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 1.1.0h and 1.1.1 [xx XXX xxxx]
12 *) SSL_MODE_AUTO_RETRY is enabled by default. Applications that use blocking
13 I/O in combination with something like select() or poll() will hang. This
14 can be turned off again using SSL_CTX_clear_mode().
15 Many applications do not properly handle non-application data records, and
16 TLS 1.3 sends more of such records. Setting SSL_MODE_AUTO_RETRY works
17 around the problems in those applications, but can also break some.
18 It's recommended to read the manpages about SSL_read(), SSL_write(),
19 SSL_get_error(), SSL_shutdown(), SSL_CTX_set_mode() and
20 SSL_CTX_set_read_ahead() again.
23 *) When unlocking a pass phrase protected PEM file or PKCS#8 container, we
24 now allow empty (zero character) pass phrases.
27 *) Apply blinding to binary field modular inversion and remove patent
28 pending (OPENSSL_SUN_GF2M_DIV) BN_GF2m_mod_div implementation.
31 *) Deprecate ec2_mult.c and unify scalar multiplication code paths for
32 binary and prime elliptic curves.
35 *) Remove ECDSA nonce padding: EC_POINT_mul is now responsible for
36 constant time fixed point multiplication.
39 *) Revise elliptic curve scalar multiplication with timing attack
40 defenses: ec_wNAF_mul redirects to a constant time implementation
41 when computing fixed point and variable point multiplication (which
42 in OpenSSL are mostly used with secret scalars in keygen, sign,
43 ECDH derive operations).
44 [Billy Bob Brumley, Nicola Tuveri, Cesar Pereida García,
47 *) Updated CONTRIBUTING
50 *) Updated DRBG / RAND to request nonce and additional low entropy
51 randomness from the system.
54 *) Updated 'openssl rehash' to use OpenSSL consistent default.
57 *) Moved the load of the ssl_conf module to libcrypto, which helps
58 loading engines that libssl uses before libssl is initialised.
61 *) Added EVP_PKEY_sign() and EVP_PKEY_verify() for EdDSA
64 *) Fixed X509_NAME_ENTRY_set to get multi-valued RDNs right in all cases.
65 [Ingo Schwarze, Rich Salz]
67 *) Added output of accepting IP address and port for 'openssl s_server'
70 *) Added a new API for TLSv1.3 ciphersuites:
71 SSL_CTX_set_ciphersuites()
72 SSL_set_ciphersuites()
75 *) Memory allocation failures consistenly add an error to the error
79 *) Don't use OPENSSL_ENGINES and OPENSSL_CONF environment values
80 in libcrypto when run as setuid/setgid.
83 *) Load any config file by default when libssl is used.
86 *) Added new public header file <openssl/rand_drbg.h> and documentation
87 for the RAND_DRBG API. See manual page RAND_DRBG(7) for an overview.
90 *) QNX support removed (cannot find contributors to get their approval
91 for the license change).
94 *) TLSv1.3 replay protection for early data has been implemented. See the
95 SSL_read_early_data() man page for further details.
98 *) Separated TLSv1.3 ciphersuite configuration out from TLSv1.2 ciphersuite
99 configuration. TLSv1.3 ciphersuites are not compatible with TLSv1.2 and
100 below. Similarly TLSv1.2 ciphersuites are not compatible with TLSv1.3.
101 In order to avoid issues where legacy TLSv1.2 ciphersuite configuration
102 would otherwise inadvertently disable all TLSv1.3 ciphersuites the
103 configuration has been separated out. See the ciphers man page or the
104 SSL_CTX_set_ciphersuites() man page for more information.
107 *) On POSIX (BSD, Linux, ...) systems the ocsp(1) command running
108 in responder mode now supports the new "-multi" option, which
109 spawns the specified number of child processes to handle OCSP
110 requests. The "-timeout" option now also limits the OCSP
111 responder's patience to wait to receive the full client request
112 on a newly accepted connection. Child processes are respawned
113 as needed, and the CA index file is automatically reloaded
114 when changed. This makes it possible to run the "ocsp" responder
115 as a long-running service, making the OpenSSL CA somewhat more
116 feature-complete. In this mode, most diagnostic messages logged
117 after entering the event loop are logged via syslog(3) rather than
121 *) Added support for X448 and Ed448. Heavily based on original work by
125 *) Extend OSSL_STORE with capabilities to search and to narrow the set of
126 objects loaded. This adds the functions OSSL_STORE_expect() and
127 OSSL_STORE_find() as well as needed tools to construct searches and
128 get the search data out of them.
131 *) Support for TLSv1.3 added. Note that users upgrading from an earlier
132 version of OpenSSL should review their configuration settings to ensure
133 that they are still appropriate for TLSv1.3. For further information see:
134 https://www.openssl.org/blog/blog/2018/02/08/tlsv1.3/
136 NOTE: In this pre-release of OpenSSL a draft version of the
137 TLSv1.3 standard has been implemented. Implementations of different draft
138 versions of the standard do not inter-operate, and this version will not
139 inter-operate with an implementation of the final standard when it is
140 eventually published. Different pre-release versions may implement
141 different versions of the draft. The final version of OpenSSL 1.1.1 will
142 implement the final version of the standard.
143 TODO(TLS1.3): Remove the above note before final release
146 *) Grand redesign of the OpenSSL random generator
148 The default RAND method now utilizes an AES-CTR DRBG according to
149 NIST standard SP 800-90Ar1. The new random generator is essentially
150 a port of the default random generator from the OpenSSL FIPS 2.0
151 object module. It is a hybrid deterministic random bit generator
152 using an AES-CTR bit stream and which seeds and reseeds itself
153 automatically using trusted system entropy sources.
155 Some of its new features are:
156 o Support for multiple DRBG instances with seed chaining.
157 o Add a public DRBG instance for the default RAND method.
158 o Add a dedicated DRBG instance for generating long term private keys.
159 o Make the DRBG instances fork-safe.
160 o Keep all global DRBG instances on the secure heap if it is enabled.
161 o Add a DRBG instance to every SSL instance for lock free operation
162 and to increase unpredictability.
163 [Paul Dale, Benjamin Kaduk, Kurt Roeckx, Rich Salz, Matthias St. Pierre]
165 *) Changed Configure so it only says what it does and doesn't dump
166 so much data. Instead, ./configdata.pm should be used as a script
167 to display all sorts of configuration data.
170 *) Added processing of "make variables" to Configure.
173 *) Added SHA512/224 and SHA512/256 algorithm support.
176 *) The last traces of Netware support, first removed in 1.1.0, have
180 *) Get rid of Makefile.shared, and in the process, make the processing
181 of certain files (rc.obj, or the .def/.map/.opt files produced from
182 the ordinal files) more visible and hopefully easier to trace and
183 debug (or make silent).
186 *) Make it possible to have environment variable assignments as
187 arguments to config / Configure.
190 *) Add multi-prime RSA (RFC 8017) support.
193 *) Add SM3 implemented according to GB/T 32905-2016
194 [ Jack Lloyd <jack.lloyd@ribose.com>,
195 Ronald Tse <ronald.tse@ribose.com>,
196 Erick Borsboom <erick.borsboom@ribose.com> ]
198 *) Add 'Maximum Fragment Length' TLS extension negotiation and support
199 as documented in RFC6066.
200 Based on a patch from Tomasz Moń
201 [Filipe Raimundo da Silva]
203 *) Add SM4 implemented according to GB/T 32907-2016.
204 [ Jack Lloyd <jack.lloyd@ribose.com>,
205 Ronald Tse <ronald.tse@ribose.com>,
206 Erick Borsboom <erick.borsboom@ribose.com> ]
208 *) Reimplement -newreq-nodes and ERR_error_string_n; the
209 original author does not agree with the license change.
212 *) Add ARIA AEAD TLS support.
215 *) Some macro definitions to support VS6 have been removed. Visual
216 Studio 6 has not worked since 1.1.0
219 *) Add ERR_clear_last_mark(), to allow callers to clear the last mark
220 without clearing the errors.
223 *) Add "atfork" functions. If building on a system that without
224 pthreads, see doc/man3/OPENSSL_fork_prepare.pod for application
225 requirements. The RAND facility now uses/requires this.
231 *) The UI API becomes a permanent and integral part of libcrypto, i.e.
232 not possible to disable entirely. However, it's still possible to
233 disable the console reading UI method, UI_OpenSSL() (use UI_null()
236 To disable, configure with 'no-ui-console'. 'no-ui' is still
237 possible to use as an alias. Check at compile time with the
238 macro OPENSSL_NO_UI_CONSOLE. The macro OPENSSL_NO_UI is still
239 possible to check and is an alias for OPENSSL_NO_UI_CONSOLE.
242 *) Add a STORE module, which implements a uniform and URI based reader of
243 stores that can contain keys, certificates, CRLs and numerous other
244 objects. The main API is loosely based on a few stdio functions,
245 and includes OSSL_STORE_open, OSSL_STORE_load, OSSL_STORE_eof,
246 OSSL_STORE_error and OSSL_STORE_close.
247 The implementation uses backends called "loaders" to implement arbitrary
248 URI schemes. There is one built in "loader" for the 'file' scheme.
251 *) Add devcrypto engine. This has been implemented against cryptodev-linux,
252 then adjusted to work on FreeBSD 8.4 as well.
253 Enable by configuring with 'enable-devcryptoeng'. This is done by default
254 on BSD implementations, as cryptodev.h is assumed to exist on all of them.
257 *) Module names can prefixed with OSSL_ or OPENSSL_. This affects
258 util/mkerr.pl, which is adapted to allow those prefixes, leading to
259 error code calls like this:
261 OSSL_FOOerr(OSSL_FOO_F_SOMETHING, OSSL_FOO_R_WHATEVER);
263 With this change, we claim the namespaces OSSL and OPENSSL in a manner
264 that can be encoded in C. For the foreseeable future, this will only
266 [Richard Levitte and Tim Hudson]
268 *) Removed BSD cryptodev engine.
271 *) Add a build target 'build_all_generated', to build all generated files
272 and only that. This can be used to prepare everything that requires
273 things like perl for a system that lacks perl and then move everything
274 to that system and do the rest of the build there.
277 *) In the UI interface, make it possible to duplicate the user data. This
278 can be used by engines that need to retain the data for a longer time
279 than just the call where this user data is passed.
282 *) Ignore the '-named_curve auto' value for compatibility of applications
284 [Tomas Mraz <tmraz@fedoraproject.org>]
286 *) Fragmented SSL/TLS alerts are no longer accepted. An alert message is 2
287 bytes long. In theory it is permissible in SSLv3 - TLSv1.2 to fragment such
288 alerts across multiple records (some of which could be empty). In practice
289 it make no sense to send an empty alert record, or to fragment one. TLSv1.3
290 prohibts this altogether and other libraries (BoringSSL, NSS) do not
291 support this at all. Supporting it adds significant complexity to the
292 record layer, and its removal is unlikely to cause inter-operability
296 *) Add the ASN.1 types INT32, UINT32, INT64, UINT64 and variants prefixed
297 with Z. These are meant to replace LONG and ZLONG and to be size safe.
298 The use of LONG and ZLONG is discouraged and scheduled for deprecation
302 *) Add the 'z' and 'j' modifiers to BIO_printf() et al formatting string,
303 'z' is to be used for [s]size_t, and 'j' - with [u]int64_t.
304 [Richard Levitte, Andy Polyakov]
306 *) Add EC_KEY_get0_engine(), which does for EC_KEY what RSA_get0_engine()
310 *) Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target
311 platform rather than 'mingw'.
314 *) The functions X509_STORE_add_cert and X509_STORE_add_crl return
315 success if they are asked to add an object which already exists
316 in the store. This change cascades to other functions which load
317 certificates and CRLs.
320 *) x86_64 assembly pack: annotate code with DWARF CFI directives to
321 facilitate stack unwinding even from assembly subroutines.
324 *) Remove VAX C specific definitions of OPENSSL_EXPORT, OPENSSL_EXTERN.
325 Also remove OPENSSL_GLOBAL entirely, as it became a no-op.
328 *) Remove the VMS-specific reimplementation of gmtime from crypto/o_times.c.
329 VMS C's RTL has a fully up to date gmtime() and gmtime_r() since V7.1,
330 which is the minimum version we support.
333 *) Certificate time validation (X509_cmp_time) enforces stricter
334 compliance with RFC 5280. Fractional seconds and timezone offsets
335 are no longer allowed.
338 *) Add support for ARIA
341 *) s_client will now send the Server Name Indication (SNI) extension by
342 default unless the new "-noservername" option is used. The server name is
343 based on the host provided to the "-connect" option unless overridden by
347 *) Add support for SipHash
350 *) OpenSSL now fails if it receives an unrecognised record type in TLS1.0
351 or TLS1.1. Previously this only happened in SSLv3 and TLS1.2. This is to
352 prevent issues where no progress is being made and the peer continually
353 sends unrecognised record types, using up resources processing them.
356 *) 'openssl passwd' can now produce SHA256 and SHA512 based output,
357 using the algorithm defined in
358 https://www.akkadia.org/drepper/SHA-crypt.txt
361 *) Heartbeat support has been removed; the ABI is changed for now.
362 [Richard Levitte, Rich Salz]
364 *) Support for SSL_OP_NO_ENCRYPT_THEN_MAC in SSL_CONF_cmd.
367 *) The RSA "null" method, which was partially supported to avoid patent
368 issues, has been replaced to always returns NULL.
372 Changes between 1.1.0h and 1.1.0i [xx XXX xxxx]
374 *) Fixed a text canonicalisation bug in CMS
376 Where a CMS detached signature is used with text content the text goes
377 through a canonicalisation process first prior to signing or verifying a
378 signature. This process strips trailing space at the end of lines, converts
379 line terminators to CRLF and removes additional trailing line terminators
380 at the end of a file. A bug in the canonicalisation process meant that
381 some characters, such as form-feed, were incorrectly treated as whitespace
382 and removed. This is contrary to the specification (RFC5485). This fix
383 could mean that detached text data signed with an earlier version of
384 OpenSSL 1.1.0 may fail to verify using the fixed version, or text data
385 signed with a fixed OpenSSL may fail to verify with an earlier version of
386 OpenSSL 1.1.0. A workaround is to only verify the canonicalised text data
387 and use the "-binary" flag (for the "cms" command line application) or set
388 the SMIME_BINARY/PKCS7_BINARY/CMS_BINARY flags (if using CMS_verify()).
391 Changes between 1.1.0g and 1.1.0h [27 Mar 2018]
393 *) Constructed ASN.1 types with a recursive definition could exceed the stack
395 Constructed ASN.1 types with a recursive definition (such as can be found
396 in PKCS7) could eventually exceed the stack given malicious input with
397 excessive recursion. This could result in a Denial Of Service attack. There
398 are no such structures used within SSL/TLS that come from untrusted sources
399 so this is considered safe.
401 This issue was reported to OpenSSL on 4th January 2018 by the OSS-fuzz
406 *) Incorrect CRYPTO_memcmp on HP-UX PA-RISC
408 Because of an implementation bug the PA-RISC CRYPTO_memcmp function is
409 effectively reduced to only comparing the least significant bit of each
410 byte. This allows an attacker to forge messages that would be considered as
411 authenticated in an amount of tries lower than that guaranteed by the
412 security claims of the scheme. The module can only be compiled by the
413 HP-UX assembler, so that only HP-UX PA-RISC targets are affected.
415 This issue was reported to OpenSSL on 2nd March 2018 by Peter Waltenberg
420 *) Add a build target 'build_all_generated', to build all generated files
421 and only that. This can be used to prepare everything that requires
422 things like perl for a system that lacks perl and then move everything
423 to that system and do the rest of the build there.
426 *) Backport SSL_OP_NO_RENGOTIATION
428 OpenSSL 1.0.2 and below had the ability to disable renegotiation using the
429 (undocumented) SSL3_FLAGS_NO_RENEGOTIATE_CIPHERS flag. Due to the opacity
430 changes this is no longer possible in 1.1.0. Therefore the new
431 SSL_OP_NO_RENEGOTIATION option from 1.1.1-dev has been backported to
432 1.1.0 to provide equivalent functionality.
434 Note that if an application built against 1.1.0h headers (or above) is run
435 using an older version of 1.1.0 (prior to 1.1.0h) then the option will be
436 accepted but nothing will happen, i.e. renegotiation will not be prevented.
439 *) Removed the OS390-Unix config target. It relied on a script that doesn't
443 *) rsaz_1024_mul_avx2 overflow bug on x86_64
445 There is an overflow bug in the AVX2 Montgomery multiplication procedure
446 used in exponentiation with 1024-bit moduli. No EC algorithms are affected.
447 Analysis suggests that attacks against RSA and DSA as a result of this
448 defect would be very difficult to perform and are not believed likely.
449 Attacks against DH1024 are considered just feasible, because most of the
450 work necessary to deduce information about a private key may be performed
451 offline. The amount of resources required for such an attack would be
452 significant. However, for an attack on TLS to be meaningful, the server
453 would have to share the DH1024 private key among multiple clients, which is
454 no longer an option since CVE-2016-0701.
456 This only affects processors that support the AVX2 but not ADX extensions
457 like Intel Haswell (4th generation).
459 This issue was reported to OpenSSL by David Benjamin (Google). The issue
460 was originally found via the OSS-Fuzz project.
464 Changes between 1.1.0f and 1.1.0g [2 Nov 2017]
466 *) bn_sqrx8x_internal carry bug on x86_64
468 There is a carry propagating bug in the x86_64 Montgomery squaring
469 procedure. No EC algorithms are affected. Analysis suggests that attacks
470 against RSA and DSA as a result of this defect would be very difficult to
471 perform and are not believed likely. Attacks against DH are considered just
472 feasible (although very difficult) because most of the work necessary to
473 deduce information about a private key may be performed offline. The amount
474 of resources required for such an attack would be very significant and
475 likely only accessible to a limited number of attackers. An attacker would
476 additionally need online access to an unpatched system using the target
477 private key in a scenario with persistent DH parameters and a private
478 key that is shared between multiple clients.
480 This only affects processors that support the BMI1, BMI2 and ADX extensions
481 like Intel Broadwell (5th generation) and later or AMD Ryzen.
483 This issue was reported to OpenSSL by the OSS-Fuzz project.
487 *) Malformed X.509 IPAddressFamily could cause OOB read
489 If an X.509 certificate has a malformed IPAddressFamily extension,
490 OpenSSL could do a one-byte buffer overread. The most likely result
491 would be an erroneous display of the certificate in text format.
493 This issue was reported to OpenSSL by the OSS-Fuzz project.
497 Changes between 1.1.0e and 1.1.0f [25 May 2017]
499 *) Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target
500 platform rather than 'mingw'.
503 *) Remove the VMS-specific reimplementation of gmtime from crypto/o_times.c.
504 VMS C's RTL has a fully up to date gmtime() and gmtime_r() since V7.1,
505 which is the minimum version we support.
508 Changes between 1.1.0d and 1.1.0e [16 Feb 2017]
510 *) Encrypt-Then-Mac renegotiation crash
512 During a renegotiation handshake if the Encrypt-Then-Mac extension is
513 negotiated where it was not in the original handshake (or vice-versa) then
514 this can cause OpenSSL to crash (dependant on ciphersuite). Both clients
515 and servers are affected.
517 This issue was reported to OpenSSL by Joe Orton (Red Hat).
521 Changes between 1.1.0c and 1.1.0d [26 Jan 2017]
523 *) Truncated packet could crash via OOB read
525 If one side of an SSL/TLS path is running on a 32-bit host and a specific
526 cipher is being used, then a truncated packet can cause that host to
527 perform an out-of-bounds read, usually resulting in a crash.
529 This issue was reported to OpenSSL by Robert Święcki of Google.
533 *) Bad (EC)DHE parameters cause a client crash
535 If a malicious server supplies bad parameters for a DHE or ECDHE key
536 exchange then this can result in the client attempting to dereference a
537 NULL pointer leading to a client crash. This could be exploited in a Denial
540 This issue was reported to OpenSSL by Guido Vranken.
544 *) BN_mod_exp may produce incorrect results on x86_64
546 There is a carry propagating bug in the x86_64 Montgomery squaring
547 procedure. No EC algorithms are affected. Analysis suggests that attacks
548 against RSA and DSA as a result of this defect would be very difficult to
549 perform and are not believed likely. Attacks against DH are considered just
550 feasible (although very difficult) because most of the work necessary to
551 deduce information about a private key may be performed offline. The amount
552 of resources required for such an attack would be very significant and
553 likely only accessible to a limited number of attackers. An attacker would
554 additionally need online access to an unpatched system using the target
555 private key in a scenario with persistent DH parameters and a private
556 key that is shared between multiple clients. For example this can occur by
557 default in OpenSSL DHE based SSL/TLS ciphersuites. Note: This issue is very
558 similar to CVE-2015-3193 but must be treated as a separate problem.
560 This issue was reported to OpenSSL by the OSS-Fuzz project.
564 Changes between 1.1.0b and 1.1.0c [10 Nov 2016]
566 *) ChaCha20/Poly1305 heap-buffer-overflow
568 TLS connections using *-CHACHA20-POLY1305 ciphersuites are susceptible to
569 a DoS attack by corrupting larger payloads. This can result in an OpenSSL
570 crash. This issue is not considered to be exploitable beyond a DoS.
572 This issue was reported to OpenSSL by Robert Święcki (Google Security Team)
576 *) CMS Null dereference
578 Applications parsing invalid CMS structures can crash with a NULL pointer
579 dereference. This is caused by a bug in the handling of the ASN.1 CHOICE
580 type in OpenSSL 1.1.0 which can result in a NULL value being passed to the
581 structure callback if an attempt is made to free certain invalid encodings.
582 Only CHOICE structures using a callback which do not handle NULL value are
585 This issue was reported to OpenSSL by Tyler Nighswander of ForAllSecure.
589 *) Montgomery multiplication may produce incorrect results
591 There is a carry propagating bug in the Broadwell-specific Montgomery
592 multiplication procedure that handles input lengths divisible by, but
593 longer than 256 bits. Analysis suggests that attacks against RSA, DSA
594 and DH private keys are impossible. This is because the subroutine in
595 question is not used in operations with the private key itself and an input
596 of the attacker's direct choice. Otherwise the bug can manifest itself as
597 transient authentication and key negotiation failures or reproducible
598 erroneous outcome of public-key operations with specially crafted input.
599 Among EC algorithms only Brainpool P-512 curves are affected and one
600 presumably can attack ECDH key negotiation. Impact was not analyzed in
601 detail, because pre-requisites for attack are considered unlikely. Namely
602 multiple clients have to choose the curve in question and the server has to
603 share the private key among them, neither of which is default behaviour.
604 Even then only clients that chose the curve will be affected.
606 This issue was publicly reported as transient failures and was not
607 initially recognized as a security issue. Thanks to Richard Morgan for
608 providing reproducible case.
612 *) Removed automatic addition of RPATH in shared libraries and executables,
613 as this was a remainder from OpenSSL 1.0.x and isn't needed any more.
616 Changes between 1.1.0a and 1.1.0b [26 Sep 2016]
618 *) Fix Use After Free for large message sizes
620 The patch applied to address CVE-2016-6307 resulted in an issue where if a
621 message larger than approx 16k is received then the underlying buffer to
622 store the incoming message is reallocated and moved. Unfortunately a
623 dangling pointer to the old location is left which results in an attempt to
624 write to the previously freed location. This is likely to result in a
625 crash, however it could potentially lead to execution of arbitrary code.
627 This issue only affects OpenSSL 1.1.0a.
629 This issue was reported to OpenSSL by Robert Święcki.
633 Changes between 1.1.0 and 1.1.0a [22 Sep 2016]
635 *) OCSP Status Request extension unbounded memory growth
637 A malicious client can send an excessively large OCSP Status Request
638 extension. If that client continually requests renegotiation, sending a
639 large OCSP Status Request extension each time, then there will be unbounded
640 memory growth on the server. This will eventually lead to a Denial Of
641 Service attack through memory exhaustion. Servers with a default
642 configuration are vulnerable even if they do not support OCSP. Builds using
643 the "no-ocsp" build time option are not affected.
645 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
649 *) SSL_peek() hang on empty record
651 OpenSSL 1.1.0 SSL/TLS will hang during a call to SSL_peek() if the peer
652 sends an empty record. This could be exploited by a malicious peer in a
653 Denial Of Service attack.
655 This issue was reported to OpenSSL by Alex Gaynor.
659 *) Excessive allocation of memory in tls_get_message_header() and
660 dtls1_preprocess_fragment()
662 A (D)TLS message includes 3 bytes for its length in the header for the
663 message. This would allow for messages up to 16Mb in length. Messages of
664 this length are excessive and OpenSSL includes a check to ensure that a
665 peer is sending reasonably sized messages in order to avoid too much memory
666 being consumed to service a connection. A flaw in the logic of version
667 1.1.0 means that memory for the message is allocated too early, prior to
668 the excessive message length check. Due to way memory is allocated in
669 OpenSSL this could mean an attacker could force up to 21Mb to be allocated
670 to service a connection. This could lead to a Denial of Service through
671 memory exhaustion. However, the excessive message length check still takes
672 place, and this would cause the connection to immediately fail. Assuming
673 that the application calls SSL_free() on the failed connection in a timely
674 manner then the 21Mb of allocated memory will then be immediately freed
675 again. Therefore the excessive memory allocation will be transitory in
676 nature. This then means that there is only a security impact if:
678 1) The application does not call SSL_free() in a timely manner in the event
679 that the connection fails
681 2) The application is working in a constrained environment where there is
682 very little free memory
684 3) The attacker initiates multiple connection attempts such that there are
685 multiple connections in a state where memory has been allocated for the
686 connection; SSL_free() has not yet been called; and there is insufficient
687 memory to service the multiple requests.
689 Except in the instance of (1) above any Denial Of Service is likely to be
690 transitory because as soon as the connection fails the memory is
691 subsequently freed again in the SSL_free() call. However there is an
692 increased risk during this period of application crashes due to the lack of
693 memory - which would then mean a more serious Denial of Service.
695 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
696 (CVE-2016-6307 and CVE-2016-6308)
699 *) solaris-x86-cc, i.e. 32-bit configuration with vendor compiler,
700 had to be removed. Primary reason is that vendor assembler can't
701 assemble our modules with -KPIC flag. As result it, assembly
702 support, was not even available as option. But its lack means
703 lack of side-channel resistant code, which is incompatible with
704 security by todays standards. Fortunately gcc is readily available
705 prepackaged option, which we firmly point at...
708 Changes between 1.0.2h and 1.1.0 [25 Aug 2016]
710 *) Windows command-line tool supports UTF-8 opt-in option for arguments
711 and console input. Setting OPENSSL_WIN32_UTF8 environment variable
712 (to any value) allows Windows user to access PKCS#12 file generated
713 with Windows CryptoAPI and protected with non-ASCII password, as well
714 as files generated under UTF-8 locale on Linux also protected with
718 *) To mitigate the SWEET32 attack (CVE-2016-2183), 3DES cipher suites
719 have been disabled by default and removed from DEFAULT, just like RC4.
720 See the RC4 item below to re-enable both.
723 *) The method for finding the storage location for the Windows RAND seed file
724 has changed. First we check %RANDFILE%. If that is not set then we check
725 the directories %HOME%, %USERPROFILE% and %SYSTEMROOT% in that order. If
726 all else fails we fall back to C:\.
729 *) The EVP_EncryptUpdate() function has had its return type changed from void
730 to int. A return of 0 indicates and error while a return of 1 indicates
734 *) The flags RSA_FLAG_NO_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME and
735 DH_FLAG_NO_EXP_CONSTTIME which previously provided the ability to switch
736 off the constant time implementation for RSA, DSA and DH have been made
737 no-ops and deprecated.
740 *) Windows RAND implementation was simplified to only get entropy by
741 calling CryptGenRandom(). Various other RAND-related tickets
743 [Joseph Wylie Yandle, Rich Salz]
745 *) The stack and lhash API's were renamed to start with OPENSSL_SK_
746 and OPENSSL_LH_, respectively. The old names are available
747 with API compatibility. They new names are now completely documented.
750 *) Unify TYPE_up_ref(obj) methods signature.
751 SSL_CTX_up_ref(), SSL_up_ref(), X509_up_ref(), EVP_PKEY_up_ref(),
752 X509_CRL_up_ref(), X509_OBJECT_up_ref_count() methods are now returning an
753 int (instead of void) like all others TYPE_up_ref() methods.
754 So now these methods also check the return value of CRYPTO_atomic_add(),
755 and the validity of object reference counter.
756 [fdasilvayy@gmail.com]
758 *) With Windows Visual Studio builds, the .pdb files are installed
759 alongside the installed libraries and executables. For a static
760 library installation, ossl_static.pdb is the associate compiler
761 generated .pdb file to be used when linking programs.
764 *) Remove openssl.spec. Packaging files belong with the packagers.
767 *) Automatic Darwin/OSX configuration has had a refresh, it will now
768 recognise x86_64 architectures automatically. You can still decide
769 to build for a different bitness with the environment variable
770 KERNEL_BITS (can be 32 or 64), for example:
772 KERNEL_BITS=32 ./config
776 *) Change default algorithms in pkcs8 utility to use PKCS#5 v2.0,
777 256 bit AES and HMAC with SHA256.
780 *) Remove support for MIPS o32 ABI on IRIX (and IRIX only).
783 *) Triple-DES ciphers have been moved from HIGH to MEDIUM.
786 *) To enable users to have their own config files and build file templates,
787 Configure looks in the directory indicated by the environment variable
788 OPENSSL_LOCAL_CONFIG_DIR as well as the in-source Configurations/
789 directory. On VMS, OPENSSL_LOCAL_CONFIG_DIR is expected to be a logical
790 name and is used as is.
793 *) The following datatypes were made opaque: X509_OBJECT, X509_STORE_CTX,
794 X509_STORE, X509_LOOKUP, and X509_LOOKUP_METHOD. The unused type
795 X509_CERT_FILE_CTX was removed.
798 *) "shared" builds are now the default. To create only static libraries use
799 the "no-shared" Configure option.
802 *) Remove the no-aes, no-hmac, no-rsa, no-sha and no-md5 Configure options.
803 All of these option have not worked for some while and are fundamental
807 *) Make various cleanup routines no-ops and mark them as deprecated. Most
808 global cleanup functions are no longer required because they are handled
809 via auto-deinit (see OPENSSL_init_crypto and OPENSSL_init_ssl man pages).
810 Explicitly de-initing can cause problems (e.g. where a library that uses
811 OpenSSL de-inits, but an application is still using it). The affected
812 functions are CONF_modules_free(), ENGINE_cleanup(), OBJ_cleanup(),
813 EVP_cleanup(), BIO_sock_cleanup(), CRYPTO_cleanup_all_ex_data(),
814 RAND_cleanup(), SSL_COMP_free_compression_methods(), ERR_free_strings() and
818 *) --strict-warnings no longer enables runtime debugging options
819 such as REF_DEBUG. Instead, debug options are automatically
820 enabled with '--debug' builds.
821 [Andy Polyakov, Emilia Käsper]
823 *) Made DH and DH_METHOD opaque. The structures for managing DH objects
824 have been moved out of the public header files. New functions for managing
825 these have been added.
828 *) Made RSA and RSA_METHOD opaque. The structures for managing RSA
829 objects have been moved out of the public header files. New
830 functions for managing these have been added.
833 *) Made DSA and DSA_METHOD opaque. The structures for managing DSA objects
834 have been moved out of the public header files. New functions for managing
835 these have been added.
838 *) Made BIO and BIO_METHOD opaque. The structures for managing BIOs have been
839 moved out of the public header files. New functions for managing these
843 *) Removed no-rijndael as a config option. Rijndael is an old name for AES.
846 *) Removed the mk1mf build scripts.
849 *) Headers are now wrapped, if necessary, with OPENSSL_NO_xxx, so
850 it is always safe to #include a header now.
853 *) Removed the aged BC-32 config and all its supporting scripts
856 *) Removed support for Ultrix, Netware, and OS/2.
859 *) Add support for HKDF.
862 *) Add support for blake2b and blake2s
865 *) Added support for "pipelining". Ciphers that have the
866 EVP_CIPH_FLAG_PIPELINE flag set have a capability to process multiple
867 encryptions/decryptions simultaneously. There are currently no built-in
868 ciphers with this property but the expectation is that engines will be able
869 to offer it to significantly improve throughput. Support has been extended
870 into libssl so that multiple records for a single connection can be
871 processed in one go (for >=TLS 1.1).
874 *) Added the AFALG engine. This is an async capable engine which is able to
875 offload work to the Linux kernel. In this initial version it only supports
876 AES128-CBC. The kernel must be version 4.1.0 or greater.
879 *) OpenSSL now uses a new threading API. It is no longer necessary to
880 set locking callbacks to use OpenSSL in a multi-threaded environment. There
881 are two supported threading models: pthreads and windows threads. It is
882 also possible to configure OpenSSL at compile time for "no-threads". The
883 old threading API should no longer be used. The functions have been
884 replaced with "no-op" compatibility macros.
885 [Alessandro Ghedini, Matt Caswell]
887 *) Modify behavior of ALPN to invoke callback after SNI/servername
888 callback, such that updates to the SSL_CTX affect ALPN.
891 *) Add SSL_CIPHER queries for authentication and key-exchange.
894 *) Changes to the DEFAULT cipherlist:
895 - Prefer (EC)DHE handshakes over plain RSA.
896 - Prefer AEAD ciphers over legacy ciphers.
897 - Prefer ECDSA over RSA when both certificates are available.
898 - Prefer TLSv1.2 ciphers/PRF.
899 - Remove DSS, SEED, IDEA, CAMELLIA, and AES-CCM from the
903 *) Change the ECC default curve list to be this, in order: x25519,
904 secp256r1, secp521r1, secp384r1.
907 *) RC4 based libssl ciphersuites are now classed as "weak" ciphers and are
908 disabled by default. They can be re-enabled using the
909 enable-weak-ssl-ciphers option to Configure.
912 *) If the server has ALPN configured, but supports no protocols that the
913 client advertises, send a fatal "no_application_protocol" alert.
914 This behaviour is SHALL in RFC 7301, though it isn't universally
915 implemented by other servers.
918 *) Add X25519 support.
919 Add ASN.1 and EVP_PKEY methods for X25519. This includes support
920 for public and private key encoding using the format documented in
921 draft-ietf-curdle-pkix-02. The corresponding EVP_PKEY method supports
922 key generation and key derivation.
924 TLS support complies with draft-ietf-tls-rfc4492bis-08 and uses
928 *) Deprecate SRP_VBASE_get_by_user.
929 SRP_VBASE_get_by_user had inconsistent memory management behaviour.
930 In order to fix an unavoidable memory leak (CVE-2016-0798),
931 SRP_VBASE_get_by_user was changed to ignore the "fake user" SRP
932 seed, even if the seed is configured.
934 Users should use SRP_VBASE_get1_by_user instead. Note that in
935 SRP_VBASE_get1_by_user, caller must free the returned value. Note
936 also that even though configuring the SRP seed attempts to hide
937 invalid usernames by continuing the handshake with fake
938 credentials, this behaviour is not constant time and no strong
939 guarantees are made that the handshake is indistinguishable from
940 that of a valid user.
943 *) Configuration change; it's now possible to build dynamic engines
944 without having to build shared libraries and vice versa. This
945 only applies to the engines in engines/, those in crypto/engine/
946 will always be built into libcrypto (i.e. "static").
948 Building dynamic engines is enabled by default; to disable, use
949 the configuration option "disable-dynamic-engine".
951 The only requirements for building dynamic engines are the
952 presence of the DSO module and building with position independent
953 code, so they will also automatically be disabled if configuring
954 with "disable-dso" or "disable-pic".
956 The macros OPENSSL_NO_STATIC_ENGINE and OPENSSL_NO_DYNAMIC_ENGINE
957 are also taken away from openssl/opensslconf.h, as they are
961 *) Configuration change; if there is a known flag to compile
962 position independent code, it will always be applied on the
963 libcrypto and libssl object files, and never on the application
964 object files. This means other libraries that use routines from
965 libcrypto / libssl can be made into shared libraries regardless
966 of how OpenSSL was configured.
968 If this isn't desirable, the configuration options "disable-pic"
969 or "no-pic" can be used to disable the use of PIC. This will
970 also disable building shared libraries and dynamic engines.
973 *) Removed JPAKE code. It was experimental and has no wide use.
976 *) The INSTALL_PREFIX Makefile variable has been renamed to
977 DESTDIR. That makes for less confusion on what this variable
978 is for. Also, the configuration option --install_prefix is
982 *) Heartbeat for TLS has been removed and is disabled by default
983 for DTLS; configure with enable-heartbeats. Code that uses the
984 old #define's might need to be updated.
985 [Emilia Käsper, Rich Salz]
987 *) Rename REF_CHECK to REF_DEBUG.
990 *) New "unified" build system
992 The "unified" build system is aimed to be a common system for all
993 platforms we support. With it comes new support for VMS.
995 This system builds supports building in a different directory tree
996 than the source tree. It produces one Makefile (for unix family
997 or lookalikes), or one descrip.mms (for VMS).
999 The source of information to make the Makefile / descrip.mms is
1000 small files called 'build.info', holding the necessary
1001 information for each directory with source to compile, and a
1002 template in Configurations, like unix-Makefile.tmpl or
1005 With this change, the library names were also renamed on Windows
1006 and on VMS. They now have names that are closer to the standard
1007 on Unix, and include the major version number, and in certain
1008 cases, the architecture they are built for. See "Notes on shared
1009 libraries" in INSTALL.
1011 We rely heavily on the perl module Text::Template.
1014 *) Added support for auto-initialisation and de-initialisation of the library.
1015 OpenSSL no longer requires explicit init or deinit routines to be called,
1016 except in certain circumstances. See the OPENSSL_init_crypto() and
1017 OPENSSL_init_ssl() man pages for further information.
1020 *) The arguments to the DTLSv1_listen function have changed. Specifically the
1021 "peer" argument is now expected to be a BIO_ADDR object.
1023 *) Rewrite of BIO networking library. The BIO library lacked consistent
1024 support of IPv6, and adding it required some more extensive
1025 modifications. This introduces the BIO_ADDR and BIO_ADDRINFO types,
1026 which hold all types of addresses and chains of address information.
1027 It also introduces a new API, with functions like BIO_socket,
1028 BIO_connect, BIO_listen, BIO_lookup and a rewrite of BIO_accept.
1029 The source/sink BIOs BIO_s_connect, BIO_s_accept and BIO_s_datagram
1030 have been adapted accordingly.
1033 *) RSA_padding_check_PKCS1_type_1 now accepts inputs with and without
1037 *) CRIME protection: disable compression by default, even if OpenSSL is
1038 compiled with zlib enabled. Applications can still enable compression
1039 by calling SSL_CTX_clear_options(ctx, SSL_OP_NO_COMPRESSION), or by
1040 using the SSL_CONF library to configure compression.
1043 *) The signature of the session callback configured with
1044 SSL_CTX_sess_set_get_cb was changed. The read-only input buffer
1045 was explicitly marked as 'const unsigned char*' instead of
1049 *) Always DPURIFY. Remove the use of uninitialized memory in the
1050 RNG, and other conditional uses of DPURIFY. This makes -DPURIFY a no-op.
1053 *) Removed many obsolete configuration items, including
1054 DES_PTR, DES_RISC1, DES_RISC2, DES_INT
1055 MD2_CHAR, MD2_INT, MD2_LONG
1057 IDEA_SHORT, IDEA_LONG
1058 RC2_SHORT, RC2_LONG, RC4_LONG, RC4_CHUNK, RC4_INDEX
1059 [Rich Salz, with advice from Andy Polyakov]
1061 *) Many BN internals have been moved to an internal header file.
1062 [Rich Salz with help from Andy Polyakov]
1064 *) Configuration and writing out the results from it has changed.
1065 Files such as Makefile include/openssl/opensslconf.h and are now
1066 produced through general templates, such as Makefile.in and
1067 crypto/opensslconf.h.in and some help from the perl module
1070 Also, the center of configuration information is no longer
1071 Makefile. Instead, Configure produces a perl module in
1072 configdata.pm which holds most of the config data (in the hash
1073 table %config), the target data that comes from the target
1074 configuration in one of the Configurations/*.conf files (in
1078 *) To clarify their intended purposes, the Configure options
1079 --prefix and --openssldir change their semantics, and become more
1080 straightforward and less interdependent.
1082 --prefix shall be used exclusively to give the location INSTALLTOP
1083 where programs, scripts, libraries, include files and manuals are
1084 going to be installed. The default is now /usr/local.
1086 --openssldir shall be used exclusively to give the default
1087 location OPENSSLDIR where certificates, private keys, CRLs are
1088 managed. This is also where the default openssl.cnf gets
1090 If the directory given with this option is a relative path, the
1091 values of both the --prefix value and the --openssldir value will
1092 be combined to become OPENSSLDIR.
1093 The default for --openssldir is INSTALLTOP/ssl.
1095 Anyone who uses --openssldir to specify where OpenSSL is to be
1096 installed MUST change to use --prefix instead.
1099 *) The GOST engine was out of date and therefore it has been removed. An up
1100 to date GOST engine is now being maintained in an external repository.
1101 See: https://wiki.openssl.org/index.php/Binaries. Libssl still retains
1102 support for GOST ciphersuites (these are only activated if a GOST engine
1106 *) EGD is no longer supported by default; use enable-egd when
1108 [Ben Kaduk and Rich Salz]
1110 *) The distribution now has Makefile.in files, which are used to
1111 create Makefile's when Configure is run. *Configure must be run
1112 before trying to build now.*
1115 *) The return value for SSL_CIPHER_description() for error conditions
1119 *) Support for RFC6698/RFC7671 DANE TLSA peer authentication.
1121 Obtaining and performing DNSSEC validation of TLSA records is
1122 the application's responsibility. The application provides
1123 the TLSA records of its choice to OpenSSL, and these are then
1124 used to authenticate the peer.
1126 The TLSA records need not even come from DNS. They can, for
1127 example, be used to implement local end-entity certificate or
1128 trust-anchor "pinning", where the "pin" data takes the form
1129 of TLSA records, which can augment or replace verification
1130 based on the usual WebPKI public certification authorities.
1133 *) Revert default OPENSSL_NO_DEPRECATED setting. Instead OpenSSL
1134 continues to support deprecated interfaces in default builds.
1135 However, applications are strongly advised to compile their
1136 source files with -DOPENSSL_API_COMPAT=0x10100000L, which hides
1137 the declarations of all interfaces deprecated in 0.9.8, 1.0.0
1138 or the 1.1.0 releases.
1140 In environments in which all applications have been ported to
1141 not use any deprecated interfaces OpenSSL's Configure script
1142 should be used with the --api=1.1.0 option to entirely remove
1143 support for the deprecated features from the library and
1144 unconditionally disable them in the installed headers.
1145 Essentially the same effect can be achieved with the "no-deprecated"
1146 argument to Configure, except that this will always restrict
1147 the build to just the latest API, rather than a fixed API
1150 As applications are ported to future revisions of the API,
1151 they should update their compile-time OPENSSL_API_COMPAT define
1152 accordingly, but in most cases should be able to continue to
1153 compile with later releases.
1155 The OPENSSL_API_COMPAT versions for 1.0.0, and 0.9.8 are
1156 0x10000000L and 0x00908000L, respectively. However those
1157 versions did not support the OPENSSL_API_COMPAT feature, and
1158 so applications are not typically tested for explicit support
1159 of just the undeprecated features of either release.
1162 *) Add support for setting the minimum and maximum supported protocol.
1163 It can bet set via the SSL_set_min_proto_version() and
1164 SSL_set_max_proto_version(), or via the SSL_CONF's MinProtocol and
1165 MaxProtocol. It's recommended to use the new APIs to disable
1166 protocols instead of disabling individual protocols using
1167 SSL_set_options() or SSL_CONF's Protocol. This change also
1168 removes support for disabling TLS 1.2 in the OpenSSL TLS
1169 client at compile time by defining OPENSSL_NO_TLS1_2_CLIENT.
1172 *) Support for ChaCha20 and Poly1305 added to libcrypto and libssl.
1175 *) New EC_KEY_METHOD, this replaces the older ECDSA_METHOD and ECDH_METHOD
1176 and integrates ECDSA and ECDH functionality into EC. Implementations can
1177 now redirect key generation and no longer need to convert to or from
1180 Note: the ecdsa.h and ecdh.h headers are now no longer needed and just
1181 include the ec.h header file instead.
1184 *) Remove support for all 40 and 56 bit ciphers. This includes all the export
1185 ciphers who are no longer supported and drops support the ephemeral RSA key
1186 exchange. The LOW ciphers currently doesn't have any ciphers in it.
1189 *) Made EVP_MD_CTX, EVP_MD, EVP_CIPHER_CTX, EVP_CIPHER and HMAC_CTX
1190 opaque. For HMAC_CTX, the following constructors and destructors
1193 HMAC_CTX *HMAC_CTX_new(void);
1194 void HMAC_CTX_free(HMAC_CTX *ctx);
1196 For EVP_MD and EVP_CIPHER, complete APIs to create, fill and
1197 destroy such methods has been added. See EVP_MD_meth_new(3) and
1198 EVP_CIPHER_meth_new(3) for documentation.
1201 1) EVP_MD_CTX_cleanup(), EVP_CIPHER_CTX_cleanup() and
1202 HMAC_CTX_cleanup() were removed. HMAC_CTX_reset() and
1203 EVP_MD_CTX_reset() should be called instead to reinitialise
1204 an already created structure.
1205 2) For consistency with the majority of our object creators and
1206 destructors, EVP_MD_CTX_(create|destroy) were renamed to
1207 EVP_MD_CTX_(new|free). The old names are retained as macros
1208 for deprecated builds.
1211 *) Added ASYNC support. Libcrypto now includes the async sub-library to enable
1212 cryptographic operations to be performed asynchronously as long as an
1213 asynchronous capable engine is used. See the ASYNC_start_job() man page for
1214 further details. Libssl has also had this capability integrated with the
1215 introduction of the new mode SSL_MODE_ASYNC and associated error
1216 SSL_ERROR_WANT_ASYNC. See the SSL_CTX_set_mode() and SSL_get_error() man
1217 pages. This work was developed in partnership with Intel Corp.
1220 *) SSL_{CTX_}set_ecdh_auto() has been removed and ECDH is support is
1221 always enabled now. If you want to disable the support you should
1222 exclude it using the list of supported ciphers. This also means that the
1223 "-no_ecdhe" option has been removed from s_server.
1226 *) SSL_{CTX}_set_tmp_ecdh() which can set 1 EC curve now internally calls
1227 SSL_{CTX_}set1_curves() which can set a list.
1230 *) Remove support for SSL_{CTX_}set_tmp_ecdh_callback(). You should set the
1231 curve you want to support using SSL_{CTX_}set1_curves().
1234 *) State machine rewrite. The state machine code has been significantly
1235 refactored in order to remove much duplication of code and solve issues
1236 with the old code (see ssl/statem/README for further details). This change
1237 does have some associated API changes. Notably the SSL_state() function
1238 has been removed and replaced by SSL_get_state which now returns an
1239 "OSSL_HANDSHAKE_STATE" instead of an int. SSL_set_state() has been removed
1240 altogether. The previous handshake states defined in ssl.h and ssl3.h have
1244 *) All instances of the string "ssleay" in the public API were replaced
1245 with OpenSSL (case-matching; e.g., OPENSSL_VERSION for #define's)
1246 Some error codes related to internal RSA_eay API's were renamed.
1249 *) The demo files in crypto/threads were moved to demo/threads.
1252 *) Removed obsolete engines: 4758cca, aep, atalla, cswift, nuron, gmp,
1254 [Matt Caswell, Rich Salz]
1256 *) New ASN.1 embed macro.
1258 New ASN.1 macro ASN1_EMBED. This is the same as ASN1_SIMPLE except the
1259 structure is not allocated: it is part of the parent. That is instead of
1267 This reduces memory fragmentation and make it impossible to accidentally
1268 set a mandatory field to NULL.
1270 This currently only works for some fields specifically a SEQUENCE, CHOICE,
1271 or ASN1_STRING type which is part of a parent SEQUENCE. Since it is
1272 equivalent to ASN1_SIMPLE it cannot be tagged, OPTIONAL, SET OF or
1276 *) Remove EVP_CHECK_DES_KEY, a compile-time option that never compiled.
1279 *) Removed DES and RC4 ciphersuites from DEFAULT. Also removed RC2 although
1280 in 1.0.2 EXPORT was already removed and the only RC2 ciphersuite is also
1281 an EXPORT one. COMPLEMENTOFDEFAULT has been updated accordingly to add
1282 DES and RC4 ciphersuites.
1285 *) Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs.
1286 This changes the decoding behaviour for some invalid messages,
1287 though the change is mostly in the more lenient direction, and
1288 legacy behaviour is preserved as much as possible.
1291 *) Fix no-stdio build.
1292 [ David Woodhouse <David.Woodhouse@intel.com> and also
1293 Ivan Nestlerode <ivan.nestlerode@sonos.com> ]
1295 *) New testing framework
1296 The testing framework has been largely rewritten and is now using
1297 perl and the perl modules Test::Harness and an extended variant of
1298 Test::More called OpenSSL::Test to do its work. All test scripts in
1299 test/ have been rewritten into test recipes, and all direct calls to
1300 executables in test/Makefile have become individual recipes using the
1301 simplified testing OpenSSL::Test::Simple.
1303 For documentation on our testing modules, do:
1305 perldoc test/testlib/OpenSSL/Test/Simple.pm
1306 perldoc test/testlib/OpenSSL/Test.pm
1310 *) Revamped memory debug; only -DCRYPTO_MDEBUG and -DCRYPTO_MDEBUG_ABORT
1311 are used; the latter aborts on memory leaks (usually checked on exit).
1312 Some undocumented "set malloc, etc., hooks" functions were removed
1313 and others were changed. All are now documented.
1316 *) In DSA_generate_parameters_ex, if the provided seed is too short,
1318 [Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>]
1320 *) Rewrite PSK to support ECDHE_PSK, DHE_PSK and RSA_PSK. Add ciphersuites
1321 from RFC4279, RFC4785, RFC5487, RFC5489.
1323 Thanks to Christian J. Dietrich and Giuseppe D'Angelo for the
1324 original RSA_PSK patch.
1327 *) Dropped support for the SSL3_FLAGS_DELAY_CLIENT_FINISHED flag. This SSLeay
1328 era flag was never set throughout the codebase (only read). Also removed
1329 SSL3_FLAGS_POP_BUFFER which was only used if
1330 SSL3_FLAGS_DELAY_CLIENT_FINISHED was also set.
1333 *) Changed the default name options in the "ca", "crl", "req" and "x509"
1334 to be "oneline" instead of "compat".
1337 *) Remove SSL_OP_TLS_BLOCK_PADDING_BUG. This is SSLeay legacy, we're
1338 not aware of clients that still exhibit this bug, and the workaround
1339 hasn't been working properly for a while.
1342 *) The return type of BIO_number_read() and BIO_number_written() as well as
1343 the corresponding num_read and num_write members in the BIO structure has
1344 changed from unsigned long to uint64_t. On platforms where an unsigned
1345 long is 32 bits (e.g. Windows) these counters could overflow if >4Gb is
1349 *) Given the pervasive nature of TLS extensions it is inadvisable to run
1350 OpenSSL without support for them. It also means that maintaining
1351 the OPENSSL_NO_TLSEXT option within the code is very invasive (and probably
1352 not well tested). Therefore the OPENSSL_NO_TLSEXT option has been removed.
1355 *) Removed support for the two export grade static DH ciphersuites
1356 EXP-DH-RSA-DES-CBC-SHA and EXP-DH-DSS-DES-CBC-SHA. These two ciphersuites
1357 were newly added (along with a number of other static DH ciphersuites) to
1358 1.0.2. However the two export ones have *never* worked since they were
1359 introduced. It seems strange in any case to be adding new export
1360 ciphersuites, and given "logjam" it also does not seem correct to fix them.
1363 *) Version negotiation has been rewritten. In particular SSLv23_method(),
1364 SSLv23_client_method() and SSLv23_server_method() have been deprecated,
1365 and turned into macros which simply call the new preferred function names
1366 TLS_method(), TLS_client_method() and TLS_server_method(). All new code
1367 should use the new names instead. Also as part of this change the ssl23.h
1368 header file has been removed.
1371 *) Support for Kerberos ciphersuites in TLS (RFC2712) has been removed. This
1372 code and the associated standard is no longer considered fit-for-purpose.
1375 *) RT2547 was closed. When generating a private key, try to make the
1376 output file readable only by the owner. This behavior change might
1377 be noticeable when interacting with other software.
1379 *) Documented all exdata functions. Added CRYPTO_free_ex_index.
1383 *) Added HTTP GET support to the ocsp command.
1386 *) Changed default digest for the dgst and enc commands from MD5 to
1390 *) RAND_pseudo_bytes has been deprecated. Users should use RAND_bytes instead.
1393 *) Added support for TLS extended master secret from
1394 draft-ietf-tls-session-hash-03.txt. Thanks for Alfredo Pironti for an
1395 initial patch which was a great help during development.
1398 *) All libssl internal structures have been removed from the public header
1399 files, and the OPENSSL_NO_SSL_INTERN option has been removed (since it is
1400 now redundant). Users should not attempt to access internal structures
1401 directly. Instead they should use the provided API functions.
1404 *) config has been changed so that by default OPENSSL_NO_DEPRECATED is used.
1405 Access to deprecated functions can be re-enabled by running config with
1406 "enable-deprecated". In addition applications wishing to use deprecated
1407 functions must define OPENSSL_USE_DEPRECATED. Note that this new behaviour
1408 will, by default, disable some transitive includes that previously existed
1409 in the header files (e.g. ec.h will no longer, by default, include bn.h)
1412 *) Added support for OCB mode. OpenSSL has been granted a patent license
1413 compatible with the OpenSSL license for use of OCB. Details are available
1414 at https://www.openssl.org/source/OCB-patent-grant-OpenSSL.pdf. Support
1415 for OCB can be removed by calling config with no-ocb.
1418 *) SSLv2 support has been removed. It still supports receiving a SSLv2
1419 compatible client hello.
1422 *) Increased the minimal RSA keysize from 256 to 512 bits [Rich Salz],
1423 done while fixing the error code for the key-too-small case.
1424 [Annie Yousar <a.yousar@informatik.hu-berlin.de>]
1426 *) CA.sh has been removed; use CA.pl instead.
1429 *) Removed old DES API.
1432 *) Remove various unsupported platforms:
1438 Sinix/ReliantUNIX RM400
1443 16-bit platforms such as WIN16
1446 *) Clean up OPENSSL_NO_xxx #define's
1447 Use setbuf() and remove OPENSSL_NO_SETVBUF_IONBF
1448 Rename OPENSSL_SYSNAME_xxx to OPENSSL_SYS_xxx
1449 OPENSSL_NO_EC{DH,DSA} merged into OPENSSL_NO_EC
1450 OPENSSL_NO_RIPEMD160, OPENSSL_NO_RIPEMD merged into OPENSSL_NO_RMD160
1451 OPENSSL_NO_FP_API merged into OPENSSL_NO_STDIO
1452 Remove OPENSSL_NO_BIO OPENSSL_NO_BUFFER OPENSSL_NO_CHAIN_VERIFY
1453 OPENSSL_NO_EVP OPENSSL_NO_FIPS_ERR OPENSSL_NO_HASH_COMP
1454 OPENSSL_NO_LHASH OPENSSL_NO_OBJECT OPENSSL_NO_SPEED OPENSSL_NO_STACK
1455 OPENSSL_NO_X509 OPENSSL_NO_X509_VERIFY
1456 Remove MS_STATIC; it's a relic from platforms <32 bits.
1459 *) Cleaned up dead code
1460 Remove all but one '#ifdef undef' which is to be looked at.
1463 *) Clean up calling of xxx_free routines.
1464 Just like free(), fix most of the xxx_free routines to accept
1465 NULL. Remove the non-null checks from callers. Save much code.
1468 *) Add secure heap for storage of private keys (when possible).
1469 Add BIO_s_secmem(), CBIGNUM, etc.
1470 Contributed by Akamai Technologies under our Corporate CLA.
1473 *) Experimental support for a new, fast, unbiased prime candidate generator,
1474 bn_probable_prime_dh_coprime(). Not currently used by any prime generator.
1475 [Felix Laurie von Massenbach <felix@erbridge.co.uk>]
1477 *) New output format NSS in the sess_id command line tool. This allows
1478 exporting the session id and the master key in NSS keylog format.
1479 [Martin Kaiser <martin@kaiser.cx>]
1481 *) Harmonize version and its documentation. -f flag is used to display
1483 [mancha <mancha1@zoho.com>]
1485 *) Fix eckey_priv_encode so it immediately returns an error upon a failure
1486 in i2d_ECPrivateKey. Thanks to Ted Unangst for feedback on this issue.
1487 [mancha <mancha1@zoho.com>]
1489 *) Fix some double frees. These are not thought to be exploitable.
1490 [mancha <mancha1@zoho.com>]
1492 *) A missing bounds check in the handling of the TLS heartbeat extension
1493 can be used to reveal up to 64k of memory to a connected client or
1496 Thanks for Neel Mehta of Google Security for discovering this bug and to
1497 Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for
1498 preparing the fix (CVE-2014-0160)
1499 [Adam Langley, Bodo Moeller]
1501 *) Fix for the attack described in the paper "Recovering OpenSSL
1502 ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack"
1503 by Yuval Yarom and Naomi Benger. Details can be obtained from:
1504 http://eprint.iacr.org/2014/140
1506 Thanks to Yuval Yarom and Naomi Benger for discovering this
1507 flaw and to Yuval Yarom for supplying a fix (CVE-2014-0076)
1508 [Yuval Yarom and Naomi Benger]
1510 *) Use algorithm specific chains in SSL_CTX_use_certificate_chain_file():
1511 this fixes a limitation in previous versions of OpenSSL.
1514 *) Experimental encrypt-then-mac support.
1516 Experimental support for encrypt then mac from
1517 draft-gutmann-tls-encrypt-then-mac-02.txt
1519 To enable it set the appropriate extension number (0x42 for the test
1520 server) using e.g. -DTLSEXT_TYPE_encrypt_then_mac=0x42
1522 For non-compliant peers (i.e. just about everything) this should have no
1525 WARNING: EXPERIMENTAL, SUBJECT TO CHANGE.
1529 *) Add EVP support for key wrapping algorithms, to avoid problems with
1530 existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in
1531 the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap
1532 algorithms and include tests cases.
1535 *) Extend CMS code to support RSA-PSS signatures and RSA-OAEP for
1539 *) Extended RSA OAEP support via EVP_PKEY API. Options to specify digest,
1540 MGF1 digest and OAEP label.
1543 *) Make openssl verify return errors.
1544 [Chris Palmer <palmer@google.com> and Ben Laurie]
1546 *) New function ASN1_TIME_diff to calculate the difference between two
1547 ASN1_TIME structures or one structure and the current time.
1550 *) Update fips_test_suite to support multiple command line options. New
1551 test to induce all self test errors in sequence and check expected
1555 *) Add FIPS_{rsa,dsa,ecdsa}_{sign,verify} functions which digest and
1556 sign or verify all in one operation.
1559 *) Add fips_algvs: a multicall fips utility incorporating all the algorithm
1560 test programs and fips_test_suite. Includes functionality to parse
1561 the minimal script output of fipsalgest.pl directly.
1564 *) Add authorisation parameter to FIPS_module_mode_set().
1567 *) Add FIPS selftest for ECDH algorithm using P-224 and B-233 curves.
1570 *) Use separate DRBG fields for internal and external flags. New function
1571 FIPS_drbg_health_check() to perform on demand health checking. Add
1572 generation tests to fips_test_suite with reduced health check interval to
1573 demonstrate periodic health checking. Add "nodh" option to
1574 fips_test_suite to skip very slow DH test.
1577 *) New function FIPS_get_cipherbynid() to lookup FIPS supported ciphers
1581 *) More extensive health check for DRBG checking many more failure modes.
1582 New function FIPS_selftest_drbg_all() to handle every possible DRBG
1583 combination: call this in fips_test_suite.
1586 *) Add support for canonical generation of DSA parameter 'g'. See
1589 *) Add support for HMAC DRBG from SP800-90. Update DRBG algorithm test and
1590 POST to handle HMAC cases.
1593 *) Add functions FIPS_module_version() and FIPS_module_version_text()
1594 to return numerical and string versions of the FIPS module number.
1597 *) Rename FIPS_mode_set and FIPS_mode to FIPS_module_mode_set and
1598 FIPS_module_mode. FIPS_mode and FIPS_mode_set will be implemented
1599 outside the validated module in the FIPS capable OpenSSL.
1602 *) Minor change to DRBG entropy callback semantics. In some cases
1603 there is no multiple of the block length between min_len and
1604 max_len. Allow the callback to return more than max_len bytes
1605 of entropy but discard any extra: it is the callback's responsibility
1606 to ensure that the extra data discarded does not impact the
1607 requested amount of entropy.
1610 *) Add PRNG security strength checks to RSA, DSA and ECDSA using
1611 information in FIPS186-3, SP800-57 and SP800-131A.
1614 *) CCM support via EVP. Interface is very similar to GCM case except we
1615 must supply all data in one chunk (i.e. no update, final) and the
1616 message length must be supplied if AAD is used. Add algorithm test
1620 *) Initial version of POST overhaul. Add POST callback to allow the status
1621 of POST to be monitored and/or failures induced. Modify fips_test_suite
1622 to use callback. Always run all selftests even if one fails.
1625 *) XTS support including algorithm test driver in the fips_gcmtest program.
1626 Note: this does increase the maximum key length from 32 to 64 bytes but
1627 there should be no binary compatibility issues as existing applications
1628 will never use XTS mode.
1631 *) Extensive reorganisation of FIPS PRNG behaviour. Remove all dependencies
1632 to OpenSSL RAND code and replace with a tiny FIPS RAND API which also
1633 performs algorithm blocking for unapproved PRNG types. Also do not
1634 set PRNG type in FIPS_mode_set(): leave this to the application.
1635 Add default OpenSSL DRBG handling: sets up FIPS PRNG and seeds with
1636 the standard OpenSSL PRNG: set additional data to a date time vector.
1639 *) Rename old X9.31 PRNG functions of the form FIPS_rand* to FIPS_x931*.
1640 This shouldn't present any incompatibility problems because applications
1641 shouldn't be using these directly and any that are will need to rethink
1642 anyway as the X9.31 PRNG is now deprecated by FIPS 140-2
1645 *) Extensive self tests and health checking required by SP800-90 DRBG.
1646 Remove strength parameter from FIPS_drbg_instantiate and always
1647 instantiate at maximum supported strength.
1650 *) Add ECDH code to fips module and fips_ecdhvs for primitives only testing.
1653 *) New algorithm test program fips_dhvs to handle DH primitives only testing.
1656 *) New function DH_compute_key_padded() to compute a DH key and pad with
1657 leading zeroes if needed: this complies with SP800-56A et al.
1660 *) Initial implementation of SP800-90 DRBGs for Hash and CTR. Not used by
1661 anything, incomplete, subject to change and largely untested at present.
1664 *) Modify fipscanisteronly build option to only build the necessary object
1665 files by filtering FIPS_EX_OBJ through a perl script in crypto/Makefile.
1668 *) Add experimental option FIPSSYMS to give all symbols in
1669 fipscanister.o and FIPS or fips prefix. This will avoid
1670 conflicts with future versions of OpenSSL. Add perl script
1671 util/fipsas.pl to preprocess assembly language source files
1672 and rename any affected symbols.
1675 *) Add selftest checks and algorithm block of non-fips algorithms in
1676 FIPS mode. Remove DES2 from selftests.
1679 *) Add ECDSA code to fips module. Add tiny fips_ecdsa_check to just
1680 return internal method without any ENGINE dependencies. Add new
1681 tiny fips sign and verify functions.
1684 *) New build option no-ec2m to disable characteristic 2 code.
1687 *) New build option "fipscanisteronly". This only builds fipscanister.o
1688 and (currently) associated fips utilities. Uses the file Makefile.fips
1689 instead of Makefile.org as the prototype.
1692 *) Add some FIPS mode restrictions to GCM. Add internal IV generator.
1693 Update fips_gcmtest to use IV generator.
1696 *) Initial, experimental EVP support for AES-GCM. AAD can be input by
1697 setting output buffer to NULL. The *Final function must be
1698 called although it will not retrieve any additional data. The tag
1699 can be set or retrieved with a ctrl. The IV length is by default 12
1700 bytes (96 bits) but can be set to an alternative value. If the IV
1701 length exceeds the maximum IV length (currently 16 bytes) it cannot be
1705 *) New flag in ciphers: EVP_CIPH_FLAG_CUSTOM_CIPHER. This means the
1706 underlying do_cipher function handles all cipher semantics itself
1707 including padding and finalisation. This is useful if (for example)
1708 an ENGINE cipher handles block padding itself. The behaviour of
1709 do_cipher is subtly changed if this flag is set: the return value
1710 is the number of characters written to the output buffer (zero is
1711 no longer an error code) or a negative error code. Also if the
1712 input buffer is NULL and length 0 finalisation should be performed.
1715 *) If a candidate issuer certificate is already part of the constructed
1716 path ignore it: new debug notification X509_V_ERR_PATH_LOOP for this case.
1719 *) Improve forward-security support: add functions
1721 void SSL_CTX_set_not_resumable_session_callback(SSL_CTX *ctx, int (*cb)(SSL *ssl, int is_forward_secure))
1722 void SSL_set_not_resumable_session_callback(SSL *ssl, int (*cb)(SSL *ssl, int is_forward_secure))
1724 for use by SSL/TLS servers; the callback function will be called whenever a
1725 new session is created, and gets to decide whether the session may be
1726 cached to make it resumable (return 0) or not (return 1). (As by the
1727 SSL/TLS protocol specifications, the session_id sent by the server will be
1728 empty to indicate that the session is not resumable; also, the server will
1729 not generate RFC 4507 (RFC 5077) session tickets.)
1731 A simple reasonable callback implementation is to return is_forward_secure.
1732 This parameter will be set to 1 or 0 depending on the ciphersuite selected
1733 by the SSL/TLS server library, indicating whether it can provide forward
1735 [Emilia Käsper <emilia.kasper@esat.kuleuven.be> (Google)]
1737 *) New -verify_name option in command line utilities to set verification
1741 *) Initial CMAC implementation. WARNING: EXPERIMENTAL, API MAY CHANGE.
1742 Add CMAC pkey methods.
1745 *) Experimental renegotiation in s_server -www mode. If the client
1746 browses /reneg connection is renegotiated. If /renegcert it is
1747 renegotiated requesting a certificate.
1750 *) Add an "external" session cache for debugging purposes to s_server. This
1751 should help trace issues which normally are only apparent in deployed
1752 multi-process servers.
1755 *) Extensive audit of libcrypto with DEBUG_UNUSED. Fix many cases where
1756 return value is ignored. NB. The functions RAND_add(), RAND_seed(),
1757 BIO_set_cipher() and some obscure PEM functions were changed so they
1758 can now return an error. The RAND changes required a change to the
1759 RAND_METHOD structure.
1762 *) New macro __owur for "OpenSSL Warn Unused Result". This makes use of
1763 a gcc attribute to warn if the result of a function is ignored. This
1764 is enable if DEBUG_UNUSED is set. Add to several functions in evp.h
1765 whose return value is often ignored.
1768 *) New -noct, -requestct, -requirect and -ctlogfile options for s_client.
1769 These allow SCTs (signed certificate timestamps) to be requested and
1770 validated when establishing a connection.
1771 [Rob Percival <robpercival@google.com>]
1773 Changes between 1.0.2g and 1.0.2h [3 May 2016]
1775 *) Prevent padding oracle in AES-NI CBC MAC check
1777 A MITM attacker can use a padding oracle attack to decrypt traffic
1778 when the connection uses an AES CBC cipher and the server support
1781 This issue was introduced as part of the fix for Lucky 13 padding
1782 attack (CVE-2013-0169). The padding check was rewritten to be in
1783 constant time by making sure that always the same bytes are read and
1784 compared against either the MAC or padding bytes. But it no longer
1785 checked that there was enough data to have both the MAC and padding
1788 This issue was reported by Juraj Somorovsky using TLS-Attacker.
1792 *) Fix EVP_EncodeUpdate overflow
1794 An overflow can occur in the EVP_EncodeUpdate() function which is used for
1795 Base64 encoding of binary data. If an attacker is able to supply very large
1796 amounts of input data then a length check can overflow resulting in a heap
1799 Internally to OpenSSL the EVP_EncodeUpdate() function is primarily used by
1800 the PEM_write_bio* family of functions. These are mainly used within the
1801 OpenSSL command line applications, so any application which processes data
1802 from an untrusted source and outputs it as a PEM file should be considered
1803 vulnerable to this issue. User applications that call these APIs directly
1804 with large amounts of untrusted data may also be vulnerable.
1806 This issue was reported by Guido Vranken.
1810 *) Fix EVP_EncryptUpdate overflow
1812 An overflow can occur in the EVP_EncryptUpdate() function. If an attacker
1813 is able to supply very large amounts of input data after a previous call to
1814 EVP_EncryptUpdate() with a partial block then a length check can overflow
1815 resulting in a heap corruption. Following an analysis of all OpenSSL
1816 internal usage of the EVP_EncryptUpdate() function all usage is one of two
1817 forms. The first form is where the EVP_EncryptUpdate() call is known to be
1818 the first called function after an EVP_EncryptInit(), and therefore that
1819 specific call must be safe. The second form is where the length passed to
1820 EVP_EncryptUpdate() can be seen from the code to be some small value and
1821 therefore there is no possibility of an overflow. Since all instances are
1822 one of these two forms, it is believed that there can be no overflows in
1823 internal code due to this problem. It should be noted that
1824 EVP_DecryptUpdate() can call EVP_EncryptUpdate() in certain code paths.
1825 Also EVP_CipherUpdate() is a synonym for EVP_EncryptUpdate(). All instances
1826 of these calls have also been analysed too and it is believed there are no
1827 instances in internal usage where an overflow could occur.
1829 This issue was reported by Guido Vranken.
1833 *) Prevent ASN.1 BIO excessive memory allocation
1835 When ASN.1 data is read from a BIO using functions such as d2i_CMS_bio()
1836 a short invalid encoding can cause allocation of large amounts of memory
1837 potentially consuming excessive resources or exhausting memory.
1839 Any application parsing untrusted data through d2i BIO functions is
1840 affected. The memory based functions such as d2i_X509() are *not* affected.
1841 Since the memory based functions are used by the TLS library, TLS
1842 applications are not affected.
1844 This issue was reported by Brian Carpenter.
1850 ASN1 Strings that are over 1024 bytes can cause an overread in applications
1851 using the X509_NAME_oneline() function on EBCDIC systems. This could result
1852 in arbitrary stack data being returned in the buffer.
1854 This issue was reported by Guido Vranken.
1858 *) Modify behavior of ALPN to invoke callback after SNI/servername
1859 callback, such that updates to the SSL_CTX affect ALPN.
1862 *) Remove LOW from the DEFAULT cipher list. This removes singles DES from the
1866 *) Only remove the SSLv2 methods with the no-ssl2-method option. When the
1867 methods are enabled and ssl2 is disabled the methods return NULL.
1870 Changes between 1.0.2f and 1.0.2g [1 Mar 2016]
1872 * Disable weak ciphers in SSLv3 and up in default builds of OpenSSL.
1873 Builds that are not configured with "enable-weak-ssl-ciphers" will not
1874 provide any "EXPORT" or "LOW" strength ciphers.
1877 * Disable SSLv2 default build, default negotiation and weak ciphers. SSLv2
1878 is by default disabled at build-time. Builds that are not configured with
1879 "enable-ssl2" will not support SSLv2. Even if "enable-ssl2" is used,
1880 users who want to negotiate SSLv2 via the version-flexible SSLv23_method()
1881 will need to explicitly call either of:
1883 SSL_CTX_clear_options(ctx, SSL_OP_NO_SSLv2);
1885 SSL_clear_options(ssl, SSL_OP_NO_SSLv2);
1887 as appropriate. Even if either of those is used, or the application
1888 explicitly uses the version-specific SSLv2_method() or its client and
1889 server variants, SSLv2 ciphers vulnerable to exhaustive search key
1890 recovery have been removed. Specifically, the SSLv2 40-bit EXPORT
1891 ciphers, and SSLv2 56-bit DES are no longer available.
1895 *) Fix a double-free in DSA code
1897 A double free bug was discovered when OpenSSL parses malformed DSA private
1898 keys and could lead to a DoS attack or memory corruption for applications
1899 that receive DSA private keys from untrusted sources. This scenario is
1902 This issue was reported to OpenSSL by Adam Langley(Google/BoringSSL) using
1907 *) Disable SRP fake user seed to address a server memory leak.
1909 Add a new method SRP_VBASE_get1_by_user that handles the seed properly.
1911 SRP_VBASE_get_by_user had inconsistent memory management behaviour.
1912 In order to fix an unavoidable memory leak, SRP_VBASE_get_by_user
1913 was changed to ignore the "fake user" SRP seed, even if the seed
1916 Users should use SRP_VBASE_get1_by_user instead. Note that in
1917 SRP_VBASE_get1_by_user, caller must free the returned value. Note
1918 also that even though configuring the SRP seed attempts to hide
1919 invalid usernames by continuing the handshake with fake
1920 credentials, this behaviour is not constant time and no strong
1921 guarantees are made that the handshake is indistinguishable from
1922 that of a valid user.
1926 *) Fix BN_hex2bn/BN_dec2bn NULL pointer deref/heap corruption
1928 In the BN_hex2bn function the number of hex digits is calculated using an
1929 int value |i|. Later |bn_expand| is called with a value of |i * 4|. For
1930 large values of |i| this can result in |bn_expand| not allocating any
1931 memory because |i * 4| is negative. This can leave the internal BIGNUM data
1932 field as NULL leading to a subsequent NULL ptr deref. For very large values
1933 of |i|, the calculation |i * 4| could be a positive value smaller than |i|.
1934 In this case memory is allocated to the internal BIGNUM data field, but it
1935 is insufficiently sized leading to heap corruption. A similar issue exists
1936 in BN_dec2bn. This could have security consequences if BN_hex2bn/BN_dec2bn
1937 is ever called by user applications with very large untrusted hex/dec data.
1938 This is anticipated to be a rare occurrence.
1940 All OpenSSL internal usage of these functions use data that is not expected
1941 to be untrusted, e.g. config file data or application command line
1942 arguments. If user developed applications generate config file data based
1943 on untrusted data then it is possible that this could also lead to security
1944 consequences. This is also anticipated to be rare.
1946 This issue was reported to OpenSSL by Guido Vranken.
1950 *) Fix memory issues in BIO_*printf functions
1952 The internal |fmtstr| function used in processing a "%s" format string in
1953 the BIO_*printf functions could overflow while calculating the length of a
1954 string and cause an OOB read when printing very long strings.
1956 Additionally the internal |doapr_outch| function can attempt to write to an
1957 OOB memory location (at an offset from the NULL pointer) in the event of a
1958 memory allocation failure. In 1.0.2 and below this could be caused where
1959 the size of a buffer to be allocated is greater than INT_MAX. E.g. this
1960 could be in processing a very long "%s" format string. Memory leaks can
1963 The first issue may mask the second issue dependent on compiler behaviour.
1964 These problems could enable attacks where large amounts of untrusted data
1965 is passed to the BIO_*printf functions. If applications use these functions
1966 in this way then they could be vulnerable. OpenSSL itself uses these
1967 functions when printing out human-readable dumps of ASN.1 data. Therefore
1968 applications that print this data could be vulnerable if the data is from
1969 untrusted sources. OpenSSL command line applications could also be
1970 vulnerable where they print out ASN.1 data, or if untrusted data is passed
1971 as command line arguments.
1973 Libssl is not considered directly vulnerable. Additionally certificates etc
1974 received via remote connections via libssl are also unlikely to be able to
1975 trigger these issues because of message size limits enforced within libssl.
1977 This issue was reported to OpenSSL Guido Vranken.
1981 *) Side channel attack on modular exponentiation
1983 A side-channel attack was found which makes use of cache-bank conflicts on
1984 the Intel Sandy-Bridge microarchitecture which could lead to the recovery
1985 of RSA keys. The ability to exploit this issue is limited as it relies on
1986 an attacker who has control of code in a thread running on the same
1987 hyper-threaded core as the victim thread which is performing decryptions.
1989 This issue was reported to OpenSSL by Yuval Yarom, The University of
1990 Adelaide and NICTA, Daniel Genkin, Technion and Tel Aviv University, and
1991 Nadia Heninger, University of Pennsylvania with more information at
1992 http://cachebleed.info.
1996 *) Change the req app to generate a 2048-bit RSA/DSA key by default,
1997 if no keysize is specified with default_bits. This fixes an
1998 omission in an earlier change that changed all RSA/DSA key generation
1999 apps to use 2048 bits by default.
2002 Changes between 1.0.2e and 1.0.2f [28 Jan 2016]
2003 *) DH small subgroups
2005 Historically OpenSSL only ever generated DH parameters based on "safe"
2006 primes. More recently (in version 1.0.2) support was provided for
2007 generating X9.42 style parameter files such as those required for RFC 5114
2008 support. The primes used in such files may not be "safe". Where an
2009 application is using DH configured with parameters based on primes that are
2010 not "safe" then an attacker could use this fact to find a peer's private
2011 DH exponent. This attack requires that the attacker complete multiple
2012 handshakes in which the peer uses the same private DH exponent. For example
2013 this could be used to discover a TLS server's private DH exponent if it's
2014 reusing the private DH exponent or it's using a static DH ciphersuite.
2016 OpenSSL provides the option SSL_OP_SINGLE_DH_USE for ephemeral DH (DHE) in
2017 TLS. It is not on by default. If the option is not set then the server
2018 reuses the same private DH exponent for the life of the server process and
2019 would be vulnerable to this attack. It is believed that many popular
2020 applications do set this option and would therefore not be at risk.
2022 The fix for this issue adds an additional check where a "q" parameter is
2023 available (as is the case in X9.42 based parameters). This detects the
2024 only known attack, and is the only possible defense for static DH
2025 ciphersuites. This could have some performance impact.
2027 Additionally the SSL_OP_SINGLE_DH_USE option has been switched on by
2028 default and cannot be disabled. This could have some performance impact.
2030 This issue was reported to OpenSSL by Antonio Sanso (Adobe).
2034 *) SSLv2 doesn't block disabled ciphers
2036 A malicious client can negotiate SSLv2 ciphers that have been disabled on
2037 the server and complete SSLv2 handshakes even if all SSLv2 ciphers have
2038 been disabled, provided that the SSLv2 protocol was not also disabled via
2041 This issue was reported to OpenSSL on 26th December 2015 by Nimrod Aviram
2042 and Sebastian Schinzel.
2046 Changes between 1.0.2d and 1.0.2e [3 Dec 2015]
2048 *) BN_mod_exp may produce incorrect results on x86_64
2050 There is a carry propagating bug in the x86_64 Montgomery squaring
2051 procedure. No EC algorithms are affected. Analysis suggests that attacks
2052 against RSA and DSA as a result of this defect would be very difficult to
2053 perform and are not believed likely. Attacks against DH are considered just
2054 feasible (although very difficult) because most of the work necessary to
2055 deduce information about a private key may be performed offline. The amount
2056 of resources required for such an attack would be very significant and
2057 likely only accessible to a limited number of attackers. An attacker would
2058 additionally need online access to an unpatched system using the target
2059 private key in a scenario with persistent DH parameters and a private
2060 key that is shared between multiple clients. For example this can occur by
2061 default in OpenSSL DHE based SSL/TLS ciphersuites.
2063 This issue was reported to OpenSSL by Hanno Böck.
2067 *) Certificate verify crash with missing PSS parameter
2069 The signature verification routines will crash with a NULL pointer
2070 dereference if presented with an ASN.1 signature using the RSA PSS
2071 algorithm and absent mask generation function parameter. Since these
2072 routines are used to verify certificate signature algorithms this can be
2073 used to crash any certificate verification operation and exploited in a
2074 DoS attack. Any application which performs certificate verification is
2075 vulnerable including OpenSSL clients and servers which enable client
2078 This issue was reported to OpenSSL by Loïc Jonas Etienne (Qnective AG).
2082 *) X509_ATTRIBUTE memory leak
2084 When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak
2085 memory. This structure is used by the PKCS#7 and CMS routines so any
2086 application which reads PKCS#7 or CMS data from untrusted sources is
2087 affected. SSL/TLS is not affected.
2089 This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using
2094 *) Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs.
2095 This changes the decoding behaviour for some invalid messages,
2096 though the change is mostly in the more lenient direction, and
2097 legacy behaviour is preserved as much as possible.
2100 *) In DSA_generate_parameters_ex, if the provided seed is too short,
2102 [Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>]
2104 Changes between 1.0.2c and 1.0.2d [9 Jul 2015]
2106 *) Alternate chains certificate forgery
2108 During certificate verification, OpenSSL will attempt to find an
2109 alternative certificate chain if the first attempt to build such a chain
2110 fails. An error in the implementation of this logic can mean that an
2111 attacker could cause certain checks on untrusted certificates to be
2112 bypassed, such as the CA flag, enabling them to use a valid leaf
2113 certificate to act as a CA and "issue" an invalid certificate.
2115 This issue was reported to OpenSSL by Adam Langley/David Benjamin
2119 Changes between 1.0.2b and 1.0.2c [12 Jun 2015]
2121 *) Fix HMAC ABI incompatibility. The previous version introduced an ABI
2122 incompatibility in the handling of HMAC. The previous ABI has now been
2126 Changes between 1.0.2a and 1.0.2b [11 Jun 2015]
2128 *) Malformed ECParameters causes infinite loop
2130 When processing an ECParameters structure OpenSSL enters an infinite loop
2131 if the curve specified is over a specially malformed binary polynomial
2134 This can be used to perform denial of service against any
2135 system which processes public keys, certificate requests or
2136 certificates. This includes TLS clients and TLS servers with
2137 client authentication enabled.
2139 This issue was reported to OpenSSL by Joseph Barr-Pixton.
2143 *) Exploitable out-of-bounds read in X509_cmp_time
2145 X509_cmp_time does not properly check the length of the ASN1_TIME
2146 string and can read a few bytes out of bounds. In addition,
2147 X509_cmp_time accepts an arbitrary number of fractional seconds in the
2150 An attacker can use this to craft malformed certificates and CRLs of
2151 various sizes and potentially cause a segmentation fault, resulting in
2152 a DoS on applications that verify certificates or CRLs. TLS clients
2153 that verify CRLs are affected. TLS clients and servers with client
2154 authentication enabled may be affected if they use custom verification
2157 This issue was reported to OpenSSL by Robert Swiecki (Google), and
2158 independently by Hanno Böck.
2162 *) PKCS7 crash with missing EnvelopedContent
2164 The PKCS#7 parsing code does not handle missing inner EncryptedContent
2165 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs
2166 with missing content and trigger a NULL pointer dereference on parsing.
2168 Applications that decrypt PKCS#7 data or otherwise parse PKCS#7
2169 structures from untrusted sources are affected. OpenSSL clients and
2170 servers are not affected.
2172 This issue was reported to OpenSSL by Michal Zalewski (Google).
2176 *) CMS verify infinite loop with unknown hash function
2178 When verifying a signedData message the CMS code can enter an infinite loop
2179 if presented with an unknown hash function OID. This can be used to perform
2180 denial of service against any system which verifies signedData messages using
2182 This issue was reported to OpenSSL by Johannes Bauer.
2186 *) Race condition handling NewSessionTicket
2188 If a NewSessionTicket is received by a multi-threaded client when attempting to
2189 reuse a previous ticket then a race condition can occur potentially leading to
2190 a double free of the ticket data.
2194 *) Only support 256-bit or stronger elliptic curves with the
2195 'ecdh_auto' setting (server) or by default (client). Of supported
2196 curves, prefer P-256 (both).
2199 Changes between 1.0.2 and 1.0.2a [19 Mar 2015]
2201 *) ClientHello sigalgs DoS fix
2203 If a client connects to an OpenSSL 1.0.2 server and renegotiates with an
2204 invalid signature algorithms extension a NULL pointer dereference will
2205 occur. This can be exploited in a DoS attack against the server.
2207 This issue was was reported to OpenSSL by David Ramos of Stanford
2210 [Stephen Henson and Matt Caswell]
2212 *) Multiblock corrupted pointer fix
2214 OpenSSL 1.0.2 introduced the "multiblock" performance improvement. This
2215 feature only applies on 64 bit x86 architecture platforms that support AES
2216 NI instructions. A defect in the implementation of "multiblock" can cause
2217 OpenSSL's internal write buffer to become incorrectly set to NULL when
2218 using non-blocking IO. Typically, when the user application is using a
2219 socket BIO for writing, this will only result in a failed connection.
2220 However if some other BIO is used then it is likely that a segmentation
2221 fault will be triggered, thus enabling a potential DoS attack.
2223 This issue was reported to OpenSSL by Daniel Danner and Rainer Mueller.
2227 *) Segmentation fault in DTLSv1_listen fix
2229 The DTLSv1_listen function is intended to be stateless and processes the
2230 initial ClientHello from many peers. It is common for user code to loop
2231 over the call to DTLSv1_listen until a valid ClientHello is received with
2232 an associated cookie. A defect in the implementation of DTLSv1_listen means
2233 that state is preserved in the SSL object from one invocation to the next
2234 that can lead to a segmentation fault. Errors processing the initial
2235 ClientHello can trigger this scenario. An example of such an error could be
2236 that a DTLS1.0 only client is attempting to connect to a DTLS1.2 only
2239 This issue was reported to OpenSSL by Per Allansson.
2243 *) Segmentation fault in ASN1_TYPE_cmp fix
2245 The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is
2246 made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check
2247 certificate signature algorithm consistency this can be used to crash any
2248 certificate verification operation and exploited in a DoS attack. Any
2249 application which performs certificate verification is vulnerable including
2250 OpenSSL clients and servers which enable client authentication.
2254 *) Segmentation fault for invalid PSS parameters fix
2256 The signature verification routines will crash with a NULL pointer
2257 dereference if presented with an ASN.1 signature using the RSA PSS
2258 algorithm and invalid parameters. Since these routines are used to verify
2259 certificate signature algorithms this can be used to crash any
2260 certificate verification operation and exploited in a DoS attack. Any
2261 application which performs certificate verification is vulnerable including
2262 OpenSSL clients and servers which enable client authentication.
2264 This issue was was reported to OpenSSL by Brian Carpenter.
2268 *) ASN.1 structure reuse memory corruption fix
2270 Reusing a structure in ASN.1 parsing may allow an attacker to cause
2271 memory corruption via an invalid write. Such reuse is and has been
2272 strongly discouraged and is believed to be rare.
2274 Applications that parse structures containing CHOICE or ANY DEFINED BY
2275 components may be affected. Certificate parsing (d2i_X509 and related
2276 functions) are however not affected. OpenSSL clients and servers are
2281 *) PKCS7 NULL pointer dereferences fix
2283 The PKCS#7 parsing code does not handle missing outer ContentInfo
2284 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with
2285 missing content and trigger a NULL pointer dereference on parsing.
2287 Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or
2288 otherwise parse PKCS#7 structures from untrusted sources are
2289 affected. OpenSSL clients and servers are not affected.
2291 This issue was reported to OpenSSL by Michal Zalewski (Google).
2295 *) DoS via reachable assert in SSLv2 servers fix
2297 A malicious client can trigger an OPENSSL_assert (i.e., an abort) in
2298 servers that both support SSLv2 and enable export cipher suites by sending
2299 a specially crafted SSLv2 CLIENT-MASTER-KEY message.
2301 This issue was discovered by Sean Burford (Google) and Emilia Käsper
2302 (OpenSSL development team).
2306 *) Empty CKE with client auth and DHE fix
2308 If client auth is used then a server can seg fault in the event of a DHE
2309 ciphersuite being selected and a zero length ClientKeyExchange message
2310 being sent by the client. This could be exploited in a DoS attack.
2314 *) Handshake with unseeded PRNG fix
2316 Under certain conditions an OpenSSL 1.0.2 client can complete a handshake
2317 with an unseeded PRNG. The conditions are:
2318 - The client is on a platform where the PRNG has not been seeded
2319 automatically, and the user has not seeded manually
2320 - A protocol specific client method version has been used (i.e. not
2321 SSL_client_methodv23)
2322 - A ciphersuite is used that does not require additional random data from
2323 the PRNG beyond the initial ClientHello client random (e.g. PSK-RC4-SHA).
2325 If the handshake succeeds then the client random that has been used will
2326 have been generated from a PRNG with insufficient entropy and therefore the
2327 output may be predictable.
2329 For example using the following command with an unseeded openssl will
2330 succeed on an unpatched platform:
2332 openssl s_client -psk 1a2b3c4d -tls1_2 -cipher PSK-RC4-SHA
2336 *) Use After Free following d2i_ECPrivatekey error fix
2338 A malformed EC private key file consumed via the d2i_ECPrivateKey function
2339 could cause a use after free condition. This, in turn, could cause a double
2340 free in several private key parsing functions (such as d2i_PrivateKey
2341 or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption
2342 for applications that receive EC private keys from untrusted
2343 sources. This scenario is considered rare.
2345 This issue was discovered by the BoringSSL project and fixed in their
2350 *) X509_to_X509_REQ NULL pointer deref fix
2352 The function X509_to_X509_REQ will crash with a NULL pointer dereference if
2353 the certificate key is invalid. This function is rarely used in practice.
2355 This issue was discovered by Brian Carpenter.
2359 *) Removed the export ciphers from the DEFAULT ciphers
2362 Changes between 1.0.1l and 1.0.2 [22 Jan 2015]
2364 *) Facilitate "universal" ARM builds targeting range of ARM ISAs, e.g.
2365 ARMv5 through ARMv8, as opposite to "locking" it to single one.
2366 So far those who have to target multiple platforms would compromise
2367 and argue that binary targeting say ARMv5 would still execute on
2368 ARMv8. "Universal" build resolves this compromise by providing
2369 near-optimal performance even on newer platforms.
2372 *) Accelerated NIST P-256 elliptic curve implementation for x86_64
2373 (other platforms pending).
2374 [Shay Gueron & Vlad Krasnov (Intel Corp), Andy Polyakov]
2376 *) Add support for the SignedCertificateTimestampList certificate and
2377 OCSP response extensions from RFC6962.
2380 *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.)
2381 for corner cases. (Certain input points at infinity could lead to
2382 bogus results, with non-infinity inputs mapped to infinity too.)
2385 *) Initial support for PowerISA 2.0.7, first implemented in POWER8.
2386 This covers AES, SHA256/512 and GHASH. "Initial" means that most
2387 common cases are optimized and there still is room for further
2388 improvements. Vector Permutation AES for Altivec is also added.
2391 *) Add support for little-endian ppc64 Linux target.
2392 [Marcelo Cerri (IBM)]
2394 *) Initial support for AMRv8 ISA crypto extensions. This covers AES,
2395 SHA1, SHA256 and GHASH. "Initial" means that most common cases
2396 are optimized and there still is room for further improvements.
2397 Both 32- and 64-bit modes are supported.
2398 [Andy Polyakov, Ard Biesheuvel (Linaro)]
2400 *) Improved ARMv7 NEON support.
2403 *) Support for SPARC Architecture 2011 crypto extensions, first
2404 implemented in SPARC T4. This covers AES, DES, Camellia, SHA1,
2405 SHA256/512, MD5, GHASH and modular exponentiation.
2406 [Andy Polyakov, David Miller]
2408 *) Accelerated modular exponentiation for Intel processors, a.k.a.
2410 [Shay Gueron & Vlad Krasnov (Intel Corp)]
2412 *) Support for new and upcoming Intel processors, including AVX2,
2413 BMI and SHA ISA extensions. This includes additional "stitched"
2414 implementations, AESNI-SHA256 and GCM, and multi-buffer support
2417 This work was sponsored by Intel Corp.
2420 *) Support for DTLS 1.2. This adds two sets of DTLS methods: DTLS_*_method()
2421 supports both DTLS 1.2 and 1.0 and should use whatever version the peer
2422 supports and DTLSv1_2_*_method() which supports DTLS 1.2 only.
2425 *) Use algorithm specific chains in SSL_CTX_use_certificate_chain_file():
2426 this fixes a limitation in previous versions of OpenSSL.
2429 *) Extended RSA OAEP support via EVP_PKEY API. Options to specify digest,
2430 MGF1 digest and OAEP label.
2433 *) Add EVP support for key wrapping algorithms, to avoid problems with
2434 existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in
2435 the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap
2436 algorithms and include tests cases.
2439 *) Add functions to allocate and set the fields of an ECDSA_METHOD
2441 [Douglas E. Engert, Steve Henson]
2443 *) New functions OPENSSL_gmtime_diff and ASN1_TIME_diff to find the
2444 difference in days and seconds between two tm or ASN1_TIME structures.
2447 *) Add -rev test option to s_server to just reverse order of characters
2448 received by client and send back to server. Also prints an abbreviated
2449 summary of the connection parameters.
2452 *) New option -brief for s_client and s_server to print out a brief summary
2453 of connection parameters.
2456 *) Add callbacks for arbitrary TLS extensions.
2457 [Trevor Perrin <trevp@trevp.net> and Ben Laurie]
2459 *) New option -crl_download in several openssl utilities to download CRLs
2460 from CRLDP extension in certificates.
2463 *) New options -CRL and -CRLform for s_client and s_server for CRLs.
2466 *) New function X509_CRL_diff to generate a delta CRL from the difference
2467 of two full CRLs. Add support to "crl" utility.
2470 *) New functions to set lookup_crls function and to retrieve
2471 X509_STORE from X509_STORE_CTX.
2474 *) Print out deprecated issuer and subject unique ID fields in
2478 *) Extend OCSP I/O functions so they can be used for simple general purpose
2479 HTTP as well as OCSP. New wrapper function which can be used to download
2480 CRLs using the OCSP API.
2483 *) Delegate command line handling in s_client/s_server to SSL_CONF APIs.
2486 *) SSL_CONF* functions. These provide a common framework for application
2487 configuration using configuration files or command lines.
2490 *) SSL/TLS tracing code. This parses out SSL/TLS records using the
2491 message callback and prints the results. Needs compile time option
2492 "enable-ssl-trace". New options to s_client and s_server to enable
2496 *) New ctrl and macro to retrieve supported points extensions.
2497 Print out extension in s_server and s_client.
2500 *) New functions to retrieve certificate signature and signature
2504 *) Add functions to retrieve and manipulate the raw cipherlist sent by a
2508 *) New Suite B modes for TLS code. These use and enforce the requirements
2509 of RFC6460: restrict ciphersuites, only permit Suite B algorithms and
2510 only use Suite B curves. The Suite B modes can be set by using the
2511 strings "SUITEB128", "SUITEB192" or "SUITEB128ONLY" for the cipherstring.
2514 *) New chain verification flags for Suite B levels of security. Check
2515 algorithms are acceptable when flags are set in X509_verify_cert.
2518 *) Make tls1_check_chain return a set of flags indicating checks passed
2519 by a certificate chain. Add additional tests to handle client
2520 certificates: checks for matching certificate type and issuer name
2524 *) If an attempt is made to use a signature algorithm not in the peer
2525 preference list abort the handshake. If client has no suitable
2526 signature algorithms in response to a certificate request do not
2527 use the certificate.
2530 *) If server EC tmp key is not in client preference list abort handshake.
2533 *) Add support for certificate stores in CERT structure. This makes it
2534 possible to have different stores per SSL structure or one store in
2535 the parent SSL_CTX. Include distinct stores for certificate chain
2536 verification and chain building. New ctrl SSL_CTRL_BUILD_CERT_CHAIN
2537 to build and store a certificate chain in CERT structure: returning
2538 an error if the chain cannot be built: this will allow applications
2539 to test if a chain is correctly configured.
2541 Note: if the CERT based stores are not set then the parent SSL_CTX
2542 store is used to retain compatibility with existing behaviour.
2546 *) New function ssl_set_client_disabled to set a ciphersuite disabled
2547 mask based on the current session, check mask when sending client
2548 hello and checking the requested ciphersuite.
2551 *) New ctrls to retrieve and set certificate types in a certificate
2552 request message. Print out received values in s_client. If certificate
2553 types is not set with custom values set sensible values based on
2554 supported signature algorithms.
2557 *) Support for distinct client and server supported signature algorithms.
2560 *) Add certificate callback. If set this is called whenever a certificate
2561 is required by client or server. An application can decide which
2562 certificate chain to present based on arbitrary criteria: for example
2563 supported signature algorithms. Add very simple example to s_server.
2564 This fixes many of the problems and restrictions of the existing client
2565 certificate callback: for example you can now clear an existing
2566 certificate and specify the whole chain.
2569 *) Add new "valid_flags" field to CERT_PKEY structure which determines what
2570 the certificate can be used for (if anything). Set valid_flags field
2571 in new tls1_check_chain function. Simplify ssl_set_cert_masks which used
2572 to have similar checks in it.
2574 Add new "cert_flags" field to CERT structure and include a "strict mode".
2575 This enforces some TLS certificate requirements (such as only permitting
2576 certificate signature algorithms contained in the supported algorithms
2577 extension) which some implementations ignore: this option should be used
2578 with caution as it could cause interoperability issues.
2581 *) Update and tidy signature algorithm extension processing. Work out
2582 shared signature algorithms based on preferences and peer algorithms
2583 and print them out in s_client and s_server. Abort handshake if no
2584 shared signature algorithms.
2587 *) Add new functions to allow customised supported signature algorithms
2588 for SSL and SSL_CTX structures. Add options to s_client and s_server
2592 *) New function SSL_certs_clear() to delete all references to certificates
2593 from an SSL structure. Before this once a certificate had been added
2594 it couldn't be removed.
2597 *) Integrate hostname, email address and IP address checking with certificate
2598 verification. New verify options supporting checking in openssl utility.
2601 *) Fixes and wildcard matching support to hostname and email checking
2602 functions. Add manual page.
2603 [Florian Weimer (Red Hat Product Security Team)]
2605 *) New functions to check a hostname email or IP address against a
2606 certificate. Add options x509 utility to print results of checks against
2610 *) Fix OCSP checking.
2611 [Rob Stradling <rob.stradling@comodo.com> and Ben Laurie]
2613 *) Initial experimental support for explicitly trusted non-root CAs.
2614 OpenSSL still tries to build a complete chain to a root but if an
2615 intermediate CA has a trust setting included that is used. The first
2616 setting is used: whether to trust (e.g., -addtrust option to the x509
2620 *) Add -trusted_first option which attempts to find certificates in the
2621 trusted store even if an untrusted chain is also supplied.
2624 *) MIPS assembly pack updates: support for MIPS32r2 and SmartMIPS ASE,
2625 platform support for Linux and Android.
2628 *) Support for linux-x32, ILP32 environment in x86_64 framework.
2631 *) Experimental multi-implementation support for FIPS capable OpenSSL.
2632 When in FIPS mode the approved implementations are used as normal,
2633 when not in FIPS mode the internal unapproved versions are used instead.
2634 This means that the FIPS capable OpenSSL isn't forced to use the
2635 (often lower performance) FIPS implementations outside FIPS mode.
2638 *) Transparently support X9.42 DH parameters when calling
2639 PEM_read_bio_DHparameters. This means existing applications can handle
2640 the new parameter format automatically.
2643 *) Initial experimental support for X9.42 DH parameter format: mainly
2644 to support use of 'q' parameter for RFC5114 parameters.
2647 *) Add DH parameters from RFC5114 including test data to dhtest.
2650 *) Support for automatic EC temporary key parameter selection. If enabled
2651 the most preferred EC parameters are automatically used instead of
2652 hardcoded fixed parameters. Now a server just has to call:
2653 SSL_CTX_set_ecdh_auto(ctx, 1) and the server will automatically
2654 support ECDH and use the most appropriate parameters.
2657 *) Enhance and tidy EC curve and point format TLS extension code. Use
2658 static structures instead of allocation if default values are used.
2659 New ctrls to set curves we wish to support and to retrieve shared curves.
2660 Print out shared curves in s_server. New options to s_server and s_client
2661 to set list of supported curves.
2664 *) New ctrls to retrieve supported signature algorithms and
2665 supported curve values as an array of NIDs. Extend openssl utility
2666 to print out received values.
2669 *) Add new APIs EC_curve_nist2nid and EC_curve_nid2nist which convert
2670 between NIDs and the more common NIST names such as "P-256". Enhance
2671 ecparam utility and ECC method to recognise the NIST names for curves.
2674 *) Enhance SSL/TLS certificate chain handling to support different
2675 chains for each certificate instead of one chain in the parent SSL_CTX.
2678 *) Support for fixed DH ciphersuite client authentication: where both
2679 server and client use DH certificates with common parameters.
2682 *) Support for fixed DH ciphersuites: those requiring DH server
2686 *) New function i2d_re_X509_tbs for re-encoding the TBS portion of
2688 Note: Related 1.0.2-beta specific macros X509_get_cert_info,
2689 X509_CINF_set_modified, X509_CINF_get_issuer, X509_CINF_get_extensions and
2690 X509_CINF_get_signature were reverted post internal team review.
2692 Changes between 1.0.1k and 1.0.1l [15 Jan 2015]
2694 *) Build fixes for the Windows and OpenVMS platforms
2695 [Matt Caswell and Richard Levitte]
2697 Changes between 1.0.1j and 1.0.1k [8 Jan 2015]
2699 *) Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS
2700 message can cause a segmentation fault in OpenSSL due to a NULL pointer
2701 dereference. This could lead to a Denial Of Service attack. Thanks to
2702 Markus Stenberg of Cisco Systems, Inc. for reporting this issue.
2706 *) Fix DTLS memory leak in dtls1_buffer_record. A memory leak can occur in the
2707 dtls1_buffer_record function under certain conditions. In particular this
2708 could occur if an attacker sent repeated DTLS records with the same
2709 sequence number but for the next epoch. The memory leak could be exploited
2710 by an attacker in a Denial of Service attack through memory exhaustion.
2711 Thanks to Chris Mueller for reporting this issue.
2715 *) Fix issue where no-ssl3 configuration sets method to NULL. When openssl is
2716 built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl
2717 method would be set to NULL which could later result in a NULL pointer
2718 dereference. Thanks to Frank Schmirler for reporting this issue.
2722 *) Abort handshake if server key exchange message is omitted for ephemeral
2725 Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for
2726 reporting this issue.
2730 *) Remove non-export ephemeral RSA code on client and server. This code
2731 violated the TLS standard by allowing the use of temporary RSA keys in
2732 non-export ciphersuites and could be used by a server to effectively
2733 downgrade the RSA key length used to a value smaller than the server
2734 certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at
2735 INRIA or reporting this issue.
2739 *) Fixed issue where DH client certificates are accepted without verification.
2740 An OpenSSL server will accept a DH certificate for client authentication
2741 without the certificate verify message. This effectively allows a client to
2742 authenticate without the use of a private key. This only affects servers
2743 which trust a client certificate authority which issues certificates
2744 containing DH keys: these are extremely rare and hardly ever encountered.
2745 Thanks for Karthikeyan Bhargavan of the PROSECCO team at INRIA or reporting
2750 *) Ensure that the session ID context of an SSL is updated when its
2751 SSL_CTX is updated via SSL_set_SSL_CTX.
2753 The session ID context is typically set from the parent SSL_CTX,
2754 and can vary with the CTX.
2757 *) Fix various certificate fingerprint issues.
2759 By using non-DER or invalid encodings outside the signed portion of a
2760 certificate the fingerprint can be changed without breaking the signature.
2761 Although no details of the signed portion of the certificate can be changed
2762 this can cause problems with some applications: e.g. those using the
2763 certificate fingerprint for blacklists.
2765 1. Reject signatures with non zero unused bits.
2767 If the BIT STRING containing the signature has non zero unused bits reject
2768 the signature. All current signature algorithms require zero unused bits.
2770 2. Check certificate algorithm consistency.
2772 Check the AlgorithmIdentifier inside TBS matches the one in the
2773 certificate signature. NB: this will result in signature failure
2774 errors for some broken certificates.
2776 Thanks to Konrad Kraszewski from Google for reporting this issue.
2778 3. Check DSA/ECDSA signatures use DER.
2780 Re-encode DSA/ECDSA signatures and compare with the original received
2781 signature. Return an error if there is a mismatch.
2783 This will reject various cases including garbage after signature
2784 (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS
2785 program for discovering this case) and use of BER or invalid ASN.1 INTEGERs
2786 (negative or with leading zeroes).
2788 Further analysis was conducted and fixes were developed by Stephen Henson
2789 of the OpenSSL core team.
2794 *) Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect
2795 results on some platforms, including x86_64. This bug occurs at random
2796 with a very low probability, and is not known to be exploitable in any
2797 way, though its exact impact is difficult to determine. Thanks to Pieter
2798 Wuille (Blockstream) who reported this issue and also suggested an initial
2799 fix. Further analysis was conducted by the OpenSSL development team and
2800 Adam Langley of Google. The final fix was developed by Andy Polyakov of
2801 the OpenSSL core team.
2805 *) Do not resume sessions on the server if the negotiated protocol
2806 version does not match the session's version. Resuming with a different
2807 version, while not strictly forbidden by the RFC, is of questionable
2808 sanity and breaks all known clients.
2809 [David Benjamin, Emilia Käsper]
2811 *) Tighten handling of the ChangeCipherSpec (CCS) message: reject
2812 early CCS messages during renegotiation. (Note that because
2813 renegotiation is encrypted, this early CCS was not exploitable.)
2816 *) Tighten client-side session ticket handling during renegotiation:
2817 ensure that the client only accepts a session ticket if the server sends
2818 the extension anew in the ServerHello. Previously, a TLS client would
2819 reuse the old extension state and thus accept a session ticket if one was
2820 announced in the initial ServerHello.
2822 Similarly, ensure that the client requires a session ticket if one
2823 was advertised in the ServerHello. Previously, a TLS client would
2824 ignore a missing NewSessionTicket message.
2827 Changes between 1.0.1i and 1.0.1j [15 Oct 2014]
2829 *) SRTP Memory Leak.
2831 A flaw in the DTLS SRTP extension parsing code allows an attacker, who
2832 sends a carefully crafted handshake message, to cause OpenSSL to fail
2833 to free up to 64k of memory causing a memory leak. This could be
2834 exploited in a Denial Of Service attack. This issue affects OpenSSL
2835 1.0.1 server implementations for both SSL/TLS and DTLS regardless of
2836 whether SRTP is used or configured. Implementations of OpenSSL that
2837 have been compiled with OPENSSL_NO_SRTP defined are not affected.
2839 The fix was developed by the OpenSSL team.
2843 *) Session Ticket Memory Leak.
2845 When an OpenSSL SSL/TLS/DTLS server receives a session ticket the
2846 integrity of that ticket is first verified. In the event of a session
2847 ticket integrity check failing, OpenSSL will fail to free memory
2848 causing a memory leak. By sending a large number of invalid session
2849 tickets an attacker could exploit this issue in a Denial Of Service
2854 *) Build option no-ssl3 is incomplete.
2856 When OpenSSL is configured with "no-ssl3" as a build option, servers
2857 could accept and complete a SSL 3.0 handshake, and clients could be
2858 configured to send them.
2860 [Akamai and the OpenSSL team]
2862 *) Add support for TLS_FALLBACK_SCSV.
2863 Client applications doing fallback retries should call
2864 SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV).
2866 [Adam Langley, Bodo Moeller]
2868 *) Add additional DigestInfo checks.
2870 Re-encode DigestInto in DER and check against the original when
2871 verifying RSA signature: this will reject any improperly encoded
2872 DigestInfo structures.
2874 Note: this is a precautionary measure and no attacks are currently known.
2878 Changes between 1.0.1h and 1.0.1i [6 Aug 2014]
2880 *) Fix SRP buffer overrun vulnerability. Invalid parameters passed to the
2881 SRP code can be overrun an internal buffer. Add sanity check that
2882 g, A, B < N to SRP code.
2884 Thanks to Sean Devlin and Watson Ladd of Cryptography Services, NCC
2885 Group for discovering this issue.
2889 *) A flaw in the OpenSSL SSL/TLS server code causes the server to negotiate
2890 TLS 1.0 instead of higher protocol versions when the ClientHello message
2891 is badly fragmented. This allows a man-in-the-middle attacker to force a
2892 downgrade to TLS 1.0 even if both the server and the client support a
2893 higher protocol version, by modifying the client's TLS records.
2895 Thanks to David Benjamin and Adam Langley (Google) for discovering and
2896 researching this issue.
2900 *) OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject
2901 to a denial of service attack. A malicious server can crash the client
2902 with a null pointer dereference (read) by specifying an anonymous (EC)DH
2903 ciphersuite and sending carefully crafted handshake messages.
2905 Thanks to Felix Gröbert (Google) for discovering and researching this
2910 *) By sending carefully crafted DTLS packets an attacker could cause openssl
2911 to leak memory. This can be exploited through a Denial of Service attack.
2912 Thanks to Adam Langley for discovering and researching this issue.
2916 *) An attacker can force openssl to consume large amounts of memory whilst
2917 processing DTLS handshake messages. This can be exploited through a
2918 Denial of Service attack.
2919 Thanks to Adam Langley for discovering and researching this issue.
2923 *) An attacker can force an error condition which causes openssl to crash
2924 whilst processing DTLS packets due to memory being freed twice. This
2925 can be exploited through a Denial of Service attack.
2926 Thanks to Adam Langley and Wan-Teh Chang for discovering and researching
2931 *) If a multithreaded client connects to a malicious server using a resumed
2932 session and the server sends an ec point format extension it could write
2933 up to 255 bytes to freed memory.
2935 Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this
2940 *) A malicious server can crash an OpenSSL client with a null pointer
2941 dereference (read) by specifying an SRP ciphersuite even though it was not
2942 properly negotiated with the client. This can be exploited through a
2943 Denial of Service attack.
2945 Thanks to Joonas Kuorilehto and Riku Hietamäki (Codenomicon) for
2946 discovering and researching this issue.
2950 *) A flaw in OBJ_obj2txt may cause pretty printing functions such as
2951 X509_name_oneline, X509_name_print_ex et al. to leak some information
2952 from the stack. Applications may be affected if they echo pretty printing
2953 output to the attacker.
2955 Thanks to Ivan Fratric (Google) for discovering this issue.
2957 [Emilia Käsper, and Steve Henson]
2959 *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.)
2960 for corner cases. (Certain input points at infinity could lead to
2961 bogus results, with non-infinity inputs mapped to infinity too.)
2964 Changes between 1.0.1g and 1.0.1h [5 Jun 2014]
2966 *) Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted
2967 handshake can force the use of weak keying material in OpenSSL
2968 SSL/TLS clients and servers.
2970 Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and
2971 researching this issue. (CVE-2014-0224)
2972 [KIKUCHI Masashi, Steve Henson]
2974 *) Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an
2975 OpenSSL DTLS client the code can be made to recurse eventually crashing
2978 Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue.
2980 [Imre Rad, Steve Henson]
2982 *) Fix DTLS invalid fragment vulnerability. A buffer overrun attack can
2983 be triggered by sending invalid DTLS fragments to an OpenSSL DTLS
2984 client or server. This is potentially exploitable to run arbitrary
2985 code on a vulnerable client or server.
2987 Thanks to Jüri Aedla for reporting this issue. (CVE-2014-0195)
2988 [Jüri Aedla, Steve Henson]
2990 *) Fix bug in TLS code where clients enable anonymous ECDH ciphersuites
2991 are subject to a denial of service attack.
2993 Thanks to Felix Gröbert and Ivan Fratric at Google for discovering
2994 this issue. (CVE-2014-3470)
2995 [Felix Gröbert, Ivan Fratric, Steve Henson]
2997 *) Harmonize version and its documentation. -f flag is used to display
2999 [mancha <mancha1@zoho.com>]
3001 *) Fix eckey_priv_encode so it immediately returns an error upon a failure
3002 in i2d_ECPrivateKey.
3003 [mancha <mancha1@zoho.com>]
3005 *) Fix some double frees. These are not thought to be exploitable.
3006 [mancha <mancha1@zoho.com>]
3008 Changes between 1.0.1f and 1.0.1g [7 Apr 2014]
3010 *) A missing bounds check in the handling of the TLS heartbeat extension
3011 can be used to reveal up to 64k of memory to a connected client or
3014 Thanks for Neel Mehta of Google Security for discovering this bug and to
3015 Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for
3016 preparing the fix (CVE-2014-0160)
3017 [Adam Langley, Bodo Moeller]
3019 *) Fix for the attack described in the paper "Recovering OpenSSL
3020 ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack"
3021 by Yuval Yarom and Naomi Benger. Details can be obtained from:
3022 http://eprint.iacr.org/2014/140
3024 Thanks to Yuval Yarom and Naomi Benger for discovering this
3025 flaw and to Yuval Yarom for supplying a fix (CVE-2014-0076)
3026 [Yuval Yarom and Naomi Benger]
3028 *) TLS pad extension: draft-agl-tls-padding-03
3030 Workaround for the "TLS hang bug" (see FAQ and PR#2771): if the
3031 TLS client Hello record length value would otherwise be > 255 and
3032 less that 512 pad with a dummy extension containing zeroes so it
3033 is at least 512 bytes long.
3035 [Adam Langley, Steve Henson]
3037 Changes between 1.0.1e and 1.0.1f [6 Jan 2014]
3039 *) Fix for TLS record tampering bug. A carefully crafted invalid
3040 handshake could crash OpenSSL with a NULL pointer exception.
3041 Thanks to Anton Johansson for reporting this issues.
3044 *) Keep original DTLS digest and encryption contexts in retransmission
3045 structures so we can use the previous session parameters if they need
3046 to be resent. (CVE-2013-6450)
3049 *) Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which
3050 avoids preferring ECDHE-ECDSA ciphers when the client appears to be
3051 Safari on OS X. Safari on OS X 10.8..10.8.3 advertises support for
3052 several ECDHE-ECDSA ciphers, but fails to negotiate them. The bug
3053 is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing
3054 10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer.
3055 [Rob Stradling, Adam Langley]
3057 Changes between 1.0.1d and 1.0.1e [11 Feb 2013]
3059 *) Correct fix for CVE-2013-0169. The original didn't work on AES-NI
3060 supporting platforms or when small records were transferred.
3061 [Andy Polyakov, Steve Henson]
3063 Changes between 1.0.1c and 1.0.1d [5 Feb 2013]
3065 *) Make the decoding of SSLv3, TLS and DTLS CBC records constant time.
3067 This addresses the flaw in CBC record processing discovered by
3068 Nadhem Alfardan and Kenny Paterson. Details of this attack can be found
3069 at: http://www.isg.rhul.ac.uk/tls/
3071 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
3072 Security Group at Royal Holloway, University of London
3073 (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and
3074 Emilia Käsper for the initial patch.
3076 [Emilia Käsper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson]
3078 *) Fix flaw in AESNI handling of TLS 1.2 and 1.1 records for CBC mode
3079 ciphersuites which can be exploited in a denial of service attack.
3080 Thanks go to and to Adam Langley <agl@chromium.org> for discovering
3081 and detecting this bug and to Wolfgang Ettlinger
3082 <wolfgang.ettlinger@gmail.com> for independently discovering this issue.
3086 *) Return an error when checking OCSP signatures when key is NULL.
3087 This fixes a DoS attack. (CVE-2013-0166)
3090 *) Make openssl verify return errors.
3091 [Chris Palmer <palmer@google.com> and Ben Laurie]
3093 *) Call OCSP Stapling callback after ciphersuite has been chosen, so
3094 the right response is stapled. Also change SSL_get_certificate()
3095 so it returns the certificate actually sent.
3096 See http://rt.openssl.org/Ticket/Display.html?id=2836.
3097 [Rob Stradling <rob.stradling@comodo.com>]
3099 *) Fix possible deadlock when decoding public keys.
3102 *) Don't use TLS 1.0 record version number in initial client hello
3106 Changes between 1.0.1b and 1.0.1c [10 May 2012]
3108 *) Sanity check record length before skipping explicit IV in TLS
3109 1.2, 1.1 and DTLS to fix DoS attack.
3111 Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic
3112 fuzzing as a service testing platform.
3116 *) Initialise tkeylen properly when encrypting CMS messages.
3117 Thanks to Solar Designer of Openwall for reporting this issue.
3120 *) In FIPS mode don't try to use composite ciphers as they are not
3124 Changes between 1.0.1a and 1.0.1b [26 Apr 2012]
3126 *) OpenSSL 1.0.0 sets SSL_OP_ALL to 0x80000FFFL and OpenSSL 1.0.1 and
3127 1.0.1a set SSL_OP_NO_TLSv1_1 to 0x00000400L which would unfortunately
3128 mean any application compiled against OpenSSL 1.0.0 headers setting
3129 SSL_OP_ALL would also set SSL_OP_NO_TLSv1_1, unintentionally disabling
3130 TLS 1.1 also. Fix this by changing the value of SSL_OP_NO_TLSv1_1 to
3131 0x10000000L Any application which was previously compiled against
3132 OpenSSL 1.0.1 or 1.0.1a headers and which cares about SSL_OP_NO_TLSv1_1
3133 will need to be recompiled as a result. Letting be results in
3134 inability to disable specifically TLS 1.1 and in client context,
3135 in unlike event, limit maximum offered version to TLS 1.0 [see below].
3138 *) In order to ensure interoperability SSL_OP_NO_protocolX does not
3139 disable just protocol X, but all protocols above X *if* there are
3140 protocols *below* X still enabled. In more practical terms it means
3141 that if application wants to disable TLS1.0 in favor of TLS1.1 and
3142 above, it's not sufficient to pass SSL_OP_NO_TLSv1, one has to pass
3143 SSL_OP_NO_TLSv1|SSL_OP_NO_SSLv3|SSL_OP_NO_SSLv2. This applies to
3147 Changes between 1.0.1 and 1.0.1a [19 Apr 2012]
3149 *) Check for potentially exploitable overflows in asn1_d2i_read_bio
3150 BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer
3151 in CRYPTO_realloc_clean.
3153 Thanks to Tavis Ormandy, Google Security Team, for discovering this
3154 issue and to Adam Langley <agl@chromium.org> for fixing it.
3156 [Adam Langley (Google), Tavis Ormandy, Google Security Team]
3158 *) Don't allow TLS 1.2 SHA-256 ciphersuites in TLS 1.0, 1.1 connections.
3161 *) Workarounds for some broken servers that "hang" if a client hello
3162 record length exceeds 255 bytes.
3164 1. Do not use record version number > TLS 1.0 in initial client
3165 hello: some (but not all) hanging servers will now work.
3166 2. If we set OPENSSL_MAX_TLS1_2_CIPHER_LENGTH this will truncate
3167 the number of ciphers sent in the client hello. This should be
3168 set to an even number, such as 50, for example by passing:
3169 -DOPENSSL_MAX_TLS1_2_CIPHER_LENGTH=50 to config or Configure.
3170 Most broken servers should now work.
3171 3. If all else fails setting OPENSSL_NO_TLS1_2_CLIENT will disable
3172 TLS 1.2 client support entirely.
3175 *) Fix SEGV in Vector Permutation AES module observed in OpenSSH.
3178 Changes between 1.0.0h and 1.0.1 [14 Mar 2012]
3180 *) Add compatibility with old MDC2 signatures which use an ASN1 OCTET
3181 STRING form instead of a DigestInfo.
3184 *) The format used for MDC2 RSA signatures is inconsistent between EVP
3185 and the RSA_sign/RSA_verify functions. This was made more apparent when
3186 OpenSSL used RSA_sign/RSA_verify for some RSA signatures in particular
3187 those which went through EVP_PKEY_METHOD in 1.0.0 and later. Detect
3188 the correct format in RSA_verify so both forms transparently work.
3191 *) Some servers which support TLS 1.0 can choke if we initially indicate
3192 support for TLS 1.2 and later renegotiate using TLS 1.0 in the RSA
3193 encrypted premaster secret. As a workaround use the maximum permitted
3194 client version in client hello, this should keep such servers happy
3195 and still work with previous versions of OpenSSL.
3198 *) Add support for TLS/DTLS heartbeats.
3199 [Robin Seggelmann <seggelmann@fh-muenster.de>]
3201 *) Add support for SCTP.
3202 [Robin Seggelmann <seggelmann@fh-muenster.de>]
3204 *) Improved PRNG seeding for VOS.
3205 [Paul Green <Paul.Green@stratus.com>]
3207 *) Extensive assembler packs updates, most notably:
3209 - x86[_64]: AES-NI, PCLMULQDQ, RDRAND support;
3210 - x86[_64]: SSSE3 support (SHA1, vector-permutation AES);
3211 - x86_64: bit-sliced AES implementation;
3212 - ARM: NEON support, contemporary platforms optimizations;
3213 - s390x: z196 support;
3214 - *: GHASH and GF(2^m) multiplication implementations;
3218 *) Make TLS-SRP code conformant with RFC 5054 API cleanup
3219 (removal of unnecessary code)
3220 [Peter Sylvester <peter.sylvester@edelweb.fr>]
3222 *) Add TLS key material exporter from RFC 5705.
3225 *) Add DTLS-SRTP negotiation from RFC 5764.
3228 *) Add Next Protocol Negotiation,
3229 http://tools.ietf.org/html/draft-agl-tls-nextprotoneg-00. Can be
3230 disabled with a no-npn flag to config or Configure. Code donated
3232 [Adam Langley <agl@google.com> and Ben Laurie]
3234 *) Add optional 64-bit optimized implementations of elliptic curves NIST-P224,
3235 NIST-P256, NIST-P521, with constant-time single point multiplication on
3236 typical inputs. Compiler support for the nonstandard type __uint128_t is
3237 required to use this (present in gcc 4.4 and later, for 64-bit builds).
3238 Code made available under Apache License version 2.0.
3240 Specify "enable-ec_nistp_64_gcc_128" on the Configure (or config) command
3241 line to include this in your build of OpenSSL, and run "make depend" (or
3242 "make update"). This enables the following EC_METHODs:
3244 EC_GFp_nistp224_method()
3245 EC_GFp_nistp256_method()
3246 EC_GFp_nistp521_method()
3248 EC_GROUP_new_by_curve_name() will automatically use these (while
3249 EC_GROUP_new_curve_GFp() currently prefers the more flexible
3251 [Emilia Käsper, Adam Langley, Bodo Moeller (Google)]
3253 *) Use type ossl_ssize_t instad of ssize_t which isn't available on
3254 all platforms. Move ssize_t definition from e_os.h to the public
3255 header file e_os2.h as it now appears in public header file cms.h
3258 *) New -sigopt option to the ca, req and x509 utilities. Additional
3259 signature parameters can be passed using this option and in
3263 *) Add RSA PSS signing function. This will generate and set the
3264 appropriate AlgorithmIdentifiers for PSS based on those in the
3265 corresponding EVP_MD_CTX structure. No application support yet.
3268 *) Support for companion algorithm specific ASN1 signing routines.
3269 New function ASN1_item_sign_ctx() signs a pre-initialised
3270 EVP_MD_CTX structure and sets AlgorithmIdentifiers based on
3271 the appropriate parameters.
3274 *) Add new algorithm specific ASN1 verification initialisation function
3275 to EVP_PKEY_ASN1_METHOD: this is not in EVP_PKEY_METHOD since the ASN1
3276 handling will be the same no matter what EVP_PKEY_METHOD is used.
3277 Add a PSS handler to support verification of PSS signatures: checked
3278 against a number of sample certificates.
3281 *) Add signature printing for PSS. Add PSS OIDs.
3282 [Steve Henson, Martin Kaiser <lists@kaiser.cx>]
3284 *) Add algorithm specific signature printing. An individual ASN1 method
3285 can now print out signatures instead of the standard hex dump.
3287 More complex signatures (e.g. PSS) can print out more meaningful
3288 information. Include DSA version that prints out the signature
3292 *) Password based recipient info support for CMS library: implementing
3296 *) Split password based encryption into PBES2 and PBKDF2 functions. This
3297 neatly separates the code into cipher and PBE sections and is required
3298 for some algorithms that split PBES2 into separate pieces (such as
3299 password based CMS).
3302 *) Session-handling fixes:
3303 - Fix handling of connections that are resuming with a session ID,
3304 but also support Session Tickets.
3305 - Fix a bug that suppressed issuing of a new ticket if the client
3306 presented a ticket with an expired session.
3307 - Try to set the ticket lifetime hint to something reasonable.
3308 - Make tickets shorter by excluding irrelevant information.
3309 - On the client side, don't ignore renewed tickets.
3310 [Adam Langley, Bodo Moeller (Google)]
3312 *) Fix PSK session representation.
3315 *) Add RC4-MD5 and AESNI-SHA1 "stitched" implementations.
3317 This work was sponsored by Intel.
3320 *) Add GCM support to TLS library. Some custom code is needed to split
3321 the IV between the fixed (from PRF) and explicit (from TLS record)
3322 portions. This adds all GCM ciphersuites supported by RFC5288 and
3323 RFC5289. Generalise some AES* cipherstrings to include GCM and
3324 add a special AESGCM string for GCM only.
3327 *) Expand range of ctrls for AES GCM. Permit setting invocation
3328 field on decrypt and retrieval of invocation field only on encrypt.
3331 *) Add HMAC ECC ciphersuites from RFC5289. Include SHA384 PRF support.
3332 As required by RFC5289 these ciphersuites cannot be used if for
3333 versions of TLS earlier than 1.2.
3336 *) For FIPS capable OpenSSL interpret a NULL default public key method
3337 as unset and return the appropriate default but do *not* set the default.
3338 This means we can return the appropriate method in applications that
3339 switch between FIPS and non-FIPS modes.
3342 *) Redirect HMAC and CMAC operations to FIPS module in FIPS mode. If an
3343 ENGINE is used then we cannot handle that in the FIPS module so we
3344 keep original code iff non-FIPS operations are allowed.
3347 *) Add -attime option to openssl utilities.
3348 [Peter Eckersley <pde@eff.org>, Ben Laurie and Steve Henson]
3350 *) Redirect DSA and DH operations to FIPS module in FIPS mode.
3353 *) Redirect ECDSA and ECDH operations to FIPS module in FIPS mode. Also use
3354 FIPS EC methods unconditionally for now.
3357 *) New build option no-ec2m to disable characteristic 2 code.
3360 *) Backport libcrypto audit of return value checking from 1.1.0-dev; not
3361 all cases can be covered as some introduce binary incompatibilities.
3364 *) Redirect RSA operations to FIPS module including keygen,
3365 encrypt, decrypt, sign and verify. Block use of non FIPS RSA methods.
3368 *) Add similar low level API blocking to ciphers.
3371 *) Low level digest APIs are not approved in FIPS mode: any attempt
3372 to use these will cause a fatal error. Applications that *really* want
3373 to use them can use the private_* version instead.
3376 *) Redirect cipher operations to FIPS module for FIPS builds.
3379 *) Redirect digest operations to FIPS module for FIPS builds.
3382 *) Update build system to add "fips" flag which will link in fipscanister.o
3383 for static and shared library builds embedding a signature if needed.
3386 *) Output TLS supported curves in preference order instead of numerical
3387 order. This is currently hardcoded for the highest order curves first.
3388 This should be configurable so applications can judge speed vs strength.
3391 *) Add TLS v1.2 server support for client authentication.
3394 *) Add support for FIPS mode in ssl library: disable SSLv3, non-FIPS ciphers
3398 *) Functions FIPS_mode_set() and FIPS_mode() which call the underlying
3399 FIPS modules versions.
3402 *) Add TLS v1.2 client side support for client authentication. Keep cache
3403 of handshake records longer as we don't know the hash algorithm to use
3404 until after the certificate request message is received.
3407 *) Initial TLS v1.2 client support. Add a default signature algorithms
3408 extension including all the algorithms we support. Parse new signature
3409 format in client key exchange. Relax some ECC signing restrictions for
3410 TLS v1.2 as indicated in RFC5246.
3413 *) Add server support for TLS v1.2 signature algorithms extension. Switch
3414 to new signature format when needed using client digest preference.
3415 All server ciphersuites should now work correctly in TLS v1.2. No client
3416 support yet and no support for client certificates.
3419 *) Initial TLS v1.2 support. Add new SHA256 digest to ssl code, switch
3420 to SHA256 for PRF when using TLS v1.2 and later. Add new SHA256 based
3421 ciphersuites. At present only RSA key exchange ciphersuites work with
3422 TLS v1.2. Add new option for TLS v1.2 replacing the old and obsolete
3423 SSL_OP_PKCS1_CHECK flags with SSL_OP_NO_TLSv1_2. New TLSv1.2 methods
3424 and version checking.
3427 *) New option OPENSSL_NO_SSL_INTERN. If an application can be compiled
3428 with this defined it will not be affected by any changes to ssl internal
3429 structures. Add several utility functions to allow openssl application
3430 to work with OPENSSL_NO_SSL_INTERN defined.
3433 *) A long standing patch to add support for SRP from EdelWeb (Peter
3434 Sylvester and Christophe Renou) was integrated.
3435 [Christophe Renou <christophe.renou@edelweb.fr>, Peter Sylvester
3436 <peter.sylvester@edelweb.fr>, Tom Wu <tjw@cs.stanford.edu>, and
3439 *) Add functions to copy EVP_PKEY_METHOD and retrieve flags and id.
3442 *) Permit abbreviated handshakes when renegotiating using the function
3443 SSL_renegotiate_abbreviated().
3444 [Robin Seggelmann <seggelmann@fh-muenster.de>]
3446 *) Add call to ENGINE_register_all_complete() to
3447 ENGINE_load_builtin_engines(), so some implementations get used
3448 automatically instead of needing explicit application support.
3451 *) Add support for TLS key exporter as described in RFC5705.
3452 [Robin Seggelmann <seggelmann@fh-muenster.de>, Steve Henson]
3454 *) Initial TLSv1.1 support. Since TLSv1.1 is very similar to TLS v1.0 only
3455 a few changes are required:
3457 Add SSL_OP_NO_TLSv1_1 flag.
3458 Add TLSv1_1 methods.
3459 Update version checking logic to handle version 1.1.
3460 Add explicit IV handling (ported from DTLS code).
3461 Add command line options to s_client/s_server.
3464 Changes between 1.0.0g and 1.0.0h [12 Mar 2012]
3466 *) Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness
3467 in CMS and PKCS7 code. When RSA decryption fails use a random key for
3468 content decryption and always return the same error. Note: this attack
3469 needs on average 2^20 messages so it only affects automated senders. The
3470 old behaviour can be re-enabled in the CMS code by setting the
3471 CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where
3472 an MMA defence is not necessary.
3473 Thanks to Ivan Nestlerode <inestlerode@us.ibm.com> for discovering
3474 this issue. (CVE-2012-0884)
3477 *) Fix CVE-2011-4619: make sure we really are receiving a
3478 client hello before rejecting multiple SGC restarts. Thanks to
3479 Ivan Nestlerode <inestlerode@us.ibm.com> for discovering this bug.
3482 Changes between 1.0.0f and 1.0.0g [18 Jan 2012]
3484 *) Fix for DTLS DoS issue introduced by fix for CVE-2011-4109.
3485 Thanks to Antonio Martin, Enterprise Secure Access Research and
3486 Development, Cisco Systems, Inc. for discovering this bug and
3487 preparing a fix. (CVE-2012-0050)
3490 Changes between 1.0.0e and 1.0.0f [4 Jan 2012]
3492 *) Nadhem Alfardan and Kenny Paterson have discovered an extension
3493 of the Vaudenay padding oracle attack on CBC mode encryption
3494 which enables an efficient plaintext recovery attack against
3495 the OpenSSL implementation of DTLS. Their attack exploits timing
3496 differences arising during decryption processing. A research
3497 paper describing this attack can be found at:
3498 http://www.isg.rhul.ac.uk/~kp/dtls.pdf
3499 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
3500 Security Group at Royal Holloway, University of London
3501 (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann
3502 <seggelmann@fh-muenster.de> and Michael Tuexen <tuexen@fh-muenster.de>
3503 for preparing the fix. (CVE-2011-4108)
3504 [Robin Seggelmann, Michael Tuexen]
3506 *) Clear bytes used for block padding of SSL 3.0 records.
3508 [Adam Langley (Google)]
3510 *) Only allow one SGC handshake restart for SSL/TLS. Thanks to George
3511 Kadianakis <desnacked@gmail.com> for discovering this issue and
3512 Adam Langley for preparing the fix. (CVE-2011-4619)
3513 [Adam Langley (Google)]
3515 *) Check parameters are not NULL in GOST ENGINE. (CVE-2012-0027)
3516 [Andrey Kulikov <amdeich@gmail.com>]
3518 *) Prevent malformed RFC3779 data triggering an assertion failure.
3519 Thanks to Andrew Chi, BBN Technologies, for discovering the flaw
3520 and Rob Austein <sra@hactrn.net> for fixing it. (CVE-2011-4577)
3521 [Rob Austein <sra@hactrn.net>]
3523 *) Improved PRNG seeding for VOS.
3524 [Paul Green <Paul.Green@stratus.com>]
3526 *) Fix ssl_ciph.c set-up race.
3527 [Adam Langley (Google)]
3529 *) Fix spurious failures in ecdsatest.c.
3530 [Emilia Käsper (Google)]
3532 *) Fix the BIO_f_buffer() implementation (which was mixing different
3533 interpretations of the '..._len' fields).
3534 [Adam Langley (Google)]
3536 *) Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than
3537 BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent
3538 threads won't reuse the same blinding coefficients.
3540 This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING
3541 lock to call BN_BLINDING_invert_ex, and avoids one use of
3542 BN_BLINDING_update for each BN_BLINDING structure (previously,
3543 the last update always remained unused).
3544 [Emilia Käsper (Google)]
3546 *) In ssl3_clear, preserve s3->init_extra along with s3->rbuf.
3547 [Bob Buckholz (Google)]
3549 Changes between 1.0.0d and 1.0.0e [6 Sep 2011]
3551 *) Fix bug where CRLs with nextUpdate in the past are sometimes accepted
3552 by initialising X509_STORE_CTX properly. (CVE-2011-3207)
3553 [Kaspar Brand <ossl@velox.ch>]
3555 *) Fix SSL memory handling for (EC)DH ciphersuites, in particular
3556 for multi-threaded use of ECDH. (CVE-2011-3210)
3557 [Adam Langley (Google)]
3559 *) Fix x509_name_ex_d2i memory leak on bad inputs.
3562 *) Remove hard coded ecdsaWithSHA1 signature tests in ssl code and check
3563 signature public key algorithm by using OID xref utilities instead.
3564 Before this you could only use some ECC ciphersuites with SHA1 only.
3567 *) Add protection against ECDSA timing attacks as mentioned in the paper
3568 by Billy Bob Brumley and Nicola Tuveri, see:
3570 http://eprint.iacr.org/2011/232.pdf
3572 [Billy Bob Brumley and Nicola Tuveri]
3574 Changes between 1.0.0c and 1.0.0d [8 Feb 2011]
3576 *) Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014
3577 [Neel Mehta, Adam Langley, Bodo Moeller (Google)]
3579 *) Fix bug in string printing code: if *any* escaping is enabled we must
3580 escape the escape character (backslash) or the resulting string is
3584 Changes between 1.0.0b and 1.0.0c [2 Dec 2010]
3586 *) Disable code workaround for ancient and obsolete Netscape browsers
3587 and servers: an attacker can use it in a ciphersuite downgrade attack.
3588 Thanks to Martin Rex for discovering this bug. CVE-2010-4180
3591 *) Fixed J-PAKE implementation error, originally discovered by
3592 Sebastien Martini, further info and confirmation from Stefan
3593 Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252
3596 Changes between 1.0.0a and 1.0.0b [16 Nov 2010]
3598 *) Fix extension code to avoid race conditions which can result in a buffer
3599 overrun vulnerability: resumed sessions must not be modified as they can
3600 be shared by multiple threads. CVE-2010-3864
3603 *) Fix WIN32 build system to correctly link an ENGINE directory into
3607 Changes between 1.0.0 and 1.0.0a [01 Jun 2010]
3609 *) Check return value of int_rsa_verify in pkey_rsa_verifyrecover
3611 [Steve Henson, Peter-Michael Hager <hager@dortmund.net>]
3613 Changes between 0.9.8n and 1.0.0 [29 Mar 2010]
3615 *) Add "missing" function EVP_CIPHER_CTX_copy(). This copies a cipher
3616 context. The operation can be customised via the ctrl mechanism in
3617 case ENGINEs want to include additional functionality.
3620 *) Tolerate yet another broken PKCS#8 key format: private key value negative.
3623 *) Add new -subject_hash_old and -issuer_hash_old options to x509 utility to
3624 output hashes compatible with older versions of OpenSSL.
3625 [Willy Weisz <weisz@vcpc.univie.ac.at>]
3627 *) Fix compression algorithm handling: if resuming a session use the
3628 compression algorithm of the resumed session instead of determining
3629 it from client hello again. Don't allow server to change algorithm.
3632 *) Add load_crls() function to apps tidying load_certs() too. Add option
3633 to verify utility to allow additional CRLs to be included.
3636 *) Update OCSP request code to permit adding custom headers to the request:
3637 some responders need this.
3640 *) The function EVP_PKEY_sign() returns <=0 on error: check return code
3642 [Julia Lawall <julia@diku.dk>]
3644 *) Update verify callback code in apps/s_cb.c and apps/verify.c, it
3645 needlessly dereferenced structures, used obsolete functions and
3646 didn't handle all updated verify codes correctly.
3649 *) Disable MD2 in the default configuration.
3652 *) In BIO_pop() and BIO_push() use the ctrl argument (which was NULL) to
3653 indicate the initial BIO being pushed or popped. This makes it possible
3654 to determine whether the BIO is the one explicitly called or as a result
3655 of the ctrl being passed down the chain. Fix BIO_pop() and SSL BIOs so
3656 it handles reference counts correctly and doesn't zero out the I/O bio
3657 when it is not being explicitly popped. WARNING: applications which
3658 included workarounds for the old buggy behaviour will need to be modified
3659 or they could free up already freed BIOs.
3662 *) Extend the uni2asc/asc2uni => OPENSSL_uni2asc/OPENSSL_asc2uni
3663 renaming to all platforms (within the 0.9.8 branch, this was
3664 done conditionally on Netware platforms to avoid a name clash).
3665 [Guenter <lists@gknw.net>]
3667 *) Add ECDHE and PSK support to DTLS.
3668 [Michael Tuexen <tuexen@fh-muenster.de>]
3670 *) Add CHECKED_STACK_OF macro to safestack.h, otherwise safestack can't
3674 *) Add "missing" function EVP_MD_flags() (without this the only way to
3675 retrieve a digest flags is by accessing the structure directly. Update
3676 EVP_MD_do_all*() and EVP_CIPHER_do_all*() to include the name a digest
3677 or cipher is registered as in the "from" argument. Print out all
3678 registered digests in the dgst usage message instead of manually
3679 attempting to work them out.
3682 *) If no SSLv2 ciphers are used don't use an SSLv2 compatible client hello:
3683 this allows the use of compression and extensions. Change default cipher
3684 string to remove SSLv2 ciphersuites. This effectively avoids ancient SSLv2
3685 by default unless an application cipher string requests it.
3688 *) Alter match criteria in PKCS12_parse(). It used to try to use local
3689 key ids to find matching certificates and keys but some PKCS#12 files
3690 don't follow the (somewhat unwritten) rules and this strategy fails.
3691 Now just gather all certificates together and the first private key
3692 then look for the first certificate that matches the key.
3695 *) Support use of registered digest and cipher names for dgst and cipher
3696 commands instead of having to add each one as a special case. So now
3703 openssl dgst -sha256 foo
3705 and this works for ENGINE based algorithms too.
3709 *) Update Gost ENGINE to support parameter files.
3710 [Victor B. Wagner <vitus@cryptocom.ru>]
3712 *) Support GeneralizedTime in ca utility.
3713 [Oliver Martin <oliver@volatilevoid.net>, Steve Henson]
3715 *) Enhance the hash format used for certificate directory links. The new
3716 form uses the canonical encoding (meaning equivalent names will work
3717 even if they aren't identical) and uses SHA1 instead of MD5. This form
3718 is incompatible with the older format and as a result c_rehash should
3719 be used to rebuild symbolic links.
3722 *) Make PKCS#8 the default write format for private keys, replacing the
3723 traditional format. This form is standardised, more secure and doesn't
3724 include an implicit MD5 dependency.
3727 *) Add a $gcc_devteam_warn option to Configure. The idea is that any code
3728 committed to OpenSSL should pass this lot as a minimum.
3731 *) Add session ticket override functionality for use by EAP-FAST.
3732 [Jouni Malinen <j@w1.fi>]
3734 *) Modify HMAC functions to return a value. Since these can be implemented
3735 in an ENGINE errors can occur.
3738 *) Type-checked OBJ_bsearch_ex.
3741 *) Type-checked OBJ_bsearch. Also some constification necessitated
3742 by type-checking. Still to come: TXT_DB, bsearch(?),
3743 OBJ_bsearch_ex, qsort, CRYPTO_EX_DATA, ASN1_VALUE, ASN1_STRING,
3747 *) New function OPENSSL_gmtime_adj() to add a specific number of days and
3748 seconds to a tm structure directly, instead of going through OS
3749 specific date routines. This avoids any issues with OS routines such
3750 as the year 2038 bug. New *_adj() functions for ASN1 time structures
3751 and X509_time_adj_ex() to cover the extended range. The existing
3752 X509_time_adj() is still usable and will no longer have any date issues.
3755 *) Delta CRL support. New use deltas option which will attempt to locate
3756 and search any appropriate delta CRLs available.
3758 This work was sponsored by Google.
3761 *) Support for CRLs partitioned by reason code. Reorganise CRL processing
3762 code and add additional score elements. Validate alternate CRL paths
3763 as part of the CRL checking and indicate a new error "CRL path validation
3764 error" in this case. Applications wanting additional details can use
3765 the verify callback and check the new "parent" field. If this is not
3766 NULL CRL path validation is taking place. Existing applications won't
3767 see this because it requires extended CRL support which is off by
3770 This work was sponsored by Google.
3773 *) Support for freshest CRL extension.
3775 This work was sponsored by Google.
3778 *) Initial indirect CRL support. Currently only supported in the CRLs
3779 passed directly and not via lookup. Process certificate issuer
3780 CRL entry extension and lookup CRL entries by bother issuer name
3781 and serial number. Check and process CRL issuer entry in IDP extension.
3783 This work was sponsored by Google.
3786 *) Add support for distinct certificate and CRL paths. The CRL issuer
3787 certificate is validated separately in this case. Only enabled if
3788 an extended CRL support flag is set: this flag will enable additional
3789 CRL functionality in future.
3791 This work was sponsored by Google.
3794 *) Add support for policy mappings extension.
3796 This work was sponsored by Google.
3799 *) Fixes to pathlength constraint, self issued certificate handling,
3800 policy processing to align with RFC3280 and PKITS tests.
3802 This work was sponsored by Google.
3805 *) Support for name constraints certificate extension. DN, email, DNS
3806 and URI types are currently supported.
3808 This work was sponsored by Google.
3811 *) To cater for systems that provide a pointer-based thread ID rather
3812 than numeric, deprecate the current numeric thread ID mechanism and
3813 replace it with a structure and associated callback type. This
3814 mechanism allows a numeric "hash" to be extracted from a thread ID in
3815 either case, and on platforms where pointers are larger than 'long',
3816 mixing is done to help ensure the numeric 'hash' is usable even if it
3817 can't be guaranteed unique. The default mechanism is to use "&errno"
3818 as a pointer-based thread ID to distinguish between threads.
3820 Applications that want to provide their own thread IDs should now use
3821 CRYPTO_THREADID_set_callback() to register a callback that will call
3822 either CRYPTO_THREADID_set_numeric() or CRYPTO_THREADID_set_pointer().
3824 Note that ERR_remove_state() is now deprecated, because it is tied
3825 to the assumption that thread IDs are numeric. ERR_remove_state(0)
3826 to free the current thread's error state should be replaced by
3827 ERR_remove_thread_state(NULL).
3829 (This new approach replaces the functions CRYPTO_set_idptr_callback(),
3830 CRYPTO_get_idptr_callback(), and CRYPTO_thread_idptr() that existed in
3831 OpenSSL 0.9.9-dev between June 2006 and August 2008. Also, if an
3832 application was previously providing a numeric thread callback that
3833 was inappropriate for distinguishing threads, then uniqueness might
3834 have been obtained with &errno that happened immediately in the
3835 intermediate development versions of OpenSSL; this is no longer the
3836 case, the numeric thread callback will now override the automatic use
3838 [Geoff Thorpe, with help from Bodo Moeller]
3840 *) Initial support for different CRL issuing certificates. This covers a
3841 simple case where the self issued certificates in the chain exist and
3842 the real CRL issuer is higher in the existing chain.
3844 This work was sponsored by Google.
3847 *) Removed effectively defunct crypto/store from the build.
3850 *) Revamp of STACK to provide stronger type-checking. Still to come:
3851 TXT_DB, bsearch(?), OBJ_bsearch, qsort, CRYPTO_EX_DATA, ASN1_VALUE,
3852 ASN1_STRING, CONF_VALUE.
3855 *) Add a new SSL_MODE_RELEASE_BUFFERS mode flag to release unused buffer
3856 RAM on SSL connections. This option can save about 34k per idle SSL.
3859 *) Revamp of LHASH to provide stronger type-checking. Still to come:
3860 STACK, TXT_DB, bsearch, qsort.
3863 *) Initial support for Cryptographic Message Syntax (aka CMS) based
3864 on RFC3850, RFC3851 and RFC3852. New cms directory and cms utility,
3865 support for data, signedData, compressedData, digestedData and
3866 encryptedData, envelopedData types included. Scripts to check against
3867 RFC4134 examples draft and interop and consistency checks of many
3868 content types and variants.
3871 *) Add options to enc utility to support use of zlib compression BIO.
3874 *) Extend mk1mf to support importing of options and assembly language
3875 files from Configure script, currently only included in VC-WIN32.
3876 The assembly language rules can now optionally generate the source
3877 files from the associated perl scripts.
3880 *) Implement remaining functionality needed to support GOST ciphersuites.
3881 Interop testing has been performed using CryptoPro implementations.
3882 [Victor B. Wagner <vitus@cryptocom.ru>]
3884 *) s390x assembler pack.
3887 *) ARMv4 assembler pack. ARMv4 refers to v4 and later ISA, not CPU
3891 *) Implement Opaque PRF Input TLS extension as specified in
3892 draft-rescorla-tls-opaque-prf-input-00.txt. Since this is not an
3893 official specification yet and no extension type assignment by
3894 IANA exists, this extension (for now) will have to be explicitly
3895 enabled when building OpenSSL by providing the extension number
3896 to use. For example, specify an option
3898 -DTLSEXT_TYPE_opaque_prf_input=0x9527
3900 to the "config" or "Configure" script to enable the extension,
3901 assuming extension number 0x9527 (which is a completely arbitrary
3902 and unofficial assignment based on the MD5 hash of the Internet
3903 Draft). Note that by doing so, you potentially lose
3904 interoperability with other TLS implementations since these might
3905 be using the same extension number for other purposes.
3907 SSL_set_tlsext_opaque_prf_input(ssl, src, len) is used to set the
3908 opaque PRF input value to use in the handshake. This will create
3909 an internal copy of the length-'len' string at 'src', and will
3910 return non-zero for success.
3912 To get more control and flexibility, provide a callback function
3915 SSL_CTX_set_tlsext_opaque_prf_input_callback(ctx, cb)
3916 SSL_CTX_set_tlsext_opaque_prf_input_callback_arg(ctx, arg)
3920 int (*cb)(SSL *, void *peerinput, size_t len, void *arg);
3923 Callback function 'cb' will be called in handshakes, and is
3924 expected to use SSL_set_tlsext_opaque_prf_input() as appropriate.
3925 Argument 'arg' is for application purposes (the value as given to
3926 SSL_CTX_set_tlsext_opaque_prf_input_callback_arg() will directly
3927 be provided to the callback function). The callback function
3928 has to return non-zero to report success: usually 1 to use opaque
3929 PRF input just if possible, or 2 to enforce use of the opaque PRF
3930 input. In the latter case, the library will abort the handshake
3931 if opaque PRF input is not successfully negotiated.
3933 Arguments 'peerinput' and 'len' given to the callback function
3934 will always be NULL and 0 in the case of a client. A server will
3935 see the client's opaque PRF input through these variables if
3936 available (NULL and 0 otherwise). Note that if the server
3937 provides an opaque PRF input, the length must be the same as the
3938 length of the client's opaque PRF input.
3940 Note that the callback function will only be called when creating
3941 a new session (session resumption can resume whatever was
3942 previously negotiated), and will not be called in SSL 2.0
3943 handshakes; thus, SSL_CTX_set_options(ctx, SSL_OP_NO_SSLv2) or
3944 SSL_set_options(ssl, SSL_OP_NO_SSLv2) is especially recommended
3945 for applications that need to enforce opaque PRF input.
3949 *) Update ssl code to support digests other than SHA1+MD5 for handshake
3952 [Victor B. Wagner <vitus@cryptocom.ru>]
3954 *) Add RFC4507 support to OpenSSL. This includes the corrections in
3955 RFC4507bis. The encrypted ticket format is an encrypted encoded
3956 SSL_SESSION structure, that way new session features are automatically
3959 If a client application caches session in an SSL_SESSION structure
3960 support is transparent because tickets are now stored in the encoded
3963 The SSL_CTX structure automatically generates keys for ticket
3964 protection in servers so again support should be possible
3965 with no application modification.
3967 If a client or server wishes to disable RFC4507 support then the option
3968 SSL_OP_NO_TICKET can be set.
3970 Add a TLS extension debugging callback to allow the contents of any client
3971 or server extensions to be examined.
3973 This work was sponsored by Google.
3976 *) Final changes to avoid use of pointer pointer casts in OpenSSL.
3977 OpenSSL should now compile cleanly on gcc 4.2
3978 [Peter Hartley <pdh@utter.chaos.org.uk>, Steve Henson]
3980 *) Update SSL library to use new EVP_PKEY MAC API. Include generic MAC
3981 support including streaming MAC support: this is required for GOST
3982 ciphersuite support.
3983 [Victor B. Wagner <vitus@cryptocom.ru>, Steve Henson]
3985 *) Add option -stream to use PKCS#7 streaming in smime utility. New
3986 function i2d_PKCS7_bio_stream() and PEM_write_PKCS7_bio_stream()
3987 to output in BER and PEM format.
3990 *) Experimental support for use of HMAC via EVP_PKEY interface. This
3991 allows HMAC to be handled via the EVP_DigestSign*() interface. The
3992 EVP_PKEY "key" in this case is the HMAC key, potentially allowing
3993 ENGINE support for HMAC keys which are unextractable. New -mac and
3994 -macopt options to dgst utility.
3997 *) New option -sigopt to dgst utility. Update dgst to use
3998 EVP_Digest{Sign,Verify}*. These two changes make it possible to use
3999 alternative signing parameters such as X9.31 or PSS in the dgst
4003 *) Change ssl_cipher_apply_rule(), the internal function that does
4004 the work each time a ciphersuite string requests enabling
4005 ("foo+bar"), moving ("+foo+bar"), disabling ("-foo+bar", or
4006 removing ("!foo+bar") a class of ciphersuites: Now it maintains
4007 the order of disabled ciphersuites such that those ciphersuites
4008 that most recently went from enabled to disabled not only stay
4009 in order with respect to each other, but also have higher priority
4010 than other disabled ciphersuites the next time ciphersuites are
4013 This means that you can now say, e.g., "PSK:-PSK:HIGH" to enable
4014 the same ciphersuites as with "HIGH" alone, but in a specific
4015 order where the PSK ciphersuites come first (since they are the
4016 most recently disabled ciphersuites when "HIGH" is parsed).
4018 Also, change ssl_create_cipher_list() (using this new
4019 functionality) such that between otherwise identical
4020 ciphersuites, ephemeral ECDH is preferred over ephemeral DH in
4024 *) Change ssl_create_cipher_list() so that it automatically
4025 arranges the ciphersuites in reasonable order before starting
4026 to process the rule string. Thus, the definition for "DEFAULT"
4027 (SSL_DEFAULT_CIPHER_LIST) now is just "ALL:!aNULL:!eNULL", but
4028 remains equivalent to "AES:ALL:!aNULL:!eNULL:+aECDH:+kRSA:+RC4:@STRENGTH".
4029 This makes it much easier to arrive at a reasonable default order
4030 in applications for which anonymous ciphers are OK (meaning
4031 that you can't actually use DEFAULT).
4032 [Bodo Moeller; suggested by Victor Duchovni]
4034 *) Split the SSL/TLS algorithm mask (as used for ciphersuite string
4035 processing) into multiple integers instead of setting
4036 "SSL_MKEY_MASK" bits, "SSL_AUTH_MASK" bits, "SSL_ENC_MASK",
4037 "SSL_MAC_MASK", and "SSL_SSL_MASK" bits all in a single integer.
4038 (These masks as well as the individual bit definitions are hidden
4039 away into the non-exported interface ssl/ssl_locl.h, so this
4040 change to the definition of the SSL_CIPHER structure shouldn't
4041 affect applications.) This give us more bits for each of these
4042 categories, so there is no longer a need to coagulate AES128 and
4043 AES256 into a single algorithm bit, and to coagulate Camellia128
4044 and Camellia256 into a single algorithm bit, which has led to all
4047 Thus, among other things, the kludge introduced in 0.9.7m and
4048 0.9.8e for masking out AES256 independently of AES128 or masking
4049 out Camellia256 independently of AES256 is not needed here in 0.9.9.
4051 With the change, we also introduce new ciphersuite aliases that
4052 so far were missing: "AES128", "AES256", "CAMELLIA128", and
4056 *) Add support for dsa-with-SHA224 and dsa-with-SHA256.
4057 Use the leftmost N bytes of the signature input if the input is
4058 larger than the prime q (with N being the size in bytes of q).
4061 *) Very *very* experimental PKCS#7 streaming encoder support. Nothing uses
4062 it yet and it is largely untested.
4065 *) Add support for the ecdsa-with-SHA224/256/384/512 signature types.
4068 *) Initial incomplete changes to avoid need for function casts in OpenSSL
4069 some compilers (gcc 4.2 and later) reject their use. Safestack is
4070 reimplemented. Update ASN1 to avoid use of legacy functions.
4073 *) Win32/64 targets are linked with Winsock2.
4076 *) Add an X509_CRL_METHOD structure to allow CRL processing to be redirected
4077 to external functions. This can be used to increase CRL handling
4078 efficiency especially when CRLs are very large by (for example) storing
4079 the CRL revoked certificates in a database.
4082 *) Overhaul of by_dir code. Add support for dynamic loading of CRLs so
4083 new CRLs added to a directory can be used. New command line option
4084 -verify_return_error to s_client and s_server. This causes real errors
4085 to be returned by the verify callback instead of carrying on no matter
4086 what. This reflects the way a "real world" verify callback would behave.
4089 *) GOST engine, supporting several GOST algorithms and public key formats.
4090 Kindly donated by Cryptocom.
4093 *) Partial support for Issuing Distribution Point CRL extension. CRLs
4094 partitioned by DP are handled but no indirect CRL or reason partitioning
4095 (yet). Complete overhaul of CRL handling: now the most suitable CRL is
4096 selected via a scoring technique which handles IDP and AKID in CRLs.
4099 *) New X509_STORE_CTX callbacks lookup_crls() and lookup_certs() which
4100 will ultimately be used for all verify operations: this will remove the
4101 X509_STORE dependency on certificate verification and allow alternative
4102 lookup methods. X509_STORE based implementations of these two callbacks.
4105 *) Allow multiple CRLs to exist in an X509_STORE with matching issuer names.
4106 Modify get_crl() to find a valid (unexpired) CRL if possible.
4109 *) New function X509_CRL_match() to check if two CRLs are identical. Normally
4110 this would be called X509_CRL_cmp() but that name is already used by
4111 a function that just compares CRL issuer names. Cache several CRL
4112 extensions in X509_CRL structure and cache CRLDP in X509.
4115 *) Store a "canonical" representation of X509_NAME structure (ASN1 Name)
4116 this maps equivalent X509_NAME structures into a consistent structure.
4117 Name comparison can then be performed rapidly using memcmp().
4120 *) Non-blocking OCSP request processing. Add -timeout option to ocsp
4124 *) Allow digests to supply their own micalg string for S/MIME type using
4125 the ctrl EVP_MD_CTRL_MICALG.
4128 *) During PKCS7 signing pass the PKCS7 SignerInfo structure to the
4129 EVP_PKEY_METHOD before and after signing via the EVP_PKEY_CTRL_PKCS7_SIGN
4130 ctrl. It can then customise the structure before and/or after signing
4134 *) New function OBJ_add_sigid() to allow application defined signature OIDs
4135 to be added to OpenSSLs internal tables. New function OBJ_sigid_free()
4136 to free up any added signature OIDs.
4139 *) New functions EVP_CIPHER_do_all(), EVP_CIPHER_do_all_sorted(),
4140 EVP_MD_do_all() and EVP_MD_do_all_sorted() to enumerate internal
4141 digest and cipher tables. New options added to openssl utility:
4142 list-message-digest-algorithms and list-cipher-algorithms.
4145 *) Change the array representation of binary polynomials: the list
4146 of degrees of non-zero coefficients is now terminated with -1.
4147 Previously it was terminated with 0, which was also part of the
4148 value; thus, the array representation was not applicable to
4149 polynomials where t^0 has coefficient zero. This change makes
4150 the array representation useful in a more general context.
4153 *) Various modifications and fixes to SSL/TLS cipher string
4154 handling. For ECC, the code now distinguishes between fixed ECDH
4155 with RSA certificates on the one hand and with ECDSA certificates
4156 on the other hand, since these are separate ciphersuites. The
4157 unused code for Fortezza ciphersuites has been removed.
4159 For consistency with EDH, ephemeral ECDH is now called "EECDH"
4160 (not "ECDHE"). For consistency with the code for DH
4161 certificates, use of ECDH certificates is now considered ECDH
4162 authentication, not RSA or ECDSA authentication (the latter is
4163 merely the CA's signing algorithm and not actively used in the
4166 The temporary ciphersuite alias "ECCdraft" is no longer
4167 available, and ECC ciphersuites are no longer excluded from "ALL"
4168 and "DEFAULT". The following aliases now exist for RFC 4492
4169 ciphersuites, most of these by analogy with the DH case:
4171 kECDHr - ECDH cert, signed with RSA
4172 kECDHe - ECDH cert, signed with ECDSA
4173 kECDH - ECDH cert (signed with either RSA or ECDSA)
4174 kEECDH - ephemeral ECDH
4175 ECDH - ECDH cert or ephemeral ECDH
4181 AECDH - anonymous ECDH
4182 EECDH - non-anonymous ephemeral ECDH (equivalent to "kEECDH:-AECDH")
4186 *) Add additional S/MIME capabilities for AES and GOST ciphers if supported.
4187 Use correct micalg parameters depending on digest(s) in signed message.
4190 *) Add engine support for EVP_PKEY_ASN1_METHOD. Add functions to process
4191 an ENGINE asn1 method. Support ENGINE lookups in the ASN1 code.
4194 *) Initial engine support for EVP_PKEY_METHOD. New functions to permit
4195 an engine to register a method. Add ENGINE lookups for methods and
4196 functional reference processing.
4199 *) New functions EVP_Digest{Sign,Verify)*. These are enhanced versions of
4200 EVP_{Sign,Verify}* which allow an application to customise the signature
4204 *) New -resign option to smime utility. This adds one or more signers
4205 to an existing PKCS#7 signedData structure. Also -md option to use an
4206 alternative message digest algorithm for signing.
4209 *) Tidy up PKCS#7 routines and add new functions to make it easier to
4210 create PKCS7 structures containing multiple signers. Update smime
4211 application to support multiple signers.
4214 *) New -macalg option to pkcs12 utility to allow setting of an alternative
4218 *) Initial support for PKCS#5 v2.0 PRFs other than default SHA1 HMAC.
4219 Reorganize PBE internals to lookup from a static table using NIDs,
4220 add support for HMAC PBE OID translation. Add a EVP_CIPHER ctrl:
4221 EVP_CTRL_PBE_PRF_NID this allows a cipher to specify an alternative
4222 PRF which will be automatically used with PBES2.
4225 *) Replace the algorithm specific calls to generate keys in "req" with the
4229 *) Update PKCS#7 enveloped data routines to use new API. This is now
4230 supported by any public key method supporting the encrypt operation. A
4231 ctrl is added to allow the public key algorithm to examine or modify
4232 the PKCS#7 RecipientInfo structure if it needs to: for RSA this is
4236 *) Add a ctrl to asn1 method to allow a public key algorithm to express
4237 a default digest type to use. In most cases this will be SHA1 but some
4238 algorithms (such as GOST) need to specify an alternative digest. The
4239 return value indicates how strong the preference is 1 means optional and
4240 2 is mandatory (that is it is the only supported type). Modify
4241 ASN1_item_sign() to accept a NULL digest argument to indicate it should
4242 use the default md. Update openssl utilities to use the default digest
4243 type for signing if it is not explicitly indicated.
4246 *) Use OID cross reference table in ASN1_sign() and ASN1_verify(). New
4247 EVP_MD flag EVP_MD_FLAG_PKEY_METHOD_SIGNATURE. This uses the relevant
4248 signing method from the key type. This effectively removes the link
4249 between digests and public key types.
4252 *) Add an OID cross reference table and utility functions. Its purpose is to
4253 translate between signature OIDs such as SHA1WithrsaEncryption and SHA1,
4254 rsaEncryption. This will allow some of the algorithm specific hackery
4255 needed to use the correct OID to be removed.
4258 *) Remove algorithm specific dependencies when setting PKCS7_SIGNER_INFO
4259 structures for PKCS7_sign(). They are now set up by the relevant public
4263 *) Add provisional EC pkey method with support for ECDSA and ECDH.
4266 *) Add support for key derivation (agreement) in the API, DH method and
4270 *) Add DSA pkey method and DH pkey methods, extend DH ASN1 method to support
4271 public and private key formats. As a side effect these add additional
4272 command line functionality not previously available: DSA signatures can be
4273 generated and verified using pkeyutl and DH key support and generation in
4278 [Oliver Tappe <zooey@hirschkaefer.de>]
4280 *) New make target "install_html_docs" installs HTML renditions of the
4282 [Oliver Tappe <zooey@hirschkaefer.de&
projects / openssl.git / blob