, have
16 *) Get rid of Makefile.shared, and in the process, make the processing
17 of certain files (rc.obj, or the .def/.map/.opt files produced from
18 the ordinal files) more visible and hopefully easier to trace and
19 debug (or make silent).
22 *) Make it possible to have environment variable assignments as
23 arguments to config / Configure.
26 *) Add multi-prime RSA (RFC 8017) support.
29 *) Add SM3 implemented according to GB/T 32905-2016
30 [ Jack Lloyd <jack.lloyd@ribose.com>,
31 Ronald Tse <ronald.tse@ribose.com>,
32 Erick Borsboom <erick.borsboom@ribose.com> ]
34 *) Add 'Maximum Fragment Length' TLS extension negotiation and support
35 as documented in RFC6066.
36 Based on a patch from Tomasz Moń
37 [Filipe Raimundo da Silva]
39 *) Add SM4 implemented according to GB/T 32907-2016.
40 [ Jack Lloyd <jack.lloyd@ribose.com>,
41 Ronald Tse <ronald.tse@ribose.com>,
42 Erick Borsboom <erick.borsboom@ribose.com> ]
44 *) Reimplement -newreq-nodes and ERR_error_string_n; the
45 original author does not agree with the license change.
48 *) Add ARIA AEAD TLS support.
51 *) Some macro definitions to support VS6 have been removed. Visual
52 Studio 6 has not worked since 1.1.0
55 *) Add ERR_clear_last_mark(), to allow callers to clear the last mark
56 without clearing the errors.
59 *) Add "atfork" functions. If building on a system that without
60 pthreads, see doc/man3/OPENSSL_fork_prepare.pod for application
61 requirements. The RAND facility now uses/requires this.
67 *) The UI API becomes a permanent and integral part of libcrypto, i.e.
68 not possible to disable entirely. However, it's still possible to
69 disable the console reading UI method, UI_OpenSSL() (use UI_null()
72 To disable, configure with 'no-ui-console'. 'no-ui' is still
73 possible to use as an alias. Check at compile time with the
74 macro OPENSSL_NO_UI_CONSOLE. The macro OPENSSL_NO_UI is still
75 possible to check and is an alias for OPENSSL_NO_UI_CONSOLE.
78 *) Add a STORE module, which implements a uniform and URI based reader of
79 stores that can contain keys, certificates, CRLs and numerous other
80 objects. The main API is loosely based on a few stdio functions,
81 and includes OSSL_STORE_open, OSSL_STORE_load, OSSL_STORE_eof,
82 OSSL_STORE_error and OSSL_STORE_close.
83 The implementation uses backends called "loaders" to implement arbitrary
84 URI schemes. There is one built in "loader" for the 'file' scheme.
87 *) Add devcrypto engine. This has been implemented against cryptodev-linux,
88 then adjusted to work on FreeBSD 8.4 as well.
89 Enable by configuring with 'enable-devcryptoeng'. This is done by default
90 on BSD implementations, as cryptodev.h is assumed to exist on all of them.
93 *) Module names can prefixed with OSSL_ or OPENSSL_. This affects
94 util/mkerr.pl, which is adapted to allow those prefixes, leading to
95 error code calls like this:
97 OSSL_FOOerr(OSSL_FOO_F_SOMETHING, OSSL_FOO_R_WHATEVER);
99 With this change, we claim the namespaces OSSL and OPENSSL in a manner
100 that can be encoded in C. For the foreseeable future, this will only
102 [Richard Levitte and Tim Hudson]
104 *) Removed BSD cryptodev engine.
107 *) Add a build target 'build_all_generated', to build all generated files
108 and only that. This can be used to prepare everything that requires
109 things like perl for a system that lacks perl and then move everything
110 to that system and do the rest of the build there.
113 *) In the UI interface, make it possible to duplicate the user data. This
114 can be used by engines that need to retain the data for a longer time
115 than just the call where this user data is passed.
118 *) Ignore the '-named_curve auto' value for compatibility of applications
120 [Tomas Mraz <tmraz@fedoraproject.org>]
122 *) Fragmented SSL/TLS alerts are no longer accepted. An alert message is 2
123 bytes long. In theory it is permissible in SSLv3 - TLSv1.2 to fragment such
124 alerts across multiple records (some of which could be empty). In practice
125 it make no sense to send an empty alert record, or to fragment one. TLSv1.3
126 prohibts this altogether and other libraries (BoringSSL, NSS) do not
127 support this at all. Supporting it adds significant complexity to the
128 record layer, and its removal is unlikely to cause inter-operability
132 *) Add the ASN.1 types INT32, UINT32, INT64, UINT64 and variants prefixed
133 with Z. These are meant to replace LONG and ZLONG and to be size safe.
134 The use of LONG and ZLONG is discouraged and scheduled for deprecation
138 *) Add the 'z' and 'j' modifiers to BIO_printf() et al formatting string,
139 'z' is to be used for [s]size_t, and 'j' - with [u]int64_t.
140 [Richard Levitte, Andy Polyakov]
142 *) Add EC_KEY_get0_engine(), which does for EC_KEY what RSA_get0_engine()
146 *) Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target
147 platform rather than 'mingw'.
150 *) The functions X509_STORE_add_cert and X509_STORE_add_crl return
151 success if they are asked to add an object which already exists
152 in the store. This change cascades to other functions which load
153 certificates and CRLs.
156 *) x86_64 assembly pack: annotate code with DWARF CFI directives to
157 facilitate stack unwinding even from assembly subroutines.
160 *) Remove VAX C specific definitions of OPENSSL_EXPORT, OPENSSL_EXTERN.
161 Also remove OPENSSL_GLOBAL entirely, as it became a no-op.
164 *) Remove the VMS-specific reimplementation of gmtime from crypto/o_times.c.
165 VMS C's RTL has a fully up to date gmtime() and gmtime_r() since V7.1,
166 which is the minimum version we support.
169 *) Certificate time validation (X509_cmp_time) enforces stricter
170 compliance with RFC 5280. Fractional seconds and timezone offsets
171 are no longer allowed.
174 *) Add support for ARIA
177 *) s_client will now send the Server Name Indication (SNI) extension by
178 default unless the new "-noservername" option is used. The server name is
179 based on the host provided to the "-connect" option unless overridden by
183 *) Add support for SipHash
186 *) OpenSSL now fails if it receives an unrecognised record type in TLS1.0
187 or TLS1.1. Previously this only happened in SSLv3 and TLS1.2. This is to
188 prevent issues where no progress is being made and the peer continually
189 sends unrecognised record types, using up resources processing them.
192 *) 'openssl passwd' can now produce SHA256 and SHA512 based output,
193 using the algorithm defined in
194 https://www.akkadia.org/drepper/SHA-crypt.txt
197 *) Heartbeat support has been removed; the ABI is changed for now.
198 [Richard Levitte, Rich Salz]
200 *) Support for SSL_OP_NO_ENCRYPT_THEN_MAC in SSL_CONF_cmd.
203 *) The RSA "null" method, which was partially supported to avoid patent
204 issues, has been replaced to always returns NULL.
207 Changes between 1.1.0g and 1.1.0h [xx XXX xxxx]
209 *) Removed the OS390-Unix config target. It relied on a script that doesn't
213 *) rsaz_1024_mul_avx2 overflow bug on x86_64
215 There is an overflow bug in the AVX2 Montgomery multiplication procedure
216 used in exponentiation with 1024-bit moduli. No EC algorithms are affected.
217 Analysis suggests that attacks against RSA and DSA as a result of this
218 defect would be very difficult to perform and are not believed likely.
219 Attacks against DH1024 are considered just feasible, because most of the
220 work necessary to deduce information about a private key may be performed
221 offline. The amount of resources required for such an attack would be
222 significant. However, for an attack on TLS to be meaningful, the server
223 would have to share the DH1024 private key among multiple clients, which is
224 no longer an option since CVE-2016-0701.
226 This only affects processors that support the AVX2 but not ADX extensions
227 like Intel Haswell (4th generation).
229 This issue was reported to OpenSSL by David Benjamin (Google). The issue
230 was originally found via the OSS-Fuzz project.
234 Changes between 1.1.0f and 1.1.0g [2 Nov 2017]
236 *) bn_sqrx8x_internal carry bug on x86_64
238 There is a carry propagating bug in the x86_64 Montgomery squaring
239 procedure. No EC algorithms are affected. Analysis suggests that attacks
240 against RSA and DSA as a result of this defect would be very difficult to
241 perform and are not believed likely. Attacks against DH are considered just
242 feasible (although very difficult) because most of the work necessary to
243 deduce information about a private key may be performed offline. The amount
244 of resources required for such an attack would be very significant and
245 likely only accessible to a limited number of attackers. An attacker would
246 additionally need online access to an unpatched system using the target
247 private key in a scenario with persistent DH parameters and a private
248 key that is shared between multiple clients.
250 This only affects processors that support the BMI1, BMI2 and ADX extensions
251 like Intel Broadwell (5th generation) and later or AMD Ryzen.
253 This issue was reported to OpenSSL by the OSS-Fuzz project.
257 *) Malformed X.509 IPAddressFamily could cause OOB read
259 If an X.509 certificate has a malformed IPAddressFamily extension,
260 OpenSSL could do a one-byte buffer overread. The most likely result
261 would be an erroneous display of the certificate in text format.
263 This issue was reported to OpenSSL by the OSS-Fuzz project.
267 Changes between 1.1.0e and 1.1.0f [25 May 2017]
269 *) Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target
270 platform rather than 'mingw'.
273 *) Remove the VMS-specific reimplementation of gmtime from crypto/o_times.c.
274 VMS C's RTL has a fully up to date gmtime() and gmtime_r() since V7.1,
275 which is the minimum version we support.
278 Changes between 1.1.0d and 1.1.0e [16 Feb 2017]
280 *) Encrypt-Then-Mac renegotiation crash
282 During a renegotiation handshake if the Encrypt-Then-Mac extension is
283 negotiated where it was not in the original handshake (or vice-versa) then
284 this can cause OpenSSL to crash (dependant on ciphersuite). Both clients
285 and servers are affected.
287 This issue was reported to OpenSSL by Joe Orton (Red Hat).
291 Changes between 1.1.0c and 1.1.0d [26 Jan 2017]
293 *) Truncated packet could crash via OOB read
295 If one side of an SSL/TLS path is running on a 32-bit host and a specific
296 cipher is being used, then a truncated packet can cause that host to
297 perform an out-of-bounds read, usually resulting in a crash.
299 This issue was reported to OpenSSL by Robert Święcki of Google.
303 *) Bad (EC)DHE parameters cause a client crash
305 If a malicious server supplies bad parameters for a DHE or ECDHE key
306 exchange then this can result in the client attempting to dereference a
307 NULL pointer leading to a client crash. This could be exploited in a Denial
310 This issue was reported to OpenSSL by Guido Vranken.
314 *) BN_mod_exp may produce incorrect results on x86_64
316 There is a carry propagating bug in the x86_64 Montgomery squaring
317 procedure. No EC algorithms are affected. Analysis suggests that attacks
318 against RSA and DSA as a result of this defect would be very difficult to
319 perform and are not believed likely. Attacks against DH are considered just
320 feasible (although very difficult) because most of the work necessary to
321 deduce information about a private key may be performed offline. The amount
322 of resources required for such an attack would be very significant and
323 likely only accessible to a limited number of attackers. An attacker would
324 additionally need online access to an unpatched system using the target
325 private key in a scenario with persistent DH parameters and a private
326 key that is shared between multiple clients. For example this can occur by
327 default in OpenSSL DHE based SSL/TLS ciphersuites. Note: This issue is very
328 similar to CVE-2015-3193 but must be treated as a separate problem.
330 This issue was reported to OpenSSL by the OSS-Fuzz project.
334 Changes between 1.1.0b and 1.1.0c [10 Nov 2016]
336 *) ChaCha20/Poly1305 heap-buffer-overflow
338 TLS connections using *-CHACHA20-POLY1305 ciphersuites are susceptible to
339 a DoS attack by corrupting larger payloads. This can result in an OpenSSL
340 crash. This issue is not considered to be exploitable beyond a DoS.
342 This issue was reported to OpenSSL by Robert Święcki (Google Security Team)
346 *) CMS Null dereference
348 Applications parsing invalid CMS structures can crash with a NULL pointer
349 dereference. This is caused by a bug in the handling of the ASN.1 CHOICE
350 type in OpenSSL 1.1.0 which can result in a NULL value being passed to the
351 structure callback if an attempt is made to free certain invalid encodings.
352 Only CHOICE structures using a callback which do not handle NULL value are
355 This issue was reported to OpenSSL by Tyler Nighswander of ForAllSecure.
359 *) Montgomery multiplication may produce incorrect results
361 There is a carry propagating bug in the Broadwell-specific Montgomery
362 multiplication procedure that handles input lengths divisible by, but
363 longer than 256 bits. Analysis suggests that attacks against RSA, DSA
364 and DH private keys are impossible. This is because the subroutine in
365 question is not used in operations with the private key itself and an input
366 of the attacker's direct choice. Otherwise the bug can manifest itself as
367 transient authentication and key negotiation failures or reproducible
368 erroneous outcome of public-key operations with specially crafted input.
369 Among EC algorithms only Brainpool P-512 curves are affected and one
370 presumably can attack ECDH key negotiation. Impact was not analyzed in
371 detail, because pre-requisites for attack are considered unlikely. Namely
372 multiple clients have to choose the curve in question and the server has to
373 share the private key among them, neither of which is default behaviour.
374 Even then only clients that chose the curve will be affected.
376 This issue was publicly reported as transient failures and was not
377 initially recognized as a security issue. Thanks to Richard Morgan for
378 providing reproducible case.
382 *) Removed automatic addition of RPATH in shared libraries and executables,
383 as this was a remainder from OpenSSL 1.0.x and isn't needed any more.
386 Changes between 1.1.0a and 1.1.0b [26 Sep 2016]
388 *) Fix Use After Free for large message sizes
390 The patch applied to address CVE-2016-6307 resulted in an issue where if a
391 message larger than approx 16k is received then the underlying buffer to
392 store the incoming message is reallocated and moved. Unfortunately a
393 dangling pointer to the old location is left which results in an attempt to
394 write to the previously freed location. This is likely to result in a
395 crash, however it could potentially lead to execution of arbitrary code.
397 This issue only affects OpenSSL 1.1.0a.
399 This issue was reported to OpenSSL by Robert Święcki.
403 Changes between 1.1.0 and 1.1.0a [22 Sep 2016]
405 *) OCSP Status Request extension unbounded memory growth
407 A malicious client can send an excessively large OCSP Status Request
408 extension. If that client continually requests renegotiation, sending a
409 large OCSP Status Request extension each time, then there will be unbounded
410 memory growth on the server. This will eventually lead to a Denial Of
411 Service attack through memory exhaustion. Servers with a default
412 configuration are vulnerable even if they do not support OCSP. Builds using
413 the "no-ocsp" build time option are not affected.
415 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
419 *) SSL_peek() hang on empty record
421 OpenSSL 1.1.0 SSL/TLS will hang during a call to SSL_peek() if the peer
422 sends an empty record. This could be exploited by a malicious peer in a
423 Denial Of Service attack.
425 This issue was reported to OpenSSL by Alex Gaynor.
429 *) Excessive allocation of memory in tls_get_message_header() and
430 dtls1_preprocess_fragment()
432 A (D)TLS message includes 3 bytes for its length in the header for the
433 message. This would allow for messages up to 16Mb in length. Messages of
434 this length are excessive and OpenSSL includes a check to ensure that a
435 peer is sending reasonably sized messages in order to avoid too much memory
436 being consumed to service a connection. A flaw in the logic of version
437 1.1.0 means that memory for the message is allocated too early, prior to
438 the excessive message length check. Due to way memory is allocated in
439 OpenSSL this could mean an attacker could force up to 21Mb to be allocated
440 to service a connection. This could lead to a Denial of Service through
441 memory exhaustion. However, the excessive message length check still takes
442 place, and this would cause the connection to immediately fail. Assuming
443 that the application calls SSL_free() on the failed connection in a timely
444 manner then the 21Mb of allocated memory will then be immediately freed
445 again. Therefore the excessive memory allocation will be transitory in
446 nature. This then means that there is only a security impact if:
448 1) The application does not call SSL_free() in a timely manner in the event
449 that the connection fails
451 2) The application is working in a constrained environment where there is
452 very little free memory
454 3) The attacker initiates multiple connection attempts such that there are
455 multiple connections in a state where memory has been allocated for the
456 connection; SSL_free() has not yet been called; and there is insufficient
457 memory to service the multiple requests.
459 Except in the instance of (1) above any Denial Of Service is likely to be
460 transitory because as soon as the connection fails the memory is
461 subsequently freed again in the SSL_free() call. However there is an
462 increased risk during this period of application crashes due to the lack of
463 memory - which would then mean a more serious Denial of Service.
465 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
466 (CVE-2016-6307 and CVE-2016-6308)
469 *) solaris-x86-cc, i.e. 32-bit configuration with vendor compiler,
470 had to be removed. Primary reason is that vendor assembler can't
471 assemble our modules with -KPIC flag. As result it, assembly
472 support, was not even available as option. But its lack means
473 lack of side-channel resistant code, which is incompatible with
474 security by todays standards. Fortunately gcc is readily available
475 prepackaged option, which we firmly point at...
478 Changes between 1.0.2h and 1.1.0 [25 Aug 2016]
480 *) Windows command-line tool supports UTF-8 opt-in option for arguments
481 and console input. Setting OPENSSL_WIN32_UTF8 environment variable
482 (to any value) allows Windows user to access PKCS#12 file generated
483 with Windows CryptoAPI and protected with non-ASCII password, as well
484 as files generated under UTF-8 locale on Linux also protected with
488 *) To mitigate the SWEET32 attack (CVE-2016-2183), 3DES cipher suites
489 have been disabled by default and removed from DEFAULT, just like RC4.
490 See the RC4 item below to re-enable both.
493 *) The method for finding the storage location for the Windows RAND seed file
494 has changed. First we check %RANDFILE%. If that is not set then we check
495 the directories %HOME%, %USERPROFILE% and %SYSTEMROOT% in that order. If
496 all else fails we fall back to C:\.
499 *) The EVP_EncryptUpdate() function has had its return type changed from void
500 to int. A return of 0 indicates and error while a return of 1 indicates
504 *) The flags RSA_FLAG_NO_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME and
505 DH_FLAG_NO_EXP_CONSTTIME which previously provided the ability to switch
506 off the constant time implementation for RSA, DSA and DH have been made
507 no-ops and deprecated.
510 *) Windows RAND implementation was simplified to only get entropy by
511 calling CryptGenRandom(). Various other RAND-related tickets
513 [Joseph Wylie Yandle, Rich Salz]
515 *) The stack and lhash API's were renamed to start with OPENSSL_SK_
516 and OPENSSL_LH_, respectively. The old names are available
517 with API compatibility. They new names are now completely documented.
520 *) Unify TYPE_up_ref(obj) methods signature.
521 SSL_CTX_up_ref(), SSL_up_ref(), X509_up_ref(), EVP_PKEY_up_ref(),
522 X509_CRL_up_ref(), X509_OBJECT_up_ref_count() methods are now returning an
523 int (instead of void) like all others TYPE_up_ref() methods.
524 So now these methods also check the return value of CRYPTO_atomic_add(),
525 and the validity of object reference counter.
526 [fdasilvayy@gmail.com]
528 *) With Windows Visual Studio builds, the .pdb files are installed
529 alongside the installed libraries and executables. For a static
530 library installation, ossl_static.pdb is the associate compiler
531 generated .pdb file to be used when linking programs.
534 *) Remove openssl.spec. Packaging files belong with the packagers.
537 *) Automatic Darwin/OSX configuration has had a refresh, it will now
538 recognise x86_64 architectures automatically. You can still decide
539 to build for a different bitness with the environment variable
540 KERNEL_BITS (can be 32 or 64), for example:
542 KERNEL_BITS=32 ./config
546 *) Change default algorithms in pkcs8 utility to use PKCS#5 v2.0,
547 256 bit AES and HMAC with SHA256.
550 *) Remove support for MIPS o32 ABI on IRIX (and IRIX only).
553 *) Triple-DES ciphers have been moved from HIGH to MEDIUM.
556 *) To enable users to have their own config files and build file templates,
557 Configure looks in the directory indicated by the environment variable
558 OPENSSL_LOCAL_CONFIG_DIR as well as the in-source Configurations/
559 directory. On VMS, OPENSSL_LOCAL_CONFIG_DIR is expected to be a logical
560 name and is used as is.
563 *) The following datatypes were made opaque: X509_OBJECT, X509_STORE_CTX,
564 X509_STORE, X509_LOOKUP, and X509_LOOKUP_METHOD. The unused type
565 X509_CERT_FILE_CTX was removed.
568 *) "shared" builds are now the default. To create only static libraries use
569 the "no-shared" Configure option.
572 *) Remove the no-aes, no-hmac, no-rsa, no-sha and no-md5 Configure options.
573 All of these option have not worked for some while and are fundamental
577 *) Make various cleanup routines no-ops and mark them as deprecated. Most
578 global cleanup functions are no longer required because they are handled
579 via auto-deinit (see OPENSSL_init_crypto and OPENSSL_init_ssl man pages).
580 Explicitly de-initing can cause problems (e.g. where a library that uses
581 OpenSSL de-inits, but an application is still using it). The affected
582 functions are CONF_modules_free(), ENGINE_cleanup(), OBJ_cleanup(),
583 EVP_cleanup(), BIO_sock_cleanup(), CRYPTO_cleanup_all_ex_data(),
584 RAND_cleanup(), SSL_COMP_free_compression_methods(), ERR_free_strings() and
588 *) --strict-warnings no longer enables runtime debugging options
589 such as REF_DEBUG. Instead, debug options are automatically
590 enabled with '--debug' builds.
591 [Andy Polyakov, Emilia Käsper]
593 *) Made DH and DH_METHOD opaque. The structures for managing DH objects
594 have been moved out of the public header files. New functions for managing
595 these have been added.
598 *) Made RSA and RSA_METHOD opaque. The structures for managing RSA
599 objects have been moved out of the public header files. New
600 functions for managing these have been added.
603 *) Made DSA and DSA_METHOD opaque. The structures for managing DSA objects
604 have been moved out of the public header files. New functions for managing
605 these have been added.
608 *) Made BIO and BIO_METHOD opaque. The structures for managing BIOs have been
609 moved out of the public header files. New functions for managing these
613 *) Removed no-rijndael as a config option. Rijndael is an old name for AES.
616 *) Removed the mk1mf build scripts.
619 *) Headers are now wrapped, if necessary, with OPENSSL_NO_xxx, so
620 it is always safe to #include a header now.
623 *) Removed the aged BC-32 config and all its supporting scripts
626 *) Removed support for Ultrix, Netware, and OS/2.
629 *) Add support for HKDF.
632 *) Add support for blake2b and blake2s
635 *) Added support for "pipelining". Ciphers that have the
636 EVP_CIPH_FLAG_PIPELINE flag set have a capability to process multiple
637 encryptions/decryptions simultaneously. There are currently no built-in
638 ciphers with this property but the expectation is that engines will be able
639 to offer it to significantly improve throughput. Support has been extended
640 into libssl so that multiple records for a single connection can be
641 processed in one go (for >=TLS 1.1).
644 *) Added the AFALG engine. This is an async capable engine which is able to
645 offload work to the Linux kernel. In this initial version it only supports
646 AES128-CBC. The kernel must be version 4.1.0 or greater.
649 *) OpenSSL now uses a new threading API. It is no longer necessary to
650 set locking callbacks to use OpenSSL in a multi-threaded environment. There
651 are two supported threading models: pthreads and windows threads. It is
652 also possible to configure OpenSSL at compile time for "no-threads". The
653 old threading API should no longer be used. The functions have been
654 replaced with "no-op" compatibility macros.
655 [Alessandro Ghedini, Matt Caswell]
657 *) Modify behavior of ALPN to invoke callback after SNI/servername
658 callback, such that updates to the SSL_CTX affect ALPN.
661 *) Add SSL_CIPHER queries for authentication and key-exchange.
664 *) Changes to the DEFAULT cipherlist:
665 - Prefer (EC)DHE handshakes over plain RSA.
666 - Prefer AEAD ciphers over legacy ciphers.
667 - Prefer ECDSA over RSA when both certificates are available.
668 - Prefer TLSv1.2 ciphers/PRF.
669 - Remove DSS, SEED, IDEA, CAMELLIA, and AES-CCM from the
673 *) Change the ECC default curve list to be this, in order: x25519,
674 secp256r1, secp521r1, secp384r1.
677 *) RC4 based libssl ciphersuites are now classed as "weak" ciphers and are
678 disabled by default. They can be re-enabled using the
679 enable-weak-ssl-ciphers option to Configure.
682 *) If the server has ALPN configured, but supports no protocols that the
683 client advertises, send a fatal "no_application_protocol" alert.
684 This behaviour is SHALL in RFC 7301, though it isn't universally
685 implemented by other servers.
688 *) Add X25519 support.
689 Add ASN.1 and EVP_PKEY methods for X25519. This includes support
690 for public and private key encoding using the format documented in
691 draft-ietf-curdle-pkix-02. The corresponding EVP_PKEY method supports
692 key generation and key derivation.
694 TLS support complies with draft-ietf-tls-rfc4492bis-08 and uses
698 *) Deprecate SRP_VBASE_get_by_user.
699 SRP_VBASE_get_by_user had inconsistent memory management behaviour.
700 In order to fix an unavoidable memory leak (CVE-2016-0798),
701 SRP_VBASE_get_by_user was changed to ignore the "fake user" SRP
702 seed, even if the seed is configured.
704 Users should use SRP_VBASE_get1_by_user instead. Note that in
705 SRP_VBASE_get1_by_user, caller must free the returned value. Note
706 also that even though configuring the SRP seed attempts to hide
707 invalid usernames by continuing the handshake with fake
708 credentials, this behaviour is not constant time and no strong
709 guarantees are made that the handshake is indistinguishable from
710 that of a valid user.
713 *) Configuration change; it's now possible to build dynamic engines
714 without having to build shared libraries and vice versa. This
715 only applies to the engines in engines/, those in crypto/engine/
716 will always be built into libcrypto (i.e. "static").
718 Building dynamic engines is enabled by default; to disable, use
719 the configuration option "disable-dynamic-engine".
721 The only requirements for building dynamic engines are the
722 presence of the DSO module and building with position independent
723 code, so they will also automatically be disabled if configuring
724 with "disable-dso" or "disable-pic".
726 The macros OPENSSL_NO_STATIC_ENGINE and OPENSSL_NO_DYNAMIC_ENGINE
727 are also taken away from openssl/opensslconf.h, as they are
731 *) Configuration change; if there is a known flag to compile
732 position independent code, it will always be applied on the
733 libcrypto and libssl object files, and never on the application
734 object files. This means other libraries that use routines from
735 libcrypto / libssl can be made into shared libraries regardless
736 of how OpenSSL was configured.
738 If this isn't desirable, the configuration options "disable-pic"
739 or "no-pic" can be used to disable the use of PIC. This will
740 also disable building shared libraries and dynamic engines.
743 *) Removed JPAKE code. It was experimental and has no wide use.
746 *) The INSTALL_PREFIX Makefile variable has been renamed to
747 DESTDIR. That makes for less confusion on what this variable
748 is for. Also, the configuration option --install_prefix is
752 *) Heartbeat for TLS has been removed and is disabled by default
753 for DTLS; configure with enable-heartbeats. Code that uses the
754 old #define's might need to be updated.
755 [Emilia Käsper, Rich Salz]
757 *) Rename REF_CHECK to REF_DEBUG.
760 *) New "unified" build system
762 The "unified" build system is aimed to be a common system for all
763 platforms we support. With it comes new support for VMS.
765 This system builds supports building in a different directory tree
766 than the source tree. It produces one Makefile (for unix family
767 or lookalikes), or one descrip.mms (for VMS).
769 The source of information to make the Makefile / descrip.mms is
770 small files called 'build.info', holding the necessary
771 information for each directory with source to compile, and a
772 template in Configurations, like unix-Makefile.tmpl or
775 With this change, the library names were also renamed on Windows
776 and on VMS. They now have names that are closer to the standard
777 on Unix, and include the major version number, and in certain
778 cases, the architecture they are built for. See "Notes on shared
779 libraries" in INSTALL.
781 We rely heavily on the perl module Text::Template.
784 *) Added support for auto-initialisation and de-initialisation of the library.
785 OpenSSL no longer requires explicit init or deinit routines to be called,
786 except in certain circumstances. See the OPENSSL_init_crypto() and
787 OPENSSL_init_ssl() man pages for further information.
790 *) The arguments to the DTLSv1_listen function have changed. Specifically the
791 "peer" argument is now expected to be a BIO_ADDR object.
793 *) Rewrite of BIO networking library. The BIO library lacked consistent
794 support of IPv6, and adding it required some more extensive
795 modifications. This introduces the BIO_ADDR and BIO_ADDRINFO types,
796 which hold all types of addresses and chains of address information.
797 It also introduces a new API, with functions like BIO_socket,
798 BIO_connect, BIO_listen, BIO_lookup and a rewrite of BIO_accept.
799 The source/sink BIOs BIO_s_connect, BIO_s_accept and BIO_s_datagram
800 have been adapted accordingly.
803 *) RSA_padding_check_PKCS1_type_1 now accepts inputs with and without
807 *) CRIME protection: disable compression by default, even if OpenSSL is
808 compiled with zlib enabled. Applications can still enable compression
809 by calling SSL_CTX_clear_options(ctx, SSL_OP_NO_COMPRESSION), or by
810 using the SSL_CONF library to configure compression.
813 *) The signature of the session callback configured with
814 SSL_CTX_sess_set_get_cb was changed. The read-only input buffer
815 was explicitly marked as 'const unsigned char*' instead of
819 *) Always DPURIFY. Remove the use of uninitialized memory in the
820 RNG, and other conditional uses of DPURIFY. This makes -DPURIFY a no-op.
823 *) Removed many obsolete configuration items, including
824 DES_PTR, DES_RISC1, DES_RISC2, DES_INT
825 MD2_CHAR, MD2_INT, MD2_LONG
827 IDEA_SHORT, IDEA_LONG
828 RC2_SHORT, RC2_LONG, RC4_LONG, RC4_CHUNK, RC4_INDEX
829 [Rich Salz, with advice from Andy Polyakov]
831 *) Many BN internals have been moved to an internal header file.
832 [Rich Salz with help from Andy Polyakov]
834 *) Configuration and writing out the results from it has changed.
835 Files such as Makefile include/openssl/opensslconf.h and are now
836 produced through general templates, such as Makefile.in and
837 crypto/opensslconf.h.in and some help from the perl module
840 Also, the center of configuration information is no longer
841 Makefile. Instead, Configure produces a perl module in
842 configdata.pm which holds most of the config data (in the hash
843 table %config), the target data that comes from the target
844 configuration in one of the Configurations/*.conf files (in
848 *) To clarify their intended purposes, the Configure options
849 --prefix and --openssldir change their semantics, and become more
850 straightforward and less interdependent.
852 --prefix shall be used exclusively to give the location INSTALLTOP
853 where programs, scripts, libraries, include files and manuals are
854 going to be installed. The default is now /usr/local.
856 --openssldir shall be used exclusively to give the default
857 location OPENSSLDIR where certificates, private keys, CRLs are
858 managed. This is also where the default openssl.cnf gets
860 If the directory given with this option is a relative path, the
861 values of both the --prefix value and the --openssldir value will
862 be combined to become OPENSSLDIR.
863 The default for --openssldir is INSTALLTOP/ssl.
865 Anyone who uses --openssldir to specify where OpenSSL is to be
866 installed MUST change to use --prefix instead.
869 *) The GOST engine was out of date and therefore it has been removed. An up
870 to date GOST engine is now being maintained in an external repository.
871 See: https://wiki.openssl.org/index.php/Binaries. Libssl still retains
872 support for GOST ciphersuites (these are only activated if a GOST engine
876 *) EGD is no longer supported by default; use enable-egd when
878 [Ben Kaduk and Rich Salz]
880 *) The distribution now has Makefile.in files, which are used to
881 create Makefile's when Configure is run. *Configure must be run
882 before trying to build now.*
885 *) The return value for SSL_CIPHER_description() for error conditions
889 *) Support for RFC6698/RFC7671 DANE TLSA peer authentication.
891 Obtaining and performing DNSSEC validation of TLSA records is
892 the application's responsibility. The application provides
893 the TLSA records of its choice to OpenSSL, and these are then
894 used to authenticate the peer.
896 The TLSA records need not even come from DNS. They can, for
897 example, be used to implement local end-entity certificate or
898 trust-anchor "pinning", where the "pin" data takes the form
899 of TLSA records, which can augment or replace verification
900 based on the usual WebPKI public certification authorities.
903 *) Revert default OPENSSL_NO_DEPRECATED setting. Instead OpenSSL
904 continues to support deprecated interfaces in default builds.
905 However, applications are strongly advised to compile their
906 source files with -DOPENSSL_API_COMPAT=0x10100000L, which hides
907 the declarations of all interfaces deprecated in 0.9.8, 1.0.0
908 or the 1.1.0 releases.
910 In environments in which all applications have been ported to
911 not use any deprecated interfaces OpenSSL's Configure script
912 should be used with the --api=1.1.0 option to entirely remove
913 support for the deprecated features from the library and
914 unconditionally disable them in the installed headers.
915 Essentially the same effect can be achieved with the "no-deprecated"
916 argument to Configure, except that this will always restrict
917 the build to just the latest API, rather than a fixed API
920 As applications are ported to future revisions of the API,
921 they should update their compile-time OPENSSL_API_COMPAT define
922 accordingly, but in most cases should be able to continue to
923 compile with later releases.
925 The OPENSSL_API_COMPAT versions for 1.0.0, and 0.9.8 are
926 0x10000000L and 0x00908000L, respectively. However those
927 versions did not support the OPENSSL_API_COMPAT feature, and
928 so applications are not typically tested for explicit support
929 of just the undeprecated features of either release.
932 *) Add support for setting the minimum and maximum supported protocol.
933 It can bet set via the SSL_set_min_proto_version() and
934 SSL_set_max_proto_version(), or via the SSL_CONF's MinProtocol and
935 MaxProtocol. It's recommended to use the new APIs to disable
936 protocols instead of disabling individual protocols using
937 SSL_set_options() or SSL_CONF's Protocol. This change also
938 removes support for disabling TLS 1.2 in the OpenSSL TLS
939 client at compile time by defining OPENSSL_NO_TLS1_2_CLIENT.
942 *) Support for ChaCha20 and Poly1305 added to libcrypto and libssl.
945 *) New EC_KEY_METHOD, this replaces the older ECDSA_METHOD and ECDH_METHOD
946 and integrates ECDSA and ECDH functionality into EC. Implementations can
947 now redirect key generation and no longer need to convert to or from
950 Note: the ecdsa.h and ecdh.h headers are now no longer needed and just
951 include the ec.h header file instead.
954 *) Remove support for all 40 and 56 bit ciphers. This includes all the export
955 ciphers who are no longer supported and drops support the ephemeral RSA key
956 exchange. The LOW ciphers currently doesn't have any ciphers in it.
959 *) Made EVP_MD_CTX, EVP_MD, EVP_CIPHER_CTX, EVP_CIPHER and HMAC_CTX
960 opaque. For HMAC_CTX, the following constructors and destructors
963 HMAC_CTX *HMAC_CTX_new(void);
964 void HMAC_CTX_free(HMAC_CTX *ctx);
966 For EVP_MD and EVP_CIPHER, complete APIs to create, fill and
967 destroy such methods has been added. See EVP_MD_meth_new(3) and
968 EVP_CIPHER_meth_new(3) for documentation.
971 1) EVP_MD_CTX_cleanup(), EVP_CIPHER_CTX_cleanup() and
972 HMAC_CTX_cleanup() were removed. HMAC_CTX_reset() and
973 EVP_MD_CTX_reset() should be called instead to reinitialise
974 an already created structure.
975 2) For consistency with the majority of our object creators and
976 destructors, EVP_MD_CTX_(create|destroy) were renamed to
977 EVP_MD_CTX_(new|free). The old names are retained as macros
978 for deprecated builds.
981 *) Added ASYNC support. Libcrypto now includes the async sub-library to enable
982 cryptographic operations to be performed asynchronously as long as an
983 asynchronous capable engine is used. See the ASYNC_start_job() man page for
984 further details. Libssl has also had this capability integrated with the
985 introduction of the new mode SSL_MODE_ASYNC and associated error
986 SSL_ERROR_WANT_ASYNC. See the SSL_CTX_set_mode() and SSL_get_error() man
987 pages. This work was developed in partnership with Intel Corp.
990 *) SSL_{CTX_}set_ecdh_auto() has been removed and ECDH is support is
991 always enabled now. If you want to disable the support you should
992 exclude it using the list of supported ciphers. This also means that the
993 "-no_ecdhe" option has been removed from s_server.
996 *) SSL_{CTX}_set_tmp_ecdh() which can set 1 EC curve now internally calls
997 SSL_{CTX_}set1_curves() which can set a list.
1000 *) Remove support for SSL_{CTX_}set_tmp_ecdh_callback(). You should set the
1001 curve you want to support using SSL_{CTX_}set1_curves().
1004 *) State machine rewrite. The state machine code has been significantly
1005 refactored in order to remove much duplication of code and solve issues
1006 with the old code (see ssl/statem/README for further details). This change
1007 does have some associated API changes. Notably the SSL_state() function
1008 has been removed and replaced by SSL_get_state which now returns an
1009 "OSSL_HANDSHAKE_STATE" instead of an int. SSL_set_state() has been removed
1010 altogether. The previous handshake states defined in ssl.h and ssl3.h have
1014 *) All instances of the string "ssleay" in the public API were replaced
1015 with OpenSSL (case-matching; e.g., OPENSSL_VERSION for #define's)
1016 Some error codes related to internal RSA_eay API's were renamed.
1019 *) The demo files in crypto/threads were moved to demo/threads.
1022 *) Removed obsolete engines: 4758cca, aep, atalla, cswift, nuron, gmp,
1024 [Matt Caswell, Rich Salz]
1026 *) New ASN.1 embed macro.
1028 New ASN.1 macro ASN1_EMBED. This is the same as ASN1_SIMPLE except the
1029 structure is not allocated: it is part of the parent. That is instead of
1037 This reduces memory fragmentation and make it impossible to accidentally
1038 set a mandatory field to NULL.
1040 This currently only works for some fields specifically a SEQUENCE, CHOICE,
1041 or ASN1_STRING type which is part of a parent SEQUENCE. Since it is
1042 equivalent to ASN1_SIMPLE it cannot be tagged, OPTIONAL, SET OF or
1046 *) Remove EVP_CHECK_DES_KEY, a compile-time option that never compiled.
1049 *) Removed DES and RC4 ciphersuites from DEFAULT. Also removed RC2 although
1050 in 1.0.2 EXPORT was already removed and the only RC2 ciphersuite is also
1051 an EXPORT one. COMPLEMENTOFDEFAULT has been updated accordingly to add
1052 DES and RC4 ciphersuites.
1055 *) Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs.
1056 This changes the decoding behaviour for some invalid messages,
1057 though the change is mostly in the more lenient direction, and
1058 legacy behaviour is preserved as much as possible.
1061 *) Fix no-stdio build.
1062 [ David Woodhouse <David.Woodhouse@intel.com> and also
1063 Ivan Nestlerode <ivan.nestlerode@sonos.com> ]
1065 *) New testing framework
1066 The testing framework has been largely rewritten and is now using
1067 perl and the perl modules Test::Harness and an extended variant of
1068 Test::More called OpenSSL::Test to do its work. All test scripts in
1069 test/ have been rewritten into test recipes, and all direct calls to
1070 executables in test/Makefile have become individual recipes using the
1071 simplified testing OpenSSL::Test::Simple.
1073 For documentation on our testing modules, do:
1075 perldoc test/testlib/OpenSSL/Test/Simple.pm
1076 perldoc test/testlib/OpenSSL/Test.pm
1080 *) Revamped memory debug; only -DCRYPTO_MDEBUG and -DCRYPTO_MDEBUG_ABORT
1081 are used; the latter aborts on memory leaks (usually checked on exit).
1082 Some undocumented "set malloc, etc., hooks" functions were removed
1083 and others were changed. All are now documented.
1086 *) In DSA_generate_parameters_ex, if the provided seed is too short,
1088 [Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>]
1090 *) Rewrite PSK to support ECDHE_PSK, DHE_PSK and RSA_PSK. Add ciphersuites
1091 from RFC4279, RFC4785, RFC5487, RFC5489.
1093 Thanks to Christian J. Dietrich and Giuseppe D'Angelo for the
1094 original RSA_PSK patch.
1097 *) Dropped support for the SSL3_FLAGS_DELAY_CLIENT_FINISHED flag. This SSLeay
1098 era flag was never set throughout the codebase (only read). Also removed
1099 SSL3_FLAGS_POP_BUFFER which was only used if
1100 SSL3_FLAGS_DELAY_CLIENT_FINISHED was also set.
1103 *) Changed the default name options in the "ca", "crl", "req" and "x509"
1104 to be "oneline" instead of "compat".
1107 *) Remove SSL_OP_TLS_BLOCK_PADDING_BUG. This is SSLeay legacy, we're
1108 not aware of clients that still exhibit this bug, and the workaround
1109 hasn't been working properly for a while.
1112 *) The return type of BIO_number_read() and BIO_number_written() as well as
1113 the corresponding num_read and num_write members in the BIO structure has
1114 changed from unsigned long to uint64_t. On platforms where an unsigned
1115 long is 32 bits (e.g. Windows) these counters could overflow if >4Gb is
1119 *) Given the pervasive nature of TLS extensions it is inadvisable to run
1120 OpenSSL without support for them. It also means that maintaining
1121 the OPENSSL_NO_TLSEXT option within the code is very invasive (and probably
1122 not well tested). Therefore the OPENSSL_NO_TLSEXT option has been removed.
1125 *) Removed support for the two export grade static DH ciphersuites
1126 EXP-DH-RSA-DES-CBC-SHA and EXP-DH-DSS-DES-CBC-SHA. These two ciphersuites
1127 were newly added (along with a number of other static DH ciphersuites) to
1128 1.0.2. However the two export ones have *never* worked since they were
1129 introduced. It seems strange in any case to be adding new export
1130 ciphersuites, and given "logjam" it also does not seem correct to fix them.
1133 *) Version negotiation has been rewritten. In particular SSLv23_method(),
1134 SSLv23_client_method() and SSLv23_server_method() have been deprecated,
1135 and turned into macros which simply call the new preferred function names
1136 TLS_method(), TLS_client_method() and TLS_server_method(). All new code
1137 should use the new names instead. Also as part of this change the ssl23.h
1138 header file has been removed.
1141 *) Support for Kerberos ciphersuites in TLS (RFC2712) has been removed. This
1142 code and the associated standard is no longer considered fit-for-purpose.
1145 *) RT2547 was closed. When generating a private key, try to make the
1146 output file readable only by the owner. This behavior change might
1147 be noticeable when interacting with other software.
1149 *) Documented all exdata functions. Added CRYPTO_free_ex_index.
1153 *) Added HTTP GET support to the ocsp command.
1156 *) Changed default digest for the dgst and enc commands from MD5 to
1160 *) RAND_pseudo_bytes has been deprecated. Users should use RAND_bytes instead.
1163 *) Added support for TLS extended master secret from
1164 draft-ietf-tls-session-hash-03.txt. Thanks for Alfredo Pironti for an
1165 initial patch which was a great help during development.
1168 *) All libssl internal structures have been removed from the public header
1169 files, and the OPENSSL_NO_SSL_INTERN option has been removed (since it is
1170 now redundant). Users should not attempt to access internal structures
1171 directly. Instead they should use the provided API functions.
1174 *) config has been changed so that by default OPENSSL_NO_DEPRECATED is used.
1175 Access to deprecated functions can be re-enabled by running config with
1176 "enable-deprecated". In addition applications wishing to use deprecated
1177 functions must define OPENSSL_USE_DEPRECATED. Note that this new behaviour
1178 will, by default, disable some transitive includes that previously existed
1179 in the header files (e.g. ec.h will no longer, by default, include bn.h)
1182 *) Added support for OCB mode. OpenSSL has been granted a patent license
1183 compatible with the OpenSSL license for use of OCB. Details are available
1184 at https://www.openssl.org/source/OCB-patent-grant-OpenSSL.pdf. Support
1185 for OCB can be removed by calling config with no-ocb.
1188 *) SSLv2 support has been removed. It still supports receiving a SSLv2
1189 compatible client hello.
1192 *) Increased the minimal RSA keysize from 256 to 512 bits [Rich Salz],
1193 done while fixing the error code for the key-too-small case.
1194 [Annie Yousar <a.yousar@informatik.hu-berlin.de>]
1196 *) CA.sh has been removed; use CA.pl instead.
1199 *) Removed old DES API.
1202 *) Remove various unsupported platforms:
1208 Sinix/ReliantUNIX RM400
1213 16-bit platforms such as WIN16
1216 *) Clean up OPENSSL_NO_xxx #define's
1217 Use setbuf() and remove OPENSSL_NO_SETVBUF_IONBF
1218 Rename OPENSSL_SYSNAME_xxx to OPENSSL_SYS_xxx
1219 OPENSSL_NO_EC{DH,DSA} merged into OPENSSL_NO_EC
1220 OPENSSL_NO_RIPEMD160, OPENSSL_NO_RIPEMD merged into OPENSSL_NO_RMD160
1221 OPENSSL_NO_FP_API merged into OPENSSL_NO_STDIO
1222 Remove OPENSSL_NO_BIO OPENSSL_NO_BUFFER OPENSSL_NO_CHAIN_VERIFY
1223 OPENSSL_NO_EVP OPENSSL_NO_FIPS_ERR OPENSSL_NO_HASH_COMP
1224 OPENSSL_NO_LHASH OPENSSL_NO_OBJECT OPENSSL_NO_SPEED OPENSSL_NO_STACK
1225 OPENSSL_NO_X509 OPENSSL_NO_X509_VERIFY
1226 Remove MS_STATIC; it's a relic from platforms <32 bits.
1229 *) Cleaned up dead code
1230 Remove all but one '#ifdef undef' which is to be looked at.
1233 *) Clean up calling of xxx_free routines.
1234 Just like free(), fix most of the xxx_free routines to accept
1235 NULL. Remove the non-null checks from callers. Save much code.
1238 *) Add secure heap for storage of private keys (when possible).
1239 Add BIO_s_secmem(), CBIGNUM, etc.
1240 Contributed by Akamai Technologies under our Corporate CLA.
1243 *) Experimental support for a new, fast, unbiased prime candidate generator,
1244 bn_probable_prime_dh_coprime(). Not currently used by any prime generator.
1245 [Felix Laurie von Massenbach <felix@erbridge.co.uk>]
1247 *) New output format NSS in the sess_id command line tool. This allows
1248 exporting the session id and the master key in NSS keylog format.
1249 [Martin Kaiser <martin@kaiser.cx>]
1251 *) Harmonize version and its documentation. -f flag is used to display
1253 [mancha <mancha1@zoho.com>]
1255 *) Fix eckey_priv_encode so it immediately returns an error upon a failure
1256 in i2d_ECPrivateKey. Thanks to Ted Unangst for feedback on this issue.
1257 [mancha <mancha1@zoho.com>]
1259 *) Fix some double frees. These are not thought to be exploitable.
1260 [mancha <mancha1@zoho.com>]
1262 *) A missing bounds check in the handling of the TLS heartbeat extension
1263 can be used to reveal up to 64k of memory to a connected client or
1266 Thanks for Neel Mehta of Google Security for discovering this bug and to
1267 Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for
1268 preparing the fix (CVE-2014-0160)
1269 [Adam Langley, Bodo Moeller]
1271 *) Fix for the attack described in the paper "Recovering OpenSSL
1272 ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack"
1273 by Yuval Yarom and Naomi Benger. Details can be obtained from:
1274 http://eprint.iacr.org/2014/140
1276 Thanks to Yuval Yarom and Naomi Benger for discovering this
1277 flaw and to Yuval Yarom for supplying a fix (CVE-2014-0076)
1278 [Yuval Yarom and Naomi Benger]
1280 *) Use algorithm specific chains in SSL_CTX_use_certificate_chain_file():
1281 this fixes a limitation in previous versions of OpenSSL.
1284 *) Experimental encrypt-then-mac support.
1286 Experimental support for encrypt then mac from
1287 draft-gutmann-tls-encrypt-then-mac-02.txt
1289 To enable it set the appropriate extension number (0x42 for the test
1290 server) using e.g. -DTLSEXT_TYPE_encrypt_then_mac=0x42
1292 For non-compliant peers (i.e. just about everything) this should have no
1295 WARNING: EXPERIMENTAL, SUBJECT TO CHANGE.
1299 *) Add EVP support for key wrapping algorithms, to avoid problems with
1300 existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in
1301 the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap
1302 algorithms and include tests cases.
1305 *) Extend CMS code to support RSA-PSS signatures and RSA-OAEP for
1309 *) Extended RSA OAEP support via EVP_PKEY API. Options to specify digest,
1310 MGF1 digest and OAEP label.
1313 *) Make openssl verify return errors.
1314 [Chris Palmer <palmer@google.com> and Ben Laurie]
1316 *) New function ASN1_TIME_diff to calculate the difference between two
1317 ASN1_TIME structures or one structure and the current time.
1320 *) Update fips_test_suite to support multiple command line options. New
1321 test to induce all self test errors in sequence and check expected
1325 *) Add FIPS_{rsa,dsa,ecdsa}_{sign,verify} functions which digest and
1326 sign or verify all in one operation.
1329 *) Add fips_algvs: a multicall fips utility incorporating all the algorithm
1330 test programs and fips_test_suite. Includes functionality to parse
1331 the minimal script output of fipsalgest.pl directly.
1334 *) Add authorisation parameter to FIPS_module_mode_set().
1337 *) Add FIPS selftest for ECDH algorithm using P-224 and B-233 curves.
1340 *) Use separate DRBG fields for internal and external flags. New function
1341 FIPS_drbg_health_check() to perform on demand health checking. Add
1342 generation tests to fips_test_suite with reduced health check interval to
1343 demonstrate periodic health checking. Add "nodh" option to
1344 fips_test_suite to skip very slow DH test.
1347 *) New function FIPS_get_cipherbynid() to lookup FIPS supported ciphers
1351 *) More extensive health check for DRBG checking many more failure modes.
1352 New function FIPS_selftest_drbg_all() to handle every possible DRBG
1353 combination: call this in fips_test_suite.
1356 *) Add support for canonical generation of DSA parameter 'g'. See
1359 *) Add support for HMAC DRBG from SP800-90. Update DRBG algorithm test and
1360 POST to handle HMAC cases.
1363 *) Add functions FIPS_module_version() and FIPS_module_version_text()
1364 to return numerical and string versions of the FIPS module number.
1367 *) Rename FIPS_mode_set and FIPS_mode to FIPS_module_mode_set and
1368 FIPS_module_mode. FIPS_mode and FIPS_mode_set will be implemented
1369 outside the validated module in the FIPS capable OpenSSL.
1372 *) Minor change to DRBG entropy callback semantics. In some cases
1373 there is no multiple of the block length between min_len and
1374 max_len. Allow the callback to return more than max_len bytes
1375 of entropy but discard any extra: it is the callback's responsibility
1376 to ensure that the extra data discarded does not impact the
1377 requested amount of entropy.
1380 *) Add PRNG security strength checks to RSA, DSA and ECDSA using
1381 information in FIPS186-3, SP800-57 and SP800-131A.
1384 *) CCM support via EVP. Interface is very similar to GCM case except we
1385 must supply all data in one chunk (i.e. no update, final) and the
1386 message length must be supplied if AAD is used. Add algorithm test
1390 *) Initial version of POST overhaul. Add POST callback to allow the status
1391 of POST to be monitored and/or failures induced. Modify fips_test_suite
1392 to use callback. Always run all selftests even if one fails.
1395 *) XTS support including algorithm test driver in the fips_gcmtest program.
1396 Note: this does increase the maximum key length from 32 to 64 bytes but
1397 there should be no binary compatibility issues as existing applications
1398 will never use XTS mode.
1401 *) Extensive reorganisation of FIPS PRNG behaviour. Remove all dependencies
1402 to OpenSSL RAND code and replace with a tiny FIPS RAND API which also
1403 performs algorithm blocking for unapproved PRNG types. Also do not
1404 set PRNG type in FIPS_mode_set(): leave this to the application.
1405 Add default OpenSSL DRBG handling: sets up FIPS PRNG and seeds with
1406 the standard OpenSSL PRNG: set additional data to a date time vector.
1409 *) Rename old X9.31 PRNG functions of the form FIPS_rand* to FIPS_x931*.
1410 This shouldn't present any incompatibility problems because applications
1411 shouldn't be using these directly and any that are will need to rethink
1412 anyway as the X9.31 PRNG is now deprecated by FIPS 140-2
1415 *) Extensive self tests and health checking required by SP800-90 DRBG.
1416 Remove strength parameter from FIPS_drbg_instantiate and always
1417 instantiate at maximum supported strength.
1420 *) Add ECDH code to fips module and fips_ecdhvs for primitives only testing.
1423 *) New algorithm test program fips_dhvs to handle DH primitives only testing.
1426 *) New function DH_compute_key_padded() to compute a DH key and pad with
1427 leading zeroes if needed: this complies with SP800-56A et al.
1430 *) Initial implementation of SP800-90 DRBGs for Hash and CTR. Not used by
1431 anything, incomplete, subject to change and largely untested at present.
1434 *) Modify fipscanisteronly build option to only build the necessary object
1435 files by filtering FIPS_EX_OBJ through a perl script in crypto/Makefile.
1438 *) Add experimental option FIPSSYMS to give all symbols in
1439 fipscanister.o and FIPS or fips prefix. This will avoid
1440 conflicts with future versions of OpenSSL. Add perl script
1441 util/fipsas.pl to preprocess assembly language source files
1442 and rename any affected symbols.
1445 *) Add selftest checks and algorithm block of non-fips algorithms in
1446 FIPS mode. Remove DES2 from selftests.
1449 *) Add ECDSA code to fips module. Add tiny fips_ecdsa_check to just
1450 return internal method without any ENGINE dependencies. Add new
1451 tiny fips sign and verify functions.
1454 *) New build option no-ec2m to disable characteristic 2 code.
1457 *) New build option "fipscanisteronly". This only builds fipscanister.o
1458 and (currently) associated fips utilities. Uses the file Makefile.fips
1459 instead of Makefile.org as the prototype.
1462 *) Add some FIPS mode restrictions to GCM. Add internal IV generator.
1463 Update fips_gcmtest to use IV generator.
1466 *) Initial, experimental EVP support for AES-GCM. AAD can be input by
1467 setting output buffer to NULL. The *Final function must be
1468 called although it will not retrieve any additional data. The tag
1469 can be set or retrieved with a ctrl. The IV length is by default 12
1470 bytes (96 bits) but can be set to an alternative value. If the IV
1471 length exceeds the maximum IV length (currently 16 bytes) it cannot be
1475 *) New flag in ciphers: EVP_CIPH_FLAG_CUSTOM_CIPHER. This means the
1476 underlying do_cipher function handles all cipher semantics itself
1477 including padding and finalisation. This is useful if (for example)
1478 an ENGINE cipher handles block padding itself. The behaviour of
1479 do_cipher is subtly changed if this flag is set: the return value
1480 is the number of characters written to the output buffer (zero is
1481 no longer an error code) or a negative error code. Also if the
1482 input buffer is NULL and length 0 finalisation should be performed.
1485 *) If a candidate issuer certificate is already part of the constructed
1486 path ignore it: new debug notification X509_V_ERR_PATH_LOOP for this case.
1489 *) Improve forward-security support: add functions
1491 void SSL_CTX_set_not_resumable_session_callback(SSL_CTX *ctx, int (*cb)(SSL *ssl, int is_forward_secure))
1492 void SSL_set_not_resumable_session_callback(SSL *ssl, int (*cb)(SSL *ssl, int is_forward_secure))
1494 for use by SSL/TLS servers; the callback function will be called whenever a
1495 new session is created, and gets to decide whether the session may be
1496 cached to make it resumable (return 0) or not (return 1). (As by the
1497 SSL/TLS protocol specifications, the session_id sent by the server will be
1498 empty to indicate that the session is not resumable; also, the server will
1499 not generate RFC 4507 (RFC 5077) session tickets.)
1501 A simple reasonable callback implementation is to return is_forward_secure.
1502 This parameter will be set to 1 or 0 depending on the ciphersuite selected
1503 by the SSL/TLS server library, indicating whether it can provide forward
1505 [Emilia Käsper <emilia.kasper@esat.kuleuven.be> (Google)]
1507 *) New -verify_name option in command line utilities to set verification
1511 *) Initial CMAC implementation. WARNING: EXPERIMENTAL, API MAY CHANGE.
1512 Add CMAC pkey methods.
1515 *) Experimental renegotiation in s_server -www mode. If the client
1516 browses /reneg connection is renegotiated. If /renegcert it is
1517 renegotiated requesting a certificate.
1520 *) Add an "external" session cache for debugging purposes to s_server. This
1521 should help trace issues which normally are only apparent in deployed
1522 multi-process servers.
1525 *) Extensive audit of libcrypto with DEBUG_UNUSED. Fix many cases where
1526 return value is ignored. NB. The functions RAND_add(), RAND_seed(),
1527 BIO_set_cipher() and some obscure PEM functions were changed so they
1528 can now return an error. The RAND changes required a change to the
1529 RAND_METHOD structure.
1532 *) New macro __owur for "OpenSSL Warn Unused Result". This makes use of
1533 a gcc attribute to warn if the result of a function is ignored. This
1534 is enable if DEBUG_UNUSED is set. Add to several functions in evp.h
1535 whose return value is often ignored.
1538 *) New -noct, -requestct, -requirect and -ctlogfile options for s_client.
1539 These allow SCTs (signed certificate timestamps) to be requested and
1540 validated when establishing a connection.
1541 [Rob Percival <robpercival@google.com>]
1543 Changes between 1.0.2g and 1.0.2h [3 May 2016]
1545 *) Prevent padding oracle in AES-NI CBC MAC check
1547 A MITM attacker can use a padding oracle attack to decrypt traffic
1548 when the connection uses an AES CBC cipher and the server support
1551 This issue was introduced as part of the fix for Lucky 13 padding
1552 attack (CVE-2013-0169). The padding check was rewritten to be in
1553 constant time by making sure that always the same bytes are read and
1554 compared against either the MAC or padding bytes. But it no longer
1555 checked that there was enough data to have both the MAC and padding
1558 This issue was reported by Juraj Somorovsky using TLS-Attacker.
1562 *) Fix EVP_EncodeUpdate overflow
1564 An overflow can occur in the EVP_EncodeUpdate() function which is used for
1565 Base64 encoding of binary data. If an attacker is able to supply very large
1566 amounts of input data then a length check can overflow resulting in a heap
1569 Internally to OpenSSL the EVP_EncodeUpdate() function is primarily used by
1570 the PEM_write_bio* family of functions. These are mainly used within the
1571 OpenSSL command line applications, so any application which processes data
1572 from an untrusted source and outputs it as a PEM file should be considered
1573 vulnerable to this issue. User applications that call these APIs directly
1574 with large amounts of untrusted data may also be vulnerable.
1576 This issue was reported by Guido Vranken.
1580 *) Fix EVP_EncryptUpdate overflow
1582 An overflow can occur in the EVP_EncryptUpdate() function. If an attacker
1583 is able to supply very large amounts of input data after a previous call to
1584 EVP_EncryptUpdate() with a partial block then a length check can overflow
1585 resulting in a heap corruption. Following an analysis of all OpenSSL
1586 internal usage of the EVP_EncryptUpdate() function all usage is one of two
1587 forms. The first form is where the EVP_EncryptUpdate() call is known to be
1588 the first called function after an EVP_EncryptInit(), and therefore that
1589 specific call must be safe. The second form is where the length passed to
1590 EVP_EncryptUpdate() can be seen from the code to be some small value and
1591 therefore there is no possibility of an overflow. Since all instances are
1592 one of these two forms, it is believed that there can be no overflows in
1593 internal code due to this problem. It should be noted that
1594 EVP_DecryptUpdate() can call EVP_EncryptUpdate() in certain code paths.
1595 Also EVP_CipherUpdate() is a synonym for EVP_EncryptUpdate(). All instances
1596 of these calls have also been analysed too and it is believed there are no
1597 instances in internal usage where an overflow could occur.
1599 This issue was reported by Guido Vranken.
1603 *) Prevent ASN.1 BIO excessive memory allocation
1605 When ASN.1 data is read from a BIO using functions such as d2i_CMS_bio()
1606 a short invalid encoding can cause allocation of large amounts of memory
1607 potentially consuming excessive resources or exhausting memory.
1609 Any application parsing untrusted data through d2i BIO functions is
1610 affected. The memory based functions such as d2i_X509() are *not* affected.
1611 Since the memory based functions are used by the TLS library, TLS
1612 applications are not affected.
1614 This issue was reported by Brian Carpenter.
1620 ASN1 Strings that are over 1024 bytes can cause an overread in applications
1621 using the X509_NAME_oneline() function on EBCDIC systems. This could result
1622 in arbitrary stack data being returned in the buffer.
1624 This issue was reported by Guido Vranken.
1628 *) Modify behavior of ALPN to invoke callback after SNI/servername
1629 callback, such that updates to the SSL_CTX affect ALPN.
1632 *) Remove LOW from the DEFAULT cipher list. This removes singles DES from the
1636 *) Only remove the SSLv2 methods with the no-ssl2-method option. When the
1637 methods are enabled and ssl2 is disabled the methods return NULL.
1640 Changes between 1.0.2f and 1.0.2g [1 Mar 2016]
1642 * Disable weak ciphers in SSLv3 and up in default builds of OpenSSL.
1643 Builds that are not configured with "enable-weak-ssl-ciphers" will not
1644 provide any "EXPORT" or "LOW" strength ciphers.
1647 * Disable SSLv2 default build, default negotiation and weak ciphers. SSLv2
1648 is by default disabled at build-time. Builds that are not configured with
1649 "enable-ssl2" will not support SSLv2. Even if "enable-ssl2" is used,
1650 users who want to negotiate SSLv2 via the version-flexible SSLv23_method()
1651 will need to explicitly call either of:
1653 SSL_CTX_clear_options(ctx, SSL_OP_NO_SSLv2);
1655 SSL_clear_options(ssl, SSL_OP_NO_SSLv2);
1657 as appropriate. Even if either of those is used, or the application
1658 explicitly uses the version-specific SSLv2_method() or its client and
1659 server variants, SSLv2 ciphers vulnerable to exhaustive search key
1660 recovery have been removed. Specifically, the SSLv2 40-bit EXPORT
1661 ciphers, and SSLv2 56-bit DES are no longer available.
1665 *) Fix a double-free in DSA code
1667 A double free bug was discovered when OpenSSL parses malformed DSA private
1668 keys and could lead to a DoS attack or memory corruption for applications
1669 that receive DSA private keys from untrusted sources. This scenario is
1672 This issue was reported to OpenSSL by Adam Langley(Google/BoringSSL) using
1677 *) Disable SRP fake user seed to address a server memory leak.
1679 Add a new method SRP_VBASE_get1_by_user that handles the seed properly.
1681 SRP_VBASE_get_by_user had inconsistent memory management behaviour.
1682 In order to fix an unavoidable memory leak, SRP_VBASE_get_by_user
1683 was changed to ignore the "fake user" SRP seed, even if the seed
1686 Users should use SRP_VBASE_get1_by_user instead. Note that in
1687 SRP_VBASE_get1_by_user, caller must free the returned value. Note
1688 also that even though configuring the SRP seed attempts to hide
1689 invalid usernames by continuing the handshake with fake
1690 credentials, this behaviour is not constant time and no strong
1691 guarantees are made that the handshake is indistinguishable from
1692 that of a valid user.
1696 *) Fix BN_hex2bn/BN_dec2bn NULL pointer deref/heap corruption
1698 In the BN_hex2bn function the number of hex digits is calculated using an
1699 int value |i|. Later |bn_expand| is called with a value of |i * 4|. For
1700 large values of |i| this can result in |bn_expand| not allocating any
1701 memory because |i * 4| is negative. This can leave the internal BIGNUM data
1702 field as NULL leading to a subsequent NULL ptr deref. For very large values
1703 of |i|, the calculation |i * 4| could be a positive value smaller than |i|.
1704 In this case memory is allocated to the internal BIGNUM data field, but it
1705 is insufficiently sized leading to heap corruption. A similar issue exists
1706 in BN_dec2bn. This could have security consequences if BN_hex2bn/BN_dec2bn
1707 is ever called by user applications with very large untrusted hex/dec data.
1708 This is anticipated to be a rare occurrence.
1710 All OpenSSL internal usage of these functions use data that is not expected
1711 to be untrusted, e.g. config file data or application command line
1712 arguments. If user developed applications generate config file data based
1713 on untrusted data then it is possible that this could also lead to security
1714 consequences. This is also anticipated to be rare.
1716 This issue was reported to OpenSSL by Guido Vranken.
1720 *) Fix memory issues in BIO_*printf functions
1722 The internal |fmtstr| function used in processing a "%s" format string in
1723 the BIO_*printf functions could overflow while calculating the length of a
1724 string and cause an OOB read when printing very long strings.
1726 Additionally the internal |doapr_outch| function can attempt to write to an
1727 OOB memory location (at an offset from the NULL pointer) in the event of a
1728 memory allocation failure. In 1.0.2 and below this could be caused where
1729 the size of a buffer to be allocated is greater than INT_MAX. E.g. this
1730 could be in processing a very long "%s" format string. Memory leaks can
1733 The first issue may mask the second issue dependent on compiler behaviour.
1734 These problems could enable attacks where large amounts of untrusted data
1735 is passed to the BIO_*printf functions. If applications use these functions
1736 in this way then they could be vulnerable. OpenSSL itself uses these
1737 functions when printing out human-readable dumps of ASN.1 data. Therefore
1738 applications that print this data could be vulnerable if the data is from
1739 untrusted sources. OpenSSL command line applications could also be
1740 vulnerable where they print out ASN.1 data, or if untrusted data is passed
1741 as command line arguments.
1743 Libssl is not considered directly vulnerable. Additionally certificates etc
1744 received via remote connections via libssl are also unlikely to be able to
1745 trigger these issues because of message size limits enforced within libssl.
1747 This issue was reported to OpenSSL Guido Vranken.
1751 *) Side channel attack on modular exponentiation
1753 A side-channel attack was found which makes use of cache-bank conflicts on
1754 the Intel Sandy-Bridge microarchitecture which could lead to the recovery
1755 of RSA keys. The ability to exploit this issue is limited as it relies on
1756 an attacker who has control of code in a thread running on the same
1757 hyper-threaded core as the victim thread which is performing decryptions.
1759 This issue was reported to OpenSSL by Yuval Yarom, The University of
1760 Adelaide and NICTA, Daniel Genkin, Technion and Tel Aviv University, and
1761 Nadia Heninger, University of Pennsylvania with more information at
1762 http://cachebleed.info.
1766 *) Change the req app to generate a 2048-bit RSA/DSA key by default,
1767 if no keysize is specified with default_bits. This fixes an
1768 omission in an earlier change that changed all RSA/DSA key generation
1769 apps to use 2048 bits by default.
1772 Changes between 1.0.2e and 1.0.2f [28 Jan 2016]
1773 *) DH small subgroups
1775 Historically OpenSSL only ever generated DH parameters based on "safe"
1776 primes. More recently (in version 1.0.2) support was provided for
1777 generating X9.42 style parameter files such as those required for RFC 5114
1778 support. The primes used in such files may not be "safe". Where an
1779 application is using DH configured with parameters based on primes that are
1780 not "safe" then an attacker could use this fact to find a peer's private
1781 DH exponent. This attack requires that the attacker complete multiple
1782 handshakes in which the peer uses the same private DH exponent. For example
1783 this could be used to discover a TLS server's private DH exponent if it's
1784 reusing the private DH exponent or it's using a static DH ciphersuite.
1786 OpenSSL provides the option SSL_OP_SINGLE_DH_USE for ephemeral DH (DHE) in
1787 TLS. It is not on by default. If the option is not set then the server
1788 reuses the same private DH exponent for the life of the server process and
1789 would be vulnerable to this attack. It is believed that many popular
1790 applications do set this option and would therefore not be at risk.
1792 The fix for this issue adds an additional check where a "q" parameter is
1793 available (as is the case in X9.42 based parameters). This detects the
1794 only known attack, and is the only possible defense for static DH
1795 ciphersuites. This could have some performance impact.
1797 Additionally the SSL_OP_SINGLE_DH_USE option has been switched on by
1798 default and cannot be disabled. This could have some performance impact.
1800 This issue was reported to OpenSSL by Antonio Sanso (Adobe).
1804 *) SSLv2 doesn't block disabled ciphers
1806 A malicious client can negotiate SSLv2 ciphers that have been disabled on
1807 the server and complete SSLv2 handshakes even if all SSLv2 ciphers have
1808 been disabled, provided that the SSLv2 protocol was not also disabled via
1811 This issue was reported to OpenSSL on 26th December 2015 by Nimrod Aviram
1812 and Sebastian Schinzel.
1816 Changes between 1.0.2d and 1.0.2e [3 Dec 2015]
1818 *) BN_mod_exp may produce incorrect results on x86_64
1820 There is a carry propagating bug in the x86_64 Montgomery squaring
1821 procedure. No EC algorithms are affected. Analysis suggests that attacks
1822 against RSA and DSA as a result of this defect would be very difficult to
1823 perform and are not believed likely. Attacks against DH are considered just
1824 feasible (although very difficult) because most of the work necessary to
1825 deduce information about a private key may be performed offline. The amount
1826 of resources required for such an attack would be very significant and
1827 likely only accessible to a limited number of attackers. An attacker would
1828 additionally need online access to an unpatched system using the target
1829 private key in a scenario with persistent DH parameters and a private
1830 key that is shared between multiple clients. For example this can occur by
1831 default in OpenSSL DHE based SSL/TLS ciphersuites.
1833 This issue was reported to OpenSSL by Hanno Böck.
1837 *) Certificate verify crash with missing PSS parameter
1839 The signature verification routines will crash with a NULL pointer
1840 dereference if presented with an ASN.1 signature using the RSA PSS
1841 algorithm and absent mask generation function parameter. Since these
1842 routines are used to verify certificate signature algorithms this can be
1843 used to crash any certificate verification operation and exploited in a
1844 DoS attack. Any application which performs certificate verification is
1845 vulnerable including OpenSSL clients and servers which enable client
1848 This issue was reported to OpenSSL by Loïc Jonas Etienne (Qnective AG).
1852 *) X509_ATTRIBUTE memory leak
1854 When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak
1855 memory. This structure is used by the PKCS#7 and CMS routines so any
1856 application which reads PKCS#7 or CMS data from untrusted sources is
1857 affected. SSL/TLS is not affected.
1859 This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using
1864 *) Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs.
1865 This changes the decoding behaviour for some invalid messages,
1866 though the change is mostly in the more lenient direction, and
1867 legacy behaviour is preserved as much as possible.
1870 *) In DSA_generate_parameters_ex, if the provided seed is too short,
1872 [Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>]
1874 Changes between 1.0.2c and 1.0.2d [9 Jul 2015]
1876 *) Alternate chains certificate forgery
1878 During certificate verification, OpenSSL will attempt to find an
1879 alternative certificate chain if the first attempt to build such a chain
1880 fails. An error in the implementation of this logic can mean that an
1881 attacker could cause certain checks on untrusted certificates to be
1882 bypassed, such as the CA flag, enabling them to use a valid leaf
1883 certificate to act as a CA and "issue" an invalid certificate.
1885 This issue was reported to OpenSSL by Adam Langley/David Benjamin
1889 Changes between 1.0.2b and 1.0.2c [12 Jun 2015]
1891 *) Fix HMAC ABI incompatibility. The previous version introduced an ABI
1892 incompatibility in the handling of HMAC. The previous ABI has now been
1896 Changes between 1.0.2a and 1.0.2b [11 Jun 2015]
1898 *) Malformed ECParameters causes infinite loop
1900 When processing an ECParameters structure OpenSSL enters an infinite loop
1901 if the curve specified is over a specially malformed binary polynomial
1904 This can be used to perform denial of service against any
1905 system which processes public keys, certificate requests or
1906 certificates. This includes TLS clients and TLS servers with
1907 client authentication enabled.
1909 This issue was reported to OpenSSL by Joseph Barr-Pixton.
1913 *) Exploitable out-of-bounds read in X509_cmp_time
1915 X509_cmp_time does not properly check the length of the ASN1_TIME
1916 string and can read a few bytes out of bounds. In addition,
1917 X509_cmp_time accepts an arbitrary number of fractional seconds in the
1920 An attacker can use this to craft malformed certificates and CRLs of
1921 various sizes and potentially cause a segmentation fault, resulting in
1922 a DoS on applications that verify certificates or CRLs. TLS clients
1923 that verify CRLs are affected. TLS clients and servers with client
1924 authentication enabled may be affected if they use custom verification
1927 This issue was reported to OpenSSL by Robert Swiecki (Google), and
1928 independently by Hanno Böck.
1932 *) PKCS7 crash with missing EnvelopedContent
1934 The PKCS#7 parsing code does not handle missing inner EncryptedContent
1935 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs
1936 with missing content and trigger a NULL pointer dereference on parsing.
1938 Applications that decrypt PKCS#7 data or otherwise parse PKCS#7
1939 structures from untrusted sources are affected. OpenSSL clients and
1940 servers are not affected.
1942 This issue was reported to OpenSSL by Michal Zalewski (Google).
1946 *) CMS verify infinite loop with unknown hash function
1948 When verifying a signedData message the CMS code can enter an infinite loop
1949 if presented with an unknown hash function OID. This can be used to perform
1950 denial of service against any system which verifies signedData messages using
1952 This issue was reported to OpenSSL by Johannes Bauer.
1956 *) Race condition handling NewSessionTicket
1958 If a NewSessionTicket is received by a multi-threaded client when attempting to
1959 reuse a previous ticket then a race condition can occur potentially leading to
1960 a double free of the ticket data.
1964 *) Only support 256-bit or stronger elliptic curves with the
1965 'ecdh_auto' setting (server) or by default (client). Of supported
1966 curves, prefer P-256 (both).
1969 Changes between 1.0.2 and 1.0.2a [19 Mar 2015]
1971 *) ClientHello sigalgs DoS fix
1973 If a client connects to an OpenSSL 1.0.2 server and renegotiates with an
1974 invalid signature algorithms extension a NULL pointer dereference will
1975 occur. This can be exploited in a DoS attack against the server.
1977 This issue was was reported to OpenSSL by David Ramos of Stanford
1980 [Stephen Henson and Matt Caswell]
1982 *) Multiblock corrupted pointer fix
1984 OpenSSL 1.0.2 introduced the "multiblock" performance improvement. This
1985 feature only applies on 64 bit x86 architecture platforms that support AES
1986 NI instructions. A defect in the implementation of "multiblock" can cause
1987 OpenSSL's internal write buffer to become incorrectly set to NULL when
1988 using non-blocking IO. Typically, when the user application is using a
1989 socket BIO for writing, this will only result in a failed connection.
1990 However if some other BIO is used then it is likely that a segmentation
1991 fault will be triggered, thus enabling a potential DoS attack.
1993 This issue was reported to OpenSSL by Daniel Danner and Rainer Mueller.
1997 *) Segmentation fault in DTLSv1_listen fix
1999 The DTLSv1_listen function is intended to be stateless and processes the
2000 initial ClientHello from many peers. It is common for user code to loop
2001 over the call to DTLSv1_listen until a valid ClientHello is received with
2002 an associated cookie. A defect in the implementation of DTLSv1_listen means
2003 that state is preserved in the SSL object from one invocation to the next
2004 that can lead to a segmentation fault. Errors processing the initial
2005 ClientHello can trigger this scenario. An example of such an error could be
2006 that a DTLS1.0 only client is attempting to connect to a DTLS1.2 only
2009 This issue was reported to OpenSSL by Per Allansson.
2013 *) Segmentation fault in ASN1_TYPE_cmp fix
2015 The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is
2016 made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check
2017 certificate signature algorithm consistency this can be used to crash any
2018 certificate verification operation and exploited in a DoS attack. Any
2019 application which performs certificate verification is vulnerable including
2020 OpenSSL clients and servers which enable client authentication.
2024 *) Segmentation fault for invalid PSS parameters fix
2026 The signature verification routines will crash with a NULL pointer
2027 dereference if presented with an ASN.1 signature using the RSA PSS
2028 algorithm and invalid parameters. Since these routines are used to verify
2029 certificate signature algorithms this can be used to crash any
2030 certificate verification operation and exploited in a DoS attack. Any
2031 application which performs certificate verification is vulnerable including
2032 OpenSSL clients and servers which enable client authentication.
2034 This issue was was reported to OpenSSL by Brian Carpenter.
2038 *) ASN.1 structure reuse memory corruption fix
2040 Reusing a structure in ASN.1 parsing may allow an attacker to cause
2041 memory corruption via an invalid write. Such reuse is and has been
2042 strongly discouraged and is believed to be rare.
2044 Applications that parse structures containing CHOICE or ANY DEFINED BY
2045 components may be affected. Certificate parsing (d2i_X509 and related
2046 functions) are however not affected. OpenSSL clients and servers are
2051 *) PKCS7 NULL pointer dereferences fix
2053 The PKCS#7 parsing code does not handle missing outer ContentInfo
2054 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with
2055 missing content and trigger a NULL pointer dereference on parsing.
2057 Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or
2058 otherwise parse PKCS#7 structures from untrusted sources are
2059 affected. OpenSSL clients and servers are not affected.
2061 This issue was reported to OpenSSL by Michal Zalewski (Google).
2065 *) DoS via reachable assert in SSLv2 servers fix
2067 A malicious client can trigger an OPENSSL_assert (i.e., an abort) in
2068 servers that both support SSLv2 and enable export cipher suites by sending
2069 a specially crafted SSLv2 CLIENT-MASTER-KEY message.
2071 This issue was discovered by Sean Burford (Google) and Emilia Käsper
2072 (OpenSSL development team).
2076 *) Empty CKE with client auth and DHE fix
2078 If client auth is used then a server can seg fault in the event of a DHE
2079 ciphersuite being selected and a zero length ClientKeyExchange message
2080 being sent by the client. This could be exploited in a DoS attack.
2084 *) Handshake with unseeded PRNG fix
2086 Under certain conditions an OpenSSL 1.0.2 client can complete a handshake
2087 with an unseeded PRNG. The conditions are:
2088 - The client is on a platform where the PRNG has not been seeded
2089 automatically, and the user has not seeded manually
2090 - A protocol specific client method version has been used (i.e. not
2091 SSL_client_methodv23)
2092 - A ciphersuite is used that does not require additional random data from
2093 the PRNG beyond the initial ClientHello client random (e.g. PSK-RC4-SHA).
2095 If the handshake succeeds then the client random that has been used will
2096 have been generated from a PRNG with insufficient entropy and therefore the
2097 output may be predictable.
2099 For example using the following command with an unseeded openssl will
2100 succeed on an unpatched platform:
2102 openssl s_client -psk 1a2b3c4d -tls1_2 -cipher PSK-RC4-SHA
2106 *) Use After Free following d2i_ECPrivatekey error fix
2108 A malformed EC private key file consumed via the d2i_ECPrivateKey function
2109 could cause a use after free condition. This, in turn, could cause a double
2110 free in several private key parsing functions (such as d2i_PrivateKey
2111 or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption
2112 for applications that receive EC private keys from untrusted
2113 sources. This scenario is considered rare.
2115 This issue was discovered by the BoringSSL project and fixed in their
2120 *) X509_to_X509_REQ NULL pointer deref fix
2122 The function X509_to_X509_REQ will crash with a NULL pointer dereference if
2123 the certificate key is invalid. This function is rarely used in practice.
2125 This issue was discovered by Brian Carpenter.
2129 *) Removed the export ciphers from the DEFAULT ciphers
2132 Changes between 1.0.1l and 1.0.2 [22 Jan 2015]
2134 *) Facilitate "universal" ARM builds targeting range of ARM ISAs, e.g.
2135 ARMv5 through ARMv8, as opposite to "locking" it to single one.
2136 So far those who have to target multiple platforms would compromise
2137 and argue that binary targeting say ARMv5 would still execute on
2138 ARMv8. "Universal" build resolves this compromise by providing
2139 near-optimal performance even on newer platforms.
2142 *) Accelerated NIST P-256 elliptic curve implementation for x86_64
2143 (other platforms pending).
2144 [Shay Gueron & Vlad Krasnov (Intel Corp), Andy Polyakov]
2146 *) Add support for the SignedCertificateTimestampList certificate and
2147 OCSP response extensions from RFC6962.
2150 *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.)
2151 for corner cases. (Certain input points at infinity could lead to
2152 bogus results, with non-infinity inputs mapped to infinity too.)
2155 *) Initial support for PowerISA 2.0.7, first implemented in POWER8.
2156 This covers AES, SHA256/512 and GHASH. "Initial" means that most
2157 common cases are optimized and there still is room for further
2158 improvements. Vector Permutation AES for Altivec is also added.
2161 *) Add support for little-endian ppc64 Linux target.
2162 [Marcelo Cerri (IBM)]
2164 *) Initial support for AMRv8 ISA crypto extensions. This covers AES,
2165 SHA1, SHA256 and GHASH. "Initial" means that most common cases
2166 are optimized and there still is room for further improvements.
2167 Both 32- and 64-bit modes are supported.
2168 [Andy Polyakov, Ard Biesheuvel (Linaro)]
2170 *) Improved ARMv7 NEON support.
2173 *) Support for SPARC Architecture 2011 crypto extensions, first
2174 implemented in SPARC T4. This covers AES, DES, Camellia, SHA1,
2175 SHA256/512, MD5, GHASH and modular exponentiation.
2176 [Andy Polyakov, David Miller]
2178 *) Accelerated modular exponentiation for Intel processors, a.k.a.
2180 [Shay Gueron & Vlad Krasnov (Intel Corp)]
2182 *) Support for new and upcoming Intel processors, including AVX2,
2183 BMI and SHA ISA extensions. This includes additional "stitched"
2184 implementations, AESNI-SHA256 and GCM, and multi-buffer support
2187 This work was sponsored by Intel Corp.
2190 *) Support for DTLS 1.2. This adds two sets of DTLS methods: DTLS_*_method()
2191 supports both DTLS 1.2 and 1.0 and should use whatever version the peer
2192 supports and DTLSv1_2_*_method() which supports DTLS 1.2 only.
2195 *) Use algorithm specific chains in SSL_CTX_use_certificate_chain_file():
2196 this fixes a limitation in previous versions of OpenSSL.
2199 *) Extended RSA OAEP support via EVP_PKEY API. Options to specify digest,
2200 MGF1 digest and OAEP label.
2203 *) Add EVP support for key wrapping algorithms, to avoid problems with
2204 existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in
2205 the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap
2206 algorithms and include tests cases.
2209 *) Add functions to allocate and set the fields of an ECDSA_METHOD
2211 [Douglas E. Engert, Steve Henson]
2213 *) New functions OPENSSL_gmtime_diff and ASN1_TIME_diff to find the
2214 difference in days and seconds between two tm or ASN1_TIME structures.
2217 *) Add -rev test option to s_server to just reverse order of characters
2218 received by client and send back to server. Also prints an abbreviated
2219 summary of the connection parameters.
2222 *) New option -brief for s_client and s_server to print out a brief summary
2223 of connection parameters.
2226 *) Add callbacks for arbitrary TLS extensions.
2227 [Trevor Perrin <trevp@trevp.net> and Ben Laurie]
2229 *) New option -crl_download in several openssl utilities to download CRLs
2230 from CRLDP extension in certificates.
2233 *) New options -CRL and -CRLform for s_client and s_server for CRLs.
2236 *) New function X509_CRL_diff to generate a delta CRL from the difference
2237 of two full CRLs. Add support to "crl" utility.
2240 *) New functions to set lookup_crls function and to retrieve
2241 X509_STORE from X509_STORE_CTX.
2244 *) Print out deprecated issuer and subject unique ID fields in
2248 *) Extend OCSP I/O functions so they can be used for simple general purpose
2249 HTTP as well as OCSP. New wrapper function which can be used to download
2250 CRLs using the OCSP API.
2253 *) Delegate command line handling in s_client/s_server to SSL_CONF APIs.
2256 *) SSL_CONF* functions. These provide a common framework for application
2257 configuration using configuration files or command lines.
2260 *) SSL/TLS tracing code. This parses out SSL/TLS records using the
2261 message callback and prints the results. Needs compile time option
2262 "enable-ssl-trace". New options to s_client and s_server to enable
2266 *) New ctrl and macro to retrieve supported points extensions.
2267 Print out extension in s_server and s_client.
2270 *) New functions to retrieve certificate signature and signature
2274 *) Add functions to retrieve and manipulate the raw cipherlist sent by a
2278 *) New Suite B modes for TLS code. These use and enforce the requirements
2279 of RFC6460: restrict ciphersuites, only permit Suite B algorithms and
2280 only use Suite B curves. The Suite B modes can be set by using the
2281 strings "SUITEB128", "SUITEB192" or "SUITEB128ONLY" for the cipherstring.
2284 *) New chain verification flags for Suite B levels of security. Check
2285 algorithms are acceptable when flags are set in X509_verify_cert.
2288 *) Make tls1_check_chain return a set of flags indicating checks passed
2289 by a certificate chain. Add additional tests to handle client
2290 certificates: checks for matching certificate type and issuer name
2294 *) If an attempt is made to use a signature algorithm not in the peer
2295 preference list abort the handshake. If client has no suitable
2296 signature algorithms in response to a certificate request do not
2297 use the certificate.
2300 *) If server EC tmp key is not in client preference list abort handshake.
2303 *) Add support for certificate stores in CERT structure. This makes it
2304 possible to have different stores per SSL structure or one store in
2305 the parent SSL_CTX. Include distinct stores for certificate chain
2306 verification and chain building. New ctrl SSL_CTRL_BUILD_CERT_CHAIN
2307 to build and store a certificate chain in CERT structure: returning
2308 an error if the chain cannot be built: this will allow applications
2309 to test if a chain is correctly configured.
2311 Note: if the CERT based stores are not set then the parent SSL_CTX
2312 store is used to retain compatibility with existing behaviour.
2316 *) New function ssl_set_client_disabled to set a ciphersuite disabled
2317 mask based on the current session, check mask when sending client
2318 hello and checking the requested ciphersuite.
2321 *) New ctrls to retrieve and set certificate types in a certificate
2322 request message. Print out received values in s_client. If certificate
2323 types is not set with custom values set sensible values based on
2324 supported signature algorithms.
2327 *) Support for distinct client and server supported signature algorithms.
2330 *) Add certificate callback. If set this is called whenever a certificate
2331 is required by client or server. An application can decide which
2332 certificate chain to present based on arbitrary criteria: for example
2333 supported signature algorithms. Add very simple example to s_server.
2334 This fixes many of the problems and restrictions of the existing client
2335 certificate callback: for example you can now clear an existing
2336 certificate and specify the whole chain.
2339 *) Add new "valid_flags" field to CERT_PKEY structure which determines what
2340 the certificate can be used for (if anything). Set valid_flags field
2341 in new tls1_check_chain function. Simplify ssl_set_cert_masks which used
2342 to have similar checks in it.
2344 Add new "cert_flags" field to CERT structure and include a "strict mode".
2345 This enforces some TLS certificate requirements (such as only permitting
2346 certificate signature algorithms contained in the supported algorithms
2347 extension) which some implementations ignore: this option should be used
2348 with caution as it could cause interoperability issues.
2351 *) Update and tidy signature algorithm extension processing. Work out
2352 shared signature algorithms based on preferences and peer algorithms
2353 and print them out in s_client and s_server. Abort handshake if no
2354 shared signature algorithms.
2357 *) Add new functions to allow customised supported signature algorithms
2358 for SSL and SSL_CTX structures. Add options to s_client and s_server
2362 *) New function SSL_certs_clear() to delete all references to certificates
2363 from an SSL structure. Before this once a certificate had been added
2364 it couldn't be removed.
2367 *) Integrate hostname, email address and IP address checking with certificate
2368 verification. New verify options supporting checking in openssl utility.
2371 *) Fixes and wildcard matching support to hostname and email checking
2372 functions. Add manual page.
2373 [Florian Weimer (Red Hat Product Security Team)]
2375 *) New functions to check a hostname email or IP address against a
2376 certificate. Add options x509 utility to print results of checks against
2380 *) Fix OCSP checking.
2381 [Rob Stradling <rob.stradling@comodo.com> and Ben Laurie]
2383 *) Initial experimental support for explicitly trusted non-root CAs.
2384 OpenSSL still tries to build a complete chain to a root but if an
2385 intermediate CA has a trust setting included that is used. The first
2386 setting is used: whether to trust (e.g., -addtrust option to the x509
2390 *) Add -trusted_first option which attempts to find certificates in the
2391 trusted store even if an untrusted chain is also supplied.
2394 *) MIPS assembly pack updates: support for MIPS32r2 and SmartMIPS ASE,
2395 platform support for Linux and Android.
2398 *) Support for linux-x32, ILP32 environment in x86_64 framework.
2401 *) Experimental multi-implementation support for FIPS capable OpenSSL.
2402 When in FIPS mode the approved implementations are used as normal,
2403 when not in FIPS mode the internal unapproved versions are used instead.
2404 This means that the FIPS capable OpenSSL isn't forced to use the
2405 (often lower performance) FIPS implementations outside FIPS mode.
2408 *) Transparently support X9.42 DH parameters when calling
2409 PEM_read_bio_DHparameters. This means existing applications can handle
2410 the new parameter format automatically.
2413 *) Initial experimental support for X9.42 DH parameter format: mainly
2414 to support use of 'q' parameter for RFC5114 parameters.
2417 *) Add DH parameters from RFC5114 including test data to dhtest.
2420 *) Support for automatic EC temporary key parameter selection. If enabled
2421 the most preferred EC parameters are automatically used instead of
2422 hardcoded fixed parameters. Now a server just has to call:
2423 SSL_CTX_set_ecdh_auto(ctx, 1) and the server will automatically
2424 support ECDH and use the most appropriate parameters.
2427 *) Enhance and tidy EC curve and point format TLS extension code. Use
2428 static structures instead of allocation if default values are used.
2429 New ctrls to set curves we wish to support and to retrieve shared curves.
2430 Print out shared curves in s_server. New options to s_server and s_client
2431 to set list of supported curves.
2434 *) New ctrls to retrieve supported signature algorithms and
2435 supported curve values as an array of NIDs. Extend openssl utility
2436 to print out received values.
2439 *) Add new APIs EC_curve_nist2nid and EC_curve_nid2nist which convert
2440 between NIDs and the more common NIST names such as "P-256". Enhance
2441 ecparam utility and ECC method to recognise the NIST names for curves.
2444 *) Enhance SSL/TLS certificate chain handling to support different
2445 chains for each certificate instead of one chain in the parent SSL_CTX.
2448 *) Support for fixed DH ciphersuite client authentication: where both
2449 server and client use DH certificates with common parameters.
2452 *) Support for fixed DH ciphersuites: those requiring DH server
2456 *) New function i2d_re_X509_tbs for re-encoding the TBS portion of
2458 Note: Related 1.0.2-beta specific macros X509_get_cert_info,
2459 X509_CINF_set_modified, X509_CINF_get_issuer, X509_CINF_get_extensions and
2460 X509_CINF_get_signature were reverted post internal team review.
2462 Changes between 1.0.1k and 1.0.1l [15 Jan 2015]
2464 *) Build fixes for the Windows and OpenVMS platforms
2465 [Matt Caswell and Richard Levitte]
2467 Changes between 1.0.1j and 1.0.1k [8 Jan 2015]
2469 *) Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS
2470 message can cause a segmentation fault in OpenSSL due to a NULL pointer
2471 dereference. This could lead to a Denial Of Service attack. Thanks to
2472 Markus Stenberg of Cisco Systems, Inc. for reporting this issue.
2476 *) Fix DTLS memory leak in dtls1_buffer_record. A memory leak can occur in the
2477 dtls1_buffer_record function under certain conditions. In particular this
2478 could occur if an attacker sent repeated DTLS records with the same
2479 sequence number but for the next epoch. The memory leak could be exploited
2480 by an attacker in a Denial of Service attack through memory exhaustion.
2481 Thanks to Chris Mueller for reporting this issue.
2485 *) Fix issue where no-ssl3 configuration sets method to NULL. When openssl is
2486 built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl
2487 method would be set to NULL which could later result in a NULL pointer
2488 dereference. Thanks to Frank Schmirler for reporting this issue.
2492 *) Abort handshake if server key exchange message is omitted for ephemeral
2495 Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for
2496 reporting this issue.
2500 *) Remove non-export ephemeral RSA code on client and server. This code
2501 violated the TLS standard by allowing the use of temporary RSA keys in
2502 non-export ciphersuites and could be used by a server to effectively
2503 downgrade the RSA key length used to a value smaller than the server
2504 certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at
2505 INRIA or reporting this issue.
2509 *) Fixed issue where DH client certificates are accepted without verification.
2510 An OpenSSL server will accept a DH certificate for client authentication
2511 without the certificate verify message. This effectively allows a client to
2512 authenticate without the use of a private key. This only affects servers
2513 which trust a client certificate authority which issues certificates
2514 containing DH keys: these are extremely rare and hardly ever encountered.
2515 Thanks for Karthikeyan Bhargavan of the PROSECCO team at INRIA or reporting
2520 *) Ensure that the session ID context of an SSL is updated when its
2521 SSL_CTX is updated via SSL_set_SSL_CTX.
2523 The session ID context is typically set from the parent SSL_CTX,
2524 and can vary with the CTX.
2527 *) Fix various certificate fingerprint issues.
2529 By using non-DER or invalid encodings outside the signed portion of a
2530 certificate the fingerprint can be changed without breaking the signature.
2531 Although no details of the signed portion of the certificate can be changed
2532 this can cause problems with some applications: e.g. those using the
2533 certificate fingerprint for blacklists.
2535 1. Reject signatures with non zero unused bits.
2537 If the BIT STRING containing the signature has non zero unused bits reject
2538 the signature. All current signature algorithms require zero unused bits.
2540 2. Check certificate algorithm consistency.
2542 Check the AlgorithmIdentifier inside TBS matches the one in the
2543 certificate signature. NB: this will result in signature failure
2544 errors for some broken certificates.
2546 Thanks to Konrad Kraszewski from Google for reporting this issue.
2548 3. Check DSA/ECDSA signatures use DER.
2550 Re-encode DSA/ECDSA signatures and compare with the original received
2551 signature. Return an error if there is a mismatch.
2553 This will reject various cases including garbage after signature
2554 (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS
2555 program for discovering this case) and use of BER or invalid ASN.1 INTEGERs
2556 (negative or with leading zeroes).
2558 Further analysis was conducted and fixes were developed by Stephen Henson
2559 of the OpenSSL core team.
2564 *) Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect
2565 results on some platforms, including x86_64. This bug occurs at random
2566 with a very low probability, and is not known to be exploitable in any
2567 way, though its exact impact is difficult to determine. Thanks to Pieter
2568 Wuille (Blockstream) who reported this issue and also suggested an initial
2569 fix. Further analysis was conducted by the OpenSSL development team and
2570 Adam Langley of Google. The final fix was developed by Andy Polyakov of
2571 the OpenSSL core team.
2575 *) Do not resume sessions on the server if the negotiated protocol
2576 version does not match the session's version. Resuming with a different
2577 version, while not strictly forbidden by the RFC, is of questionable
2578 sanity and breaks all known clients.
2579 [David Benjamin, Emilia Käsper]
2581 *) Tighten handling of the ChangeCipherSpec (CCS) message: reject
2582 early CCS messages during renegotiation. (Note that because
2583 renegotiation is encrypted, this early CCS was not exploitable.)
2586 *) Tighten client-side session ticket handling during renegotiation:
2587 ensure that the client only accepts a session ticket if the server sends
2588 the extension anew in the ServerHello. Previously, a TLS client would
2589 reuse the old extension state and thus accept a session ticket if one was
2590 announced in the initial ServerHello.
2592 Similarly, ensure that the client requires a session ticket if one
2593 was advertised in the ServerHello. Previously, a TLS client would
2594 ignore a missing NewSessionTicket message.
2597 Changes between 1.0.1i and 1.0.1j [15 Oct 2014]
2599 *) SRTP Memory Leak.
2601 A flaw in the DTLS SRTP extension parsing code allows an attacker, who
2602 sends a carefully crafted handshake message, to cause OpenSSL to fail
2603 to free up to 64k of memory causing a memory leak. This could be
2604 exploited in a Denial Of Service attack. This issue affects OpenSSL
2605 1.0.1 server implementations for both SSL/TLS and DTLS regardless of
2606 whether SRTP is used or configured. Implementations of OpenSSL that
2607 have been compiled with OPENSSL_NO_SRTP defined are not affected.
2609 The fix was developed by the OpenSSL team.
2613 *) Session Ticket Memory Leak.
2615 When an OpenSSL SSL/TLS/DTLS server receives a session ticket the
2616 integrity of that ticket is first verified. In the event of a session
2617 ticket integrity check failing, OpenSSL will fail to free memory
2618 causing a memory leak. By sending a large number of invalid session
2619 tickets an attacker could exploit this issue in a Denial Of Service
2624 *) Build option no-ssl3 is incomplete.
2626 When OpenSSL is configured with "no-ssl3" as a build option, servers
2627 could accept and complete a SSL 3.0 handshake, and clients could be
2628 configured to send them.
2630 [Akamai and the OpenSSL team]
2632 *) Add support for TLS_FALLBACK_SCSV.
2633 Client applications doing fallback retries should call
2634 SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV).
2636 [Adam Langley, Bodo Moeller]
2638 *) Add additional DigestInfo checks.
2640 Re-encode DigestInto in DER and check against the original when
2641 verifying RSA signature: this will reject any improperly encoded
2642 DigestInfo structures.
2644 Note: this is a precautionary measure and no attacks are currently known.
2648 Changes between 1.0.1h and 1.0.1i [6 Aug 2014]
2650 *) Fix SRP buffer overrun vulnerability. Invalid parameters passed to the
2651 SRP code can be overrun an internal buffer. Add sanity check that
2652 g, A, B < N to SRP code.
2654 Thanks to Sean Devlin and Watson Ladd of Cryptography Services, NCC
2655 Group for discovering this issue.
2659 *) A flaw in the OpenSSL SSL/TLS server code causes the server to negotiate
2660 TLS 1.0 instead of higher protocol versions when the ClientHello message
2661 is badly fragmented. This allows a man-in-the-middle attacker to force a
2662 downgrade to TLS 1.0 even if both the server and the client support a
2663 higher protocol version, by modifying the client's TLS records.
2665 Thanks to David Benjamin and Adam Langley (Google) for discovering and
2666 researching this issue.
2670 *) OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject
2671 to a denial of service attack. A malicious server can crash the client
2672 with a null pointer dereference (read) by specifying an anonymous (EC)DH
2673 ciphersuite and sending carefully crafted handshake messages.
2675 Thanks to Felix Gröbert (Google) for discovering and researching this
2680 *) By sending carefully crafted DTLS packets an attacker could cause openssl
2681 to leak memory. This can be exploited through a Denial of Service attack.
2682 Thanks to Adam Langley for discovering and researching this issue.
2686 *) An attacker can force openssl to consume large amounts of memory whilst
2687 processing DTLS handshake messages. This can be exploited through a
2688 Denial of Service attack.
2689 Thanks to Adam Langley for discovering and researching this issue.
2693 *) An attacker can force an error condition which causes openssl to crash
2694 whilst processing DTLS packets due to memory being freed twice. This
2695 can be exploited through a Denial of Service attack.
2696 Thanks to Adam Langley and Wan-Teh Chang for discovering and researching
2701 *) If a multithreaded client connects to a malicious server using a resumed
2702 session and the server sends an ec point format extension it could write
2703 up to 255 bytes to freed memory.
2705 Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this
2710 *) A malicious server can crash an OpenSSL client with a null pointer
2711 dereference (read) by specifying an SRP ciphersuite even though it was not
2712 properly negotiated with the client. This can be exploited through a
2713 Denial of Service attack.
2715 Thanks to Joonas Kuorilehto and Riku Hietamäki (Codenomicon) for
2716 discovering and researching this issue.
2720 *) A flaw in OBJ_obj2txt may cause pretty printing functions such as
2721 X509_name_oneline, X509_name_print_ex et al. to leak some information
2722 from the stack. Applications may be affected if they echo pretty printing
2723 output to the attacker.
2725 Thanks to Ivan Fratric (Google) for discovering this issue.
2727 [Emilia Käsper, and Steve Henson]
2729 *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.)
2730 for corner cases. (Certain input points at infinity could lead to
2731 bogus results, with non-infinity inputs mapped to infinity too.)
2734 Changes between 1.0.1g and 1.0.1h [5 Jun 2014]
2736 *) Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted
2737 handshake can force the use of weak keying material in OpenSSL
2738 SSL/TLS clients and servers.
2740 Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and
2741 researching this issue. (CVE-2014-0224)
2742 [KIKUCHI Masashi, Steve Henson]
2744 *) Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an
2745 OpenSSL DTLS client the code can be made to recurse eventually crashing
2748 Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue.
2750 [Imre Rad, Steve Henson]
2752 *) Fix DTLS invalid fragment vulnerability. A buffer overrun attack can
2753 be triggered by sending invalid DTLS fragments to an OpenSSL DTLS
2754 client or server. This is potentially exploitable to run arbitrary
2755 code on a vulnerable client or server.
2757 Thanks to Jüri Aedla for reporting this issue. (CVE-2014-0195)
2758 [Jüri Aedla, Steve Henson]
2760 *) Fix bug in TLS code where clients enable anonymous ECDH ciphersuites
2761 are subject to a denial of service attack.
2763 Thanks to Felix Gröbert and Ivan Fratric at Google for discovering
2764 this issue. (CVE-2014-3470)
2765 [Felix Gröbert, Ivan Fratric, Steve Henson]
2767 *) Harmonize version and its documentation. -f flag is used to display
2769 [mancha <mancha1@zoho.com>]
2771 *) Fix eckey_priv_encode so it immediately returns an error upon a failure
2772 in i2d_ECPrivateKey.
2773 [mancha <mancha1@zoho.com>]
2775 *) Fix some double frees. These are not thought to be exploitable.
2776 [mancha <mancha1@zoho.com>]
2778 Changes between 1.0.1f and 1.0.1g [7 Apr 2014]
2780 *) A missing bounds check in the handling of the TLS heartbeat extension
2781 can be used to reveal up to 64k of memory to a connected client or
2784 Thanks for Neel Mehta of Google Security for discovering this bug and to
2785 Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for
2786 preparing the fix (CVE-2014-0160)
2787 [Adam Langley, Bodo Moeller]
2789 *) Fix for the attack described in the paper "Recovering OpenSSL
2790 ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack"
2791 by Yuval Yarom and Naomi Benger. Details can be obtained from:
2792 http://eprint.iacr.org/2014/140
2794 Thanks to Yuval Yarom and Naomi Benger for discovering this
2795 flaw and to Yuval Yarom for supplying a fix (CVE-2014-0076)
2796 [Yuval Yarom and Naomi Benger]
2798 *) TLS pad extension: draft-agl-tls-padding-03
2800 Workaround for the "TLS hang bug" (see FAQ and PR#2771): if the
2801 TLS client Hello record length value would otherwise be > 255 and
2802 less that 512 pad with a dummy extension containing zeroes so it
2803 is at least 512 bytes long.
2805 [Adam Langley, Steve Henson]
2807 Changes between 1.0.1e and 1.0.1f [6 Jan 2014]
2809 *) Fix for TLS record tampering bug. A carefully crafted invalid
2810 handshake could crash OpenSSL with a NULL pointer exception.
2811 Thanks to Anton Johansson for reporting this issues.
2814 *) Keep original DTLS digest and encryption contexts in retransmission
2815 structures so we can use the previous session parameters if they need
2816 to be resent. (CVE-2013-6450)
2819 *) Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which
2820 avoids preferring ECDHE-ECDSA ciphers when the client appears to be
2821 Safari on OS X. Safari on OS X 10.8..10.8.3 advertises support for
2822 several ECDHE-ECDSA ciphers, but fails to negotiate them. The bug
2823 is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing
2824 10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer.
2825 [Rob Stradling, Adam Langley]
2827 Changes between 1.0.1d and 1.0.1e [11 Feb 2013]
2829 *) Correct fix for CVE-2013-0169. The original didn't work on AES-NI
2830 supporting platforms or when small records were transferred.
2831 [Andy Polyakov, Steve Henson]
2833 Changes between 1.0.1c and 1.0.1d [5 Feb 2013]
2835 *) Make the decoding of SSLv3, TLS and DTLS CBC records constant time.
2837 This addresses the flaw in CBC record processing discovered by
2838 Nadhem Alfardan and Kenny Paterson. Details of this attack can be found
2839 at: http://www.isg.rhul.ac.uk/tls/
2841 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
2842 Security Group at Royal Holloway, University of London
2843 (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and
2844 Emilia Käsper for the initial patch.
2846 [Emilia Käsper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson]
2848 *) Fix flaw in AESNI handling of TLS 1.2 and 1.1 records for CBC mode
2849 ciphersuites which can be exploited in a denial of service attack.
2850 Thanks go to and to Adam Langley <agl@chromium.org> for discovering
2851 and detecting this bug and to Wolfgang Ettlinger
2852 <wolfgang.ettlinger@gmail.com> for independently discovering this issue.
2856 *) Return an error when checking OCSP signatures when key is NULL.
2857 This fixes a DoS attack. (CVE-2013-0166)
2860 *) Make openssl verify return errors.
2861 [Chris Palmer <palmer@google.com> and Ben Laurie]
2863 *) Call OCSP Stapling callback after ciphersuite has been chosen, so
2864 the right response is stapled. Also change SSL_get_certificate()
2865 so it returns the certificate actually sent.
2866 See http://rt.openssl.org/Ticket/Display.html?id=2836.
2867 [Rob Stradling <rob.stradling@comodo.com>]
2869 *) Fix possible deadlock when decoding public keys.
2872 *) Don't use TLS 1.0 record version number in initial client hello
2876 Changes between 1.0.1b and 1.0.1c [10 May 2012]
2878 *) Sanity check record length before skipping explicit IV in TLS
2879 1.2, 1.1 and DTLS to fix DoS attack.
2881 Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic
2882 fuzzing as a service testing platform.
2886 *) Initialise tkeylen properly when encrypting CMS messages.
2887 Thanks to Solar Designer of Openwall for reporting this issue.
2890 *) In FIPS mode don't try to use composite ciphers as they are not
2894 Changes between 1.0.1a and 1.0.1b [26 Apr 2012]
2896 *) OpenSSL 1.0.0 sets SSL_OP_ALL to 0x80000FFFL and OpenSSL 1.0.1 and
2897 1.0.1a set SSL_OP_NO_TLSv1_1 to 0x00000400L which would unfortunately
2898 mean any application compiled against OpenSSL 1.0.0 headers setting
2899 SSL_OP_ALL would also set SSL_OP_NO_TLSv1_1, unintentionally disabling
2900 TLS 1.1 also. Fix this by changing the value of SSL_OP_NO_TLSv1_1 to
2901 0x10000000L Any application which was previously compiled against
2902 OpenSSL 1.0.1 or 1.0.1a headers and which cares about SSL_OP_NO_TLSv1_1
2903 will need to be recompiled as a result. Letting be results in
2904 inability to disable specifically TLS 1.1 and in client context,
2905 in unlike event, limit maximum offered version to TLS 1.0 [see below].
2908 *) In order to ensure interoperability SSL_OP_NO_protocolX does not
2909 disable just protocol X, but all protocols above X *if* there are
2910 protocols *below* X still enabled. In more practical terms it means
2911 that if application wants to disable TLS1.0 in favor of TLS1.1 and
2912 above, it's not sufficient to pass SSL_OP_NO_TLSv1, one has to pass
2913 SSL_OP_NO_TLSv1|SSL_OP_NO_SSLv3|SSL_OP_NO_SSLv2. This applies to
2917 Changes between 1.0.1 and 1.0.1a [19 Apr 2012]
2919 *) Check for potentially exploitable overflows in asn1_d2i_read_bio
2920 BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer
2921 in CRYPTO_realloc_clean.
2923 Thanks to Tavis Ormandy, Google Security Team, for discovering this
2924 issue and to Adam Langley <agl@chromium.org> for fixing it.
2926 [Adam Langley (Google), Tavis Ormandy, Google Security Team]
2928 *) Don't allow TLS 1.2 SHA-256 ciphersuites in TLS 1.0, 1.1 connections.
2931 *) Workarounds for some broken servers that "hang" if a client hello
2932 record length exceeds 255 bytes.
2934 1. Do not use record version number > TLS 1.0 in initial client
2935 hello: some (but not all) hanging servers will now work.
2936 2. If we set OPENSSL_MAX_TLS1_2_CIPHER_LENGTH this will truncate
2937 the number of ciphers sent in the client hello. This should be
2938 set to an even number, such as 50, for example by passing:
2939 -DOPENSSL_MAX_TLS1_2_CIPHER_LENGTH=50 to config or Configure.
2940 Most broken servers should now work.
2941 3. If all else fails setting OPENSSL_NO_TLS1_2_CLIENT will disable
2942 TLS 1.2 client support entirely.
2945 *) Fix SEGV in Vector Permutation AES module observed in OpenSSH.
2948 Changes between 1.0.0h and 1.0.1 [14 Mar 2012]
2950 *) Add compatibility with old MDC2 signatures which use an ASN1 OCTET
2951 STRING form instead of a DigestInfo.
2954 *) The format used for MDC2 RSA signatures is inconsistent between EVP
2955 and the RSA_sign/RSA_verify functions. This was made more apparent when
2956 OpenSSL used RSA_sign/RSA_verify for some RSA signatures in particular
2957 those which went through EVP_PKEY_METHOD in 1.0.0 and later. Detect
2958 the correct format in RSA_verify so both forms transparently work.
2961 *) Some servers which support TLS 1.0 can choke if we initially indicate
2962 support for TLS 1.2 and later renegotiate using TLS 1.0 in the RSA
2963 encrypted premaster secret. As a workaround use the maximum permitted
2964 client version in client hello, this should keep such servers happy
2965 and still work with previous versions of OpenSSL.
2968 *) Add support for TLS/DTLS heartbeats.
2969 [Robin Seggelmann <seggelmann@fh-muenster.de>]
2971 *) Add support for SCTP.
2972 [Robin Seggelmann <seggelmann@fh-muenster.de>]
2974 *) Improved PRNG seeding for VOS.
2975 [Paul Green <Paul.Green@stratus.com>]
2977 *) Extensive assembler packs updates, most notably:
2979 - x86[_64]: AES-NI, PCLMULQDQ, RDRAND support;
2980 - x86[_64]: SSSE3 support (SHA1, vector-permutation AES);
2981 - x86_64: bit-sliced AES implementation;
2982 - ARM: NEON support, contemporary platforms optimizations;
2983 - s390x: z196 support;
2984 - *: GHASH and GF(2^m) multiplication implementations;
2988 *) Make TLS-SRP code conformant with RFC 5054 API cleanup
2989 (removal of unnecessary code)
2990 [Peter Sylvester <peter.sylvester@edelweb.fr>]
2992 *) Add TLS key material exporter from RFC 5705.
2995 *) Add DTLS-SRTP negotiation from RFC 5764.
2998 *) Add Next Protocol Negotiation,
2999 http://tools.ietf.org/html/draft-agl-tls-nextprotoneg-00. Can be
3000 disabled with a no-npn flag to config or Configure. Code donated
3002 [Adam Langley <agl@google.com> and Ben Laurie]
3004 *) Add optional 64-bit optimized implementations of elliptic curves NIST-P224,
3005 NIST-P256, NIST-P521, with constant-time single point multiplication on
3006 typical inputs. Compiler support for the nonstandard type __uint128_t is
3007 required to use this (present in gcc 4.4 and later, for 64-bit builds).
3008 Code made available under Apache License version 2.0.
3010 Specify "enable-ec_nistp_64_gcc_128" on the Configure (or config) command
3011 line to include this in your build of OpenSSL, and run "make depend" (or
3012 "make update"). This enables the following EC_METHODs:
3014 EC_GFp_nistp224_method()
3015 EC_GFp_nistp256_method()
3016 EC_GFp_nistp521_method()
3018 EC_GROUP_new_by_curve_name() will automatically use these (while
3019 EC_GROUP_new_curve_GFp() currently prefers the more flexible
3021 [Emilia Käsper, Adam Langley, Bodo Moeller (Google)]
3023 *) Use type ossl_ssize_t instad of ssize_t which isn't available on
3024 all platforms. Move ssize_t definition from e_os.h to the public
3025 header file e_os2.h as it now appears in public header file cms.h
3028 *) New -sigopt option to the ca, req and x509 utilities. Additional
3029 signature parameters can be passed using this option and in
3033 *) Add RSA PSS signing function. This will generate and set the
3034 appropriate AlgorithmIdentifiers for PSS based on those in the
3035 corresponding EVP_MD_CTX structure. No application support yet.
3038 *) Support for companion algorithm specific ASN1 signing routines.
3039 New function ASN1_item_sign_ctx() signs a pre-initialised
3040 EVP_MD_CTX structure and sets AlgorithmIdentifiers based on
3041 the appropriate parameters.
3044 *) Add new algorithm specific ASN1 verification initialisation function
3045 to EVP_PKEY_ASN1_METHOD: this is not in EVP_PKEY_METHOD since the ASN1
3046 handling will be the same no matter what EVP_PKEY_METHOD is used.
3047 Add a PSS handler to support verification of PSS signatures: checked
3048 against a number of sample certificates.
3051 *) Add signature printing for PSS. Add PSS OIDs.
3052 [Steve Henson, Martin Kaiser <lists@kaiser.cx>]
3054 *) Add algorithm specific signature printing. An individual ASN1 method
3055 can now print out signatures instead of the standard hex dump.
3057 More complex signatures (e.g. PSS) can print out more meaningful
3058 information. Include DSA version that prints out the signature
3062 *) Password based recipient info support for CMS library: implementing
3066 *) Split password based encryption into PBES2 and PBKDF2 functions. This
3067 neatly separates the code into cipher and PBE sections and is required
3068 for some algorithms that split PBES2 into separate pieces (such as
3069 password based CMS).
3072 *) Session-handling fixes:
3073 - Fix handling of connections that are resuming with a session ID,
3074 but also support Session Tickets.
3075 - Fix a bug that suppressed issuing of a new ticket if the client
3076 presented a ticket with an expired session.
3077 - Try to set the ticket lifetime hint to something reasonable.
3078 - Make tickets shorter by excluding irrelevant information.
3079 - On the client side, don't ignore renewed tickets.
3080 [Adam Langley, Bodo Moeller (Google)]
3082 *) Fix PSK session representation.
3085 *) Add RC4-MD5 and AESNI-SHA1 "stitched" implementations.
3087 This work was sponsored by Intel.
3090 *) Add GCM support to TLS library. Some custom code is needed to split
3091 the IV between the fixed (from PRF) and explicit (from TLS record)
3092 portions. This adds all GCM ciphersuites supported by RFC5288 and
3093 RFC5289. Generalise some AES* cipherstrings to include GCM and
3094 add a special AESGCM string for GCM only.
3097 *) Expand range of ctrls for AES GCM. Permit setting invocation
3098 field on decrypt and retrieval of invocation field only on encrypt.
3101 *) Add HMAC ECC ciphersuites from RFC5289. Include SHA384 PRF support.
3102 As required by RFC5289 these ciphersuites cannot be used if for
3103 versions of TLS earlier than 1.2.
3106 *) For FIPS capable OpenSSL interpret a NULL default public key method
3107 as unset and return the appropriate default but do *not* set the default.
3108 This means we can return the appropriate method in applications that
3109 switch between FIPS and non-FIPS modes.
3112 *) Redirect HMAC and CMAC operations to FIPS module in FIPS mode. If an
3113 ENGINE is used then we cannot handle that in the FIPS module so we
3114 keep original code iff non-FIPS operations are allowed.
3117 *) Add -attime option to openssl utilities.
3118 [Peter Eckersley <pde@eff.org>, Ben Laurie and Steve Henson]
3120 *) Redirect DSA and DH operations to FIPS module in FIPS mode.
3123 *) Redirect ECDSA and ECDH operations to FIPS module in FIPS mode. Also use
3124 FIPS EC methods unconditionally for now.
3127 *) New build option no-ec2m to disable characteristic 2 code.
3130 *) Backport libcrypto audit of return value checking from 1.1.0-dev; not
3131 all cases can be covered as some introduce binary incompatibilities.
3134 *) Redirect RSA operations to FIPS module including keygen,
3135 encrypt, decrypt, sign and verify. Block use of non FIPS RSA methods.
3138 *) Add similar low level API blocking to ciphers.
3141 *) Low level digest APIs are not approved in FIPS mode: any attempt
3142 to use these will cause a fatal error. Applications that *really* want
3143 to use them can use the private_* version instead.
3146 *) Redirect cipher operations to FIPS module for FIPS builds.
3149 *) Redirect digest operations to FIPS module for FIPS builds.
3152 *) Update build system to add "fips" flag which will link in fipscanister.o
3153 for static and shared library builds embedding a signature if needed.
3156 *) Output TLS supported curves in preference order instead of numerical
3157 order. This is currently hardcoded for the highest order curves first.
3158 This should be configurable so applications can judge speed vs strength.
3161 *) Add TLS v1.2 server support for client authentication.
3164 *) Add support for FIPS mode in ssl library: disable SSLv3, non-FIPS ciphers
3168 *) Functions FIPS_mode_set() and FIPS_mode() which call the underlying
3169 FIPS modules versions.
3172 *) Add TLS v1.2 client side support for client authentication. Keep cache
3173 of handshake records longer as we don't know the hash algorithm to use
3174 until after the certificate request message is received.
3177 *) Initial TLS v1.2 client support. Add a default signature algorithms
3178 extension including all the algorithms we support. Parse new signature
3179 format in client key exchange. Relax some ECC signing restrictions for
3180 TLS v1.2 as indicated in RFC5246.
3183 *) Add server support for TLS v1.2 signature algorithms extension. Switch
3184 to new signature format when needed using client digest preference.
3185 All server ciphersuites should now work correctly in TLS v1.2. No client
3186 support yet and no support for client certificates.
3189 *) Initial TLS v1.2 support. Add new SHA256 digest to ssl code, switch
3190 to SHA256 for PRF when using TLS v1.2 and later. Add new SHA256 based
3191 ciphersuites. At present only RSA key exchange ciphersuites work with
3192 TLS v1.2. Add new option for TLS v1.2 replacing the old and obsolete
3193 SSL_OP_PKCS1_CHECK flags with SSL_OP_NO_TLSv1_2. New TLSv1.2 methods
3194 and version checking.
3197 *) New option OPENSSL_NO_SSL_INTERN. If an application can be compiled
3198 with this defined it will not be affected by any changes to ssl internal
3199 structures. Add several utility functions to allow openssl application
3200 to work with OPENSSL_NO_SSL_INTERN defined.
3204 [Tom Wu <tjw@cs.stanford.edu> and Ben Laurie]
3206 *) Add functions to copy EVP_PKEY_METHOD and retrieve flags and id.
3209 *) Permit abbreviated handshakes when renegotiating using the function
3210 SSL_renegotiate_abbreviated().
3211 [Robin Seggelmann <seggelmann@fh-muenster.de>]
3213 *) Add call to ENGINE_register_all_complete() to
3214 ENGINE_load_builtin_engines(), so some implementations get used
3215 automatically instead of needing explicit application support.
3218 *) Add support for TLS key exporter as described in RFC5705.
3219 [Robin Seggelmann <seggelmann@fh-muenster.de>, Steve Henson]
3221 *) Initial TLSv1.1 support. Since TLSv1.1 is very similar to TLS v1.0 only
3222 a few changes are required:
3224 Add SSL_OP_NO_TLSv1_1 flag.
3225 Add TLSv1_1 methods.
3226 Update version checking logic to handle version 1.1.
3227 Add explicit IV handling (ported from DTLS code).
3228 Add command line options to s_client/s_server.
3231 Changes between 1.0.0g and 1.0.0h [12 Mar 2012]
3233 *) Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness
3234 in CMS and PKCS7 code. When RSA decryption fails use a random key for
3235 content decryption and always return the same error. Note: this attack
3236 needs on average 2^20 messages so it only affects automated senders. The
3237 old behaviour can be re-enabled in the CMS code by setting the
3238 CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where
3239 an MMA defence is not necessary.
3240 Thanks to Ivan Nestlerode <inestlerode@us.ibm.com> for discovering
3241 this issue. (CVE-2012-0884)
3244 *) Fix CVE-2011-4619: make sure we really are receiving a
3245 client hello before rejecting multiple SGC restarts. Thanks to
3246 Ivan Nestlerode <inestlerode@us.ibm.com> for discovering this bug.
3249 Changes between 1.0.0f and 1.0.0g [18 Jan 2012]
3251 *) Fix for DTLS DoS issue introduced by fix for CVE-2011-4109.
3252 Thanks to Antonio Martin, Enterprise Secure Access Research and
3253 Development, Cisco Systems, Inc. for discovering this bug and
3254 preparing a fix. (CVE-2012-0050)
3257 Changes between 1.0.0e and 1.0.0f [4 Jan 2012]
3259 *) Nadhem Alfardan and Kenny Paterson have discovered an extension
3260 of the Vaudenay padding oracle attack on CBC mode encryption
3261 which enables an efficient plaintext recovery attack against
3262 the OpenSSL implementation of DTLS. Their attack exploits timing
3263 differences arising during decryption processing. A research
3264 paper describing this attack can be found at:
3265 http://www.isg.rhul.ac.uk/~kp/dtls.pdf
3266 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
3267 Security Group at Royal Holloway, University of London
3268 (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann
3269 <seggelmann@fh-muenster.de> and Michael Tuexen <tuexen@fh-muenster.de>
3270 for preparing the fix. (CVE-2011-4108)
3271 [Robin Seggelmann, Michael Tuexen]
3273 *) Clear bytes used for block padding of SSL 3.0 records.
3275 [Adam Langley (Google)]
3277 *) Only allow one SGC handshake restart for SSL/TLS. Thanks to George
3278 Kadianakis <desnacked@gmail.com> for discovering this issue and
3279 Adam Langley for preparing the fix. (CVE-2011-4619)
3280 [Adam Langley (Google)]
3282 *) Check parameters are not NULL in GOST ENGINE. (CVE-2012-0027)
3283 [Andrey Kulikov <amdeich@gmail.com>]
3285 *) Prevent malformed RFC3779 data triggering an assertion failure.
3286 Thanks to Andrew Chi, BBN Technologies, for discovering the flaw
3287 and Rob Austein <sra@hactrn.net> for fixing it. (CVE-2011-4577)
3288 [Rob Austein <sra@hactrn.net>]
3290 *) Improved PRNG seeding for VOS.
3291 [Paul Green <Paul.Green@stratus.com>]
3293 *) Fix ssl_ciph.c set-up race.
3294 [Adam Langley (Google)]
3296 *) Fix spurious failures in ecdsatest.c.
3297 [Emilia Käsper (Google)]
3299 *) Fix the BIO_f_buffer() implementation (which was mixing different
3300 interpretations of the '..._len' fields).
3301 [Adam Langley (Google)]
3303 *) Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than
3304 BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent
3305 threads won't reuse the same blinding coefficients.
3307 This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING
3308 lock to call BN_BLINDING_invert_ex, and avoids one use of
3309 BN_BLINDING_update for each BN_BLINDING structure (previously,
3310 the last update always remained unused).
3311 [Emilia Käsper (Google)]
3313 *) In ssl3_clear, preserve s3->init_extra along with s3->rbuf.
3314 [Bob Buckholz (Google)]
3316 Changes between 1.0.0d and 1.0.0e [6 Sep 2011]
3318 *) Fix bug where CRLs with nextUpdate in the past are sometimes accepted
3319 by initialising X509_STORE_CTX properly. (CVE-2011-3207)
3320 [Kaspar Brand <ossl@velox.ch>]
3322 *) Fix SSL memory handling for (EC)DH ciphersuites, in particular
3323 for multi-threaded use of ECDH. (CVE-2011-3210)
3324 [Adam Langley (Google)]
3326 *) Fix x509_name_ex_d2i memory leak on bad inputs.
3329 *) Remove hard coded ecdsaWithSHA1 signature tests in ssl code and check
3330 signature public key algorithm by using OID xref utilities instead.
3331 Before this you could only use some ECC ciphersuites with SHA1 only.
3334 *) Add protection against ECDSA timing attacks as mentioned in the paper
3335 by Billy Bob Brumley and Nicola Tuveri, see:
3337 http://eprint.iacr.org/2011/232.pdf
3339 [Billy Bob Brumley and Nicola Tuveri]
3341 Changes between 1.0.0c and 1.0.0d [8 Feb 2011]
3343 *) Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014
3344 [Neel Mehta, Adam Langley, Bodo Moeller (Google)]
3346 *) Fix bug in string printing code: if *any* escaping is enabled we must
3347 escape the escape character (backslash) or the resulting string is
3351 Changes between 1.0.0b and 1.0.0c [2 Dec 2010]
3353 *) Disable code workaround for ancient and obsolete Netscape browsers
3354 and servers: an attacker can use it in a ciphersuite downgrade attack.
3355 Thanks to Martin Rex for discovering this bug. CVE-2010-4180
3358 *) Fixed J-PAKE implementation error, originally discovered by
3359 Sebastien Martini, further info and confirmation from Stefan
3360 Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252
3363 Changes between 1.0.0a and 1.0.0b [16 Nov 2010]
3365 *) Fix extension code to avoid race conditions which can result in a buffer
3366 overrun vulnerability: resumed sessions must not be modified as they can
3367 be shared by multiple threads. CVE-2010-3864
3370 *) Fix WIN32 build system to correctly link an ENGINE directory into
3374 Changes between 1.0.0 and 1.0.0a [01 Jun 2010]
3376 *) Check return value of int_rsa_verify in pkey_rsa_verifyrecover
3378 [Steve Henson, Peter-Michael Hager <hager@dortmund.net>]
3380 Changes between 0.9.8n and 1.0.0 [29 Mar 2010]
3382 *) Add "missing" function EVP_CIPHER_CTX_copy(). This copies a cipher
3383 context. The operation can be customised via the ctrl mechanism in
3384 case ENGINEs want to include additional functionality.
3387 *) Tolerate yet another broken PKCS#8 key format: private key value negative.
3390 *) Add new -subject_hash_old and -issuer_hash_old options to x509 utility to
3391 output hashes compatible with older versions of OpenSSL.
3392 [Willy Weisz <weisz@vcpc.univie.ac.at>]
3394 *) Fix compression algorithm handling: if resuming a session use the
3395 compression algorithm of the resumed session instead of determining
3396 it from client hello again. Don't allow server to change algorithm.
3399 *) Add load_crls() function to apps tidying load_certs() too. Add option
3400 to verify utility to allow additional CRLs to be included.
3403 *) Update OCSP request code to permit adding custom headers to the request:
3404 some responders need this.
3407 *) The function EVP_PKEY_sign() returns <=0 on error: check return code
3409 [Julia Lawall <julia@diku.dk>]
3411 *) Update verify callback code in apps/s_cb.c and apps/verify.c, it
3412 needlessly dereferenced structures, used obsolete functions and
3413 didn't handle all updated verify codes correctly.
3416 *) Disable MD2 in the default configuration.
3419 *) In BIO_pop() and BIO_push() use the ctrl argument (which was NULL) to
3420 indicate the initial BIO being pushed or popped. This makes it possible
3421 to determine whether the BIO is the one explicitly called or as a result
3422 of the ctrl being passed down the chain. Fix BIO_pop() and SSL BIOs so
3423 it handles reference counts correctly and doesn't zero out the I/O bio
3424 when it is not being explicitly popped. WARNING: applications which
3425 included workarounds for the old buggy behaviour will need to be modified
3426 or they could free up already freed BIOs.
3429 *) Extend the uni2asc/asc2uni => OPENSSL_uni2asc/OPENSSL_asc2uni
3430 renaming to all platforms (within the 0.9.8 branch, this was
3431 done conditionally on Netware platforms to avoid a name clash).
3432 [Guenter <lists@gknw.net>]
3434 *) Add ECDHE and PSK support to DTLS.
3435 [Michael Tuexen <tuexen@fh-muenster.de>]
3437 *) Add CHECKED_STACK_OF macro to safestack.h, otherwise safestack can't
3441 *) Add "missing" function EVP_MD_flags() (without this the only way to
3442 retrieve a digest flags is by accessing the structure directly. Update
3443 EVP_MD_do_all*() and EVP_CIPHER_do_all*() to include the name a digest
3444 or cipher is registered as in the "from" argument. Print out all
3445 registered digests in the dgst usage message instead of manually
3446 attempting to work them out.
3449 *) If no SSLv2 ciphers are used don't use an SSLv2 compatible client hello:
3450 this allows the use of compression and extensions. Change default cipher
3451 string to remove SSLv2 ciphersuites. This effectively avoids ancient SSLv2
3452 by default unless an application cipher string requests it.
3455 *) Alter match criteria in PKCS12_parse(). It used to try to use local
3456 key ids to find matching certificates and keys but some PKCS#12 files
3457 don't follow the (somewhat unwritten) rules and this strategy fails.
3458 Now just gather all certificates together and the first private key
3459 then look for the first certificate that matches the key.
3462 *) Support use of registered digest and cipher names for dgst and cipher
3463 commands instead of having to add each one as a special case. So now
3470 openssl dgst -sha256 foo
3472 and this works for ENGINE based algorithms too.
3476 *) Update Gost ENGINE to support parameter files.
3477 [Victor B. Wagner <vitus@cryptocom.ru>]
3479 *) Support GeneralizedTime in ca utility.
3480 [Oliver Martin <oliver@volatilevoid.net>, Steve Henson]
3482 *) Enhance the hash format used for certificate directory links. The new
3483 form uses the canonical encoding (meaning equivalent names will work
3484 even if they aren't identical) and uses SHA1 instead of MD5. This form
3485 is incompatible with the older format and as a result c_rehash should
3486 be used to rebuild symbolic links.
3489 *) Make PKCS#8 the default write format for private keys, replacing the
3490 traditional format. This form is standardised, more secure and doesn't
3491 include an implicit MD5 dependency.
3494 *) Add a $gcc_devteam_warn option to Configure. The idea is that any code
3495 committed to OpenSSL should pass this lot as a minimum.
3498 *) Add session ticket override functionality for use by EAP-FAST.
3499 [Jouni Malinen <j@w1.fi>]
3501 *) Modify HMAC functions to return a value. Since these can be implemented
3502 in an ENGINE errors can occur.
3505 *) Type-checked OBJ_bsearch_ex.
3508 *) Type-checked OBJ_bsearch. Also some constification necessitated
3509 by type-checking. Still to come: TXT_DB, bsearch(?),
3510 OBJ_bsearch_ex, qsort, CRYPTO_EX_DATA, ASN1_VALUE, ASN1_STRING,
3514 *) New function OPENSSL_gmtime_adj() to add a specific number of days and
3515 seconds to a tm structure directly, instead of going through OS
3516 specific date routines. This avoids any issues with OS routines such
3517 as the year 2038 bug. New *_adj() functions for ASN1 time structures
3518 and X509_time_adj_ex() to cover the extended range. The existing
3519 X509_time_adj() is still usable and will no longer have any date issues.
3522 *) Delta CRL support. New use deltas option which will attempt to locate
3523 and search any appropriate delta CRLs available.
3525 This work was sponsored by Google.
3528 *) Support for CRLs partitioned by reason code. Reorganise CRL processing
3529 code and add additional score elements. Validate alternate CRL paths
3530 as part of the CRL checking and indicate a new error "CRL path validation
3531 error" in this case. Applications wanting additional details can use
3532 the verify callback and check the new "parent" field. If this is not
3533 NULL CRL path validation is taking place. Existing applications won't
3534 see this because it requires extended CRL support which is off by
3537 This work was sponsored by Google.
3540 *) Support for freshest CRL extension.
3542 This work was sponsored by Google.
3545 *) Initial indirect CRL support. Currently only supported in the CRLs
3546 passed directly and not via lookup. Process certificate issuer
3547 CRL entry extension and lookup CRL entries by bother issuer name
3548 and serial number. Check and process CRL issuer entry in IDP extension.
3550 This work was sponsored by Google.
3553 *) Add support for distinct certificate and CRL paths. The CRL issuer
3554 certificate is validated separately in this case. Only enabled if
3555 an extended CRL support flag is set: this flag will enable additional
3556 CRL functionality in future.
3558 This work was sponsored by Google.
3561 *) Add support for policy mappings extension.
3563 This work was sponsored by Google.
3566 *) Fixes to pathlength constraint, self issued certificate handling,
3567 policy processing to align with RFC3280 and PKITS tests.
3569 This work was sponsored by Google.
3572 *) Support for name constraints certificate extension. DN, email, DNS
3573 and URI types are currently supported.
3575 This work was sponsored by Google.
3578 *) To cater for systems that provide a pointer-based thread ID rather
3579 than numeric, deprecate the current numeric thread ID mechanism and
3580 replace it with a structure and associated callback type. This
3581 mechanism allows a numeric "hash" to be extracted from a thread ID in
3582 either case, and on platforms where pointers are larger than 'long',
3583 mixing is done to help ensure the numeric 'hash' is usable even if it
3584 can't be guaranteed unique. The default mechanism is to use "&errno"
3585 as a pointer-based thread ID to distinguish between threads.
3587 Applications that want to provide their own thread IDs should now use
3588 CRYPTO_THREADID_set_callback() to register a callback that will call
3589 either CRYPTO_THREADID_set_numeric() or CRYPTO_THREADID_set_pointer().
3591 Note that ERR_remove_state() is now deprecated, because it is tied
3592 to the assumption that thread IDs are numeric. ERR_remove_state(0)
3593 to free the current thread's error state should be replaced by
3594 ERR_remove_thread_state(NULL).
3596 (This new approach replaces the functions CRYPTO_set_idptr_callback(),
3597 CRYPTO_get_idptr_callback(), and CRYPTO_thread_idptr() that existed in
3598 OpenSSL 0.9.9-dev between June 2006 and August 2008. Also, if an
3599 application was previously providing a numeric thread callback that
3600 was inappropriate for distinguishing threads, then uniqueness might
3601 have been obtained with &errno that happened immediately in the
3602 intermediate development versions of OpenSSL; this is no longer the
3603 case, the numeric thread callback will now override the automatic use
3605 [Geoff Thorpe, with help from Bodo Moeller]
3607 *) Initial support for different CRL issuing certificates. This covers a
3608 simple case where the self issued certificates in the chain exist and
3609 the real CRL issuer is higher in the existing chain.
3611 This work was sponsored by Google.
3614 *) Removed effectively defunct crypto/store from the build.
3617 *) Revamp of STACK to provide stronger type-checking. Still to come:
3618 TXT_DB, bsearch(?), OBJ_bsearch, qsort, CRYPTO_EX_DATA, ASN1_VALUE,
3619 ASN1_STRING, CONF_VALUE.
3622 *) Add a new SSL_MODE_RELEASE_BUFFERS mode flag to release unused buffer
3623 RAM on SSL connections. This option can save about 34k per idle SSL.
3626 *) Revamp of LHASH to provide stronger type-checking. Still to come:
3627 STACK, TXT_DB, bsearch, qsort.
3630 *) Initial support for Cryptographic Message Syntax (aka CMS) based
3631 on RFC3850, RFC3851 and RFC3852. New cms directory and cms utility,
3632 support for data, signedData, compressedData, digestedData and
3633 encryptedData, envelopedData types included. Scripts to check against
3634 RFC4134 examples draft and interop and consistency checks of many
3635 content types and variants.
projects / openssl.git / blob