4 This is a detailed breakdown of significant changes. For a high-level overview
5 of changes in each release, see [NEWS.md](./NEWS.md).
7 For a full list of changes, see the [git commit log][log] and pick the
8 appropriate release branch.
10 [log]: https://github.com/openssl/openssl/commits/
15 - [OpenSSL 3.3](#openssl-33)
16 - [OpenSSL 3.2](#openssl-32)
17 - [OpenSSL 3.1](#openssl-31)
18 - [OpenSSL 3.0](#openssl-30)
19 - [OpenSSL 1.1.1](#openssl-111)
20 - [OpenSSL 1.1.0](#openssl-110)
21 - [OpenSSL 1.0.2](#openssl-102)
22 - [OpenSSL 1.0.1](#openssl-101)
23 - [OpenSSL 1.0.0](#openssl-100)
24 - [OpenSSL 0.9.x](#openssl-09x)
29 ### Changes between 3.2 and 3.3 [xx XXX xxxx]
31 * The EVP_PKEY_fromdata function has been augmented to allow for the derivation
32 of CRT (Chinese Remainder Theorem) parameters when requested. See the
33 OSSL_PKEY_PARAM_RSA_DERIVE_FROM_PQ param in the EVP_PKEY-RSA documentation.
37 * The activate and soft_load configuration settings for providers in
38 openssl.cnf have been updated to require a value of [1|yes|true|on]
39 (in lower or UPPER case) to enable the setting. Conversely a value
40 of [0|no|false|off] will disable the setting. All other values, or the
41 omission of a value for these settings will result in an error.
45 * Added `-set_issuer` and `-set_subject` options to `openssl x509` to
46 override the Issuer and Subject when creating a certificate. The `-subj`
47 option now is an alias for `-set_subject`.
49 *Job Snijders, George Michaelson*
51 * OPENSSL_sk_push() and sk_<TYPE>_push() functions now return 0 instead of -1
52 if called with a NULL stack argument.
56 * In `openssl speed`, changed the default hash function used with `hmac` from
61 * Added several new features of CMPv3 defined in RFC 9480 and RFC 9483:
62 - `certProfile` request message header and respective `-profile` CLI option
63 - support for delayed delivery of all types of response messages
67 * The build of exporters (such as `.pc` files for pkg-config) cleaned up to
68 be less hard coded in the build file templates, and to allow easier
69 addition of more exporters. With that, an exporter for CMake is also
74 * The BLAKE2s hash algorithm matches BLAKE2b's support
75 for configurable output length.
79 * New option `SSL_OP_PREFER_NO_DHE_KEX`, which allows configuring a TLS1.3
80 server to prefer session resumption using PSK-only key exchange over PSK
81 with DHE, if both are available.
83 *Markus Minichmayr, Tapkey GmbH*
85 * New API `SSL_write_ex2`, which can be used to send an end-of-stream (FIN)
86 condition in an optimised way when using QUIC.
90 * New atexit configuration switch, which controls whether the OPENSSL_cleanup
91 is registered when libcrypto is unloaded. This is turned off on NonStop
92 configurations because of loader differences on that platform compared to
97 * Support for qlog for tracing QUIC connections has been added.
99 The qlog output from OpenSSL currently uses a pre-standard draft version of
100 qlog. The output from OpenSSL will change in incompatible ways in future
101 releases, and is not subject to any format stability or compatibility
102 guarantees at this time; therefore this functionality must currently be
103 enabled with the build-time option `enable-unstable-qlog`. See the
104 openssl-qlog(7) manpage for details.
108 * Added APIs to allow configuring the negotiated idle timeout for QUIC
109 connections, and to allow determining the number of additional streams
110 that can currently be created for a QUIC connection.
114 * Added APIs to allow disabling implicit QUIC event processing for
115 QUIC SSL objects, allowing applications to control when event handling
116 occurs. Refer to the SSL_get_value_uint(3) manpage for details.
123 ### Changes between 3.2.1 and 3.2.2 [xx XXX xxxx]
125 * Fixed bug where SSL_export_keying_material() could not be used with QUIC
126 connections. (#23560)
130 ### Changes between 3.2.0 and 3.2.1 [30 Jan 2024]
132 * A file in PKCS12 format can contain certificates and keys and may come from
133 an untrusted source. The PKCS12 specification allows certain fields to be
134 NULL, but OpenSSL did not correctly check for this case. A fix has been
135 applied to prevent a NULL pointer dereference that results in OpenSSL
136 crashing. If an application processes PKCS12 files from an untrusted source
137 using the OpenSSL APIs then that application will be vulnerable to this
138 issue prior to this fix.
140 OpenSSL APIs that were vulnerable to this are: PKCS12_parse(),
141 PKCS12_unpack_p7data(), PKCS12_unpack_p7encdata(), PKCS12_unpack_authsafes()
142 and PKCS12_newpass().
144 We have also fixed a similar issue in SMIME_write_PKCS7(). However since this
145 function is related to writing data we do not consider it security
152 * When function EVP_PKEY_public_check() is called on RSA public keys,
153 a computation is done to confirm that the RSA modulus, n, is composite.
154 For valid RSA keys, n is a product of two or more large primes and this
155 computation completes quickly. However, if n is an overly large prime,
156 then this computation would take a long time.
158 An application that calls EVP_PKEY_public_check() and supplies an RSA key
159 obtained from an untrusted source could be vulnerable to a Denial of Service
162 The function EVP_PKEY_public_check() is not called from other OpenSSL
163 functions however it is called from the OpenSSL pkey command line
164 application. For that reason that application is also vulnerable if used
165 with the "-pubin" and "-check" options on untrusted data.
167 To resolve this issue RSA keys larger than OPENSSL_RSA_MAX_MODULUS_BITS will
168 now fail the check immediately with an RSA_R_MODULUS_TOO_LARGE error reason.
174 * Restore the encoding of SM2 PrivateKeyInfo and SubjectPublicKeyInfo to
175 have the contained AlgorithmIdentifier.algorithm set to id-ecPublicKey
180 * The POLY1305 MAC (message authentication code) implementation in OpenSSL
181 for PowerPC CPUs saves the contents of vector registers in different
182 order than they are restored. Thus the contents of some of these vector
183 registers is corrupted when returning to the caller. The vulnerable code is
184 used only on newer PowerPC processors supporting the PowerISA 2.07
187 The consequences of this kind of internal application state corruption can
188 be various - from no consequences, if the calling application does not
189 depend on the contents of non-volatile XMM registers at all, to the worst
190 consequences, where the attacker could get complete control of the
191 application process. However unless the compiler uses the vector registers
192 for storing pointers, the most likely consequence, if any, would be an
193 incorrect result of some application dependent calculations or a crash
194 leading to a denial of service.
200 * Disable building QUIC server utility when OpenSSL is configured with
205 ### Changes between 3.1 and 3.2.0 [23 Nov 2023]
207 * Fix excessive time spent in DH check / generation with large Q parameter
210 Applications that use the functions DH_generate_key() to generate an
211 X9.42 DH key may experience long delays. Likewise, applications that use
212 DH_check_pub_key(), DH_check_pub_key_ex() or EVP_PKEY_public_check()
213 to check an X9.42 DH key or X9.42 DH parameters may experience long delays.
214 Where the key or parameters that are being checked have been obtained from
215 an untrusted source this may lead to a Denial of Service.
221 * The BLAKE2b hash algorithm supports a configurable output length
222 by setting the "size" parameter.
224 *Čestmír Kalina and Tomáš Mráz*
226 * Enable extra Arm64 optimization on Windows for GHASH, RAND and AES.
230 * Added a function to delete objects from store by URI - OSSL_STORE_delete()
231 and the corresponding provider-storemgmt API function
232 OSSL_FUNC_store_delete().
236 * Added OSSL_FUNC_store_open_ex() provider-storemgmt API function to pass
237 a passphrase callback when opening a store.
241 * Changed the default salt length used by PBES2 KDF's (PBKDF2 and scrypt)
242 from 8 bytes to 16 bytes.
243 The PKCS5 (RFC 8018) standard uses a 64 bit salt length for PBE, and
244 recommends a minimum of 64 bits for PBES2. For FIPS compliance PBKDF2
245 requires a salt length of 128 bits. This affects OpenSSL command line
246 applications such as "genrsa" and "pkcs8" and API's such as
247 PEM_write_bio_PrivateKey() that are reliant on the default value.
248 The additional commandline option 'saltlen' has been added to the
249 OpenSSL command line applications for "pkcs8" and "enc" to allow the
250 salt length to be set to a non default value.
254 * Changed the default value of the `ess_cert_id_alg` configuration
255 option which is used to calculate the TSA's public key certificate
256 identifier. The default algorithm is updated to be sha256 instead
259 *Małgorzata Olszówka*
261 * Added optimization for SM2 algorithm on aarch64. It uses a huge precomputed
262 table for point multiplication of the base point, which increases the size of
263 libcrypto from 4.4 MB to 4.9 MB. A new configure option `no-sm2-precomp` has
264 been added to disable the precomputed table.
268 * Added client side support for QUIC
270 *Hugo Landau, Matt Caswell, Paul Dale, Tomáš Mráz, Richard Levitte*
272 * Added multiple tutorials on the OpenSSL library and in particular
273 on writing various clients (using TLS and QUIC protocols) with libssl.
277 * Added secp384r1 implementation using Solinas' reduction to improve
278 speed of the NIST P-384 elliptic curve. To enable the implementation
279 the build option `enable-ec_nistp_64_gcc_128` must be used.
283 * Improved RFC7468 compliance of the asn1parse command.
285 *Matthias St. Pierre*
287 * Added SHA256/192 algorithm support.
291 * Added support for securely getting root CA certificate update in
296 * Improved contention on global write locks by using more read locks where
301 * Improved performance of OSSL_PARAM lookups in performance critical
306 * Added the SSL_get0_group_name() function to provide access to the
307 name of the group used for the TLS key exchange.
311 * Provide a new configure option `no-http` that can be used to disable the
312 HTTP support. Provide new configure options `no-apps` and `no-docs` to
313 disable building the openssl command line application and the documentation.
317 * Provide a new configure option `no-ecx` that can be used to disable the
318 X25519, X448, and EdDSA support.
322 * When multiple OSSL_KDF_PARAM_INFO parameters are passed to
323 the EVP_KDF_CTX_set_params() function they are now concatenated not just
324 for the HKDF algorithm but also for SSKDF and X9.63 KDF algorithms.
328 * Added OSSL_FUNC_keymgmt_im/export_types_ex() provider functions that get
329 the provider context as a parameter.
333 * TLS round-trip time calculation was added by a Brigham Young University
334 Capstone team partnering with Sandia National Laboratories. A new function
335 in ssl_lib titled SSL_get_handshake_rtt will calculate and retrieve this
340 * Added the "-quic" option to s_client to enable connectivity to QUIC servers.
341 QUIC requires the use of ALPN, so this must be specified via the "-alpn"
342 option. Use of the "advanced" s_client command command via the "-adv" option
347 * Added an "advanced" command mode to s_client. Use this with the "-adv"
348 option. The old "basic" command mode recognises certain letters that must
349 always appear at the start of a line and cannot be escaped. The advanced
350 command mode enables commands to be entered anywhere and there is an
351 escaping mechanism. After starting s_client with "-adv" type "{help}"
352 to show a list of available commands.
356 * Add Raw Public Key (RFC7250) support. Authentication is supported
357 by matching keys against either local policy (TLSA records synthesised
358 from the expected keys) or DANE (TLSA records obtained by the
359 application from DNS). TLSA records will also match the same key in
360 the server certificate, should RPK use not happen to be negotiated.
364 * Added support for modular exponentiation and CRT offloading for the
369 * Added further assembler code for the RISC-V architecture.
373 * Added EC_GROUP_to_params() which creates an OSSL_PARAM array
374 from a given EC_GROUP.
378 * Improved support for non-default library contexts and property queries
379 when parsing PKCS#12 files.
383 * Implemented support for all five instances of EdDSA from RFC8032:
384 Ed25519, Ed25519ctx, Ed25519ph, Ed448, and Ed448ph.
385 The streaming is not yet supported for the HashEdDSA variants
386 (Ed25519ph and Ed448ph).
390 * Added SM4 optimization for ARM processors using ASIMD and AES HW
395 * Implemented SM4-XTS support.
399 * Added platform-agnostic OSSL_sleep() function.
403 * Implemented deterministic ECDSA signatures (RFC6979) support.
407 * Implemented AES-GCM-SIV (RFC8452) support.
411 * Added support for pluggable (provider-based) TLS signature algorithms.
412 This enables TLS 1.3 authentication operations with algorithms embedded
413 in providers not included by default in OpenSSL. In combination with
414 the already available pluggable KEM and X.509 support, this enables
415 for example suitable providers to deliver post-quantum or quantum-safe
416 cryptography to OpenSSL users.
420 * Added support for pluggable (provider-based) CMS signature algorithms.
421 This enables CMS sign and verify operations with algorithms embedded
422 in providers not included by default in OpenSSL.
426 * Added support for Hybrid Public Key Encryption (HPKE) as defined
427 in RFC9180. HPKE is required for TLS Encrypted ClientHello (ECH),
428 Message Layer Security (MLS) and other IETF specifications.
429 HPKE can also be used by other applications that require
430 encrypting "to" an ECDH public key. External APIs are defined in
431 include/openssl/hpke.h and documented in doc/man3/OSSL_HPKE_CTX_new.pod
435 * Implemented HPKE DHKEM support in providers used by HPKE (RFC9180)
440 * Add support for certificate compression (RFC8879), including
441 library support for Brotli and Zstandard compression.
445 * Add the ability to add custom attributes to PKCS12 files. Add a new API
446 PKCS12_create_ex2, identical to the existing PKCS12_create_ex but allows
447 for a user specified callback and optional argument.
448 Added a new PKCS12_SAFEBAG_set0_attr, which allows for a new attr to be
449 added to the existing STACK_OF attrs.
453 * Major refactor of the libssl record layer.
457 * Add a mac salt length option for the pkcs12 command.
461 * Add more SRTP protection profiles from RFC8723 and RFC8269.
465 * Extended Kernel TLS (KTLS) to support TLS 1.3 receive offload.
467 *Daiki Ueno, John Baldwin and Dmitry Podgorny*
469 * Add support for TCP Fast Open (RFC7413) to macOS, Linux, and FreeBSD where
470 supported and enabled.
474 * Add ciphersuites based on DHE_PSK (RFC 4279) and ECDHE_PSK (RFC 5489)
475 to the list of ciphersuites providing Perfect Forward Secrecy as
476 required by SECLEVEL >= 3.
478 *Dmitry Belyavskiy, Nicola Tuveri*
480 * Add new SSL APIs to aid in efficiently implementing TLS/SSL fingerprinting.
481 The SSL_CTRL_GET_IANA_GROUPS control code, exposed as the
482 SSL_get0_iana_groups() function-like macro, retrieves the list of
483 supported groups sent by the peer.
484 The function SSL_client_hello_get_extension_order() populates
485 a caller-supplied array with the list of extension types present in the
486 ClientHello, in order of appearance.
490 * Fixed PEM_write_bio_PKCS8PrivateKey() and PEM_write_bio_PKCS8PrivateKey_nid()
491 to make it possible to use empty passphrase strings.
495 * The PKCS12_parse() function now supports MAC-less PKCS12 files.
499 * Added ASYNC_set_mem_functions() and ASYNC_get_mem_functions() calls to be able
500 to change functions used for allocating the memory of asynchronous call stack.
504 * Added support for signed BIGNUMs in the OSSL_PARAM APIs.
508 * A failure exit code is returned when using the openssl x509 command to check
509 certificate attributes and the checks fail.
513 * The default SSL/TLS security level has been changed from 1 to 2. RSA,
514 DSA and DH keys of 1024 bits and above and less than 2048 bits and ECC keys
515 of 160 bits and above and less than 224 bits were previously accepted by
516 default but are now no longer allowed. By default TLS compression was
517 already disabled in previous OpenSSL versions. At security level 2 it cannot
522 * The SSL_CTX_set_cipher_list family functions now accept ciphers using their
527 * The PVK key derivation function has been moved from b2i_PVK_bio_ex() into
528 the legacy crypto provider as an EVP_KDF. Applications requiring this KDF
529 will need to load the legacy crypto provider.
533 * CCM8 cipher suites in TLS have been downgraded to security level zero
534 because they use a short authentication tag which lowers their strength.
538 * Subject or issuer names in X.509 objects are now displayed as UTF-8 strings
539 by default. Also spaces surrounding `=` in DN output are removed.
543 * Add X.509 certificate codeSigning purpose and related checks on key usage and
544 extended key usage of the leaf certificate according to the CA/Browser Forum.
548 * The `x509`, `ca`, and `req` apps now produce X.509 v3 certificates.
549 The `-x509v1` option of `req` prefers generation of X.509 v1 certificates.
550 `X509_sign()` and `X509_sign_ctx()` make sure that the certificate has
551 X.509 version 3 if the certificate information includes X.509 extensions.
555 * Fix and extend certificate handling and the apps `x509`, `verify` etc.
556 such as adding a trace facility for debugging certificate chain building.
560 * Various fixes and extensions to the CMP+CRMF implementation and the `cmp` app
561 in particular supporting requests for central key generation, generalized
562 polling, and various types of genm/genp exchanges defined in CMP Updates.
566 * Fixes and extensions to the HTTP client and to the HTTP server in `apps/`
567 like correcting the TLS and proxy support and adding tracing for debugging.
571 * Extended the CMS API for handling `CMS_SignedData` and `CMS_EnvelopedData`.
575 * `CMS_add0_cert()` and `CMS_add1_cert()` no longer throw an error if
576 a certificate to be added is already present. `CMS_sign_ex()` and
577 `CMS_sign()` now ignore any duplicate certificates in their `certs` argument
578 and no longer throw an error for them.
582 * Fixed and extended `util/check-format.pl` for checking adherence to the
583 coding style <https://www.openssl.org/policies/technical/coding-style.html>.
584 The checks are meanwhile more complete and yield fewer false positives.
588 * Added BIO_s_dgram_pair() and BIO_s_dgram_mem() that provide memory-based
589 BIOs with datagram semantics and support for BIO_sendmmsg() and BIO_recvmmsg()
590 calls. They can be used as the transport BIOs for QUIC.
592 *Hugo Landau, Matt Caswell and Tomáš Mráz*
594 * Add new BIO_sendmmsg() and BIO_recvmmsg() BIO methods which allow
595 sending and receiving multiple messages in a single call. An implementation
596 is provided for BIO_dgram. For further details, see BIO_sendmmsg(3).
600 * Support for loading root certificates from the Windows certificate store
601 has been added. The support is in the form of a store which recognises the
602 URI string of `org.openssl.winstore://`. This URI scheme currently takes no
603 arguments. This store is built by default and can be disabled using the new
604 compile-time option `no-winstore`. This store is not currently used by
605 default and must be loaded explicitly using the above store URI. It is
606 expected to be loaded by default in the future.
610 * Enable KTLS with the TLS 1.3 CCM mode ciphersuites. Note that some linux
611 kernel versions that support KTLS have a known bug in CCM processing. That
612 has been fixed in stable releases starting from 5.4.164, 5.10.84, 5.15.7,
613 and all releases since 5.16. KTLS with CCM ciphersuites should be only used
618 * Added `-ktls` option to `s_server` and `s_client` commands to enable the
623 * Zerocopy KTLS sendfile() support on Linux.
627 * The OBJ_ calls are now thread safe using a global lock.
631 * New parameter `-digest` for openssl cms command allowing signing
632 pre-computed digests and new CMS API functions supporting that
637 * OPENSSL_malloc() and other allocation functions now raise errors on
638 allocation failures. The callers do not need to explicitly raise errors
639 unless they want to for tracing purposes.
643 * Added and enabled by default implicit rejection in RSA PKCS#1 v1.5
644 decryption as a protection against Bleichenbacher-like attacks.
645 The RSA decryption API will now return a randomly generated deterministic
646 message instead of an error in case it detects an error when checking
647 padding during PKCS#1 v1.5 decryption. This is a general protection against
648 issues like CVE-2020-25659 and CVE-2020-25657. This protection can be
650 `EVP_PKEY_CTX_ctrl_str(ctx, "rsa_pkcs1_implicit_rejection". "0")`
651 on the RSA decryption context.
655 * Added support for Brainpool curves in TLS-1.3.
657 *Bernd Edlinger and Matt Caswell*
659 * Added OpenBSD specific build targets.
663 * Support for Argon2d, Argon2i, Argon2id KDFs has been added along with
664 a basic thread pool implementation for select platforms.
671 ### Changes between 3.1.3 and 3.1.4 [24 Oct 2023]
673 * Fix incorrect key and IV resizing issues when calling EVP_EncryptInit_ex2(),
674 EVP_DecryptInit_ex2() or EVP_CipherInit_ex2() with OSSL_PARAM parameters
675 that alter the key or IV length ([CVE-2023-5363]).
679 ### Changes between 3.1.2 and 3.1.3 [19 Sep 2023]
681 * Fix POLY1305 MAC implementation corrupting XMM registers on Windows.
683 The POLY1305 MAC (message authentication code) implementation in OpenSSL
684 does not save the contents of non-volatile XMM registers on Windows 64
685 platform when calculating the MAC of data larger than 64 bytes. Before
686 returning to the caller all the XMM registers are set to zero rather than
687 restoring their previous content. The vulnerable code is used only on newer
688 x86_64 processors supporting the AVX512-IFMA instructions.
690 The consequences of this kind of internal application state corruption can
691 be various - from no consequences, if the calling application does not
692 depend on the contents of non-volatile XMM registers at all, to the worst
693 consequences, where the attacker could get complete control of the
694 application process. However given the contents of the registers are just
695 zeroized so the attacker cannot put arbitrary values inside, the most likely
696 consequence, if any, would be an incorrect result of some application
697 dependent calculations or a crash leading to a denial of service.
703 ### Changes between 3.1.1 and 3.1.2 [1 Aug 2023]
705 * Fix excessive time spent checking DH q parameter value.
707 The function DH_check() performs various checks on DH parameters. After
708 fixing CVE-2023-3446 it was discovered that a large q parameter value can
709 also trigger an overly long computation during some of these checks.
710 A correct q value, if present, cannot be larger than the modulus p
711 parameter, thus it is unnecessary to perform these checks if q is larger
714 If DH_check() is called with such q parameter value,
715 DH_CHECK_INVALID_Q_VALUE return flag is set and the computationally
716 intensive checks are skipped.
722 * Fix DH_check() excessive time with over sized modulus.
724 The function DH_check() performs various checks on DH parameters. One of
725 those checks confirms that the modulus ("p" parameter) is not too large.
726 Trying to use a very large modulus is slow and OpenSSL will not normally use
727 a modulus which is over 10,000 bits in length.
729 However the DH_check() function checks numerous aspects of the key or
730 parameters that have been supplied. Some of those checks use the supplied
731 modulus value even if it has already been found to be too large.
733 A new limit has been added to DH_check of 32,768 bits. Supplying a
734 key/parameters with a modulus over this size will simply cause DH_check() to
741 * Do not ignore empty associated data entries with AES-SIV.
743 The AES-SIV algorithm allows for authentication of multiple associated
744 data entries along with the encryption. To authenticate empty data the
745 application has to call `EVP_EncryptUpdate()` (or `EVP_CipherUpdate()`)
746 with NULL pointer as the output buffer and 0 as the input buffer length.
747 The AES-SIV implementation in OpenSSL just returns success for such call
748 instead of performing the associated data authentication operation.
749 The empty data thus will not be authenticated. ([CVE-2023-2975])
751 Thanks to Juerg Wullschleger (Google) for discovering the issue.
753 The fix changes the authentication tag value and the ciphertext for
754 applications that use empty associated data entries with AES-SIV.
755 To decrypt data encrypted with previous versions of OpenSSL the application
756 has to skip calls to `EVP_DecryptUpdate()` for empty associated data
761 * When building with the `enable-fips` option and using the resulting
762 FIPS provider, TLS 1.2 will, by default, mandate the use of an extended
763 master secret (FIPS 140-3 IG G.Q) and the Hash and HMAC DRBGs will
764 not operate with truncated digests (FIPS 140-3 IG G.R).
768 ### Changes between 3.1.0 and 3.1.1 [30 May 2023]
770 * Mitigate for the time it takes for `OBJ_obj2txt` to translate gigantic
771 OBJECT IDENTIFIER sub-identifiers to canonical numeric text form.
773 OBJ_obj2txt() would translate any size OBJECT IDENTIFIER to canonical
774 numeric text form. For gigantic sub-identifiers, this would take a very
775 long time, the time complexity being O(n^2) where n is the size of that
776 sub-identifier. ([CVE-2023-2650])
778 To mitigitate this, `OBJ_obj2txt()` will only translate an OBJECT
779 IDENTIFIER to canonical numeric text form if the size of that OBJECT
780 IDENTIFIER is 586 bytes or less, and fail otherwise.
782 The basis for this restriction is [RFC 2578 (STD 58), section 3.5]. OBJECT
783 IDENTIFIER values, which stipulates that OBJECT IDENTIFIERS may have at
784 most 128 sub-identifiers, and that the maximum value that each sub-
785 identifier may have is 2^32-1 (4294967295 decimal).
787 For each byte of every sub-identifier, only the 7 lower bits are part of
788 the value, so the maximum amount of bytes that an OBJECT IDENTIFIER with
789 these restrictions may occupy is 32 * 128 / 7, which is approximately 586
794 * Multiple algorithm implementation fixes for ARM BE platforms.
798 * Added a -pedantic option to fipsinstall that adjusts the various
799 settings to ensure strict FIPS compliance rather than backwards
804 * Fixed buffer overread in AES-XTS decryption on ARM 64 bit platforms which
805 happens if the buffer size is 4 mod 5 in 16 byte AES blocks. This can
806 trigger a crash of an application using AES-XTS decryption if the memory
807 just after the buffer being decrypted is not mapped.
808 Thanks to Anton Romanov (Amazon) for discovering the issue.
813 * Reworked the Fix for the Timing Oracle in RSA Decryption ([CVE-2022-4304]).
814 The previous fix for this timing side channel turned out to cause
815 a severe 2-3x performance regression in the typical use case
816 compared to 3.0.7. The new fix uses existing constant time
817 code paths, and restores the previous performance level while
818 fully eliminating all existing timing side channels.
819 The fix was developed by Bernd Edlinger with testing support
824 * Add FIPS provider configuration option to disallow the use of
825 truncated digests with Hash and HMAC DRBGs (q.v. FIPS 140-3 IG D.R.).
826 The option '-no_drbg_truncated_digests' can optionally be
827 supplied to 'openssl fipsinstall'.
831 * Corrected documentation of X509_VERIFY_PARAM_add0_policy() to mention
832 that it does not enable policy checking. Thanks to David Benjamin for
833 discovering this issue.
838 * Fixed an issue where invalid certificate policies in leaf certificates are
839 silently ignored by OpenSSL and other certificate policy checks are skipped
840 for that certificate. A malicious CA could use this to deliberately assert
841 invalid certificate policies in order to circumvent policy checking on the
842 certificate altogether.
847 * Limited the number of nodes created in a policy tree to mitigate
848 against CVE-2023-0464. The default limit is set to 1000 nodes, which
849 should be sufficient for most installations. If required, the limit
850 can be adjusted by setting the OPENSSL_POLICY_TREE_NODES_MAX build
851 time define to a desired maximum number of nodes or zero to allow
857 ### Changes between 3.0 and 3.1.0 [14 Mar 2023]
859 * Add FIPS provider configuration option to enforce the
860 Extended Master Secret (EMS) check during the TLS1_PRF KDF.
861 The option '-ems-check' can optionally be supplied to
862 'openssl fipsinstall'.
866 * The FIPS provider includes a few non-approved algorithms for
867 backward compatibility purposes and the "fips=yes" property query
868 must be used for all algorithm fetches to ensure FIPS compliance.
870 The algorithms that are included but not approved are Triple DES ECB,
871 Triple DES CBC and EdDSA.
875 * Added support for KMAC in KBKDF.
879 * RNDR and RNDRRS support in provider functions to provide
880 random number generation for Arm CPUs (aarch64).
884 * s_client and s_server apps now explicitly say when the TLS version
885 does not include the renegotiation mechanism. This avoids confusion
886 between that scenario versus when the TLS version includes secure
887 renegotiation but the peer lacks support for it.
891 * AES-GCM enabled with AVX512 vAES and vPCLMULQDQ.
893 *Tomasz Kantecki, Andrey Matyukov*
895 * The various OBJ_* functions have been made thread safe.
899 * Parallel dual-prime 1536/2048-bit modular exponentiation for
900 AVX512_IFMA capable processors.
902 *Sergey Kirillov, Andrey Matyukov (Intel Corp)*
904 * The functions `OPENSSL_LH_stats`, `OPENSSL_LH_node_stats`,
905 `OPENSSL_LH_node_usage_stats`, `OPENSSL_LH_stats_bio`,
906 `OPENSSL_LH_node_stats_bio` and `OPENSSL_LH_node_usage_stats_bio` are now
907 marked deprecated from OpenSSL 3.1 onwards and can be disabled by defining
908 `OPENSSL_NO_DEPRECATED_3_1`.
910 The macro `DEFINE_LHASH_OF` is now deprecated in favour of the macro
911 `DEFINE_LHASH_OF_EX`, which omits the corresponding type-specific function
912 definitions for these functions regardless of whether
913 `OPENSSL_NO_DEPRECATED_3_1` is defined.
915 Users of `DEFINE_LHASH_OF` may start receiving deprecation warnings for these
916 functions regardless of whether they are using them. It is recommended that
917 users transition to the new macro, `DEFINE_LHASH_OF_EX`.
921 * When generating safe-prime DH parameters set the recommended private key
922 length equivalent to minimum key lengths as in RFC 7919.
926 * Change the default salt length for PKCS#1 RSASSA-PSS signatures to the
927 maximum size that is smaller or equal to the digest length to comply with
928 FIPS 186-4 section 5. This is implemented by a new option
929 `OSSL_PKEY_RSA_PSS_SALT_LEN_AUTO_DIGEST_MAX` ("auto-digestmax") for the
930 `rsa_pss_saltlen` parameter, which is now the default. Signature
931 verification is not affected by this change and continues to work as before.
938 For OpenSSL 3.0 a [Migration guide][] has been added, so the CHANGES entries
939 listed here are only a brief description.
940 The migration guide contains more detailed information related to new features,
941 breaking changes, and mappings for the large list of deprecated functions.
943 [Migration guide]: https://github.com/openssl/openssl/tree/master/doc/man7/migration_guide.pod
945 ### Changes between 3.0.7 and 3.0.8 [7 Feb 2023]
947 * Fixed NULL dereference during PKCS7 data verification.
949 A NULL pointer can be dereferenced when signatures are being
950 verified on PKCS7 signed or signedAndEnveloped data. In case the hash
951 algorithm used for the signature is known to the OpenSSL library but
952 the implementation of the hash algorithm is not available the digest
953 initialization will fail. There is a missing check for the return
954 value from the initialization function which later leads to invalid
955 usage of the digest API most likely leading to a crash.
958 PKCS7 data is processed by the SMIME library calls and also by the
959 time stamp (TS) library calls. The TLS implementation in OpenSSL does
960 not call these functions however third party applications would be
961 affected if they call these functions to verify signatures on untrusted
966 * Fixed X.400 address type confusion in X.509 GeneralName.
968 There is a type confusion vulnerability relating to X.400 address processing
969 inside an X.509 GeneralName. X.400 addresses were parsed as an ASN1_STRING
970 but the public structure definition for GENERAL_NAME incorrectly specified
971 the type of the x400Address field as ASN1_TYPE. This field is subsequently
972 interpreted by the OpenSSL function GENERAL_NAME_cmp as an ASN1_TYPE rather
975 When CRL checking is enabled (i.e. the application sets the
976 X509_V_FLAG_CRL_CHECK flag), this vulnerability may allow an attacker to
977 pass arbitrary pointers to a memcmp call, enabling them to read memory
978 contents or enact a denial of service.
983 * Fixed NULL dereference validating DSA public key.
985 An invalid pointer dereference on read can be triggered when an
986 application tries to check a malformed DSA public key by the
987 EVP_PKEY_public_check() function. This will most likely lead
988 to an application crash. This function can be called on public
989 keys supplied from untrusted sources which could allow an attacker
990 to cause a denial of service attack.
992 The TLS implementation in OpenSSL does not call this function
993 but applications might call the function if there are additional
994 security requirements imposed by standards such as FIPS 140-3.
997 *Shane Lontis, Tomáš Mráz*
999 * Fixed Invalid pointer dereference in d2i_PKCS7 functions.
1001 An invalid pointer dereference on read can be triggered when an
1002 application tries to load malformed PKCS7 data with the
1003 d2i_PKCS7(), d2i_PKCS7_bio() or d2i_PKCS7_fp() functions.
1005 The result of the dereference is an application crash which could
1006 lead to a denial of service attack. The TLS implementation in OpenSSL
1007 does not call this function however third party applications might
1008 call these functions on untrusted data.
1013 * Fixed Use-after-free following BIO_new_NDEF.
1015 The public API function BIO_new_NDEF is a helper function used for
1016 streaming ASN.1 data via a BIO. It is primarily used internally to OpenSSL
1017 to support the SMIME, CMS and PKCS7 streaming capabilities, but may also
1018 be called directly by end user applications.
1020 The function receives a BIO from the caller, prepends a new BIO_f_asn1
1021 filter BIO onto the front of it to form a BIO chain, and then returns
1022 the new head of the BIO chain to the caller. Under certain conditions,
1023 for example if a CMS recipient public key is invalid, the new filter BIO
1024 is freed and the function returns a NULL result indicating a failure.
1025 However, in this case, the BIO chain is not properly cleaned up and the
1026 BIO passed by the caller still retains internal pointers to the previously
1027 freed filter BIO. If the caller then goes on to call BIO_pop() on the BIO
1028 then a use-after-free will occur. This will most likely result in a crash.
1031 *Viktor Dukhovni, Matt Caswell*
1033 * Fixed Double free after calling PEM_read_bio_ex.
1035 The function PEM_read_bio_ex() reads a PEM file from a BIO and parses and
1036 decodes the "name" (e.g. "CERTIFICATE"), any header data and the payload
1037 data. If the function succeeds then the "name_out", "header" and "data"
1038 arguments are populated with pointers to buffers containing the relevant
1039 decoded data. The caller is responsible for freeing those buffers. It is
1040 possible to construct a PEM file that results in 0 bytes of payload data.
1041 In this case PEM_read_bio_ex() will return a failure code but will populate
1042 the header argument with a pointer to a buffer that has already been freed.
1043 If the caller also frees this buffer then a double free will occur. This
1044 will most likely lead to a crash.
1046 The functions PEM_read_bio() and PEM_read() are simple wrappers around
1047 PEM_read_bio_ex() and therefore these functions are also directly affected.
1049 These functions are also called indirectly by a number of other OpenSSL
1050 functions including PEM_X509_INFO_read_bio_ex() and
1051 SSL_CTX_use_serverinfo_file() which are also vulnerable. Some OpenSSL
1052 internal uses of these functions are not vulnerable because the caller does
1053 not free the header argument if PEM_read_bio_ex() returns a failure code.
1056 *Kurt Roeckx, Matt Caswell*
1058 * Fixed Timing Oracle in RSA Decryption.
1060 A timing based side channel exists in the OpenSSL RSA Decryption
1061 implementation which could be sufficient to recover a plaintext across
1062 a network in a Bleichenbacher style attack. To achieve a successful
1063 decryption an attacker would have to be able to send a very large number
1064 of trial messages for decryption. The vulnerability affects all RSA padding
1065 modes: PKCS#1 v1.5, RSA-OEAP and RSASVE.
1068 *Dmitry Belyavsky, Hubert Kario*
1070 * Fixed X.509 Name Constraints Read Buffer Overflow.
1072 A read buffer overrun can be triggered in X.509 certificate verification,
1073 specifically in name constraint checking. The read buffer overrun might
1074 result in a crash which could lead to a denial of service attack.
1075 In a TLS client, this can be triggered by connecting to a malicious
1076 server. In a TLS server, this can be triggered if the server requests
1077 client authentication and a malicious client connects.
1082 * Fixed X.509 Policy Constraints Double Locking security issue.
1084 If an X.509 certificate contains a malformed policy constraint and
1085 policy processing is enabled, then a write lock will be taken twice
1086 recursively. On some operating systems (most widely: Windows) this
1087 results in a denial of service when the affected process hangs. Policy
1088 processing being enabled on a publicly facing server is not considered
1089 to be a common setup.
1094 * Our provider implementations of `OSSL_FUNC_KEYMGMT_EXPORT` and
1095 `OSSL_FUNC_KEYMGMT_GET_PARAMS` for EC and SM2 keys now honor
1096 `OSSL_PKEY_PARAM_EC_POINT_CONVERSION_FORMAT` as set (and
1097 default to `POINT_CONVERSION_UNCOMPRESSED`) when exporting
1098 `OSSL_PKEY_PARAM_PUB_KEY`, instead of unconditionally using
1099 `POINT_CONVERSION_COMPRESSED` as in previous 3.x releases.
1100 For symmetry, our implementation of `EVP_PKEY_ASN1_METHOD->export_to`
1101 for legacy EC and SM2 keys is also changed similarly to honor the
1102 equivalent conversion format flag as specified in the underlying
1103 `EC_KEY` object being exported to a provider, when this function is
1104 called through `EVP_PKEY_export()`.
1108 ### Changes between 3.0.6 and 3.0.7 [1 Nov 2022]
1110 * Fixed two buffer overflows in punycode decoding functions.
1112 A buffer overrun can be triggered in X.509 certificate verification,
1113 specifically in name constraint checking. Note that this occurs after
1114 certificate chain signature verification and requires either a CA to
1115 have signed the malicious certificate or for the application to continue
1116 certificate verification despite failure to construct a path to a trusted
1119 In a TLS client, this can be triggered by connecting to a malicious
1120 server. In a TLS server, this can be triggered if the server requests
1121 client authentication and a malicious client connects.
1123 An attacker can craft a malicious email address to overflow
1124 an arbitrary number of bytes containing the `.` character (decimal 46)
1125 on the stack. This buffer overflow could result in a crash (causing a
1129 An attacker can craft a malicious email address to overflow four
1130 attacker-controlled bytes on the stack. This buffer overflow could
1131 result in a crash (causing a denial of service) or potentially remote code
1132 execution depending on stack layout for any given platform/compiler.
1137 * Removed all references to invalid OSSL_PKEY_PARAM_RSA names for CRT
1138 parameters in OpenSSL code.
1139 Applications should not use the names OSSL_PKEY_PARAM_RSA_FACTOR,
1140 OSSL_PKEY_PARAM_RSA_EXPONENT and OSSL_PKEY_PARAM_RSA_COEFFICIENT.
1141 Use the numbered names such as OSSL_PKEY_PARAM_RSA_FACTOR1 instead.
1142 Using these invalid names may cause algorithms to use slower methods
1143 that ignore the CRT parameters.
1147 * Fixed a regression introduced in 3.0.6 version raising errors on some stack
1152 * Fixed a regression introduced in 3.0.6 version not refreshing the certificate
1153 data to be signed before signing the certificate.
1157 * Added RIPEMD160 to the default provider.
1161 * Ensured that the key share group sent or accepted for the key exchange
1162 is allowed for the protocol version.
1166 ### Changes between 3.0.5 and 3.0.6 [11 Oct 2022]
1168 * OpenSSL supports creating a custom cipher via the legacy
1169 EVP_CIPHER_meth_new() function and associated function calls. This function
1170 was deprecated in OpenSSL 3.0 and application authors are instead encouraged
1171 to use the new provider mechanism in order to implement custom ciphers.
1173 OpenSSL versions 3.0.0 to 3.0.5 incorrectly handle legacy custom ciphers
1174 passed to the EVP_EncryptInit_ex2(), EVP_DecryptInit_ex2() and
1175 EVP_CipherInit_ex2() functions (as well as other similarly named encryption
1176 and decryption initialisation functions). Instead of using the custom cipher
1177 directly it incorrectly tries to fetch an equivalent cipher from the
1178 available providers. An equivalent cipher is found based on the NID passed to
1179 EVP_CIPHER_meth_new(). This NID is supposed to represent the unique NID for a
1180 given cipher. However it is possible for an application to incorrectly pass
1181 NID_undef as this value in the call to EVP_CIPHER_meth_new(). When NID_undef
1182 is used in this way the OpenSSL encryption/decryption initialisation function
1183 will match the NULL cipher as being equivalent and will fetch this from the
1184 available providers. This will succeed if the default provider has been
1185 loaded (or if a third party provider has been loaded that offers this
1186 cipher). Using the NULL cipher means that the plaintext is emitted as the
1189 Applications are only affected by this issue if they call
1190 EVP_CIPHER_meth_new() using NID_undef and subsequently use it in a call to an
1191 encryption/decryption initialisation function. Applications that only use
1192 SSL/TLS are not impacted by this issue.
1197 * Fix LLVM vs Apple LLVM version numbering confusion that caused build failures
1202 * Fixed the linux-mips64 Configure target which was missing the
1203 SIXTY_FOUR_BIT bn_ops flag. This was causing heap corruption on that
1208 * Fix handling of a ticket key callback that returns 0 in TLSv1.3 to not send a
1213 * Correctly handle a retransmitted ClientHello in DTLS
1217 * Fixed detection of ktls support in cross-compile environment on Linux
1221 * Fixed some regressions and test failures when running the 3.0.0 FIPS provider
1226 * Fixed SSL_pending() and SSL_has_pending() with DTLS which were failing to
1227 report correct results in some cases
1231 * Fix UWP builds by defining VirtualLock
1235 * For known safe primes use the minimum key length according to RFC 7919.
1236 Longer private key sizes unnecessarily raise the cycles needed to compute the
1237 shared secret without any increase of the real security. This fixes a
1238 regression from 1.1.1 where these shorter keys were generated for the known
1243 * Added the loongarch64 target
1247 * Fixed EC ASM flag passing. Flags for ASM implementations of EC curves were
1248 only passed to the FIPS provider and not to the default or legacy provider.
1252 * Fixed reported performance degradation on aarch64. Restored the
1253 implementation prior to commit 2621751 ("aes/asm/aesv8-armx.pl: avoid
1254 32-bit lane assignment in CTR mode") for 64bit targets only, since it is
1255 reportedly 2-17% slower and the silicon errata only affects 32bit targets.
1256 The new algorithm is still used for 32 bit targets.
1260 * Added a missing header for memcmp that caused compilation failure on some
1265 ### Changes between 3.0.4 and 3.0.5 [5 Jul 2022]
1267 * The OpenSSL 3.0.4 release introduced a serious bug in the RSA
1268 implementation for X86_64 CPUs supporting the AVX512IFMA instructions.
1269 This issue makes the RSA implementation with 2048 bit private keys
1270 incorrect on such machines and memory corruption will happen during
1271 the computation. As a consequence of the memory corruption an attacker
1272 may be able to trigger a remote code execution on the machine performing
1275 SSL/TLS servers or other servers using 2048 bit RSA private keys running
1276 on machines supporting AVX512IFMA instructions of the X86_64 architecture
1277 are affected by this issue.
1282 * AES OCB mode for 32-bit x86 platforms using the AES-NI assembly optimised
1283 implementation would not encrypt the entirety of the data under some
1284 circumstances. This could reveal sixteen bytes of data that was
1285 preexisting in the memory that wasn't written. In the special case of
1286 "in place" encryption, sixteen bytes of the plaintext would be revealed.
1288 Since OpenSSL does not support OCB based cipher suites for TLS and DTLS,
1289 they are both unaffected.
1292 *Alex Chernyakhovsky, David Benjamin, Alejandro Sedeño*
1294 ### Changes between 3.0.3 and 3.0.4 [21 Jun 2022]
1296 * In addition to the c_rehash shell command injection identified in
1297 CVE-2022-1292, further bugs where the c_rehash script does not
1298 properly sanitise shell metacharacters to prevent command injection have been
1301 When the CVE-2022-1292 was fixed it was not discovered that there
1302 are other places in the script where the file names of certificates
1303 being hashed were possibly passed to a command executed through the shell.
1305 This script is distributed by some operating systems in a manner where
1306 it is automatically executed. On such operating systems, an attacker
1307 could execute arbitrary commands with the privileges of the script.
1309 Use of the c_rehash script is considered obsolete and should be replaced
1310 by the OpenSSL rehash command line tool.
1313 *Daniel Fiala, Tomáš Mráz*
1315 * Case insensitive string comparison no longer uses locales. It has instead
1316 been directly implemented.
1320 ### Changes between 3.0.2 and 3.0.3 [3 May 2022]
1322 * Case insensitive string comparison is reimplemented via new locale-agnostic
1323 comparison functions OPENSSL_str[n]casecmp always using the POSIX locale for
1324 comparison. The previous implementation had problems when the Turkish locale
1329 * Fixed a bug in the c_rehash script which was not properly sanitising shell
1330 metacharacters to prevent command injection. This script is distributed by
1331 some operating systems in a manner where it is automatically executed. On
1332 such operating systems, an attacker could execute arbitrary commands with the
1333 privileges of the script.
1335 Use of the c_rehash script is considered obsolete and should be replaced
1336 by the OpenSSL rehash command line tool.
1341 * Fixed a bug in the function `OCSP_basic_verify` that verifies the signer
1342 certificate on an OCSP response. The bug caused the function in the case
1343 where the (non-default) flag OCSP_NOCHECKS is used to return a postivie
1344 response (meaning a successful verification) even in the case where the
1345 response signing certificate fails to verify.
1347 It is anticipated that most users of `OCSP_basic_verify` will not use the
1348 OCSP_NOCHECKS flag. In this case the `OCSP_basic_verify` function will return
1349 a negative value (indicating a fatal error) in the case of a certificate
1350 verification failure. The normal expected return value in this case would be
1353 This issue also impacts the command line OpenSSL "ocsp" application. When
1354 verifying an ocsp response with the "-no_cert_checks" option the command line
1355 application will report that the verification is successful even though it
1356 has in fact failed. In this case the incorrect successful response will also
1357 be accompanied by error messages showing the failure and contradicting the
1358 apparently successful result.
1363 * Fixed a bug where the RC4-MD5 ciphersuite incorrectly used the
1364 AAD data as the MAC key. This made the MAC key trivially predictable.
1366 An attacker could exploit this issue by performing a man-in-the-middle attack
1367 to modify data being sent from one endpoint to an OpenSSL 3.0 recipient such
1368 that the modified data would still pass the MAC integrity check.
1370 Note that data sent from an OpenSSL 3.0 endpoint to a non-OpenSSL 3.0
1371 endpoint will always be rejected by the recipient and the connection will
1372 fail at that point. Many application protocols require data to be sent from
1373 the client to the server first. Therefore, in such a case, only an OpenSSL
1374 3.0 server would be impacted when talking to a non-OpenSSL 3.0 client.
1376 If both endpoints are OpenSSL 3.0 then the attacker could modify data being
1377 sent in both directions. In this case both clients and servers could be
1378 affected, regardless of the application protocol.
1380 Note that in the absence of an attacker this bug means that an OpenSSL 3.0
1381 endpoint communicating with a non-OpenSSL 3.0 endpoint will fail to complete
1382 the handshake when using this ciphersuite.
1384 The confidentiality of data is not impacted by this issue, i.e. an attacker
1385 cannot decrypt data that has been encrypted using this ciphersuite - they can
1388 In order for this attack to work both endpoints must legitimately negotiate
1389 the RC4-MD5 ciphersuite. This ciphersuite is not compiled by default in
1390 OpenSSL 3.0, and is not available within the default provider or the default
1391 ciphersuite list. This ciphersuite will never be used if TLSv1.3 has been
1392 negotiated. In order for an OpenSSL 3.0 endpoint to use this ciphersuite the
1393 following must have occurred:
1395 1) OpenSSL must have been compiled with the (non-default) compile time option
1396 enable-weak-ssl-ciphers
1398 2) OpenSSL must have had the legacy provider explicitly loaded (either
1399 through application code or via configuration)
1401 3) The ciphersuite must have been explicitly added to the ciphersuite list
1403 4) The libssl security level must have been set to 0 (default is 1)
1405 5) A version of SSL/TLS below TLSv1.3 must have been negotiated
1407 6) Both endpoints must negotiate the RC4-MD5 ciphersuite in preference to any
1408 others that both endpoints have in common
1413 * Fix a bug in the OPENSSL_LH_flush() function that breaks reuse of the memory
1414 occupied by the removed hash table entries.
1416 This function is used when decoding certificates or keys. If a long lived
1417 process periodically decodes certificates or keys its memory usage will
1418 expand without bounds and the process might be terminated by the operating
1419 system causing a denial of service. Also traversing the empty hash table
1420 entries will take increasingly more time.
1422 Typically such long lived processes might be TLS clients or TLS servers
1423 configured to accept client certificate authentication.
1426 *Hugo Landau, Aliaksei Levin*
1428 * The functions `OPENSSL_LH_stats` and `OPENSSL_LH_stats_bio` now only report
1429 the `num_items`, `num_nodes` and `num_alloc_nodes` statistics. All other
1430 statistics are no longer supported. For compatibility, these statistics are
1431 still listed in the output but are now always reported as zero.
1435 ### Changes between 3.0.1 and 3.0.2 [15 Mar 2022]
1437 * Fixed a bug in the BN_mod_sqrt() function that can cause it to loop forever
1438 for non-prime moduli.
1440 Internally this function is used when parsing certificates that contain
1441 elliptic curve public keys in compressed form or explicit elliptic curve
1442 parameters with a base point encoded in compressed form.
1444 It is possible to trigger the infinite loop by crafting a certificate that
1445 has invalid explicit curve parameters.
1447 Since certificate parsing happens prior to verification of the certificate
1448 signature, any process that parses an externally supplied certificate may thus
1449 be subject to a denial of service attack. The infinite loop can also be
1450 reached when parsing crafted private keys as they can contain explicit
1451 elliptic curve parameters.
1453 Thus vulnerable situations include:
1455 - TLS clients consuming server certificates
1456 - TLS servers consuming client certificates
1457 - Hosting providers taking certificates or private keys from customers
1458 - Certificate authorities parsing certification requests from subscribers
1459 - Anything else which parses ASN.1 elliptic curve parameters
1461 Also any other applications that use the BN_mod_sqrt() where the attacker
1462 can control the parameter values are vulnerable to this DoS issue.
1467 * Add ciphersuites based on DHE_PSK (RFC 4279) and ECDHE_PSK (RFC 5489)
1468 to the list of ciphersuites providing Perfect Forward Secrecy as
1469 required by SECLEVEL >= 3.
1471 *Dmitry Belyavskiy, Nicola Tuveri*
1473 * Made the AES constant time code for no-asm configurations
1474 optional due to the resulting 95% performance degradation.
1475 The AES constant time code can be enabled, for no assembly
1476 builds, with: ./config no-asm -DOPENSSL_AES_CONST_TIME
1480 * Fixed PEM_write_bio_PKCS8PrivateKey() to make it possible to use empty
1485 * The negative return value handling of the certificate verification callback
1486 was reverted. The replacement is to set the verification retry state with
1487 the SSL_set_retry_verify() function.
1491 ### Changes between 3.0.0 and 3.0.1 [14 Dec 2021]
1493 * Fixed invalid handling of X509_verify_cert() internal errors in libssl
1494 Internally libssl in OpenSSL calls X509_verify_cert() on the client side to
1495 verify a certificate supplied by a server. That function may return a
1496 negative return value to indicate an internal error (for example out of
1497 memory). Such a negative return value is mishandled by OpenSSL and will cause
1498 an IO function (such as SSL_connect() or SSL_do_handshake()) to not indicate
1499 success and a subsequent call to SSL_get_error() to return the value
1500 SSL_ERROR_WANT_RETRY_VERIFY. This return value is only supposed to be
1501 returned by OpenSSL if the application has previously called
1502 SSL_CTX_set_cert_verify_callback(). Since most applications do not do this
1503 the SSL_ERROR_WANT_RETRY_VERIFY return value from SSL_get_error() will be
1504 totally unexpected and applications may not behave correctly as a result. The
1505 exact behaviour will depend on the application but it could result in
1506 crashes, infinite loops or other similar incorrect responses.
1508 This issue is made more serious in combination with a separate bug in OpenSSL
1509 3.0 that will cause X509_verify_cert() to indicate an internal error when
1510 processing a certificate chain. This will occur where a certificate does not
1511 include the Subject Alternative Name extension but where a Certificate
1512 Authority has enforced name constraints. This issue can occur even with valid
1518 * Corrected a few file name and file reference bugs in the build,
1519 installation and setup scripts, which lead to installation verification
1520 failures. Slightly enhanced the installation verification script.
1524 * Fixed EVP_PKEY_eq() to make it possible to use it with strictly private
1529 * Fixed PVK encoder to properly query for the passphrase.
1533 * Multiple fixes in the OSSL_HTTP API functions.
1537 * Allow sign extension in OSSL_PARAM_allocate_from_text() for the
1538 OSSL_PARAM_INTEGER data type and return error on negative numbers
1539 used with the OSSL_PARAM_UNSIGNED_INTEGER data type. Make
1540 OSSL_PARAM_BLD_push_BN{,_pad}() return an error on negative numbers.
1544 * Allow copying uninitialized digest contexts with EVP_MD_CTX_copy_ex.
1548 * Fixed detection of ARMv7 and ARM64 CPU features on FreeBSD.
1552 * Multiple threading fixes.
1556 * Added NULL digest implementation to keep compatibility with 1.1.1 version.
1560 * Allow fetching an operation from the provider that owns an unexportable key
1561 as a fallback if that is still allowed by the property query.
1565 ### Changes between 1.1.1 and 3.0.0 [7 Sep 2021]
1567 * TLS_MAX_VERSION, DTLS_MAX_VERSION and DTLS_MIN_VERSION constants are now
1572 * The `OPENSSL_s390xcap` environment variable can be used to set bits in the
1573 S390X capability vector to zero. This simplifies testing of different code
1574 paths on S390X architecture.
1578 * Encrypting more than 2^64 TLS records with AES-GCM is disallowed
1579 as per FIPS 140-2 IG A.5 "Key/IV Pair Uniqueness Requirements from
1580 SP 800-38D". The communication will fail at this point.
1584 * The EC_GROUP_clear_free() function is deprecated as there is nothing
1585 confidential in EC_GROUP data.
1589 * The byte order mark (BOM) character is ignored if encountered at the
1590 beginning of a PEM-formatted file.
1594 * Added CMS support for the Russian GOST algorithms.
1598 * Due to move of the implementation of cryptographic operations
1599 to the providers, validation of various operation parameters can
1600 be postponed until the actual operation is executed where previously
1601 it happened immediately when an operation parameter was set.
1603 For example when setting an unsupported curve with
1604 EVP_PKEY_CTX_set_ec_paramgen_curve_nid() this function call will not
1605 fail but later keygen operations with the EVP_PKEY_CTX will fail.
1607 *OpenSSL team members and many third party contributors*
1609 * The EVP_get_cipherbyname() function will return NULL for algorithms such as
1610 "AES-128-SIV", "AES-128-CBC-CTS" and "CAMELLIA-128-CBC-CTS" which were
1611 previously only accessible via low-level interfaces. Use EVP_CIPHER_fetch()
1612 instead to retrieve these algorithms from a provider.
1616 * On build targets where the multilib postfix is set in the build
1617 configuration the libdir directory was changing based on whether
1618 the lib directory with the multilib postfix exists on the system
1619 or not. This unpredictable behavior was removed and eventual
1620 multilib postfix is now always added to the default libdir. Use
1621 `--libdir=lib` to override the libdir if adding the postfix is
1626 * The triple DES key wrap functionality now conforms to RFC 3217 but is
1627 no longer interoperable with OpenSSL 1.1.1.
1631 * The ERR_GET_FUNC() function was removed. With the loss of meaningful
1632 function codes, this function can only cause problems for calling
1637 * Add a configurable flag to output date formats as ISO 8601. Does not
1638 change the default date format.
1642 * Version of MSVC earlier than 1300 could get link warnings, which could
1643 be suppressed if the undocumented -DI_CAN_LIVE_WITH_LNK4049 was set.
1644 Support for this flag has been removed.
1648 * Rework and make DEBUG macros consistent. Remove unused -DCONF_DEBUG,
1649 -DBN_CTX_DEBUG, and REF_PRINT. Add a new tracing category and use it for
1650 printing reference counts. Rename -DDEBUG_UNUSED to -DUNUSED_RESULT_DEBUG
1651 Fix BN_DEBUG_RAND so it compiles and, when set, force DEBUG_RAND to be set
1652 also. Rename engine_debug_ref to be ENGINE_REF_PRINT also for consistency.
1656 * The signatures of the functions to get and set options on SSL and
1657 SSL_CTX objects changed from "unsigned long" to "uint64_t" type.
1658 Some source code changes may be required.
1662 * The public definitions of conf_method_st and conf_st have been
1663 deprecated. They will be made opaque in a future release.
1665 *Rich Salz and Tomáš Mráz*
1667 * Client-initiated renegotiation is disabled by default. To allow it, use
1668 the -client_renegotiation option, the SSL_OP_ALLOW_CLIENT_RENEGOTIATION
1669 flag, or the "ClientRenegotiation" config parameter as appropriate.
1673 * Add "abspath" and "includedir" pragma's to config files, to prevent,
1674 or modify relative pathname inclusion.
1678 * OpenSSL includes a cryptographic module that is intended to be FIPS 140-2
1679 validated. Please consult the README-FIPS and
1680 README-PROVIDERS files, as well as the migration guide.
1682 *OpenSSL team members and many third party contributors*
1684 * For the key types DH and DHX the allowed settable parameters are now different.
1688 * The openssl commands that read keys, certificates, and CRLs now
1689 automatically detect the PEM or DER format of the input files.
1691 *David von Oheimb, Richard Levitte, and Tomáš Mráz*
1693 * Added enhanced PKCS#12 APIs which accept a library context.
1697 * The default manual page suffix ($MANSUFFIX) has been changed to "ossl"
1701 * Added support for Kernel TLS (KTLS).
1703 *Boris Pismenny, John Baldwin and Andrew Gallatin*
1705 * Support for RFC 5746 secure renegotiation is now required by default for
1706 SSL or TLS connections to succeed.
1710 * The signature of the `copy` functional parameter of the
1711 EVP_PKEY_meth_set_copy() function has changed so its `src` argument is
1712 now `const EVP_PKEY_CTX *` instead of `EVP_PKEY_CTX *`. Similarly
1713 the signature of the `pub_decode` functional parameter of the
1714 EVP_PKEY_asn1_set_public() function has changed so its `pub` argument is
1715 now `const X509_PUBKEY *` instead of `X509_PUBKEY *`.
1719 * The error return values from some control calls (ctrl) have changed.
1723 * A public key check is now performed during EVP_PKEY_derive_set_peer().
1727 * Many functions in the EVP_ namespace that are getters of values from
1728 implementations or contexts were renamed to include get or get0 in their
1729 names. Old names are provided as macro aliases for compatibility and
1734 * The EVP_PKEY_CTRL_PKCS7_ENCRYPT, EVP_PKEY_CTRL_PKCS7_DECRYPT,
1735 EVP_PKEY_CTRL_PKCS7_SIGN, EVP_PKEY_CTRL_CMS_ENCRYPT,
1736 EVP_PKEY_CTRL_CMS_DECRYPT, and EVP_PKEY_CTRL_CMS_SIGN control operations
1741 * The EVP_PKEY_public_check() and EVP_PKEY_param_check() functions now work for
1744 * The output from the command line applications may have minor
1749 * The output from numerous "printing" may have minor changes.
1753 * Windows thread synchronization uses read/write primitives (SRWLock) when
1754 supported by the OS, otherwise CriticalSection continues to be used.
1758 * Add filter BIO BIO_f_readbuffer() that allows BIO_tell() and BIO_seek() to
1759 work on read only BIO source/sinks that do not support these functions.
1760 This allows piping or redirection of a file BIO using stdin to be buffered
1761 into memory. This is used internally in OSSL_DECODER_from_bio().
1765 * OSSL_STORE_INFO_get_type() may now return an additional value. In 1.1.1
1766 this function would return one of the values OSSL_STORE_INFO_NAME,
1767 OSSL_STORE_INFO_PKEY, OSSL_STORE_INFO_PARAMS, OSSL_STORE_INFO_CERT or
1768 OSSL_STORE_INFO_CRL. Decoded public keys would previously have been reported
1769 as type OSSL_STORE_INFO_PKEY in 1.1.1. In 3.0 decoded public keys are now
1770 reported as having the new type OSSL_STORE_INFO_PUBKEY. Applications
1771 using this function should be amended to handle the changed return value.
1775 * Improved adherence to Enhanced Security Services (ESS, RFC 2634 and RFC 5035)
1776 for the TSP and CMS Advanced Electronic Signatures (CAdES) implementations.
1777 As required by RFC 5035 check both ESSCertID and ESSCertIDv2 if both present.
1778 Correct the semantics of checking the validation chain in case ESSCertID{,v2}
1779 contains more than one certificate identifier: This means that all
1780 certificates referenced there MUST be part of the validation chain.
1784 * The implementation of older EVP ciphers related to CAST, IDEA, SEED, RC2, RC4,
1785 RC5, DESX and DES have been moved to the legacy provider.
1789 * The implementation of the EVP digests MD2, MD4, MDC2, WHIRLPOOL and
1790 RIPEMD-160 have been moved to the legacy provider.
1794 * The deprecated function EVP_PKEY_get0() now returns NULL being called for a
1799 * The deprecated functions EVP_PKEY_get0_RSA(),
1800 EVP_PKEY_get0_DSA(), EVP_PKEY_get0_EC_KEY(), EVP_PKEY_get0_DH(),
1801 EVP_PKEY_get0_hmac(), EVP_PKEY_get0_poly1305() and EVP_PKEY_get0_siphash() as
1802 well as the similarly named "get1" functions behave differently in
1807 * A number of functions handling low-level keys or engines were deprecated
1808 including EVP_PKEY_set1_engine(), EVP_PKEY_get0_engine(), EVP_PKEY_assign(),
1809 EVP_PKEY_get0(), EVP_PKEY_get0_hmac(), EVP_PKEY_get0_poly1305() and
1810 EVP_PKEY_get0_siphash().
1814 * PKCS#5 PBKDF1 key derivation has been moved from PKCS5_PBE_keyivgen() into
1815 the legacy crypto provider as an EVP_KDF. Applications requiring this KDF
1816 will need to load the legacy crypto provider. This includes these PBE
1817 algorithms which use this KDF:
1818 - NID_pbeWithMD2AndDES_CBC
1819 - NID_pbeWithMD5AndDES_CBC
1820 - NID_pbeWithSHA1AndRC2_CBC
1821 - NID_pbeWithMD2AndRC2_CBC
1822 - NID_pbeWithMD5AndRC2_CBC
1823 - NID_pbeWithSHA1AndDES_CBC
1827 * Deprecated obsolete BIO_set_callback(), BIO_get_callback(), and
1828 BIO_debug_callback() functions.
1832 * Deprecated obsolete EVP_PKEY_CTX_get0_dh_kdf_ukm() and
1833 EVP_PKEY_CTX_get0_ecdh_kdf_ukm() functions.
1837 * The RAND_METHOD APIs have been deprecated.
1841 * The SRP APIs have been deprecated.
1845 * Add a compile time option to prevent the caching of provider fetched
1846 algorithms. This is enabled by including the no-cached-fetch option
1847 at configuration time.
1851 * pkcs12 now uses defaults of PBKDF2, AES and SHA-256, with a MAC iteration
1852 count of PKCS12_DEFAULT_ITER.
1854 *Tomáš Mráz and Sahana Prasad*
1856 * The openssl speed command does not use low-level API calls anymore.
1860 * Parallel dual-prime 1024-bit modular exponentiation for AVX512_IFMA
1863 *Ilya Albrekht, Sergey Kirillov, Andrey Matyukov (Intel Corp)*
1865 * Combining the Configure options no-ec and no-dh no longer disables TLSv1.3.
1869 * Implemented support for fully "pluggable" TLSv1.3 groups. This means that
1870 providers may supply their own group implementations (using either the "key
1871 exchange" or the "key encapsulation" methods) which will automatically be
1872 detected and used by libssl.
1874 *Matt Caswell, Nicola Tuveri*
1876 * The undocumented function X509_certificate_type() has been deprecated;
1880 * Deprecated the obsolete BN_pseudo_rand() and BN_pseudo_rand_range().
1884 * Removed RSA padding mode for SSLv23 (which was only used for
1885 SSLv2). This includes the functions RSA_padding_check_SSLv23() and
1886 RSA_padding_add_SSLv23() and the `-ssl` option in the deprecated
1891 * Deprecated the obsolete X9.31 RSA key generation related functions.
1893 * While a callback function set via `SSL_CTX_set_cert_verify_callback()`
1894 is not allowed to return a value > 1, this is no more taken as failure.
1896 *Viktor Dukhovni and David von Oheimb*
1898 * Deprecated the obsolete X9.31 RSA key generation related functions
1899 BN_X931_generate_Xpq(), BN_X931_derive_prime_ex(), and
1900 BN_X931_generate_prime_ex().
1904 * The default key generation method for the regular 2-prime RSA keys was
1905 changed to the FIPS 186-4 B.3.6 method.
1909 * Deprecated the BN_is_prime_ex() and BN_is_prime_fasttest_ex() functions.
1913 * Deprecated EVP_MD_CTX_set_update_fn() and EVP_MD_CTX_update_fn().
1917 * Deprecated the type OCSP_REQ_CTX and the functions OCSP_REQ_CTX_*() and
1918 replaced with OSSL_HTTP_REQ_CTX and the functions OSSL_HTTP_REQ_CTX_*().
1920 *Rich Salz, Richard Levitte, and David von Oheimb*
1922 * Deprecated `X509_http_nbio()` and `X509_CRL_http_nbio()`.
1926 * Deprecated `OCSP_parse_url()`.
1930 * Validation of SM2 keys has been separated from the validation of regular EC
1935 * Behavior of the `pkey` app is changed, when using the `-check` or `-pubcheck`
1936 switches: a validation failure triggers an early exit, returning a failure
1937 exit status to the parent process.
1941 * Changed behavior of SSL_CTX_set_ciphersuites() and SSL_set_ciphersuites()
1942 to ignore unknown ciphers.
1946 * The `-cipher-commands` and `-digest-commands` options
1947 of the command line utility `list` have been deprecated.
1948 Instead use the `-cipher-algorithms` and `-digest-algorithms` options.
1952 * Added convenience functions for generating asymmetric key pairs:
1953 The 'quick' one-shot (yet somewhat limited) function L<EVP_PKEY_Q_keygen(3)>
1954 and macros for the most common cases: <EVP_RSA_gen(3)> and L<EVP_EC_gen(3)>.
1958 * All of the low-level EC_KEY functions have been deprecated.
1960 *Shane Lontis, Paul Dale, Richard Levitte, and Tomáš Mráz*
1962 * Deprecated all the libcrypto and libssl error string loading
1967 * The functions SSL_CTX_set_tmp_dh_callback and SSL_set_tmp_dh_callback, as
1968 well as the macros SSL_CTX_set_tmp_dh() and SSL_set_tmp_dh() have been
1973 * The `-crypt` option to the `passwd` command line tool has been removed.
1977 * The -C option to the `x509`, `dhparam`, `dsaparam`, and `ecparam` commands
1982 * Add support for AES Key Wrap inverse ciphers to the EVP layer.
1986 * Deprecated EVP_PKEY_set1_tls_encodedpoint() and
1987 EVP_PKEY_get1_tls_encodedpoint().
1991 * The security callback, which can be customised by application code, supports
1992 the security operation SSL_SECOP_TMP_DH. One location of the "other" parameter
1993 was incorrectly passing a DH object. It now passed an EVP_PKEY in all cases.
1997 * Add PKCS7_get_octet_string() and PKCS7_type_is_other() to the public
1998 interface. Their functionality remains unchanged.
2002 * Added new option for 'openssl list', '-providers', which will display the
2003 list of loaded providers, their names, version and status. It optionally
2004 displays their gettable parameters.
2008 * Removed EVP_PKEY_set_alias_type().
2012 * Deprecated `EVP_PKEY_CTX_set_rsa_keygen_pubexp()` and introduced
2013 `EVP_PKEY_CTX_set1_rsa_keygen_pubexp()`, which is now preferred.
2017 * Changed all "STACK" functions to be macros instead of inline functions. Macro
2018 parameters are still checked for type safety at compile time via helper
2023 * Remove the RAND_DRBG API
2025 *Paul Dale and Matthias St. Pierre*
2027 * Allow `SSL_set1_host()` and `SSL_add1_host()` to take IP literal addresses
2028 as well as actual hostnames.
2032 * The 'MinProtocol' and 'MaxProtocol' configuration commands now silently
2033 ignore TLS protocol version bounds when configuring DTLS-based contexts, and
2034 conversely, silently ignore DTLS protocol version bounds when configuring
2035 TLS-based contexts. The commands can be repeated to set bounds of both
2036 types. The same applies with the corresponding "min_protocol" and
2037 "max_protocol" command-line switches, in case some application uses both TLS
2040 SSL_CTX instances that are created for a fixed protocol version (e.g.
2041 `TLSv1_server_method()`) also silently ignore version bounds. Previously
2042 attempts to apply bounds to these protocol versions would result in an
2043 error. Now only the "version-flexible" SSL_CTX instances are subject to
2044 limits in configuration files in command-line options.
2048 * Deprecated the `ENGINE` API. Engines should be replaced with providers
2053 * Reworked the recorded ERR codes to make better space for system errors.
2054 To distinguish them, the macro `ERR_SYSTEM_ERROR()` indicates if the
2055 given code is a system error (true) or an OpenSSL error (false).
2059 * Reworked the test perl framework to better allow parallel testing.
2061 *Nicola Tuveri and David von Oheimb*
2063 * Added ciphertext stealing algorithms AES-128-CBC-CTS, AES-192-CBC-CTS and
2064 AES-256-CBC-CTS to the providers. CS1, CS2 and CS3 variants are supported.
2068 * 'Configure' has been changed to figure out the configuration target if
2069 none is given on the command line. Consequently, the 'config' script is
2070 now only a mere wrapper. All documentation is changed to only mention
2073 *Rich Salz and Richard Levitte*
2075 * Added a library context `OSSL_LIB_CTX` that applications as well as
2076 other libraries can use to form a separate context within which
2077 libcrypto operations are performed.
2081 * Added various `_ex` functions to the OpenSSL API that support using
2082 a non-default `OSSL_LIB_CTX`.
2086 * Handshake now fails if Extended Master Secret extension is dropped
2091 * Dropped interactive mode from the `openssl` program.
2095 * Deprecated `EVP_PKEY_cmp()` and `EVP_PKEY_cmp_parameters()`.
2097 *David von Oheimb and Shane Lontis*
2099 * Deprecated `EC_METHOD_get_field_type()`.
2103 * Deprecated EC_GFp_simple_method(), EC_GFp_mont_method(),
2104 EC_GF2m_simple_method(), EC_GFp_nist_method(), EC_GFp_nistp224_method()
2105 EC_GFp_nistp256_method(), and EC_GFp_nistp521_method().
2109 * Deprecated EC_GROUP_new(), EC_GROUP_method_of(), and EC_POINT_method_of().
2113 * Add CAdES-BES signature verification support, mostly derived
2114 from ESSCertIDv2 TS (RFC 5816) contribution by Marek Klein.
2116 *Filipe Raimundo da Silva*
2118 * Add CAdES-BES signature scheme and attributes support (RFC 5126) to CMS API.
2122 * Added the AuthEnvelopedData content type structure (RFC 5083) with AES-GCM
2123 parameter (RFC 5084) for the Cryptographic Message Syntax (CMS).
2127 * Deprecated EC_POINT_make_affine() and EC_POINTs_make_affine().
2131 * Deprecated EC_GROUP_precompute_mult(), EC_GROUP_have_precompute_mult(), and
2132 EC_KEY_precompute_mult().
2136 * Deprecated EC_POINTs_mul().
2140 * Removed FIPS_mode() and FIPS_mode_set().
2144 * The SSL option SSL_OP_IGNORE_UNEXPECTED_EOF is introduced.
2148 * Deprecated EC_POINT_set_Jprojective_coordinates_GFp() and
2149 EC_POINT_get_Jprojective_coordinates_GFp().
2153 * Added OSSL_PARAM_BLD to the public interface. This allows OSSL_PARAM
2154 arrays to be more easily constructed via a series of utility functions.
2155 Create a parameter builder using OSSL_PARAM_BLD_new(), add parameters using
2156 the various push functions and finally convert to a passable OSSL_PARAM
2157 array using OSSL_PARAM_BLD_to_param().
2161 * The security strength of SHA1 and MD5 based signatures in TLS has been
2166 * Added EVP_PKEY_set_type_by_keymgmt(), to initialise an EVP_PKEY to
2167 contain a provider side internal key.
2171 * ASN1_verify(), ASN1_digest() and ASN1_sign() have been deprecated.
2175 * Project text documents not yet having a proper file name extension
2176 (`HACKING`, `LICENSE`, `NOTES*`, `README*`, `VERSION`) have been renamed to
2177 `*.md` as far as reasonable, else `*.txt`, for better use with file managers.
2181 * The main project documents (README, NEWS, CHANGES, INSTALL, SUPPORT)
2182 have been converted to Markdown with the goal to produce documents
2183 which not only look pretty when viewed online in the browser, but
2184 remain well readable inside a plain text editor.
2186 To achieve this goal, a 'minimalistic' Markdown style has been applied
2187 which avoids formatting elements that interfere too much with the
2188 reading flow in the text file. For example, it
2190 * avoids [ATX headings][] and uses [setext headings][] instead
2191 (which works for `<h1>` and `<h2>` headings only).
2192 * avoids [inline links][] and uses [reference links][] instead.
2193 * avoids [fenced code blocks][] and uses [indented code blocks][] instead.
2195 [ATX headings]: https://github.github.com/gfm/#atx-headings
2196 [setext headings]: https://github.github.com/gfm/#setext-headings
2197 [inline links]: https://github.github.com/gfm/#inline-link
2198 [reference links]: https://github.github.com/gfm/#reference-link
2199 [fenced code blocks]: https://github.github.com/gfm/#fenced-code-blocks
2200 [indented code blocks]: https://github.github.com/gfm/#indented-code-blocks
2202 *Matthias St. Pierre*
2204 * The test suite is changed to preserve results of each test recipe.
2205 A new directory test-runs/ with subdirectories named like the
2206 test recipes are created in the build tree for this purpose.
2210 * Added an implementation of CMP and CRMF (RFC 4210, RFC 4211 RFC 6712).
2211 This adds `crypto/cmp/`, `crpyto/crmf/`, `apps/cmp.c`, and `test/cmp_*`.
2212 See L<openssl-cmp(1)> and L<OSSL_CMP_exec_IR_ses(3)> as starting points.
2214 *David von Oheimb, Martin Peylo*
2216 * Generalized the HTTP client code from `crypto/ocsp/` into `crpyto/http/`.
2217 It supports arbitrary request and response content types, GET redirection,
2218 TLS, connections via HTTP(S) proxies, connections and exchange via
2219 user-defined BIOs (allowing implicit connections), persistent connections,
2220 and timeout checks. See L<OSSL_HTTP_transfer(3)> etc. for details.
2221 The legacy OCSP-focused (and only partly documented) API
2222 is retained for backward compatibility, while most of it is deprecated.
2226 * Added `util/check-format.pl`, a tool for checking adherence to the
2227 OpenSSL coding style <https://www.openssl.org/policies/codingstyle.html>.
2228 The checks performed are incomplete and yield some false positives.
2229 Still the tool should be useful for detecting most typical glitches.
2233 * `BIO_do_connect()` and `BIO_do_handshake()` have been extended:
2234 If domain name resolution yields multiple IP addresses all of them are tried
2235 after `connect()` failures.
2239 * All of the low-level RSA functions have been deprecated.
2243 * X509 certificates signed using SHA1 are no longer allowed at security
2248 * The command line utilities dhparam, dsa, gendsa and dsaparam have been
2249 modified to use PKEY APIs. These commands are now in maintenance mode
2250 and no new features will be added to them.
2254 * The command line utility rsautl has been deprecated.
2258 * The command line utilities genrsa and rsa have been modified to use PKEY
2259 APIs. They now write PKCS#8 keys by default. These commands are now in
2260 maintenance mode and no new features will be added to them.
2264 * All of the low-level DH functions have been deprecated.
2266 *Paul Dale and Matt Caswell*
2268 * All of the low-level DSA functions have been deprecated.
2272 * Reworked the treatment of EC EVP_PKEYs with the SM2 curve to
2273 automatically become EVP_PKEY_SM2 rather than EVP_PKEY_EC.
2277 * Deprecated low-level ECDH and ECDSA functions.
2281 * Deprecated EVP_PKEY_decrypt_old() and EVP_PKEY_encrypt_old().
2285 * Enhanced the documentation of EVP_PKEY_get_size(), EVP_PKEY_get_bits()
2286 and EVP_PKEY_get_security_bits(). Especially EVP_PKEY_get_size() needed
2287 a new formulation to include all the things it can be used for,
2288 as well as words of caution.
2292 * The SSL_CTX_set_tlsext_ticket_key_cb(3) function has been deprecated.
2296 * All of the low-level HMAC functions have been deprecated.
2298 *Paul Dale and David von Oheimb*
2300 * Over two thousand fixes were made to the documentation, including:
2301 - Common options (such as -rand/-writerand, TLS version control, etc)
2302 were refactored and point to newly-enhanced descriptions in openssl.pod.
2303 - Added style conformance for all options (with help from Richard Levitte),
2304 documented all reported missing options, added a CI build to check
2305 that all options are documented and that no unimplemented options
2307 - Documented some internals, such as all use of environment variables.
2308 - Addressed all internal broken L<> references.
2312 * All of the low-level CMAC functions have been deprecated.
2316 * The low-level MD2, MD4, MD5, MDC2, RIPEMD160 and Whirlpool digest
2317 functions have been deprecated.
2319 *Paul Dale and David von Oheimb*
2321 * Corrected the documentation of the return values from the `EVP_DigestSign*`
2322 set of functions. The documentation mentioned negative values for some
2323 errors, but this was never the case, so the mention of negative values
2326 Code that followed the documentation and thereby check with something
2327 like `EVP_DigestSignInit(...) <= 0` will continue to work undisturbed.
2331 * All of the low-level cipher functions have been deprecated.
2333 *Matt Caswell and Paul Dale*
2335 * Removed include/openssl/opensslconf.h.in and replaced it with
2336 include/openssl/configuration.h.in, which differs in not including
2337 <openssl/macros.h>. A short header include/openssl/opensslconf.h
2338 was added to include both.
2340 This allows internal hacks where one might need to modify the set
2341 of configured macros, for example this if deprecated symbols are
2342 still supposed to be available internally:
2344 #include <openssl/configuration.h>
2346 #undef OPENSSL_NO_DEPRECATED
2347 #define OPENSSL_SUPPRESS_DEPRECATED
2349 #include <openssl/macros.h>
2351 This should not be used by applications that use the exported
2352 symbols, as that will lead to linking errors.
2356 * Fixed an overflow bug in the x64_64 Montgomery squaring procedure
2357 used in exponentiation with 512-bit moduli. No EC algorithms are
2358 affected. Analysis suggests that attacks against 2-prime RSA1024,
2359 3-prime RSA1536, and DSA1024 as a result of this defect would be very
2360 difficult to perform and are not believed likely. Attacks against DH512
2361 are considered just feasible. However, for an attack the target would
2362 have to reuse the DH512 private key, which is not recommended anyway.
2363 Also applications directly using the low-level API BN_mod_exp may be
2364 affected if they use BN_FLG_CONSTTIME.
2369 * Most memory-debug features have been deprecated, and the functionality
2370 replaced with no-ops.
2374 * Added documentation for the STACK API.
2378 * Introduced a new method type and API, OSSL_ENCODER, to represent
2379 generic encoders. These do the same sort of job that PEM writers
2380 and d2i functions do, but with support for methods supplied by
2381 providers, and the possibility for providers to support other
2386 * Introduced a new method type and API, OSSL_DECODER, to represent
2387 generic decoders. These do the same sort of job that PEM readers
2388 and i2d functions do, but with support for methods supplied by
2389 providers, and the possibility for providers to support other
2394 * Added a .pragma directive to the syntax of configuration files, to
2395 allow varying behavior in a supported and predictable manner.
2396 Currently added pragma:
2400 This allows dollar signs to be a keyword character unless it's
2401 followed by a opening brace or parenthesis. This is useful for
2402 platforms where dollar signs are commonly used in names, such as
2403 volume names and system directory names on VMS.
2407 * Added functionality to create an EVP_PKEY from user data.
2411 * Change the interpretation of the '--api' configuration option to
2412 mean that this is a desired API compatibility level with no
2413 further meaning. The previous interpretation, that this would
2414 also mean to remove all deprecated symbols up to and including
2415 the given version, no requires that 'no-deprecated' is also used
2416 in the configuration.
2418 When building applications, the desired API compatibility level
2419 can be set with the OPENSSL_API_COMPAT macro like before. For
2420 API compatibility version below 3.0, the old style numerical
2421 value is valid as before, such as -DOPENSSL_API_COMPAT=0x10100000L.
2422 For version 3.0 and on, the value is expected to be the decimal
2423 value calculated from the major and minor version like this:
2425 MAJOR * 10000 + MINOR * 100
2429 -DOPENSSL_API_COMPAT=30000 For 3.0
2430 -DOPENSSL_API_COMPAT=30200 For 3.2
2432 To hide declarations that are deprecated up to and including the
2433 given API compatibility level, -DOPENSSL_NO_DEPRECATED must be
2434 given when building the application as well.
2438 * Added the X509_LOOKUP_METHOD called X509_LOOKUP_store, to allow
2439 access to certificate and CRL stores via URIs and OSSL_STORE
2442 This adds the following functions:
2444 - X509_LOOKUP_store()
2445 - X509_STORE_load_file()
2446 - X509_STORE_load_path()
2447 - X509_STORE_load_store()
2448 - SSL_add_store_cert_subjects_to_stack()
2449 - SSL_CTX_set_default_verify_store()
2450 - SSL_CTX_load_verify_file()
2451 - SSL_CTX_load_verify_dir()
2452 - SSL_CTX_load_verify_store()
2456 * Added a new method to gather entropy on VMS, based on SYS$GET_ENTROPY.
2457 The presence of this system service is determined at run-time.
2461 * Added functionality to create an EVP_PKEY context based on data
2462 for methods from providers. This takes an algorithm name and a
2463 property query string and simply stores them, with the intent
2464 that any operation that uses this context will use those strings
2465 to fetch the needed methods implicitly, thereby making the port
2466 of application written for pre-3.0 OpenSSL easier.
2470 * The undocumented function NCONF_WIN32() has been deprecated; for
2471 conversion details see the HISTORY section of doc/man5/config.pod
2475 * Introduced the new functions EVP_DigestSignInit_ex() and
2476 EVP_DigestVerifyInit_ex(). The macros EVP_DigestSignUpdate() and
2477 EVP_DigestVerifyUpdate() have been converted to functions. See the man
2478 pages for further details.
2482 * Over two thousand fixes were made to the documentation, including:
2483 adding missing command flags, better style conformance, documentation
2486 *Rich Salz, Richard Levitte*
2488 * s390x assembly pack: add hardware-support for P-256, P-384, P-521,
2489 X25519, X448, Ed25519 and Ed448.
2493 * Print all values for a PKCS#12 attribute with 'openssl pkcs12', not just
2498 * Deprecated the public definition of `ERR_STATE` as well as the function
2499 `ERR_get_state()`. This is done in preparation of making `ERR_STATE` an
2504 * Added ERR functionality to give callers access to the stored function
2505 names that have replaced the older function code based functions.
2507 New functions are ERR_peek_error_func(), ERR_peek_last_error_func(),
2508 ERR_peek_error_data(), ERR_peek_last_error_data(), ERR_get_error_all(),
2509 ERR_peek_error_all() and ERR_peek_last_error_all().
2511 Deprecate ERR functions ERR_get_error_line(), ERR_get_error_line_data(),
2512 ERR_peek_error_line_data(), ERR_peek_last_error_line_data() and
2513 ERR_func_error_string().
2517 * Extended testing to be verbose for failing tests only. The make variables
2518 VERBOSE_FAILURE or VF can be used to enable this:
2520 $ make VF=1 test # Unix
2521 $ mms /macro=(VF=1) test ! OpenVMS
2522 $ nmake VF=1 test # Windows
2526 * Added the `-copy_extensions` option to the `x509` command for use with
2527 `-req` and `-x509toreq`. When given with the `copy` or `copyall` argument,
2528 all extensions in the request are copied to the certificate or vice versa.
2530 *David von Oheimb*, *Kirill Stefanenkov <kirill_stefanenkov@rambler.ru>*
2532 * Added the `-copy_extensions` option to the `req` command for use with
2533 `-x509`. When given with the `copy` or `copyall` argument,
2534 all extensions in the certification request are copied to the certificate.
2538 * The `x509`, `req`, and `ca` commands now make sure that X.509v3 certificates
2539 they generate are by default RFC 5280 compliant in the following sense:
2540 There is a subjectKeyIdentifier extension with a hash value of the public key
2541 and for not self-signed certs there is an authorityKeyIdentifier extension
2542 with a keyIdentifier field or issuer information identifying the signing key.
2543 This is done unless some configuration overrides the new default behavior,
2544 such as `subjectKeyIdentifier = none` and `authorityKeyIdentifier = none`.
2548 * Added several checks to `X509_verify_cert()` according to requirements in
2549 RFC 5280 in case `X509_V_FLAG_X509_STRICT` is set
2550 (which may be done by using the CLI option `-x509_strict`):
2551 * The basicConstraints of CA certificates must be marked critical.
2552 * CA certificates must explicitly include the keyUsage extension.
2553 * If a pathlenConstraint is given the key usage keyCertSign must be allowed.
2554 * The issuer name of any certificate must not be empty.
2555 * The subject name of CA certs, certs with keyUsage crlSign,
2556 and certs without subjectAlternativeName must not be empty.
2557 * If a subjectAlternativeName extension is given it must not be empty.
2558 * The signatureAlgorithm field and the cert signature must be consistent.
2559 * Any given authorityKeyIdentifier and any given subjectKeyIdentifier
2560 must not be marked critical.
2561 * The authorityKeyIdentifier must be given for X.509v3 certs
2562 unless they are self-signed.
2563 * The subjectKeyIdentifier must be given for all X.509v3 CA certs.
2567 * Certificate verification using `X509_verify_cert()` meanwhile rejects EC keys
2568 with explicit curve parameters (specifiedCurve) as required by RFC 5480.
2572 * For built-in EC curves, ensure an EC_GROUP built from the curve name is
2573 used even when parsing explicit parameters, when loading a encoded key
2574 or calling `EC_GROUP_new_from_ecpkparameters()`/
2575 `EC_GROUP_new_from_ecparameters()`.
2576 This prevents bypass of security hardening and performance gains,
2577 especially for curves with specialized EC_METHODs.
2578 By default, if a key encoded with explicit parameters is loaded and later
2579 encoded, the output is still encoded with explicit parameters, even if
2580 internally a "named" EC_GROUP is used for computation.
2584 * Compute ECC cofactors if not provided during EC_GROUP construction. Before
2585 this change, EC_GROUP_set_generator would accept order and/or cofactor as
2586 NULL. After this change, only the cofactor parameter can be NULL. It also
2587 does some minimal sanity checks on the passed order.
2592 * Fixed a padding oracle in PKCS7_dataDecode and CMS_decrypt_set1_pkey.
2593 An attack is simple, if the first CMS_recipientInfo is valid but the
2594 second CMS_recipientInfo is chosen ciphertext. If the second
2595 recipientInfo decodes to PKCS #1 v1.5 form plaintext, the correct
2596 encryption key will be replaced by garbage, and the message cannot be
2597 decoded, but if the RSA decryption fails, the correct encryption key is
2598 used and the recipient will not notice the attack.
2599 As a work around for this potential attack the length of the decrypted
2600 key must be equal to the cipher default key length, in case the
2601 certificate is not given and all recipientInfo are tried out.
2602 The old behaviour can be re-enabled in the CMS code by setting the
2603 CMS_DEBUG_DECRYPT flag.
2607 * Early start up entropy quality from the DEVRANDOM seed source has been
2608 improved for older Linux systems. The RAND subsystem will wait for
2609 /dev/random to be producing output before seeding from /dev/urandom.
2610 The seeded state is stored for future library initialisations using
2611 a system global shared memory segment. The shared memory identifier
2612 can be configured by defining OPENSSL_RAND_SEED_DEVRANDOM_SHM_ID to
2613 the desired value. The default identifier is 114.
2617 * Revised BN_generate_prime_ex to not avoid factors 2..17863 in p-1
2618 when primes for RSA keys are computed.
2619 Since we previously always generated primes == 2 (mod 3) for RSA keys,
2620 the 2-prime and 3-prime RSA modules were easy to distinguish, since
2621 `N = p*q = 1 (mod 3)`, but `N = p*q*r = 2 (mod 3)`. Therefore, fingerprinting
2622 2-prime vs. 3-prime RSA keys was possible by computing N mod 3.
2623 This avoids possible fingerprinting of newly generated RSA modules.
2627 * Correct the extended master secret constant on EBCDIC systems. Without this
2628 fix TLS connections between an EBCDIC system and a non-EBCDIC system that
2629 negotiate EMS will fail. Unfortunately this also means that TLS connections
2630 between EBCDIC systems with this fix, and EBCDIC systems without this
2631 fix will fail if they negotiate EMS.
2635 * Changed the library initialisation so that the config file is now loaded
2636 by default. This was already the case for libssl. It now occurs for both
2637 libcrypto and libssl. Use the OPENSSL_INIT_NO_LOAD_CONFIG option to
2638 `OPENSSL_init_crypto()` to suppress automatic loading of a config file.
2642 * Introduced new error raising macros, `ERR_raise()` and `ERR_raise_data()`,
2643 where the former acts as a replacement for `ERR_put_error()`, and the
2644 latter replaces the combination `ERR_put_error()` + `ERR_add_error_data()`.
2645 `ERR_raise_data()` adds more flexibility by taking a format string and
2646 an arbitrary number of arguments following it, to be processed with
2651 * Introduced a new function, `OSSL_PROVIDER_available()`, which can be used
2652 to check if a named provider is loaded and available. When called, it
2653 will also activate all fallback providers if such are still present.
2657 * Enforce a minimum DH modulus size of 512 bits.
2661 * Changed DH parameters to generate the order q subgroup instead of 2q.
2662 Previously generated DH parameters are still accepted by DH_check
2663 but DH_generate_key works around that by clearing bit 0 of the
2664 private key for those. This avoids leaking bit 0 of the private key.
2668 * Significantly reduce secure memory usage by the randomness pools.
2672 * `{CRYPTO,OPENSSL}_mem_debug_{push,pop}` are now no-ops and have been
2677 * A new type, EVP_KEYEXCH, has been introduced to represent key exchange
2678 algorithms. An implementation of a key exchange algorithm can be obtained
2679 by using the function EVP_KEYEXCH_fetch(). An EVP_KEYEXCH algorithm can be
2680 used in a call to EVP_PKEY_derive_init_ex() which works in a similar way to
2681 the older EVP_PKEY_derive_init() function. See the man pages for the new
2682 functions for further details.
2686 * The EVP_PKEY_CTX_set_dh_pad() macro has now been converted to a function.
2690 * Removed the function names from error messages and deprecated the
2695 * Removed NextStep support and the macro OPENSSL_UNISTD
2699 * Removed DES_check_key. Also removed OPENSSL_IMPLEMENT_GLOBAL,
2700 OPENSSL_GLOBAL_REF, OPENSSL_DECLARE_GLOBAL.
2701 Also removed "export var as function" capability; we do not export
2702 variables, only functions.
2706 * RC5_32_set_key has been changed to return an int type, with 0 indicating
2707 an error and 1 indicating success. In previous versions of OpenSSL this
2708 was a void type. If a key was set longer than the maximum possible this
2713 * Support SM2 signing and verification schemes with X509 certificate.
2717 * Use SHA256 as the default digest for TS query in the `ts` app.
2721 * Change PBKDF2 to conform to SP800-132 instead of the older PKCS5 RFC2898.
2725 * Default cipher lists/suites are now available via a function, the
2726 #defines are deprecated.
2730 * Add target VC-WIN32-UWP, VC-WIN64A-UWP, VC-WIN32-ARM-UWP and
2731 VC-WIN64-ARM-UWP in Windows OneCore target for making building libraries
2732 for Windows Store apps easier. Also, the "no-uplink" option has been added.
2736 * Join the directories crypto/x509 and crypto/x509v3
2740 * Added command 'openssl kdf' that uses the EVP_KDF API.
2744 * Added command 'openssl mac' that uses the EVP_MAC API.
2748 * Added OPENSSL_info() to get diverse built-in OpenSSL data, such
2749 as default directories. Also added the command 'openssl info'
2750 for scripting purposes.
2754 * The functions AES_ige_encrypt() and AES_bi_ige_encrypt() have been
2759 * Add prediction resistance to the DRBG reseeding process.
2763 * Limit the number of blocks in a data unit for AES-XTS to 2^20 as
2764 mandated by IEEE Std 1619-2018.
2768 * Added newline escaping functionality to a filename when using openssl dgst.
2769 This output format is to replicate the output format found in the `*sum`
2770 checksum programs. This aims to preserve backward compatibility.
2772 *Matt Eaton, Richard Levitte, and Paul Dale*
2774 * Removed the heartbeat message in DTLS feature, as it has very
2775 little usage and doesn't seem to fulfill a valuable purpose.
2776 The configuration option is now deprecated.
2780 * Changed the output of 'openssl {digestname} < file' to display the
2781 digest name in its output.
2785 * Added a new generic trace API which provides support for enabling
2786 instrumentation through trace output.
2788 *Richard Levitte & Matthias St. Pierre*
2790 * Added build tests for C++. These are generated files that only do one
2791 thing, to include one public OpenSSL head file each. This tests that
2792 the public header files can be usefully included in a C++ application.
2794 This test isn't enabled by default. It can be enabled with the option
2795 'enable-buildtest-c++'.
2799 * Added KB KDF (EVP_KDF_KB) to EVP_KDF.
2803 * Added SSH KDF (EVP_KDF_SSHKDF) and KRB5 KDF (EVP_KDF_KRB5KDF) to EVP_KDF.
2807 * Added Single Step KDF (EVP_KDF_SS), X963 KDF, and X942 KDF to EVP_KDF.
2811 * Added KMAC to EVP_MAC.
2815 * Added property based algorithm implementation selection framework to
2820 * Added SCA hardening for modular field inversion in EC_GROUP through
2821 a new dedicated field_inv() pointer in EC_METHOD.
2822 This also addresses a leakage affecting conversions from projective
2823 to affine coordinates.
2825 *Billy Bob Brumley, Nicola Tuveri*
2827 * Added EVP_KDF, an EVP layer KDF API, to simplify adding KDF and PRF
2828 implementations. This includes an EVP_PKEY to EVP_KDF bridge for
2829 those algorithms that were already supported through the EVP_PKEY API
2830 (scrypt, TLS1 PRF and HKDF). The low-level KDF functions for PBKDF2
2831 and scrypt are now wrappers that call EVP_KDF.
2835 * Build devcrypto engine as a dynamic engine.
2837 *Eneas U de Queiroz*
2839 * Add keyed BLAKE2 to EVP_MAC.
2843 * Fix a bug in the computation of the endpoint-pair shared secret used
2844 by DTLS over SCTP. This breaks interoperability with older versions
2845 of OpenSSL like OpenSSL 1.1.0 and OpenSSL 1.0.2. There is a runtime
2846 switch SSL_MODE_DTLS_SCTP_LABEL_LENGTH_BUG (off by default) enabling
2847 interoperability with such broken implementations. However, enabling
2848 this switch breaks interoperability with correct implementations.
2850 * Fix a use after free bug in d2i_X509_PUBKEY when overwriting a
2851 re-used X509_PUBKEY object if the second PUBKEY is malformed.
2855 * Move strictness check from EVP_PKEY_asn1_new() to EVP_PKEY_asn1_add0().
2859 * Changed the license to the Apache License v2.0.
2863 * Switch to a new version scheme using three numbers MAJOR.MINOR.PATCH.
2865 - Major releases (indicated by incrementing the MAJOR release number)
2866 may introduce incompatible API/ABI changes.
2867 - Minor releases (indicated by incrementing the MINOR release number)
2868 may introduce new features but retain API/ABI compatibility.
2869 - Patch releases (indicated by incrementing the PATCH number)
2870 are intended for bug fixes and other improvements of existing
2871 features only (like improving performance or adding documentation)
2872 and retain API/ABI compatibility.
2876 * Add support for RFC5297 SIV mode (siv128), including AES-SIV.
2880 * Remove the 'dist' target and add a tarball building script. The
2881 'dist' target has fallen out of use, and it shouldn't be
2882 necessary to configure just to create a source distribution.
2886 * Recreate the OS390-Unix config target. It no longer relies on a
2887 special script like it did for OpenSSL pre-1.1.0.
2891 * Instead of having the source directories listed in Configure, add
2892 a 'build.info' keyword SUBDIRS to indicate what sub-directories to
2897 * Add GMAC to EVP_MAC.
2901 * Ported the HMAC, CMAC and SipHash EVP_PKEY_METHODs to EVP_MAC.
2905 * Added EVP_MAC, an EVP layer MAC API, to simplify adding MAC
2906 implementations. This includes a generic EVP_PKEY to EVP_MAC bridge,
2907 to facilitate the continued use of MACs through raw private keys in
2908 functionality such as `EVP_DigestSign*` and `EVP_DigestVerify*`.
2912 * Deprecate ECDH_KDF_X9_62().
2916 * Added EVP_PKEY_ECDH_KDF_X9_63 and ecdh_KDF_X9_63() as replacements for
2917 the EVP_PKEY_ECDH_KDF_X9_62 KDF type and ECDH_KDF_X9_62(). The old names
2918 are retained for backwards compatibility.
2922 * AES-XTS mode now enforces that its two keys are different to mitigate
2923 the attacked described in "Efficient Instantiations of Tweakable
2924 Blockciphers and Refinements to Modes OCB and PMAC" by Phillip Rogaway.
2925 Details of this attack can be obtained from:
2926 <http://web.cs.ucdavis.edu/%7Erogaway/papers/offsets.pdf>
2930 * Rename the object files, i.e. give them other names than in previous
2931 versions. Their names now include the name of the final product, as
2932 well as its type mnemonic (bin, lib, shlib).
2936 * Added new option for 'openssl list', '-objects', which will display the
2937 list of built in objects, i.e. OIDs with names.
2941 * Added the options `-crl_lastupdate` and `-crl_nextupdate` to `openssl ca`,
2942 allowing the `lastUpdate` and `nextUpdate` fields in the generated CRL to
2947 * Added support for Linux Kernel TLS data-path. The Linux Kernel data-path
2948 improves application performance by removing data copies and providing
2949 applications with zero-copy system calls such as sendfile and splice.
2953 * The SSL option SSL_OP_CLEANSE_PLAINTEXT is introduced.
2957 * `PKCS12_parse` now maintains the order of the parsed certificates
2958 when outputting them via `*ca` (rather than reversing it).
2962 * Deprecated pthread fork support methods.
2966 * Added support for FFDHE key exchange in TLS 1.3.
2970 * Added a new concept for OpenSSL plugability: providers. This
2971 functionality is designed to replace the ENGINE API and ENGINE
2972 implementations, and to be much more dynamic, allowing provider
2973 authors to introduce new algorithms among other things, as long as
2974 there's an API that supports the algorithm type.
2976 With this concept comes a new core API for interaction between
2977 libcrypto and provider implementations. Public libcrypto functions
2978 that want to use providers do so through this core API.
2980 The main documentation for this core API is found in
2981 doc/man7/provider.pod, doc/man7/provider-base.pod, and they in turn
2982 refer to other manuals describing the API specific for supported
2983 algorithm types (also called operations).
2990 ### Changes between 1.1.1m and 1.1.1n [xx XXX xxxx]
2992 ### Changes between 1.1.1l and 1.1.1m [14 Dec 2021]
2994 * Avoid loading of a dynamic engine twice.
2998 * Prioritise DANE TLSA issuer certs over peer certs
3002 * Fixed random API for MacOS prior to 10.12
3004 These MacOS versions don't support the CommonCrypto APIs
3008 ### Changes between 1.1.1k and 1.1.1l [24 Aug 2021]
3010 * Fixed an SM2 Decryption Buffer Overflow.
3012 In order to decrypt SM2 encrypted data an application is expected to
3013 call the API function EVP_PKEY_decrypt(). Typically an application will
3014 call this function twice. The first time, on entry, the "out" parameter
3015 can be NULL and, on exit, the "outlen" parameter is populated with the
3016 buffer size required to hold the decrypted plaintext. The application
3017 can then allocate a sufficiently sized buffer and call EVP_PKEY_decrypt()
3018 again, but this time passing a non-NULL value for the "out" parameter.
3020 A bug in the implementation of the SM2 decryption code means that the
3021 calculation of the buffer size required to hold the plaintext returned
3022 by the first call to EVP_PKEY_decrypt() can be smaller than the actual
3023 size required by the second call. This can lead to a buffer overflow
3024 when EVP_PKEY_decrypt() is called by the application a second time with
3025 a buffer that is too small.
3027 A malicious attacker who is able present SM2 content for decryption to
3028 an application could cause attacker chosen data to overflow the buffer
3029 by up to a maximum of 62 bytes altering the contents of other data held
3030 after the buffer, possibly changing application behaviour or causing
3031 the application to crash. The location of the buffer is application
3032 dependent but is typically heap allocated.
3037 * Fixed various read buffer overruns processing ASN.1 strings
3039 ASN.1 strings are represented internally within OpenSSL as an ASN1_STRING
3040 structure which contains a buffer holding the string data and a field
3041 holding the buffer length. This contrasts with normal C strings which
3042 are represented as a buffer for the string data which is terminated
3043 with a NUL (0) byte.
3045 Although not a strict requirement, ASN.1 strings that are parsed using
3046 OpenSSL's own "d2i" functions (and other similar parsing functions) as
3047 well as any string whose value has been set with the ASN1_STRING_set()
3048 function will additionally NUL terminate the byte array in the
3049 ASN1_STRING structure.
3051 However, it is possible for applications to directly construct valid
3052 ASN1_STRING structures which do not NUL terminate the byte array by
3053 directly setting the "data" and "length" fields in the ASN1_STRING
3054 array. This can also happen by using the ASN1_STRING_set0() function.
3056 Numerous OpenSSL functions that print ASN.1 data have been found to
3057 assume that the ASN1_STRING byte array will be NUL terminated, even
3058 though this is not guaranteed for strings that have been directly
3059 constructed. Where an application requests an ASN.1 structure to be
3060 printed, and where that ASN.1 structure contains ASN1_STRINGs that have
3061 been directly constructed by the application without NUL terminating
3062 the "data" field, then a read buffer overrun can occur.
3064 The same thing can also occur during name constraints processing
3065 of certificates (for example if a certificate has been directly
3066 constructed by the application instead of loading it via the OpenSSL
3067 parsing functions, and the certificate contains non NUL terminated
3068 ASN1_STRING structures). It can also occur in the X509_get1_email(),
3069 X509_REQ_get1_email() and X509_get1_ocsp() functions.
3071 If a malicious actor can cause an application to directly construct an
3072 ASN1_STRING and then process it through one of the affected OpenSSL
3073 functions then this issue could be hit. This might result in a crash
3074 (causing a Denial of Service attack). It could also result in the
3075 disclosure of private memory contents (such as private keys, or
3076 sensitive plaintext).
3081 ### Changes between 1.1.1j and 1.1.1k [25 Mar 2021]
3083 * Fixed a problem with verifying a certificate chain when using the
3084 X509_V_FLAG_X509_STRICT flag. This flag enables additional security checks of
3085 the certificates present in a certificate chain. It is not set by default.
3087 Starting from OpenSSL version 1.1.1h a check to disallow certificates in
3088 the chain that have explicitly encoded elliptic curve parameters was added
3089 as an additional strict check.
3091 An error in the implementation of this check meant that the result of a
3092 previous check to confirm that certificates in the chain are valid CA
3093 certificates was overwritten. This effectively bypasses the check
3094 that non-CA certificates must not be able to issue other certificates.
3096 If a "purpose" has been configured then there is a subsequent opportunity
3097 for checks that the certificate is a valid CA. All of the named "purpose"
3098 values implemented in libcrypto perform this check. Therefore, where
3099 a purpose is set the certificate chain will still be rejected even when the
3100 strict flag has been used. A purpose is set by default in libssl client and
3101 server certificate verification routines, but it can be overridden or
3102 removed by an application.
3104 In order to be affected, an application must explicitly set the
3105 X509_V_FLAG_X509_STRICT verification flag and either not set a purpose
3106 for the certificate verification or, in the case of TLS client or server
3107 applications, override the default purpose.
3112 * Fixed an issue where an OpenSSL TLS server may crash if sent a maliciously
3113 crafted renegotiation ClientHello message from a client. If a TLSv1.2
3114 renegotiation ClientHello omits the signature_algorithms extension (where it
3115 was present in the initial ClientHello), but includes a
3116 signature_algorithms_cert extension then a NULL pointer dereference will
3117 result, leading to a crash and a denial of service attack.
3119 A server is only vulnerable if it has TLSv1.2 and renegotiation enabled
3120 (which is the default configuration). OpenSSL TLS clients are not impacted by
3124 *Peter Kästle and Samuel Sapalski*
3126 ### Changes between 1.1.1i and 1.1.1j [16 Feb 2021]
3128 * Fixed the X509_issuer_and_serial_hash() function. It attempts to
3129 create a unique hash value based on the issuer and serial number data
3130 contained within an X509 certificate. However, it was failing to correctly
3131 handle any errors that may occur while parsing the issuer field (which might
3132 occur if the issuer field is maliciously constructed). This may subsequently
3133 result in a NULL pointer deref and a crash leading to a potential denial of
3139 * Fixed the RSA_padding_check_SSLv23() function and the RSA_SSLV23_PADDING
3140 padding mode to correctly check for rollback attacks. This is considered a
3141 bug in OpenSSL 1.1.1 because it does not support SSLv2. In 1.0.2 this is
3146 Fixed the EVP_CipherUpdate, EVP_EncryptUpdate and EVP_DecryptUpdate
3147 functions. Previously they could overflow the output length argument in some
3148 cases where the input length is close to the maximum permissible length for
3149 an integer on the platform. In such cases the return value from the function
3150 call would be 1 (indicating success), but the output length value would be
3151 negative. This could cause applications to behave incorrectly or crash.
3156 * Fixed SRP_Calc_client_key so that it runs in constant time. The previous
3157 implementation called BN_mod_exp without setting BN_FLG_CONSTTIME. This
3158 could be exploited in a side channel attack to recover the password. Since
3159 the attack is local host only this is outside of the current OpenSSL
3160 threat model and therefore no CVE is assigned.
3162 Thanks to Mohammed Sabt and Daniel De Almeida Braga for reporting this
3167 ### Changes between 1.1.1h and 1.1.1i [8 Dec 2020]
3169 * Fixed NULL pointer deref in the GENERAL_NAME_cmp function
3170 This function could crash if both GENERAL_NAMEs contain an EDIPARTYNAME.
3171 If an attacker can control both items being compared then this could lead
3172 to a possible denial of service attack. OpenSSL itself uses the
3173 GENERAL_NAME_cmp function for two purposes:
3174 1) Comparing CRL distribution point names between an available CRL and a
3175 CRL distribution point embedded in an X509 certificate
3176 2) When verifying that a timestamp response token signer matches the
3177 timestamp authority name (exposed via the API functions
3178 TS_RESP_verify_response and TS_RESP_verify_token)
3183 ### Changes between 1.1.1g and 1.1.1h [22 Sep 2020]
3185 * Certificates with explicit curve parameters are now disallowed in
3186 verification chains if the X509_V_FLAG_X509_STRICT flag is used.
3190 * The 'MinProtocol' and 'MaxProtocol' configuration commands now silently
3191 ignore TLS protocol version bounds when configuring DTLS-based contexts, and
3192 conversely, silently ignore DTLS protocol version bounds when configuring
3193 TLS-based contexts. The commands can be repeated to set bounds of both
3194 types. The same applies with the corresponding "min_protocol" and
3195 "max_protocol" command-line switches, in case some application uses both TLS
3198 SSL_CTX instances that are created for a fixed protocol version (e.g.
3199 TLSv1_server_method()) also silently ignore version bounds. Previously
3200 attempts to apply bounds to these protocol versions would result in an
3201 error. Now only the "version-flexible" SSL_CTX instances are subject to
3202 limits in configuration files in command-line options.
3206 * Handshake now fails if Extended Master Secret extension is dropped
3211 * The Oracle Developer Studio compiler will start reporting deprecated APIs
3213 ### Changes between 1.1.1f and 1.1.1g [21 Apr 2020]
3215 * Fixed segmentation fault in SSL_check_chain()
3216 Server or client applications that call the SSL_check_chain() function
3217 during or after a TLS 1.3 handshake may crash due to a NULL pointer
3218 dereference as a result of incorrect handling of the
3219 "signature_algorithms_cert" TLS extension. The crash occurs if an invalid
3220 or unrecognised signature algorithm is received from the peer. This could
3221 be exploited by a malicious peer in a Denial of Service attack.
3226 * Added AES consttime code for no-asm configurations
3227 an optional constant time support for AES was added
3228 when building openssl for no-asm.
3229 Enable with: ./config no-asm -DOPENSSL_AES_CONST_TIME
3230 Disable with: ./config no-asm -DOPENSSL_NO_AES_CONST_TIME
3231 At this time this feature is by default disabled.
3232 It will be enabled by default in 3.0.
3236 ### Changes between 1.1.1e and 1.1.1f [31 Mar 2020]
3238 * Revert the change of EOF detection while reading in libssl to avoid
3239 regressions in applications depending on the current way of reporting
3240 the EOF. As the existing method is not fully accurate the change to
3241 reporting the EOF via SSL_ERROR_SSL is kept on the current development
3242 branch and will be present in the 3.0 release.
3246 * Revised BN_generate_prime_ex to not avoid factors 3..17863 in p-1
3247 when primes for RSA keys are computed.
3248 Since we previously always generated primes == 2 (mod 3) for RSA keys,
3249 the 2-prime and 3-prime RSA modules were easy to distinguish, since
3250 N = p*q = 1 (mod 3), but N = p*q*r = 2 (mod 3). Therefore, fingerprinting
3251 2-prime vs. 3-prime RSA keys was possible by computing N mod 3.
3252 This avoids possible fingerprinting of newly generated RSA modules.
3256 ### Changes between 1.1.1d and 1.1.1e [17 Mar 2020]
3258 * Properly detect EOF while reading in libssl. Previously if we hit an EOF
3259 while reading in libssl then we would report an error back to the
3260 application (SSL_ERROR_SYSCALL) but errno would be 0. We now add
3261 an error to the stack (which means we instead return SSL_ERROR_SSL) and
3262 therefore give a hint as to what went wrong.
3266 * Check that ed25519 and ed448 are allowed by the security level. Previously
3267 signature algorithms not using an MD were not being checked that they were
3268 allowed by the security level.
3272 * Fixed SSL_get_servername() behaviour. The behaviour of SSL_get_servername()
3273 was not quite right. The behaviour was not consistent between resumption
3274 and normal handshakes, and also not quite consistent with historical
3275 behaviour. The behaviour in various scenarios has been clarified and
3276 it has been updated to make it match historical behaviour as closely as
3281 * *[VMS only]* The header files that the VMS compilers include automatically,
3282 `__DECC_INCLUDE_PROLOGUE.H` and `__DECC_INCLUDE_EPILOGUE.H`, use pragmas
3283 that the C++ compiler doesn't understand. This is a shortcoming in the
3284 compiler, but can be worked around with `__cplusplus` guards.
3286 C++ applications that use OpenSSL libraries must be compiled using the
3287 qualifier `/NAMES=(AS_IS,SHORTENED)` to be able to use all the OpenSSL
3288 functions. Otherwise, only functions with symbols of less than 31
3289 characters can be used, as the linker will not be able to successfully
3290 resolve symbols with longer names.
3294 * Added a new method to gather entropy on VMS, based on SYS$GET_ENTROPY.
3295 The presence of this system service is determined at run-time.
3299 * Print all values for a PKCS#12 attribute with 'openssl pkcs12', not just
3304 ### Changes between 1.1.1c and 1.1.1d [10 Sep 2019]
3306 * Fixed a fork protection issue. OpenSSL 1.1.1 introduced a rewritten random
3307 number generator (RNG). This was intended to include protection in the
3308 event of a fork() system call in order to ensure that the parent and child
3309 processes did not share the same RNG state. However, this protection was not
3310 being used in the default case.
3312 A partial mitigation for this issue is that the output from a high
3313 precision timer is mixed into the RNG state so the likelihood of a parent
3314 and child process sharing state is significantly reduced.
3316 If an application already calls OPENSSL_init_crypto() explicitly using
3317 OPENSSL_INIT_ATFORK then this problem does not occur at all.
3320 *Matthias St. Pierre*
3322 * For built-in EC curves, ensure an EC_GROUP built from the curve name is
3323 used even when parsing explicit parameters, when loading a encoded key
3324 or calling `EC_GROUP_new_from_ecpkparameters()`/
3325 `EC_GROUP_new_from_ecparameters()`.
3326 This prevents bypass of security hardening and performance gains,
3327 especially for curves with specialized EC_METHODs.
3328 By default, if a key encoded with explicit parameters is loaded and later
3329 encoded, the output is still encoded with explicit parameters, even if
3330 internally a "named" EC_GROUP is used for computation.
3334 * Compute ECC cofactors if not provided during EC_GROUP construction. Before
3335 this change, EC_GROUP_set_generator would accept order and/or cofactor as
3336 NULL. After this change, only the cofactor parameter can be NULL. It also
3337 does some minimal sanity checks on the passed order.
3342 * Fixed a padding oracle in PKCS7_dataDecode and CMS_decrypt_set1_pkey.
3343 An attack is simple, if the first CMS_recipientInfo is valid but the
3344 second CMS_recipientInfo is chosen ciphertext. If the second
3345 recipientInfo decodes to PKCS #1 v1.5 form plaintext, the correct
3346 encryption key will be replaced by garbage, and the message cannot be
3347 decoded, but if the RSA decryption fails, the correct encryption key is
3348 used and the recipient will not notice the attack.
3349 As a work around for this potential attack the length of the decrypted
3350 key must be equal to the cipher default key length, in case the
3351 certificate is not given and all recipientInfo are tried out.
3352 The old behaviour can be re-enabled in the CMS code by setting the
3353 CMS_DEBUG_DECRYPT flag.
3358 * Early start up entropy quality from the DEVRANDOM seed source has been
3359 improved for older Linux systems. The RAND subsystem will wait for
3360 /dev/random to be producing output before seeding from /dev/urandom.
3361 The seeded state is stored for future library initialisations using
3362 a system global shared memory segment. The shared memory identifier
3363 can be configured by defining OPENSSL_RAND_SEED_DEVRANDOM_SHM_ID to
3364 the desired value. The default identifier is 114.
3368 * Correct the extended master secret constant on EBCDIC systems. Without this
3369 fix TLS connections between an EBCDIC system and a non-EBCDIC system that
3370 negotiate EMS will fail. Unfortunately this also means that TLS connections
3371 between EBCDIC systems with this fix, and EBCDIC systems without this
3372 fix will fail if they negotiate EMS.
3376 * Use Windows installation paths in the mingw builds
3378 Mingw isn't a POSIX environment per se, which means that Windows
3379 paths should be used for installation.
3384 * Changed DH_check to accept parameters with order q and 2q subgroups.
3385 With order 2q subgroups the bit 0 of the private key is not secret
3386 but DH_generate_key works around that by clearing bit 0 of the
3387 private key for those. This avoids leaking bit 0 of the private key.
3391 * Significantly reduce secure memory usage by the randomness pools.
3395 * Revert the DEVRANDOM_WAIT feature for Linux systems
3397 The DEVRANDOM_WAIT feature added a select() call to wait for the
3398 /dev/random device to become readable before reading from the
3399 /dev/urandom device.
3401 It turned out that this change had negative side effects on
3402 performance which were not acceptable. After some discussion it
3403 was decided to revert this feature and leave it up to the OS
3404 resp. the platform maintainer to ensure a proper initialization
3405 during early boot time.
3407 *Matthias St. Pierre*
3409 ### Changes between 1.1.1b and 1.1.1c [28 May 2019]
3411 * Add build tests for C++. These are generated files that only do one
3412 thing, to include one public OpenSSL head file each. This tests that
3413 the public header files can be usefully included in a C++ application.
3415 This test isn't enabled by default. It can be enabled with the option
3416 'enable-buildtest-c++'.
3420 * Enable SHA3 pre-hashing for ECDSA and DSA.
3424 * Change the default RSA, DSA and DH size to 2048 bit instead of 1024.
3425 This changes the size when using the `genpkey` command when no size is given.
3426 It fixes an omission in earlier changes that changed all RSA, DSA and DH
3427 generation commands to use 2048 bits by default.
3431 * Reorganize the manual pages to consistently have RETURN VALUES,
3432 EXAMPLES, SEE ALSO and HISTORY come in that order, and adjust
3433 util/fix-doc-nits accordingly.
3435 *Paul Yang, Joshua Lock*
3437 * Add the missing accessor EVP_PKEY_get0_engine()
3441 * Have commands like `s_client` and `s_server` output the signature scheme
3442 along with other cipher suite parameters when debugging.
3446 * Make OPENSSL_config() error agnostic again.
3450 * Do the error handling in RSA decryption constant time.
3454 * Prevent over long nonces in ChaCha20-Poly1305.
3456 ChaCha20-Poly1305 is an AEAD cipher, and requires a unique nonce input
3457 for every encryption operation. RFC 7539 specifies that the nonce value
3458 (IV) should be 96 bits (12 bytes). OpenSSL allows a variable nonce length
3459 and front pads the nonce with 0 bytes if it is less than 12
3460 bytes. However it also incorrectly allows a nonce to be set of up to 16
3461 bytes. In this case only the last 12 bytes are significant and any
3462 additional leading bytes are ignored.
3464 It is a requirement of using this cipher that nonce values are
3465 unique. Messages encrypted using a reused nonce value are susceptible to
3466 serious confidentiality and integrity attacks. If an application changes
3467 the default nonce length to be longer than 12 bytes and then makes a
3468 change to the leading bytes of the nonce expecting the new value to be a
3469 new unique nonce then such an application could inadvertently encrypt
3470 messages with a reused nonce.
3472 Additionally the ignored bytes in a long nonce are not covered by the
3473 integrity guarantee of this cipher. Any application that relies on the
3474 integrity of these ignored leading bytes of a long nonce may be further
3475 affected. Any OpenSSL internal use of this cipher, including in SSL/TLS,
3476 is safe because no such use sets such a long nonce value. However user
3477 applications that use this cipher directly and set a non-default nonce
3478 length to be longer than 12 bytes may be vulnerable.
3480 This issue was reported to OpenSSL on 16th of March 2019 by Joran Dirk
3486 * Add DEVRANDOM_WAIT feature for Linux systems
3488 On older Linux systems where the getrandom() system call is not available,
3489 OpenSSL normally uses the /dev/urandom device for seeding its CSPRNG.
3490 Contrary to getrandom(), the /dev/urandom device will not block during
3491 early boot when the kernel CSPRNG has not been seeded yet.
3493 To mitigate this known weakness, use select() to wait for /dev/random to
3494 become readable before reading from /dev/urandom.
3496 * Ensure that SM2 only uses SM3 as digest algorithm
3500 ### Changes between 1.1.1a and 1.1.1b [26 Feb 2019]
3502 * Change the info callback signals for the start and end of a post-handshake
3503 message exchange in TLSv1.3. In 1.1.1/1.1.1a we used SSL_CB_HANDSHAKE_START
3504 and SSL_CB_HANDSHAKE_DONE. Experience has shown that many applications get
3505 confused by this and assume that a TLSv1.2 renegotiation has started. This
3506 can break KeyUpdate handling. Instead we no longer signal the start and end
3507 of a post handshake message exchange (although the messages themselves are
3508 still signalled). This could break some applications that were expecting
3509 the old signals. However without this KeyUpdate is not usable for many
3514 ### Changes between 1.1.1 and 1.1.1a [20 Nov 2018]
3516 * Timing vulnerability in DSA signature generation
3518 The OpenSSL DSA signature algorithm has been shown to be vulnerable to a
3519 timing side channel attack. An attacker could use variations in the signing
3520 algorithm to recover the private key.
3522 This issue was reported to OpenSSL on 16th October 2018 by Samuel Weiser.
3527 * Timing vulnerability in ECDSA signature generation
3529 The OpenSSL ECDSA signature algorithm has been shown to be vulnerable to a
3530 timing side channel attack. An attacker could use variations in the signing
3531 algorithm to recover the private key.
3533 This issue was reported to OpenSSL on 25th October 2018 by Samuel Weiser.
3538 * Fixed the issue that RAND_add()/RAND_seed() silently discards random input
3539 if its length exceeds 4096 bytes. The limit has been raised to a buffer size
3540 of two gigabytes and the error handling improved.
3542 This issue was reported to OpenSSL by Dr. Falko Strenzke. It has been
3543 categorized as a normal bug, not a security issue, because the DRBG reseeds
3544 automatically and is fully functional even without additional randomness
3545 provided by the application.
3547 ### Changes between 1.1.0i and 1.1.1 [11 Sep 2018]
3549 * Add a new ClientHello callback. Provides a callback interface that gives
3550 the application the ability to adjust the nascent SSL object at the
3551 earliest stage of ClientHello processing, immediately after extensions have
3552 been collected but before they have been processed. In particular, this
3553 callback can adjust the supported TLS versions in response to the contents
3558 * Add SM2 base algorithm support.
3562 * s390x assembly pack: add (improved) hardware-support for the following
3563 cryptographic primitives: sha3, shake, aes-gcm, aes-ccm, aes-ctr, aes-ofb,
3564 aes-cfb/cfb8, aes-ecb.
3568 * Make EVP_PKEY_asn1_new() a bit stricter about its input. A NULL pem_str
3569 parameter is no longer accepted, as it leads to a corrupt table. NULL
3570 pem_str is reserved for alias entries only.
3574 * Use the new ec_scalar_mul_ladder scaffold to implement a specialized ladder
3575 step for prime curves. The new implementation is based on formulae from
3576 differential addition-and-doubling in homogeneous projective coordinates
3577 from Izu-Takagi "A fast parallel elliptic curve multiplication resistant
3578 against side channel attacks" and Brier-Joye "Weierstrass Elliptic Curves
3579 and Side-Channel Attacks" Eq. (8) for y-coordinate recovery, modified
3580 to work in projective coordinates.
3582 *Billy Bob Brumley, Nicola Tuveri*
3584 * Change generating and checking of primes so that the error rate of not
3585 being prime depends on the intended use based on the size of the input.
3586 For larger primes this will result in more rounds of Miller-Rabin.
3587 The maximal error rate for primes with more than 1080 bits is lowered
3590 *Kurt Roeckx, Annie Yousar*
3592 * Increase the number of Miller-Rabin rounds for DSA key generating to 64.
3596 * The 'tsget' script is renamed to 'tsget.pl', to avoid confusion when
3597 moving between systems, and to avoid confusion when a Windows build is
3598 done with mingw vs with MSVC. For POSIX installs, there's still a
3599 symlink or copy named 'tsget' to avoid that confusion as well.
3603 * Revert blinding in ECDSA sign and instead make problematic addition
3604 length-invariant. Switch even to fixed-length Montgomery multiplication.
3608 * Use the new ec_scalar_mul_ladder scaffold to implement a specialized ladder
3609 step for binary curves. The new implementation is based on formulae from
3610 differential addition-and-doubling in mixed Lopez-Dahab projective
3611 coordinates, modified to independently blind the operands.
3613 *Billy Bob Brumley, Sohaib ul Hassan, Nicola Tuveri*
3615 * Add a scaffold to optionally enhance the Montgomery ladder implementation
3616 for `ec_scalar_mul_ladder` (formerly `ec_mul_consttime`) allowing
3617 EC_METHODs to implement their own specialized "ladder step", to take
3618 advantage of more favorable coordinate systems or more efficient
3619 differential addition-and-doubling algorithms.
3621 *Billy Bob Brumley, Sohaib ul Hassan, Nicola Tuveri*
3623 * Modified the random device based seed sources to keep the relevant
3624 file descriptors open rather than reopening them on each access.
3625 This allows such sources to operate in a chroot() jail without
3626 the associated device nodes being available. This behaviour can be
3627 controlled using RAND_keep_random_devices_open().
3631 * Numerous side-channel attack mitigations have been applied. This may have
3632 performance impacts for some algorithms for the benefit of improved
3633 security. Specific changes are noted in this change log by their respective
3638 * AIX shared library support overhaul. Switch to AIX "natural" way of
3639 handling shared libraries, which means collecting shared objects of
3640 different versions and bitnesses in one common archive. This allows to
3641 mitigate conflict between 1.0 and 1.1 side-by-side installations. It
3642 doesn't affect the way 3rd party applications are linked, only how
3643 multi-version installation is managed.
3647 * Make ec_group_do_inverse_ord() more robust and available to other
3648 EC cryptosystems, so that irrespective of BN_FLG_CONSTTIME, SCA
3649 mitigations are applied to the fallback BN_mod_inverse().
3650 When using this function rather than BN_mod_inverse() directly, new
3651 EC cryptosystem implementations are then safer-by-default.
3655 * Add coordinate blinding for EC_POINT and implement projective
3656 coordinate blinding for generic prime curves as a countermeasure to
3657 chosen point SCA attacks.
3659 *Sohaib ul Hassan, Nicola Tuveri, Billy Bob Brumley*
3661 * Add blinding to ECDSA and DSA signatures to protect against side channel
3662 attacks discovered by Keegan Ryan (NCC Group).
3666 * Enforce checking in the `pkeyutl` command to ensure that the input
3667 length does not exceed the maximum supported digest length when performing
3668 a sign, verify or verifyrecover operation.
3672 * SSL_MODE_AUTO_RETRY is enabled by default. Applications that use blocking
3673 I/O in combination with something like select() or poll() will hang. This
3674 can be turned off again using SSL_CTX_clear_mode().
3675 Many applications do not properly handle non-application data records, and
3676 TLS 1.3 sends more of such records. Setting SSL_MODE_AUTO_RETRY works
3677 around the problems in those applications, but can also break some.
3678 It's recommended to read the manpages about SSL_read(), SSL_write(),
3679 SSL_get_error(), SSL_shutdown(), SSL_CTX_set_mode() and
3680 SSL_CTX_set_read_ahead() again.
3684 * When unlocking a pass phrase protected PEM file or PKCS#8 container, we
3685 now allow empty (zero character) pass phrases.
3689 * Apply blinding to binary field modular inversion and remove patent
3690 pending (OPENSSL_SUN_GF2M_DIV) BN_GF2m_mod_div implementation.
3694 * Deprecate ec2_mult.c and unify scalar multiplication code paths for
3695 binary and prime elliptic curves.
3699 * Remove ECDSA nonce padding: EC_POINT_mul is now responsible for
3700 constant time fixed point multiplication.
3704 * Revise elliptic curve scalar multiplication with timing attack
3705 defenses: ec_wNAF_mul redirects to a constant time implementation
3706 when computing fixed point and variable point multiplication (which
3707 in OpenSSL are mostly used with secret scalars in keygen, sign,
3708 ECDH derive operations).
3709 *Billy Bob Brumley, Nicola Tuveri, Cesar Pereida García,
3712 * Updated CONTRIBUTING
3716 * Updated DRBG / RAND to request nonce and additional low entropy
3717 randomness from the system.
3719 *Matthias St. Pierre*
3721 * Updated 'openssl rehash' to use OpenSSL consistent default.
3725 * Moved the load of the ssl_conf module to libcrypto, which helps
3726 loading engines that libssl uses before libssl is initialised.
3730 * Added EVP_PKEY_sign() and EVP_PKEY_verify() for EdDSA
3734 * Fixed X509_NAME_ENTRY_set to get multi-valued RDNs right in all cases.
3736 *Ingo Schwarze, Rich Salz*
3738 * Added output of accepting IP address and port for 'openssl s_server'
3742 * Added a new API for TLSv1.3 ciphersuites:
3743 SSL_CTX_set_ciphersuites()
3744 SSL_set_ciphersuites()
3748 * Memory allocation failures consistently add an error to the error
3753 * Don't use OPENSSL_ENGINES and OPENSSL_CONF environment values
3754 in libcrypto when run as setuid/setgid.
3758 * Load any config file by default when libssl is used.
3762 * Added new public header file <openssl/rand_drbg.h> and documentation
3763 for the RAND_DRBG API. See manual page RAND_DRBG(7) for an overview.
3765 *Matthias St. Pierre*
3767 * QNX support removed (cannot find contributors to get their approval
3768 for the license change).
3772 * TLSv1.3 replay protection for early data has been implemented. See the
3773 SSL_read_early_data() man page for further details.
3777 * Separated TLSv1.3 ciphersuite configuration out from TLSv1.2 ciphersuite
3778 configuration. TLSv1.3 ciphersuites are not compatible with TLSv1.2 and
3779 below. Similarly TLSv1.2 ciphersuites are not compatible with TLSv1.3.
3780 In order to avoid issues where legacy TLSv1.2 ciphersuite configuration
3781 would otherwise inadvertently disable all TLSv1.3 ciphersuites the
3782 configuration has been separated out. See the ciphers man page or the
3783 SSL_CTX_set_ciphersuites() man page for more information.
3787 * On POSIX (BSD, Linux, ...) systems the ocsp(1) command running
3788 in responder mode now supports the new "-multi" option, which
3789 spawns the specified number of child processes to handle OCSP
3790 requests. The "-timeout" option now also limits the OCSP
3791 responder's patience to wait to receive the full client request
3792 on a newly accepted connection. Child processes are respawned
3793 as needed, and the CA index file is automatically reloaded
3794 when changed. This makes it possible to run the "ocsp" responder
3795 as a long-running service, making the OpenSSL CA somewhat more
3796 feature-complete. In this mode, most diagnostic messages logged
3797 after entering the event loop are logged via syslog(3) rather than
3802 * Added support for X448 and Ed448. Heavily based on original work by
3807 * Extend OSSL_STORE with capabilities to search and to narrow the set of
3808 objects loaded. This adds the functions OSSL_STORE_expect() and
3809 OSSL_STORE_find() as well as needed tools to construct searches and
3810 get the search data out of them.
3814 * Support for TLSv1.3 added. Note that users upgrading from an earlier
3815 version of OpenSSL should review their configuration settings to ensure
3816 that they are still appropriate for TLSv1.3. For further information see:
3817 <https://wiki.openssl.org/index.php/TLS1.3>
3821 * Grand redesign of the OpenSSL random generator
3823 The default RAND method now utilizes an AES-CTR DRBG according to
3824 NIST standard SP 800-90Ar1. The new random generator is essentially
3825 a port of the default random generator from the OpenSSL FIPS 2.0
3826 object module. It is a hybrid deterministic random bit generator
3827 using an AES-CTR bit stream and which seeds and reseeds itself
3828 automatically using trusted system entropy sources.
3830 Some of its new features are:
3831 - Support for multiple DRBG instances with seed chaining.
3832 - The default RAND method makes use of a DRBG.
3833 - There is a public and private DRBG instance.
3834 - The DRBG instances are fork-safe.
3835 - Keep all global DRBG instances on the secure heap if it is enabled.
3836 - The public and private DRBG instance are per thread for lock free
3839 *Paul Dale, Benjamin Kaduk, Kurt Roeckx, Rich Salz, Matthias St. Pierre*
3841 * Changed Configure so it only says what it does and doesn't dump
3842 so much data. Instead, ./configdata.pm should be used as a script
3843 to display all sorts of configuration data.
3847 * Added processing of "make variables" to Configure.
3851 * Added SHA512/224 and SHA512/256 algorithm support.
3855 * The last traces of Netware support, first removed in 1.1.0, have
3860 * Get rid of Makefile.shared, and in the process, make the processing
3861 of certain files (rc.obj, or the .def/.map/.opt files produced from
3862 the ordinal files) more visible and hopefully easier to trace and
3863 debug (or make silent).
3867 * Make it possible to have environment variable assignments as
3868 arguments to config / Configure.
3872 * Add multi-prime RSA (RFC 8017) support.
3876 * Add SM3 implemented according to GB/T 32905-2016
3877 *Jack Lloyd <jack.lloyd@ribose.com>,*
3878 *Ronald Tse <ronald.tse@ribose.com>,*
3879 *Erick Borsboom <erick.borsboom@ribose.com>*
3881 * Add 'Maximum Fragment Length' TLS extension negotiation and support
3882 as documented in RFC6066.
3883 Based on a patch from Tomasz Moń
3885 *Filipe Raimundo da Silva*
3887 * Add SM4 implemented according to GB/T 32907-2016.
3888 *Jack Lloyd <jack.lloyd@ribose.com>,*
3889 *Ronald Tse <ronald.tse@ribose.com>,*
3890 *Erick Borsboom <erick.borsboom@ribose.com>*
3892 * Reimplement -newreq-nodes and ERR_error_string_n; the
3893 original author does not agree with the license change.
3897 * Add ARIA AEAD TLS support.
3901 * Some macro definitions to support VS6 have been removed. Visual
3902 Studio 6 has not worked since 1.1.0
3906 * Add ERR_clear_last_mark(), to allow callers to clear the last mark
3907 without clearing the errors.
3911 * Add "atfork" functions. If building on a system that without
3912 pthreads, see doc/man3/OPENSSL_fork_prepare.pod for application
3913 requirements. The RAND facility now uses/requires this.
3921 * The UI API becomes a permanent and integral part of libcrypto, i.e.
3922 not possible to disable entirely. However, it's still possible to
3923 disable the console reading UI method, UI_OpenSSL() (use UI_null()
3926 To disable, configure with 'no-ui-console'. 'no-ui' is still
3927 possible to use as an alias. Check at compile time with the
3928 macro OPENSSL_NO_UI_CONSOLE. The macro OPENSSL_NO_UI is still
3929 possible to check and is an alias for OPENSSL_NO_UI_CONSOLE.
3933 * Add a STORE module, which implements a uniform and URI based reader of
3934 stores that can contain keys, certificates, CRLs and numerous other
3935 objects. The main API is loosely based on a few stdio functions,
3936 and includes OSSL_STORE_open, OSSL_STORE_load, OSSL_STORE_eof,
3937 OSSL_STORE_error and OSSL_STORE_close.
3938 The implementation uses backends called "loaders" to implement arbitrary
3939 URI schemes. There is one built in "loader" for the 'file' scheme.
3943 * Add devcrypto engine. This has been implemented against cryptodev-linux,
3944 then adjusted to work on FreeBSD 8.4 as well.
3945 Enable by configuring with 'enable-devcryptoeng'. This is done by default
3946 on BSD implementations, as cryptodev.h is assumed to exist on all of them.
3950 * Module names can prefixed with OSSL_ or OPENSSL_. This affects
3951 util/mkerr.pl, which is adapted to allow those prefixes, leading to
3952 error code calls like this:
3954 OSSL_FOOerr(OSSL_FOO_F_SOMETHING, OSSL_FOO_R_WHATEVER);
3956 With this change, we claim the namespaces OSSL and OPENSSL in a manner
3957 that can be encoded in C. For the foreseeable future, this will only
3960 *Richard Levitte and Tim Hudson*
3962 * Removed BSD cryptodev engine.
3966 * Add a build target 'build_all_generated', to build all generated files
3967 and only that. This can be used to prepare everything that requires
3968 things like perl for a system that lacks perl and then move everything
3969 to that system and do the rest of the build there.
3973 * In the UI interface, make it possible to duplicate the user data. This
3974 can be used by engines that need to retain the data for a longer time
3975 than just the call where this user data is passed.
3979 * Ignore the '-named_curve auto' value for compatibility of applications
3982 *Tomáš Mráz <tmraz@fedoraproject.org>*
3984 * Fragmented SSL/TLS alerts are no longer accepted. An alert message is 2
3985 bytes long. In theory it is permissible in SSLv3 - TLSv1.2 to fragment such
3986 alerts across multiple records (some of which could be empty). In practice
3987 it make no sense to send an empty alert record, or to fragment one. TLSv1.3
3988 prohibits this altogether and other libraries (BoringSSL, NSS) do not
3989 support this at all. Supporting it adds significant complexity to the
3990 record layer, and its removal is unlikely to cause interoperability
3995 * Add the ASN.1 types INT32, UINT32, INT64, UINT64 and variants prefixed
3996 with Z. These are meant to replace LONG and ZLONG and to be size safe.
3997 The use of LONG and ZLONG is discouraged and scheduled for deprecation
4002 * Add the 'z' and 'j' modifiers to BIO_printf() et al formatting string,
4003 'z' is to be used for [s]size_t, and 'j' - with [u]int64_t.
4005 *Richard Levitte, Andy Polyakov*
4007 * Add EC_KEY_get0_engine(), which does for EC_KEY what RSA_get0_engine()
4012 * Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target
4013 platform rather than 'mingw'.
4017 * The functions X509_STORE_add_cert and X509_STORE_add_crl return
4018 success if they are asked to add an object which already exists
4019 in the store. This change cascades to other functions which load
4020 certificates and CRLs.
4024 * x86_64 assembly pack: annotate code with DWARF CFI directives to
4025 facilitate stack unwinding even from assembly subroutines.
4029 * Remove VAX C specific definitions of OPENSSL_EXPORT, OPENSSL_EXTERN.
4030 Also remove OPENSSL_GLOBAL entirely, as it became a no-op.
4034 * Remove the VMS-specific reimplementation of gmtime from crypto/o_times.c.
4035 VMS C's RTL has a fully up to date gmtime() and gmtime_r() since V7.1,
4036 which is the minimum version we support.
4040 * Certificate time validation (X509_cmp_time) enforces stricter
4041 compliance with RFC 5280. Fractional seconds and timezone offsets
4042 are no longer allowed.
4046 * Add support for ARIA
4050 * s_client will now send the Server Name Indication (SNI) extension by
4051 default unless the new "-noservername" option is used. The server name is
4052 based on the host provided to the "-connect" option unless overridden by
4053 using "-servername".
4057 * Add support for SipHash
4061 * OpenSSL now fails if it receives an unrecognised record type in TLS1.0
4062 or TLS1.1. Previously this only happened in SSLv3 and TLS1.2. This is to
4063 prevent issues where no progress is being made and the peer continually
4064 sends unrecognised record types, using up resources processing them.
4068 * 'openssl passwd' can now produce SHA256 and SHA512 based output,
4069 using the algorithm defined in
4070 <https://www.akkadia.org/drepper/SHA-crypt.txt>
4074 * Heartbeat support has been removed; the ABI is changed for now.
4076 *Richard Levitte, Rich Salz*
4078 * Support for SSL_OP_NO_ENCRYPT_THEN_MAC in SSL_CONF_cmd.
4082 * The RSA "null" method, which was partially supported to avoid patent
4083 issues, has been replaced to always returns NULL.
4090 ### Changes between 1.1.0k and 1.1.0l [10 Sep 2019]
4092 * For built-in EC curves, ensure an EC_GROUP built from the curve name is
4093 used even when parsing explicit parameters, when loading a encoded key
4094 or calling `EC_GROUP_new_from_ecpkparameters()`/
4095 `EC_GROUP_new_from_ecparameters()`.
4096 This prevents bypass of security hardening and performance gains,
4097 especially for curves with specialized EC_METHODs.
4098 By default, if a key encoded with explicit parameters is loaded and later
4099 encoded, the output is still encoded with explicit parameters, even if
4100 internally a "named" EC_GROUP is used for computation.
4104 * Compute ECC cofactors if not provided during EC_GROUP construction. Before
4105 this change, EC_GROUP_set_generator would accept order and/or cofactor as
4106 NULL. After this change, only the cofactor parameter can be NULL. It also
4107 does some minimal sanity checks on the passed order.
4112 * Fixed a padding oracle in PKCS7_dataDecode and CMS_decrypt_set1_pkey.
4113 An attack is simple, if the first CMS_recipientInfo is valid but the
4114 second CMS_recipientInfo is chosen ciphertext. If the second
4115 recipientInfo decodes to PKCS #1 v1.5 form plaintext, the correct
4116 encryption key will be replaced by garbage, and the message cannot be
4117 decoded, but if the RSA decryption fails, the correct encryption key is
4118 used and the recipient will not notice the attack.
4119 As a work around for this potential attack the length of the decrypted
4120 key must be equal to the cipher default key length, in case the
4121 certificate is not given and all recipientInfo are tried out.
4122 The old behaviour can be re-enabled in the CMS code by setting the
4123 CMS_DEBUG_DECRYPT flag.
4128 * Use Windows installation paths in the mingw builds
4130 Mingw isn't a POSIX environment per se, which means that Windows
4131 paths should be used for installation.
4136 ### Changes between 1.1.0j and 1.1.0k [28 May 2019]
4138 * Change the default RSA, DSA and DH size to 2048 bit instead of 1024.
4139 This changes the size when using the `genpkey` command when no size is given.
4140 It fixes an omission in earlier changes that changed all RSA, DSA and DH
4141 generation commands to use 2048 bits by default.
4145 * Prevent over long nonces in ChaCha20-Poly1305.
4147 ChaCha20-Poly1305 is an AEAD cipher, and requires a unique nonce input
4148 for every encryption operation. RFC 7539 specifies that the nonce value
4149 (IV) should be 96 bits (12 bytes). OpenSSL allows a variable nonce length
4150 and front pads the nonce with 0 bytes if it is less than 12
4151 bytes. However it also incorrectly allows a nonce to be set of up to 16
4152 bytes. In this case only the last 12 bytes are significant and any
4153 additional leading bytes are ignored.
4155 It is a requirement of using this cipher that nonce values are
4156 unique. Messages encrypted using a reused nonce value are susceptible to
4157 serious confidentiality and integrity attacks. If an application changes
4158 the default nonce length to be longer than 12 bytes and then makes a
4159 change to the leading bytes of the nonce expecting the new value to be a
4160 new unique nonce then such an application could inadvertently encrypt
4161 messages with a reused nonce.
4163 Additionally the ignored bytes in a long nonce are not covered by the
4164 integrity guarantee of this cipher. Any application that relies on the
4165 integrity of these ignored leading bytes of a long nonce may be further
4166 affected. Any OpenSSL internal use of this cipher, including in SSL/TLS,
4167 is safe because no such use sets such a long nonce value. However user
4168 applications that use this cipher directly and set a non-default nonce
4169 length to be longer than 12 bytes may be vulnerable.
4171 This issue was reported to OpenSSL on 16th of March 2019 by Joran Dirk
4177 * Added SCA hardening for modular field inversion in EC_GROUP through
4178 a new dedicated field_inv() pointer in EC_METHOD.
4179 This also addresses a leakage affecting conversions from projective
4180 to affine coordinates.
4182 *Billy Bob Brumley, Nicola Tuveri*
4184 * Fix a use after free bug in d2i_X509_PUBKEY when overwriting a
4185 re-used X509_PUBKEY object if the second PUBKEY is malformed.
4189 * Move strictness check from EVP_PKEY_asn1_new() to EVP_PKEY_asn1_add0().
4193 * Remove the 'dist' target and add a tarball building script. The
4194 'dist' target has fallen out of use, and it shouldn't be
4195 necessary to configure just to create a source distribution.
4199 ### Changes between 1.1.0i and 1.1.0j [20 Nov 2018]
4201 * Timing vulnerability in DSA signature generation
4203 The OpenSSL DSA signature algorithm has been shown to be vulnerable to a
4204 timing side channel attack. An attacker could use variations in the signing
4205 algorithm to recover the private key.
4207 This issue was reported to OpenSSL on 16th October 2018 by Samuel Weiser.
4212 * Timing vulnerability in ECDSA signature generation
4214 The OpenSSL ECDSA signature algorithm has been shown to be vulnerable to a
4215 timing side channel attack. An attacker could use variations in the signing
4216 algorithm to recover the private key.
4218 This issue was reported to OpenSSL on 25th October 2018 by Samuel Weiser.
4223 * Add coordinate blinding for EC_POINT and implement projective
4224 coordinate blinding for generic prime curves as a countermeasure to
4225 chosen point SCA attacks.
4227 *Sohaib ul Hassan, Nicola Tuveri, Billy Bob Brumley*
4229 ### Changes between 1.1.0h and 1.1.0i [14 Aug 2018]
4231 * Client DoS due to large DH parameter
4233 During key agreement in a TLS handshake using a DH(E) based ciphersuite a
4234 malicious server can send a very large prime value to the client. This will
4235 cause the client to spend an unreasonably long period of time generating a
4236 key for this prime resulting in a hang until the client has finished. This
4237 could be exploited in a Denial Of Service attack.
4239 This issue was reported to OpenSSL on 5th June 2018 by Guido Vranken
4244 * Cache timing vulnerability in RSA Key Generation
4246 The OpenSSL RSA Key generation algorithm has been shown to be vulnerable to
4247 a cache timing side channel attack. An attacker with sufficient access to
4248 mount cache timing attacks during the RSA key generation process could
4249 recover the private key.
4251 This issue was reported to OpenSSL on 4th April 2018 by Alejandro Cabrera
4252 Aldaya, Billy Brumley, Cesar Pereida Garcia and Luis Manuel Alvarez Tapia.
4257 * Make EVP_PKEY_asn1_new() a bit stricter about its input. A NULL pem_str
4258 parameter is no longer accepted, as it leads to a corrupt table. NULL
4259 pem_str is reserved for alias entries only.
4263 * Revert blinding in ECDSA sign and instead make problematic addition
4264 length-invariant. Switch even to fixed-length Montgomery multiplication.
4268 * Change generating and checking of primes so that the error rate of not
4269 being prime depends on the intended use based on the size of the input.
4270 For larger primes this will result in more rounds of Miller-Rabin.
4271 The maximal error rate for primes with more than 1080 bits is lowered
4274 *Kurt Roeckx, Annie Yousar*
4276 * Increase the number of Miller-Rabin rounds for DSA key generating to 64.
4280 * Add blinding to ECDSA and DSA signatures to protect against side channel
4281 attacks discovered by Keegan Ryan (NCC Group).
4285 * When unlocking a pass phrase protected PEM file or PKCS#8 container, we
4286 now allow empty (zero character) pass phrases.
4290 * Certificate time validation (X509_cmp_time) enforces stricter
4291 compliance with RFC 5280. Fractional seconds and timezone offsets
4292 are no longer allowed.
4296 * Fixed a text canonicalisation bug in CMS
4298 Where a CMS detached signature is used with text content the text goes
4299 through a canonicalisation process first prior to signing or verifying a
4300 signature. This process strips trailing space at the end of lines, converts
4301 line terminators to CRLF and removes additional trailing line terminators
4302 at the end of a file. A bug in the canonicalisation process meant that
4303 some characters, such as form-feed, were incorrectly treated as whitespace
4304 and removed. This is contrary to the specification (RFC5485). This fix
4305 could mean that detached text data signed with an earlier version of
4306 OpenSSL 1.1.0 may fail to verify using the fixed version, or text data
4307 signed with a fixed OpenSSL may fail to verify with an earlier version of
4308 OpenSSL 1.1.0. A workaround is to only verify the canonicalised text data
4309 and use the "-binary" flag (for the "cms" command line application) or set
4310 the SMIME_BINARY/PKCS7_BINARY/CMS_BINARY flags (if using CMS_verify()).
4314 ### Changes between 1.1.0g and 1.1.0h [27 Mar 2018]
4316 * Constructed ASN.1 types with a recursive definition could exceed the stack
4318 Constructed ASN.1 types with a recursive definition (such as can be found
4319 in PKCS7) could eventually exceed the stack given malicious input with
4320 excessive recursion. This could result in a Denial Of Service attack. There
4321 are no such structures used within SSL/TLS that come from untrusted sources
4322 so this is considered safe.
4324 This issue was reported to OpenSSL on 4th January 2018 by the OSS-fuzz
4330 * Incorrect CRYPTO_memcmp on HP-UX PA-RISC
4332 Because of an implementation bug the PA-RISC CRYPTO_memcmp function is
4333 effectively reduced to only comparing the least significant bit of each
4334 byte. This allows an attacker to forge messages that would be considered as
4335 authenticated in an amount of tries lower than that guaranteed by the
4336 security claims of the scheme. The module can only be compiled by the
4337 HP-UX assembler, so that only HP-UX PA-RISC targets are affected.
4339 This issue was reported to OpenSSL on 2nd March 2018 by Peter Waltenberg
4345 * Add a build target 'build_all_generated', to build all generated files
4346 and only that. This can be used to prepare everything that requires
4347 things like perl for a system that lacks perl and then move everything
4348 to that system and do the rest of the build there.
4352 * Backport SSL_OP_NO_RENGOTIATION
4354 OpenSSL 1.0.2 and below had the ability to disable renegotiation using the
4355 (undocumented) SSL3_FLAGS_NO_RENEGOTIATE_CIPHERS flag. Due to the opacity
4356 changes this is no longer possible in 1.1.0. Therefore, the new
4357 SSL_OP_NO_RENEGOTIATION option from 1.1.1-dev has been backported to
4358 1.1.0 to provide equivalent functionality.
4360 Note that if an application built against 1.1.0h headers (or above) is run
4361 using an older version of 1.1.0 (prior to 1.1.0h) then the option will be
4362 accepted but nothing will happen, i.e. renegotiation will not be prevented.
4366 * Removed the OS390-Unix config target. It relied on a script that doesn't
4371 * rsaz_1024_mul_avx2 overflow bug on x86_64
4373 There is an overflow bug in the AVX2 Montgomery multiplication procedure
4374 used in exponentiation with 1024-bit moduli. No EC algorithms are affected.
4375 Analysis suggests that attacks against RSA and DSA as a result of this
4376 defect would be very difficult to perform and are not believed likely.
4377 Attacks against DH1024 are considered just feasible, because most of the
4378 work necessary to deduce information about a private key may be performed
4379 offline. The amount of resources required for such an attack would be
4380 significant. However, for an attack on TLS to be meaningful, the server
4381 would have to share the DH1024 private key among multiple clients, which is
4382 no longer an option since CVE-2016-0701.
4384 This only affects processors that support the AVX2 but not ADX extensions
4385 like Intel Haswell (4th generation).
4387 This issue was reported to OpenSSL by David Benjamin (Google). The issue
4388 was originally found via the OSS-Fuzz project.
4393 ### Changes between 1.1.0f and 1.1.0g [2 Nov 2017]
4395 * bn_sqrx8x_internal carry bug on x86_64
4397 There is a carry propagating bug in the x86_64 Montgomery squaring
4398 procedure. No EC algorithms are affected. Analysis suggests that attacks
4399 against RSA and DSA as a result of this defect would be very difficult to
4400 perform and are not believed likely. Attacks against DH are considered just
4401 feasible (although very difficult) because most of the work necessary to
4402 deduce information about a private key may be performed offline. The amount
4403 of resources required for such an attack would be very significant and
4404 likely only accessible to a limited number of attackers. An attacker would
4405 additionally need online access to an unpatched system using the target
4406 private key in a scenario with persistent DH parameters and a private
4407 key that is shared between multiple clients.
4409 This only affects processors that support the BMI1, BMI2 and ADX extensions
4410 like Intel Broadwell (5th generation) and later or AMD Ryzen.
4412 This issue was reported to OpenSSL by the OSS-Fuzz project.
4417 * Malformed X.509 IPAddressFamily could cause OOB read
4419 If an X.509 certificate has a malformed IPAddressFamily extension,
4420 OpenSSL could do a one-byte buffer overread. The most likely result
4421 would be an erroneous display of the certificate in text format.
4423 This issue was reported to OpenSSL by the OSS-Fuzz project.
4428 ### Changes between 1.1.0e and 1.1.0f [25 May 2017]
4430 * Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target
4431 platform rather than 'mingw'.
4435 * Remove the VMS-specific reimplementation of gmtime from crypto/o_times.c.
4436 VMS C's RTL has a fully up to date gmtime() and gmtime_r() since V7.1,
4437 which is the minimum version we support.
4441 ### Changes between 1.1.0d and 1.1.0e [16 Feb 2017]
4443 * Encrypt-Then-Mac renegotiation crash
4445 During a renegotiation handshake if the Encrypt-Then-Mac extension is
4446 negotiated where it was not in the original handshake (or vice-versa) then
4447 this can cause OpenSSL to crash (dependent on ciphersuite). Both clients
4448 and servers are affected.
4450 This issue was reported to OpenSSL by Joe Orton (Red Hat).
4455 ### Changes between 1.1.0c and 1.1.0d [26 Jan 2017]
4457 * Truncated packet could crash via OOB read
4459 If one side of an SSL/TLS path is running on a 32-bit host and a specific
4460 cipher is being used, then a truncated packet can cause that host to
4461 perform an out-of-bounds read, usually resulting in a crash.
4463 This issue was reported to OpenSSL by Robert Święcki of Google.
4468 * Bad (EC)DHE parameters cause a client crash
4470 If a malicious server supplies bad parameters for a DHE or ECDHE key
4471 exchange then this can result in the client attempting to dereference a
4472 NULL pointer leading to a client crash. This could be exploited in a Denial
4475 This issue was reported to OpenSSL by Guido Vranken.
4480 * BN_mod_exp may produce incorrect results on x86_64
4482 There is a carry propagating bug in the x86_64 Montgomery squaring
4483 procedure. No EC algorithms are affected. Analysis suggests that attacks
4484 against RSA and DSA as a result of this defect would be very difficult to
4485 perform and are not believed likely. Attacks against DH are considered just
4486 feasible (although very difficult) because most of the work necessary to
4487 deduce information about a private key may be performed offline. The amount
4488 of resources required for such an attack would be very significant and
4489 likely only accessible to a limited number of attackers. An attacker would
4490 additionally need online access to an unpatched system using the target
4491 private key in a scenario with persistent DH parameters and a private
4492 key that is shared between multiple clients. For example this can occur by
4493 default in OpenSSL DHE based SSL/TLS ciphersuites. Note: This issue is very
4494 similar to CVE-2015-3193 but must be treated as a separate problem.
4496 This issue was reported to OpenSSL by the OSS-Fuzz project.
4501 ### Changes between 1.1.0b and 1.1.0c [10 Nov 2016]
4503 * ChaCha20/Poly1305 heap-buffer-overflow
4505 TLS connections using `*-CHACHA20-POLY1305` ciphersuites are susceptible to
4506 a DoS attack by corrupting larger payloads. This can result in an OpenSSL
4507 crash. This issue is not considered to be exploitable beyond a DoS.
4509 This issue was reported to OpenSSL by Robert Święcki (Google Security Team)
4514 * CMS Null dereference
4516 Applications parsing invalid CMS structures can crash with a NULL pointer
4517 dereference. This is caused by a bug in the handling of the ASN.1 CHOICE
4518 type in OpenSSL 1.1.0 which can result in a NULL value being passed to the
4519 structure callback if an attempt is made to free certain invalid encodings.
4520 Only CHOICE structures using a callback which do not handle NULL value are
4523 This issue was reported to OpenSSL by Tyler Nighswander of ForAllSecure.
4528 * Montgomery multiplication may produce incorrect results
4530 There is a carry propagating bug in the Broadwell-specific Montgomery
4531 multiplication procedure that handles input lengths divisible by, but
4532 longer than 256 bits. Analysis suggests that attacks against RSA, DSA
4533 and DH private keys are impossible. This is because the subroutine in
4534 question is not used in operations with the private key itself and an input
4535 of the attacker's direct choice. Otherwise the bug can manifest itself as
4536 transient authentication and key negotiation failures or reproducible
4537 erroneous outcome of public-key operations with specially crafted input.
4538 Among EC algorithms only Brainpool P-512 curves are affected and one
4539 presumably can attack ECDH key negotiation. Impact was not analyzed in
4540 detail, because pre-requisites for attack are considered unlikely. Namely
4541 multiple clients have to choose the curve in question and the server has to
4542 share the private key among them, neither of which is default behaviour.
4543 Even then only clients that chose the curve will be affected.
4545 This issue was publicly reported as transient failures and was not
4546 initially recognized as a security issue. Thanks to Richard Morgan for
4547 providing reproducible case.
4552 * Removed automatic addition of RPATH in shared libraries and executables,
4553 as this was a remainder from OpenSSL 1.0.x and isn't needed any more.
4557 ### Changes between 1.1.0a and 1.1.0b [26 Sep 2016]
4559 * Fix Use After Free for large message sizes
4561 The patch applied to address CVE-2016-6307 resulted in an issue where if a
4562 message larger than approx 16k is received then the underlying buffer to
4563 store the incoming message is reallocated and moved. Unfortunately a
4564 dangling pointer to the old location is left which results in an attempt to
4565 write to the previously freed location. This is likely to result in a
4566 crash, however it could potentially lead to execution of arbitrary code.
4568 This issue only affects OpenSSL 1.1.0a.
4570 This issue was reported to OpenSSL by Robert Święcki.
4575 ### Changes between 1.1.0 and 1.1.0a [22 Sep 2016]
4577 * OCSP Status Request extension unbounded memory growth
4579 A malicious client can send an excessively large OCSP Status Request
4580 extension. If that client continually requests renegotiation, sending a
4581 large OCSP Status Request extension each time, then there will be unbounded
4582 memory growth on the server. This will eventually lead to a Denial Of
4583 Service attack through memory exhaustion. Servers with a default
4584 configuration are vulnerable even if they do not support OCSP. Builds using
4585 the "no-ocsp" build time option are not affected.
4587 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
4592 * SSL_peek() hang on empty record
4594 OpenSSL 1.1.0 SSL/TLS will hang during a call to SSL_peek() if the peer
4595 sends an empty record. This could be exploited by a malicious peer in a
4596 Denial Of Service attack.
4598 This issue was reported to OpenSSL by Alex Gaynor.
4603 * Excessive allocation of memory in tls_get_message_header() and
4604 dtls1_preprocess_fragment()
4606 A (D)TLS message includes 3 bytes for its length in the header for the
4607 message. This would allow for messages up to 16Mb in length. Messages of
4608 this length are excessive and OpenSSL includes a check to ensure that a
4609 peer is sending reasonably sized messages in order to avoid too much memory
4610 being consumed to service a connection. A flaw in the logic of version
4611 1.1.0 means that memory for the message is allocated too early, prior to
4612 the excessive message length check. Due to way memory is allocated in
4613 OpenSSL this could mean an attacker could force up to 21Mb to be allocated
4614 to service a connection. This could lead to a Denial of Service through
4615 memory exhaustion. However, the excessive message length check still takes
4616 place, and this would cause the connection to immediately fail. Assuming
4617 that the application calls SSL_free() on the failed connection in a timely
4618 manner then the 21Mb of allocated memory will then be immediately freed
4619 again. Therefore, the excessive memory allocation will be transitory in
4620 nature. This then means that there is only a security impact if:
4622 1) The application does not call SSL_free() in a timely manner in the event
4623 that the connection fails
4625 2) The application is working in a constrained environment where there is
4626 very little free memory
4628 3) The attacker initiates multiple connection attempts such that there are
4629 multiple connections in a state where memory has been allocated for the
4630 connection; SSL_free() has not yet been called; and there is insufficient
4631 memory to service the multiple requests.
4633 Except in the instance of (1) above any Denial Of Service is likely to be
4634 transitory because as soon as the connection fails the memory is
4635 subsequently freed again in the SSL_free() call. However there is an
4636 increased risk during this period of application crashes due to the lack of
4637 memory - which would then mean a more serious Denial of Service.
4639 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
4640 (CVE-2016-6307 and CVE-2016-6308)
4644 * solaris-x86-cc, i.e. 32-bit configuration with vendor compiler,
4645 had to be removed. Primary reason is that vendor assembler can't
4646 assemble our modules with -KPIC flag. As result it, assembly
4647 support, was not even available as option. But its lack means
4648 lack of side-channel resistant code, which is incompatible with
4649 security by todays standards. Fortunately gcc is readily available
4650 prepackaged option, which we firmly point at...
4654 ### Changes between 1.0.2h and 1.1.0 [25 Aug 2016]
4656 * Windows command-line tool supports UTF-8 opt-in option for arguments
4657 and console input. Setting OPENSSL_WIN32_UTF8 environment variable
4658 (to any value) allows Windows user to access PKCS#12 file generated
4659 with Windows CryptoAPI and protected with non-ASCII password, as well
4660 as files generated under UTF-8 locale on Linux also protected with
4665 * To mitigate the SWEET32 attack ([CVE-2016-2183]), 3DES cipher suites
4666 have been disabled by default and removed from DEFAULT, just like RC4.
4667 See the RC4 item below to re-enable both.
4671 * The method for finding the storage location for the Windows RAND seed file
4672 has changed. First we check %RANDFILE%. If that is not set then we check
4673 the directories %HOME%, %USERPROFILE% and %SYSTEMROOT% in that order. If
4674 all else fails we fall back to C:\.
4678 * The EVP_EncryptUpdate() function has had its return type changed from void
4679 to int. A return of 0 indicates and error while a return of 1 indicates
4684 * The flags RSA_FLAG_NO_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME and
4685 DH_FLAG_NO_EXP_CONSTTIME which previously provided the ability to switch
4686 off the constant time implementation for RSA, DSA and DH have been made
4687 no-ops and deprecated.
4691 * Windows RAND implementation was simplified to only get entropy by
4692 calling CryptGenRandom(). Various other RAND-related tickets
4695 *Joseph Wylie Yandle, Rich Salz*
4697 * The stack and lhash API's were renamed to start with `OPENSSL_SK_`
4698 and `OPENSSL_LH_`, respectively. The old names are available
4699 with API compatibility. They new names are now completely documented.
4703 * Unify TYPE_up_ref(obj) methods signature.
4704 SSL_CTX_up_ref(), SSL_up_ref(), X509_up_ref(), EVP_PKEY_up_ref(),
4705 X509_CRL_up_ref(), X509_OBJECT_up_ref_count() methods are now returning an
4706 int (instead of void) like all others TYPE_up_ref() methods.
4707 So now these methods also check the return value of CRYPTO_atomic_add(),
4708 and the validity of object reference counter.
4710 *fdasilvayy@gmail.com*
4712 * With Windows Visual Studio builds, the .pdb files are installed
4713 alongside the installed libraries and executables. For a static
4714 library installation, ossl_static.pdb is the associate compiler
4715 generated .pdb file to be used when linking programs.
4719 * Remove openssl.spec. Packaging files belong with the packagers.
4723 * Automatic Darwin/OSX configuration has had a refresh, it will now
4724 recognise x86_64 architectures automatically. You can still decide
4725 to build for a different bitness with the environment variable
4726 KERNEL_BITS (can be 32 or 64), for example:
4728 KERNEL_BITS=32 ./config
4732 * Change default algorithms in pkcs8 utility to use PKCS#5 v2.0,
4733 256 bit AES and HMAC with SHA256.
4737 * Remove support for MIPS o32 ABI on IRIX (and IRIX only).
4741 * Triple-DES ciphers have been moved from HIGH to MEDIUM.
4745 * To enable users to have their own config files and build file templates,
4746 Configure looks in the directory indicated by the environment variable
4747 OPENSSL_LOCAL_CONFIG_DIR as well as the in-source Configurations/
4748 directory. On VMS, OPENSSL_LOCAL_CONFIG_DIR is expected to be a logical
4749 name and is used as is.
4753 * The following datatypes were made opaque: X509_OBJECT, X509_STORE_CTX,
4754 X509_STORE, X509_LOOKUP, and X509_LOOKUP_METHOD. The unused type
4755 X509_CERT_FILE_CTX was removed.
4759 * "shared" builds are now the default. To create only static libraries use
4760 the "no-shared" Configure option.
4764 * Remove the no-aes, no-hmac, no-rsa, no-sha and no-md5 Configure options.
4765 All of these option have not worked for some while and are fundamental
4770 * Make various cleanup routines no-ops and mark them as deprecated. Most
4771 global cleanup functions are no longer required because they are handled
4772 via auto-deinit (see OPENSSL_init_crypto and OPENSSL_init_ssl man pages).
4773 Explicitly de-initing can cause problems (e.g. where a library that uses
4774 OpenSSL de-inits, but an application is still using it). The affected
4775 functions are CONF_modules_free(), ENGINE_cleanup(), OBJ_cleanup(),
4776 EVP_cleanup(), BIO_sock_cleanup(), CRYPTO_cleanup_all_ex_data(),
4777 RAND_cleanup(), SSL_COMP_free_compression_methods(), ERR_free_strings() and
4778 COMP_zlib_cleanup().
4782 * --strict-warnings no longer enables runtime debugging options
4783 such as REF_DEBUG. Instead, debug options are automatically
4784 enabled with '--debug' builds.
4786 *Andy Polyakov, Emilia Käsper*
4788 * Made DH and DH_METHOD opaque. The structures for managing DH objects
4789 have been moved out of the public header files. New functions for managing
4790 these have been added.
4794 * Made RSA and RSA_METHOD opaque. The structures for managing RSA
4795 objects have been moved out of the public header files. New
4796 functions for managing these have been added.
4800 * Made DSA and DSA_METHOD opaque. The structures for managing DSA objects
4801 have been moved out of the public header files. New functions for managing
4802 these have been added.
4806 * Made BIO and BIO_METHOD opaque. The structures for managing BIOs have been
4807 moved out of the public header files. New functions for managing these
4812 * Removed no-rijndael as a config option. Rijndael is an old name for AES.
4816 * Removed the mk1mf build scripts.
4820 * Headers are now wrapped, if necessary, with OPENSSL_NO_xxx, so
4821 it is always safe to #include a header now.
4825 * Removed the aged BC-32 config and all its supporting scripts
4829 * Removed support for Ultrix, Netware, and OS/2.
4833 * Add support for HKDF.
4835 *Alessandro Ghedini*
4837 * Add support for blake2b and blake2s
4841 * Added support for "pipelining". Ciphers that have the
4842 EVP_CIPH_FLAG_PIPELINE flag set have a capability to process multiple
4843 encryptions/decryptions simultaneously. There are currently no built-in
4844 ciphers with this property but the expectation is that engines will be able
4845 to offer it to significantly improve throughput. Support has been extended
4846 into libssl so that multiple records for a single connection can be
4847 processed in one go (for >=TLS 1.1).
4851 * Added the AFALG engine. This is an async capable engine which is able to
4852 offload work to the Linux kernel. In this initial version it only supports
4853 AES128-CBC. The kernel must be version 4.1.0 or greater.
4857 * OpenSSL now uses a new threading API. It is no longer necessary to
4858 set locking callbacks to use OpenSSL in a multi-threaded environment. There
4859 are two supported threading models: pthreads and windows threads. It is
4860 also possible to configure OpenSSL at compile time for "no-threads". The
4861 old threading API should no longer be used. The functions have been
4862 replaced with "no-op" compatibility macros.
4864 *Alessandro Ghedini, Matt Caswell*
4866 * Modify behavior of ALPN to invoke callback after SNI/servername
4867 callback, such that updates to the SSL_CTX affect ALPN.
4871 * Add SSL_CIPHER queries for authentication and key-exchange.
4875 * Changes to the DEFAULT cipherlist:
4876 - Prefer (EC)DHE handshakes over plain RSA.
4877 - Prefer AEAD ciphers over legacy ciphers.
4878 - Prefer ECDSA over RSA when both certificates are available.
4879 - Prefer TLSv1.2 ciphers/PRF.
4880 - Remove DSS, SEED, IDEA, CAMELLIA, and AES-CCM from the
4885 * Change the ECC default curve list to be this, in order: x25519,
4886 secp256r1, secp521r1, secp384r1.
4890 * RC4 based libssl ciphersuites are now classed as "weak" ciphers and are
4891 disabled by default. They can be re-enabled using the
4892 enable-weak-ssl-ciphers option to Configure.
4896 * If the server has ALPN configured, but supports no protocols that the
4897 client advertises, send a fatal "no_application_protocol" alert.
4898 This behaviour is SHALL in RFC 7301, though it isn't universally
4899 implemented by other servers.
4903 * Add X25519 support.
4904 Add ASN.1 and EVP_PKEY methods for X25519. This includes support
4905 for public and private key encoding using the format documented in
4906 draft-ietf-curdle-pkix-02. The corresponding EVP_PKEY method supports
4907 key generation and key derivation.
4909 TLS support complies with draft-ietf-tls-rfc4492bis-08 and uses
4914 * Deprecate SRP_VBASE_get_by_user.
4915 SRP_VBASE_get_by_user had inconsistent memory management behaviour.
4916 In order to fix an unavoidable memory leak ([CVE-2016-0798]),
4917 SRP_VBASE_get_by_user was changed to ignore the "fake user" SRP
4918 seed, even if the seed is configured.
4920 Users should use SRP_VBASE_get1_by_user instead. Note that in
4921 SRP_VBASE_get1_by_user, caller must free the returned value. Note
4922 also that even though configuring the SRP seed attempts to hide
4923 invalid usernames by continuing the handshake with fake
4924 credentials, this behaviour is not constant time and no strong
4925 guarantees are made that the handshake is indistinguishable from
4926 that of a valid user.
4930 * Configuration change; it's now possible to build dynamic engines
4931 without having to build shared libraries and vice versa. This
4932 only applies to the engines in `engines/`, those in `crypto/engine/`
4933 will always be built into libcrypto (i.e. "static").
4935 Building dynamic engines is enabled by default; to disable, use
4936 the configuration option "disable-dynamic-engine".
4938 The only requirements for building dynamic engines are the
4939 presence of the DSO module and building with position independent
4940 code, so they will also automatically be disabled if configuring
4941 with "disable-dso" or "disable-pic".
4943 The macros OPENSSL_NO_STATIC_ENGINE and OPENSSL_NO_DYNAMIC_ENGINE
4944 are also taken away from openssl/opensslconf.h, as they are
4949 * Configuration change; if there is a known flag to compile
4950 position independent code, it will always be applied on the
4951 libcrypto and libssl object files, and never on the application
4952 object files. This means other libraries that use routines from
4953 libcrypto / libssl can be made into shared libraries regardless
4954 of how OpenSSL was configured.
4956 If this isn't desirable, the configuration options "disable-pic"
4957 or "no-pic" can be used to disable the use of PIC. This will
4958 also disable building shared libraries and dynamic engines.
4962 * Removed JPAKE code. It was experimental and has no wide use.
4966 * The INSTALL_PREFIX Makefile variable has been renamed to
4967 DESTDIR. That makes for less confusion on what this variable
4968 is for. Also, the configuration option --install_prefix is
4973 * Heartbeat for TLS has been removed and is disabled by default
4974 for DTLS; configure with enable-heartbeats. Code that uses the
4975 old #define's might need to be updated.
4977 *Emilia Käsper, Rich Salz*
4979 * Rename REF_CHECK to REF_DEBUG.
4983 * New "unified" build system
4985 The "unified" build system is aimed to be a common system for all
4986 platforms we support. With it comes new support for VMS.
4988 This system builds supports building in a different directory tree
4989 than the source tree. It produces one Makefile (for unix family
4990 or lookalikes), or one descrip.mms (for VMS).
4992 The source of information to make the Makefile / descrip.mms is
4993 small files called 'build.info', holding the necessary
4994 information for each directory with source to compile, and a
4995 template in Configurations, like unix-Makefile.tmpl or
4998 With this change, the library names were also renamed on Windows
4999 and on VMS. They now have names that are closer to the standard
5000 on Unix, and include the major version number, and in certain
5001 cases, the architecture they are built for. See "Notes on shared
5002 libraries" in INSTALL.
5004 We rely heavily on the perl module Text::Template.
5008 * Added support for auto-initialisation and de-initialisation of the library.
5009 OpenSSL no longer requires explicit init or deinit routines to be called,
5010 except in certain circumstances. See the OPENSSL_init_crypto() and
5011 OPENSSL_init_ssl() man pages for further information.
5015 * The arguments to the DTLSv1_listen function have changed. Specifically the
5016 "peer" argument is now expected to be a BIO_ADDR object.
5018 * Rewrite of BIO networking library. The BIO library lacked consistent
5019 support of IPv6, and adding it required some more extensive
5020 modifications. This introduces the BIO_ADDR and BIO_ADDRINFO types,
5021 which hold all types of addresses and chains of address information.
5022 It also introduces a new API, with functions like BIO_socket,
5023 BIO_connect, BIO_listen, BIO_lookup and a rewrite of BIO_accept.
5024 The source/sink BIOs BIO_s_connect, BIO_s_accept and BIO_s_datagram
5025 have been adapted accordingly.
5029 * RSA_padding_check_PKCS1_type_1 now accepts inputs with and without
5034 * CRIME protection: disable compression by default, even if OpenSSL is
5035 compiled with zlib enabled. Applications can still enable compression
5036 by calling SSL_CTX_clear_options(ctx, SSL_OP_NO_COMPRESSION), or by
5037 using the SSL_CONF library to configure compression.
5041 * The signature of the session callback configured with
5042 SSL_CTX_sess_set_get_cb was changed. The read-only input buffer
5043 was explicitly marked as `const unsigned char*` instead of
5048 * Always DPURIFY. Remove the use of uninitialized memory in the
5049 RNG, and other conditional uses of DPURIFY. This makes -DPURIFY a no-op.
5053 * Removed many obsolete configuration items, including
5054 DES_PTR, DES_RISC1, DES_RISC2, DES_INT
5055 MD2_CHAR, MD2_INT, MD2_LONG
5057 IDEA_SHORT, IDEA_LONG
5058 RC2_SHORT, RC2_LONG, RC4_LONG, RC4_CHUNK, RC4_INDEX
5060 *Rich Salz, with advice from Andy Polyakov*
5062 * Many BN internals have been moved to an internal header file.
5064 *Rich Salz with help from Andy Polyakov*
5066 * Configuration and writing out the results from it has changed.
5067 Files such as Makefile include/openssl/opensslconf.h and are now
5068 produced through general templates, such as Makefile.in and
5069 crypto/opensslconf.h.in and some help from the perl module
5072 Also, the center of configuration information is no longer
5073 Makefile. Instead, Configure produces a perl module in
5074 configdata.pm which holds most of the config data (in the hash
5075 table %config), the target data that comes from the target
5076 configuration in one of the `Configurations/*.conf` files (in
5081 * To clarify their intended purposes, the Configure options
5082 --prefix and --openssldir change their semantics, and become more
5083 straightforward and less interdependent.
5085 --prefix shall be used exclusively to give the location INSTALLTOP
5086 where programs, scripts, libraries, include files and manuals are
5087 going to be installed. The default is now /usr/local.
5089 --openssldir shall be used exclusively to give the default
5090 location OPENSSLDIR where certificates, private keys, CRLs are
5091 managed. This is also where the default openssl.cnf gets
5093 If the directory given with this option is a relative path, the
5094 values of both the --prefix value and the --openssldir value will
5095 be combined to become OPENSSLDIR.
5096 The default for --openssldir is INSTALLTOP/ssl.
5098 Anyone who uses --openssldir to specify where OpenSSL is to be
5099 installed MUST change to use --prefix instead.
5103 * The GOST engine was out of date and therefore it has been removed. An up
5104 to date GOST engine is now being maintained in an external repository.
5105 See: <https://wiki.openssl.org/index.php/Binaries>. Libssl still retains
5106 support for GOST ciphersuites (these are only activated if a GOST engine
5111 * EGD is no longer supported by default; use enable-egd when
5114 *Ben Kaduk and Rich Salz*
5116 * The distribution now has Makefile.in files, which are used to
5117 create Makefile's when Configure is run. *Configure must be run
5118 before trying to build now.*
5122 * The return value for SSL_CIPHER_description() for error conditions
5127 * Support for RFC6698/RFC7671 DANE TLSA peer authentication.
5129 Obtaining and performing DNSSEC validation of TLSA records is
5130 the application's responsibility. The application provides
5131 the TLSA records of its choice to OpenSSL, and these are then
5132 used to authenticate the peer.
5134 The TLSA records need not even come from DNS. They can, for
5135 example, be used to implement local end-entity certificate or
5136 trust-anchor "pinning", where the "pin" data takes the form
5137 of TLSA records, which can augment or replace verification
5138 based on the usual WebPKI public certification authorities.
5142 * Revert default OPENSSL_NO_DEPRECATED setting. Instead OpenSSL
5143 continues to support deprecated interfaces in default builds.
5144 However, applications are strongly advised to compile their
5145 source files with -DOPENSSL_API_COMPAT=0x10100000L, which hides
5146 the declarations of all interfaces deprecated in 0.9.8, 1.0.0
5147 or the 1.1.0 releases.
5149 In environments in which all applications have been ported to
5150 not use any deprecated interfaces OpenSSL's Configure script
5151 should be used with the --api=1.1.0 option to entirely remove
5152 support for the deprecated features from the library and
5153 unconditionally disable them in the installed headers.
5154 Essentially the same effect can be achieved with the "no-deprecated"
5155 argument to Configure, except that this will always restrict
5156 the build to just the latest API, rather than a fixed API
5159 As applications are ported to future revisions of the API,
5160 they should update their compile-time OPENSSL_API_COMPAT define
5161 accordingly, but in most cases should be able to continue to
5162 compile with later releases.
5164 The OPENSSL_API_COMPAT versions for 1.0.0, and 0.9.8 are
5165 0x10000000L and 0x00908000L, respectively. However those
5166 versions did not support the OPENSSL_API_COMPAT feature, and
5167 so applications are not typically tested for explicit support
5168 of just the undeprecated features of either release.
5172 * Add support for setting the minimum and maximum supported protocol.
5173 It can bet set via the SSL_set_min_proto_version() and
5174 SSL_set_max_proto_version(), or via the SSL_CONF's MinProtocol and
5175 MaxProtocol. It's recommended to use the new APIs to disable
5176 protocols instead of disabling individual protocols using
5177 SSL_set_options() or SSL_CONF's Protocol. This change also
5178 removes support for disabling TLS 1.2 in the OpenSSL TLS
5179 client at compile time by defining OPENSSL_NO_TLS1_2_CLIENT.
5183 * Support for ChaCha20 and Poly1305 added to libcrypto and libssl.
5187 * New EC_KEY_METHOD, this replaces the older ECDSA_METHOD and ECDH_METHOD
5188 and integrates ECDSA and ECDH functionality into EC. Implementations can
5189 now redirect key generation and no longer need to convert to or from
5192 Note: the ecdsa.h and ecdh.h headers are now no longer needed and just
5193 include the ec.h header file instead.
5197 * Remove support for all 40 and 56 bit ciphers. This includes all the export
5198 ciphers who are no longer supported and drops support the ephemeral RSA key
5199 exchange. The LOW ciphers currently doesn't have any ciphers in it.
5203 * Made EVP_MD_CTX, EVP_MD, EVP_CIPHER_CTX, EVP_CIPHER and HMAC_CTX
5204 opaque. For HMAC_CTX, the following constructors and destructors
5207 HMAC_CTX *HMAC_CTX_new(void);
5208 void HMAC_CTX_free(HMAC_CTX *ctx);
5210 For EVP_MD and EVP_CIPHER, complete APIs to create, fill and
5211 destroy such methods has been added. See EVP_MD_meth_new(3) and
5212 EVP_CIPHER_meth_new(3) for documentation.
5215 1) `EVP_MD_CTX_cleanup()`, `EVP_CIPHER_CTX_cleanup()` and
5216 `HMAC_CTX_cleanup()` were removed. `HMAC_CTX_reset()` and
5217 `EVP_MD_CTX_reset()` should be called instead to reinitialise
5218 an already created structure.
5219 2) For consistency with the majority of our object creators and
5220 destructors, `EVP_MD_CTX_(create|destroy)` were renamed to
5221 `EVP_MD_CTX_(new|free)`. The old names are retained as macros
5222 for deprecated builds.
5226 * Added ASYNC support. Libcrypto now includes the async sub-library to enable
5227 cryptographic operations to be performed asynchronously as long as an
5228 asynchronous capable engine is used. See the ASYNC_start_job() man page for
5229 further details. Libssl has also had this capability integrated with the
5230 introduction of the new mode SSL_MODE_ASYNC and associated error
5231 SSL_ERROR_WANT_ASYNC. See the SSL_CTX_set_mode() and SSL_get_error() man
5232 pages. This work was developed in partnership with Intel Corp.
5236 * SSL_{CTX_}set_ecdh_auto() has been removed and ECDH is support is
5237 always enabled now. If you want to disable the support you should
5238 exclude it using the list of supported ciphers. This also means that the
5239 "-no_ecdhe" option has been removed from s_server.
5243 * SSL_{CTX}_set_tmp_ecdh() which can set 1 EC curve now internally calls
5244 SSL_{CTX_}set1_curves() which can set a list.
5248 * Remove support for SSL_{CTX_}set_tmp_ecdh_callback(). You should set the
5249 curve you want to support using SSL_{CTX_}set1_curves().
5253 * State machine rewrite. The state machine code has been significantly
5254 refactored in order to remove much duplication of code and solve issues
5255 with the old code (see [ssl/statem/README.md](ssl/statem/README.md) for
5256 further details). This change does have some associated API changes.
5257 Notably the SSL_state() function has been removed and replaced by
5258 SSL_get_state which now returns an "OSSL_HANDSHAKE_STATE" instead of an int.
5259 SSL_set_state() has been removed altogether. The previous handshake states
5260 defined in ssl.h and ssl3.h have also been removed.
5264 * All instances of the string "ssleay" in the public API were replaced
5265 with OpenSSL (case-matching; e.g., OPENSSL_VERSION for #define's)
5266 Some error codes related to internal RSA_eay API's were renamed.
5270 * The demo files in crypto/threads were moved to demo/threads.
5274 * Removed obsolete engines: 4758cca, aep, atalla, cswift, nuron, gmp,
5277 *Matt Caswell, Rich Salz*
5279 * New ASN.1 embed macro.
5281 New ASN.1 macro ASN1_EMBED. This is the same as ASN1_SIMPLE except the
5282 structure is not allocated: it is part of the parent. That is instead of
5290 This reduces memory fragmentation and make it impossible to accidentally
5291 set a mandatory field to NULL.
5293 This currently only works for some fields specifically a SEQUENCE, CHOICE,
5294 or ASN1_STRING type which is part of a parent SEQUENCE. Since it is
5295 equivalent to ASN1_SIMPLE it cannot be tagged, OPTIONAL, SET OF or
5300 * Remove EVP_CHECK_DES_KEY, a compile-time option that never compiled.
5304 * Removed DES and RC4 ciphersuites from DEFAULT. Also removed RC2 although
5305 in 1.0.2 EXPORT was already removed and the only RC2 ciphersuite is also
5306 an EXPORT one. COMPLEMENTOFDEFAULT has been updated accordingly to add
5307 DES and RC4 ciphersuites.
5311 * Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs.
5312 This changes the decoding behaviour for some invalid messages,
5313 though the change is mostly in the more lenient direction, and
5314 legacy behaviour is preserved as much as possible.
5318 * Fix no-stdio build.
5319 *David Woodhouse <David.Woodhouse@intel.com> and also*
5320 *Ivan Nestlerode <ivan.nestlerode@sonos.com>*
5322 * New testing framework
5323 The testing framework has been largely rewritten and is now using
5324 perl and the perl modules Test::Harness and an extended variant of
5325 Test::More called OpenSSL::Test to do its work. All test scripts in
5326 test/ have been rewritten into test recipes, and all direct calls to
5327 executables in test/Makefile have become individual recipes using the
5328 simplified testing OpenSSL::Test::Simple.
5330 For documentation on our testing modules, do:
5332 perldoc test/testlib/OpenSSL/Test/Simple.pm
5333 perldoc test/testlib/OpenSSL/Test.pm
5337 * Revamped memory debug; only -DCRYPTO_MDEBUG and -DCRYPTO_MDEBUG_ABORT
5338 are used; the latter aborts on memory leaks (usually checked on exit).
5339 Some undocumented "set malloc, etc., hooks" functions were removed
5340 and others were changed. All are now documented.
5344 * In DSA_generate_parameters_ex, if the provided seed is too short,
5347 *Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>*
5349 * Rewrite PSK to support ECDHE_PSK, DHE_PSK and RSA_PSK. Add ciphersuites
5350 from RFC4279, RFC4785, RFC5487, RFC5489.
5352 Thanks to Christian J. Dietrich and Giuseppe D'Angelo for the
5353 original RSA_PSK patch.
5357 * Dropped support for the SSL3_FLAGS_DELAY_CLIENT_FINISHED flag. This SSLeay
5358 era flag was never set throughout the codebase (only read). Also removed
5359 SSL3_FLAGS_POP_BUFFER which was only used if
5360 SSL3_FLAGS_DELAY_CLIENT_FINISHED was also set.
5364 * Changed the default name options in the "ca", "crl", "req" and "x509"
5365 to be "oneline" instead of "compat".
5369 * Remove SSL_OP_TLS_BLOCK_PADDING_BUG. This is SSLeay legacy, we're
5370 not aware of clients that still exhibit this bug, and the workaround
5371 hasn't been working properly for a while.
5375 * The return type of BIO_number_read() and BIO_number_written() as well as
5376 the corresponding num_read and num_write members in the BIO structure has
5377 changed from unsigned long to uint64_t. On platforms where an unsigned
5378 long is 32 bits (e.g. Windows) these counters could overflow if >4Gb is
5383 * Given the pervasive nature of TLS extensions it is inadvisable to run
5384 OpenSSL without support for them. It also means that maintaining
5385 the OPENSSL_NO_TLSEXT option within the code is very invasive (and probably
5386 not well tested). Therefore, the OPENSSL_NO_TLSEXT option has been removed.
5390 * Removed support for the two export grade static DH ciphersuites
5391 EXP-DH-RSA-DES-CBC-SHA and EXP-DH-DSS-DES-CBC-SHA. These two ciphersuites
5392 were newly added (along with a number of other static DH ciphersuites) to
5393 1.0.2. However the two export ones have *never* worked since they were
5394 introduced. It seems strange in any case to be adding new export
5395 ciphersuites, and given "logjam" it also does not seem correct to fix them.
5399 * Version negotiation has been rewritten. In particular SSLv23_method(),
5400 SSLv23_client_method() and SSLv23_server_method() have been deprecated,
5401 and turned into macros which simply call the new preferred function names
5402 TLS_method(), TLS_client_method() and TLS_server_method(). All new code
5403 should use the new names instead. Also as part of this change the ssl23.h
5404 header file has been removed.
5408 * Support for Kerberos ciphersuites in TLS (RFC2712) has been removed. This
5409 code and the associated standard is no longer considered fit-for-purpose.
5413 * RT2547 was closed. When generating a private key, try to make the
5414 output file readable only by the owner. This behavior change might
5415 be noticeable when interacting with other software.
5417 * Documented all exdata functions. Added CRYPTO_free_ex_index.
5422 * Added HTTP GET support to the ocsp command.
5426 * Changed default digest for the dgst and enc commands from MD5 to
5431 * RAND_pseudo_bytes has been deprecated. Users should use RAND_bytes instead.
5435 * Added support for TLS extended master secret from
5436 draft-ietf-tls-session-hash-03.txt. Thanks for Alfredo Pironti for an
5437 initial patch which was a great help during development.
5441 * All libssl internal structures have been removed from the public header
5442 files, and the OPENSSL_NO_SSL_INTERN option has been removed (since it is
5443 now redundant). Users should not attempt to access internal structures
5444 directly. Instead they should use the provided API functions.
5448 * config has been changed so that by default OPENSSL_NO_DEPRECATED is used.
5449 Access to deprecated functions can be re-enabled by running config with
5450 "enable-deprecated". In addition applications wishing to use deprecated
5451 functions must define OPENSSL_USE_DEPRECATED. Note that this new behaviour
5452 will, by default, disable some transitive includes that previously existed
5453 in the header files (e.g. ec.h will no longer, by default, include bn.h)
5457 * Added support for OCB mode. OpenSSL has been granted a patent license
5458 compatible with the OpenSSL license for use of OCB. Details are available
5459 at <https://www.openssl.org/source/OCB-patent-grant-OpenSSL.pdf>. Support
5460 for OCB can be removed by calling config with no-ocb.
5464 * SSLv2 support has been removed. It still supports receiving an SSLv2
5465 compatible client hello.
5469 * Increased the minimal RSA keysize from 256 to 512 bits [Rich Salz],
5470 done while fixing the error code for the key-too-small case.
5472 *Annie Yousar <a.yousar@informatik.hu-berlin.de>*
5474 * CA.sh has been removed; use CA.pl instead.
5478 * Removed old DES API.
5482 * Remove various unsupported platforms:
5488 Sinix/ReliantUNIX RM400
5493 16-bit platforms such as WIN16
5497 * Clean up OPENSSL_NO_xxx #define's
5498 - Use setbuf() and remove OPENSSL_NO_SETVBUF_IONBF
5499 - Rename OPENSSL_SYSNAME_xxx to OPENSSL_SYS_xxx
5500 - OPENSSL_NO_EC{DH,DSA} merged into OPENSSL_NO_EC
5501 - OPENSSL_NO_RIPEMD160, OPENSSL_NO_RIPEMD merged into OPENSSL_NO_RMD160
5502 - OPENSSL_NO_FP_API merged into OPENSSL_NO_STDIO
5503 - Remove OPENSSL_NO_BIO OPENSSL_NO_BUFFER OPENSSL_NO_CHAIN_VERIFY
5504 OPENSSL_NO_EVP OPENSSL_NO_FIPS_ERR OPENSSL_NO_HASH_COMP
5505 OPENSSL_NO_LHASH OPENSSL_NO_OBJECT OPENSSL_NO_SPEED OPENSSL_NO_STACK
5506 OPENSSL_NO_X509 OPENSSL_NO_X509_VERIFY
5507 - Remove MS_STATIC; it's a relic from platforms <32 bits.
5511 * Cleaned up dead code
5512 Remove all but one '#ifdef undef' which is to be looked at.
5516 * Clean up calling of xxx_free routines.
5517 Just like free(), fix most of the xxx_free routines to accept
5518 NULL. Remove the non-null checks from callers. Save much code.
5522 * Add secure heap for storage of private keys (when possible).
5523 Add BIO_s_secmem(), CBIGNUM, etc.
5524 Contributed by Akamai Technologies under our Corporate CLA.
5528 * Experimental support for a new, fast, unbiased prime candidate generator,
5529 bn_probable_prime_dh_coprime(). Not currently used by any prime generator.
5531 *Felix Laurie von Massenbach <felix@erbridge.co.uk>*
5533 * New output format NSS in the sess_id command line tool. This allows
5534 exporting the session id and the master key in NSS keylog format.
5536 *Martin Kaiser <martin@kaiser.cx>*
5538 * Harmonize version and its documentation. -f flag is used to display
5541 *mancha <mancha1@zoho.com>*
5543 * Fix eckey_priv_encode so it immediately returns an error upon a failure
5544 in i2d_ECPrivateKey. Thanks to Ted Unangst for feedback on this issue.
5546 *mancha <mancha1@zoho.com>*
5548 * Fix some double frees. These are not thought to be exploitable.
5550 *mancha <mancha1@zoho.com>*
5552 * A missing bounds check in the handling of the TLS heartbeat extension
5553 can be used to reveal up to 64k of memory to a connected client or
5556 Thanks for Neel Mehta of Google Security for discovering this bug and to
5557 Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for
5558 preparing the fix ([CVE-2014-0160])
5560 *Adam Langley, Bodo Moeller*
5562 * Fix for the attack described in the paper "Recovering OpenSSL
5563 ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack"
5564 by Yuval Yarom and Naomi Benger. Details can be obtained from:
5565 <http://eprint.iacr.org/2014/140>
5567 Thanks to Yuval Yarom and Naomi Benger for discovering this
5568 flaw and to Yuval Yarom for supplying a fix ([CVE-2014-0076])
5570 *Yuval Yarom and Naomi Benger*
5572 * Use algorithm specific chains in SSL_CTX_use_certificate_chain_file():
5573 this fixes a limitation in previous versions of OpenSSL.
5577 * Experimental encrypt-then-mac support.
5579 Experimental support for encrypt then mac from
5580 draft-gutmann-tls-encrypt-then-mac-02.txt
5582 To enable it set the appropriate extension number (0x42 for the test
5583 server) using e.g. -DTLSEXT_TYPE_encrypt_then_mac=0x42
5585 For non-compliant peers (i.e. just about everything) this should have no
5588 WARNING: EXPERIMENTAL, SUBJECT TO CHANGE.
5592 * Add EVP support for key wrapping algorithms, to avoid problems with
5593 existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in
5594 the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap
5595 algorithms and include tests cases.
5599 * Extend CMS code to support RSA-PSS signatures and RSA-OAEP for
5604 * Extended RSA OAEP support via EVP_PKEY API. Options to specify digest,
5605 MGF1 digest and OAEP label.
5609 * Make openssl verify return errors.
5611 *Chris Palmer <palmer@google.com> and Ben Laurie*
5613 * New function ASN1_TIME_diff to calculate the difference between two
5614 ASN1_TIME structures or one structure and the current time.
5618 * Update fips_test_suite to support multiple command line options. New
5619 test to induce all self test errors in sequence and check expected
5624 * Add FIPS_{rsa,dsa,ecdsa}_{sign,verify} functions which digest and
5625 sign or verify all in one operation.
5629 * Add fips_algvs: a multicall fips utility incorporating all the algorithm
5630 test programs and fips_test_suite. Includes functionality to parse
5631 the minimal script output of fipsalgest.pl directly.
5635 * Add authorisation parameter to FIPS_module_mode_set().
5639 * Add FIPS selftest for ECDH algorithm using P-224 and B-233 curves.
5643 * Use separate DRBG fields for internal and external flags. New function
5644 FIPS_drbg_health_check() to perform on demand health checking. Add
5645 generation tests to fips_test_suite with reduced health check interval to
5646 demonstrate periodic health checking. Add "nodh" option to
5647 fips_test_suite to skip very slow DH test.
5651 * New function FIPS_get_cipherbynid() to lookup FIPS supported ciphers
5656 * More extensive health check for DRBG checking many more failure modes.
5657 New function FIPS_selftest_drbg_all() to handle every possible DRBG
5658 combination: call this in fips_test_suite.
5662 * Add support for canonical generation of DSA parameter 'g'. See
5665 * Add support for HMAC DRBG from SP800-90. Update DRBG algorithm test and
5666 POST to handle HMAC cases.
5670 * Add functions FIPS_module_version() and FIPS_module_version_text()
5671 to return numerical and string versions of the FIPS module number.
5675 * Rename FIPS_mode_set and FIPS_mode to FIPS_module_mode_set and
5676 FIPS_module_mode. FIPS_mode and FIPS_mode_set will be implemented
5677 outside the validated module in the FIPS capable OpenSSL.
5681 * Minor change to DRBG entropy callback semantics. In some cases
5682 there is no multiple of the block length between min_len and
5683 max_len. Allow the callback to return more than max_len bytes
5684 of entropy but discard any extra: it is the callback's responsibility
5685 to ensure that the extra data discarded does not impact the
5686 requested amount of entropy.
5690 * Add PRNG security strength checks to RSA, DSA and ECDSA using
5691 information in FIPS186-3, SP800-57 and SP800-131A.
5695 * CCM support via EVP. Interface is very similar to GCM case except we
5696 must supply all data in one chunk (i.e. no update, final) and the
5697 message length must be supplied if AAD is used. Add algorithm test
5702 * Initial version of POST overhaul. Add POST callback to allow the status
5703 of POST to be monitored and/or failures induced. Modify fips_test_suite
5704 to use callback. Always run all selftests even if one fails.
5708 * XTS support including algorithm test driver in the fips_gcmtest program.
5709 Note: this does increase the maximum key length from 32 to 64 bytes but
5710 there should be no binary compatibility issues as existing applications
5711 will never use XTS mode.
5715 * Extensive reorganisation of FIPS PRNG behaviour. Remove all dependencies
5716 to OpenSSL RAND code and replace with a tiny FIPS RAND API which also
5717 performs algorithm blocking for unapproved PRNG types. Also do not
5718 set PRNG type in FIPS_mode_set(): leave this to the application.
5719 Add default OpenSSL DRBG handling: sets up FIPS PRNG and seeds with
5720 the standard OpenSSL PRNG: set additional data to a date time vector.
5724 * Rename old X9.31 PRNG functions of the form `FIPS_rand*` to `FIPS_x931*`.
5725 This shouldn't present any incompatibility problems because applications
5726 shouldn't be using these directly and any that are will need to rethink
5727 anyway as the X9.31 PRNG is now deprecated by FIPS 140-2
5731 * Extensive self tests and health checking required by SP800-90 DRBG.
5732 Remove strength parameter from FIPS_drbg_instantiate and always
5733 instantiate at maximum supported strength.
5737 * Add ECDH code to fips module and fips_ecdhvs for primitives only testing.
5741 * New algorithm test program fips_dhvs to handle DH primitives only testing.
5745 * New function DH_compute_key_padded() to compute a DH key and pad with
5746 leading zeroes if needed: this complies with SP800-56A et al.
5750 * Initial implementation of SP800-90 DRBGs for Hash and CTR. Not used by
5751 anything, incomplete, subject to change and largely untested at present.
5755 * Modify fipscanisteronly build option to only build the necessary object
5756 files by filtering FIPS_EX_OBJ through a perl script in crypto/Makefile.
5760 * Add experimental option FIPSSYMS to give all symbols in
5761 fipscanister.o and FIPS or fips prefix. This will avoid
5762 conflicts with future versions of OpenSSL. Add perl script
5763 util/fipsas.pl to preprocess assembly language source files
5764 and rename any affected symbols.
5768 * Add selftest checks and algorithm block of non-fips algorithms in
5769 FIPS mode. Remove DES2 from selftests.
5773 * Add ECDSA code to fips module. Add tiny fips_ecdsa_check to just
5774 return internal method without any ENGINE dependencies. Add new
5775 tiny fips sign and verify functions.
5779 * New build option no-ec2m to disable characteristic 2 code.
5783 * New build option "fipscanisteronly". This only builds fipscanister.o
5784 and (currently) associated fips utilities. Uses the file Makefile.fips
5785 instead of Makefile.org as the prototype.
5789 * Add some FIPS mode restrictions to GCM. Add internal IV generator.
5790 Update fips_gcmtest to use IV generator.
5794 * Initial, experimental EVP support for AES-GCM. AAD can be input by
5795 setting output buffer to NULL. The `*Final` function must be
5796 called although it will not retrieve any additional data. The tag
5797 can be set or retrieved with a ctrl. The IV length is by default 12
5798 bytes (96 bits) but can be set to an alternative value. If the IV
5799 length exceeds the maximum IV length (currently 16 bytes) it cannot be
5804 * New flag in ciphers: EVP_CIPH_FLAG_CUSTOM_CIPHER. This means the
5805 underlying do_cipher function handles all cipher semantics itself
5806 including padding and finalisation. This is useful if (for example)
5807 an ENGINE cipher handles block padding itself. The behaviour of
5808 do_cipher is subtly changed if this flag is set: the return value
5809 is the number of characters written to the output buffer (zero is
5810 no longer an error code) or a negative error code. Also if the
5811 input buffer is NULL and length 0 finalisation should be performed.
5815 * If a candidate issuer certificate is already part of the constructed
5816 path ignore it: new debug notification X509_V_ERR_PATH_LOOP for this case.
5820 * Improve forward-security support: add functions
5822 void SSL_CTX_set_not_resumable_session_callback(
5823 SSL_CTX *ctx, int (*cb)(SSL *ssl, int is_forward_secure))
5824 void SSL_set_not_resumable_session_callback(
5825 SSL *ssl, int (*cb)(SSL *ssl, int is_forward_secure))
5827 for use by SSL/TLS servers; the callback function will be called whenever a
5828 new session is created, and gets to decide whether the session may be
5829 cached to make it resumable (return 0) or not (return 1). (As by the
5830 SSL/TLS protocol specifications, the session_id sent by the server will be
5831 empty to indicate that the session is not resumable; also, the server will
5832 not generate RFC 4507 (RFC 5077) session tickets.)
5834 A simple reasonable callback implementation is to return is_forward_secure.
5835 This parameter will be set to 1 or 0 depending on the ciphersuite selected
5836 by the SSL/TLS server library, indicating whether it can provide forward
5839 *Emilia Käsper <emilia.kasper@esat.kuleuven.be> (Google)*
5841 * New -verify_name option in command line utilities to set verification
5846 * Initial CMAC implementation. WARNING: EXPERIMENTAL, API MAY CHANGE.
5847 Add CMAC pkey methods.
5851 * Experimental renegotiation in s_server -www mode. If the client
5852 browses /reneg connection is renegotiated. If /renegcert it is
5853 renegotiated requesting a certificate.
5857 * Add an "external" session cache for debugging purposes to s_server. This
5858 should help trace issues which normally are only apparent in deployed
5859 multi-process servers.
5863 * Extensive audit of libcrypto with DEBUG_UNUSED. Fix many cases where
5864 return value is ignored. NB. The functions RAND_add(), RAND_seed(),
5865 BIO_set_cipher() and some obscure PEM functions were changed so they
5866 can now return an error. The RAND changes required a change to the
5867 RAND_METHOD structure.
5871 * New macro `__owur` for "OpenSSL Warn Unused Result". This makes use of
5872 a gcc attribute to warn if the result of a function is ignored. This
5873 is enable if DEBUG_UNUSED is set. Add to several functions in evp.h
5874 whose return value is often ignored.
5878 * New -noct, -requestct, -requirect and -ctlogfile options for s_client.
5879 These allow SCTs (signed certificate timestamps) to be requested and
5880 validated when establishing a connection.
5882 *Rob Percival <robpercival@google.com>*
5887 ### Changes between 1.0.2s and 1.0.2t [10 Sep 2019]
5889 * For built-in EC curves, ensure an EC_GROUP built from the curve name is
5890 used even when parsing explicit parameters, when loading a encoded key
5891 or calling `EC_GROUP_new_from_ecpkparameters()`/
5892 `EC_GROUP_new_from_ecparameters()`.
5893 This prevents bypass of security hardening and performance gains,
5894 especially for curves with specialized EC_METHODs.
5895 By default, if a key encoded with explicit parameters is loaded and later
5896 encoded, the output is still encoded with explicit parameters, even if
5897 internally a "named" EC_GROUP is used for computation.
5901 * Compute ECC cofactors if not provided during EC_GROUP construction. Before
5902 this change, EC_GROUP_set_generator would accept order and/or cofactor as
5903 NULL. After this change, only the cofactor parameter can be NULL. It also
5904 does some minimal sanity checks on the passed order.
5909 * Fixed a padding oracle in PKCS7_dataDecode and CMS_decrypt_set1_pkey.
5910 An attack is simple, if the first CMS_recipientInfo is valid but the
5911 second CMS_recipientInfo is chosen ciphertext. If the second
5912 recipientInfo decodes to PKCS #1 v1.5 form plaintext, the correct
5913 encryption key will be replaced by garbage, and the message cannot be
5914 decoded, but if the RSA decryption fails, the correct encryption key is
5915 used and the recipient will not notice the attack.
5916 As a work around for this potential attack the length of the decrypted
5917 key must be equal to the cipher default key length, in case the
5918 certificate is not given and all recipientInfo are tried out.
5919 The old behaviour can be re-enabled in the CMS code by setting the
5920 CMS_DEBUG_DECRYPT flag.
5925 * Document issue with installation paths in diverse Windows builds
5927 '/usr/local/ssl' is an unsafe prefix for location to install OpenSSL
5928 binaries and run-time config file.
5933 ### Changes between 1.0.2r and 1.0.2s [28 May 2019]
5935 * Change the default RSA, DSA and DH size to 2048 bit instead of 1024.
5936 This changes the size when using the `genpkey` command when no size is given.
5937 It fixes an omission in earlier changes that changed all RSA, DSA and DH
5938 generation commands to use 2048 bits by default.
5942 * Add FIPS support for Android Arm 64-bit
5944 Support for Android Arm 64-bit was added to the OpenSSL FIPS Object
5945 Module in Version 2.0.10. For some reason, the corresponding target
5946 'android64-aarch64' was missing OpenSSL 1.0.2, whence it could not be
5947 built with FIPS support on Android Arm 64-bit. This omission has been
5950 *Matthias St. Pierre*
5952 ### Changes between 1.0.2q and 1.0.2r [26 Feb 2019]
5954 * 0-byte record padding oracle
5956 If an application encounters a fatal protocol error and then calls
5957 SSL_shutdown() twice (once to send a close_notify, and once to receive one)
5958 then OpenSSL can respond differently to the calling application if a 0 byte
5959 record is received with invalid padding compared to if a 0 byte record is
5960 received with an invalid MAC. If the application then behaves differently
5961 based on that in a way that is detectable to the remote peer, then this
5962 amounts to a padding oracle that could be used to decrypt data.
5964 In order for this to be exploitable "non-stitched" ciphersuites must be in
5965 use. Stitched ciphersuites are optimised implementations of certain
5966 commonly used ciphersuites. Also the application must call SSL_shutdown()
5967 twice even if a protocol error has occurred (applications should not do
5968 this but some do anyway).
5970 This issue was discovered by Juraj Somorovsky, Robert Merget and Nimrod
5971 Aviram, with additional investigation by Steven Collison and Andrew
5972 Hourselt. It was reported to OpenSSL on 10th December 2018.
5977 * Move strictness check from EVP_PKEY_asn1_new() to EVP_PKEY_asn1_add0().
5981 ### Changes between 1.0.2p and 1.0.2q [20 Nov 2018]
5983 * Microarchitecture timing vulnerability in ECC scalar multiplication
5985 OpenSSL ECC scalar multiplication, used in e.g. ECDSA and ECDH, has been
5986 shown to be vulnerable to a microarchitecture timing side channel attack.
5987 An attacker with sufficient access to mount local timing attacks during
5988 ECDSA signature generation could recover the private key.
5990 This issue was reported to OpenSSL on 26th October 2018 by Alejandro
5991 Cabrera Aldaya, Billy Brumley, Sohaib ul Hassan, Cesar Pereida Garcia and
5997 * Timing vulnerability in DSA signature generation
5999 The OpenSSL DSA signature algorithm has been shown to be vulnerable to a
6000 timing side channel attack. An attacker could use variations in the signing
6001 algorithm to recover the private key.
6003 This issue was reported to OpenSSL on 16th October 2018 by Samuel Weiser.
6008 * Resolve a compatibility issue in EC_GROUP handling with the FIPS Object
6009 Module, accidentally introduced while backporting security fixes from the
6010 development branch and hindering the use of ECC in FIPS mode.
6014 ### Changes between 1.0.2o and 1.0.2p [14 Aug 2018]
6016 * Client DoS due to large DH parameter
6018 During key agreement in a TLS handshake using a DH(E) based ciphersuite a
6019 malicious server can send a very large prime value to the client. This will
6020 cause the client to spend an unreasonably long period of time generating a
6021 key for this prime resulting in a hang until the client has finished. This
6022 could be exploited in a Denial Of Service attack.
6024 This issue was reported to OpenSSL on 5th June 2018 by Guido Vranken
6029 * Cache timing vulnerability in RSA Key Generation
6031 The OpenSSL RSA Key generation algorithm has been shown to be vulnerable to
6032 a cache timing side channel attack. An attacker with sufficient access to
6033 mount cache timing attacks during the RSA key generation process could
6034 recover the private key.
6036 This issue was reported to OpenSSL on 4th April 2018 by Alejandro Cabrera
6037 Aldaya, Billy Brumley, Cesar Pereida Garcia and Luis Manuel Alvarez Tapia.
6042 * Make EVP_PKEY_asn1_new() a bit stricter about its input. A NULL pem_str
6043 parameter is no longer accepted, as it leads to a corrupt table. NULL
6044 pem_str is reserved for alias entries only.
6048 * Revert blinding in ECDSA sign and instead make problematic addition
6049 length-invariant. Switch even to fixed-length Montgomery multiplication.
6053 * Change generating and checking of primes so that the error rate of not
6054 being prime depends on the intended use based on the size of the input.
6055 For larger primes this will result in more rounds of Miller-Rabin.
6056 The maximal error rate for primes with more than 1080 bits is lowered
6059 *Kurt Roeckx, Annie Yousar*
6061 * Increase the number of Miller-Rabin rounds for DSA key generating to 64.
6065 * Add blinding to ECDSA and DSA signatures to protect against side channel
6066 attacks discovered by Keegan Ryan (NCC Group).
6070 * When unlocking a pass phrase protected PEM file or PKCS#8 container, we
6071 now allow empty (zero character) pass phrases.
6075 * Certificate time validation (X509_cmp_time) enforces stricter
6076 compliance with RFC 5280. Fractional seconds and timezone offsets
6077 are no longer allowed.
6081 ### Changes between 1.0.2n and 1.0.2o [27 Mar 2018]
6083 * Constructed ASN.1 types with a recursive definition could exceed the stack
6085 Constructed ASN.1 types with a recursive definition (such as can be found
6086 in PKCS7) could eventually exceed the stack given malicious input with
6087 excessive recursion. This could result in a Denial Of Service attack. There
6088 are no such structures used within SSL/TLS that come from untrusted sources
6089 so this is considered safe.
6091 This issue was reported to OpenSSL on 4th January 2018 by the OSS-fuzz
6097 ### Changes between 1.0.2m and 1.0.2n [7 Dec 2017]
6099 * Read/write after SSL object in error state
6101 OpenSSL 1.0.2 (starting from version 1.0.2b) introduced an "error state"
6102 mechanism. The intent was that if a fatal error occurred during a handshake
6103 then OpenSSL would move into the error state and would immediately fail if
6104 you attempted to continue the handshake. This works as designed for the
6105 explicit handshake functions (SSL_do_handshake(), SSL_accept() and
6106 SSL_connect()), however due to a bug it does not work correctly if
6107 SSL_read() or SSL_write() is called directly. In that scenario, if the
6108 handshake fails then a fatal error will be returned in the initial function
6109 call. If SSL_read()/SSL_write() is subsequently called by the application
6110 for the same SSL object then it will succeed and the data is passed without
6111 being decrypted/encrypted directly from the SSL/TLS record layer.
6113 In order to exploit this issue an application bug would have to be present
6114 that resulted in a call to SSL_read()/SSL_write() being issued after having
6115 already received a fatal error.
6117 This issue was reported to OpenSSL by David Benjamin (Google).
6122 * rsaz_1024_mul_avx2 overflow bug on x86_64
6124 There is an overflow bug in the AVX2 Montgomery multiplication procedure
6125 used in exponentiation with 1024-bit moduli. No EC algorithms are affected.
6126 Analysis suggests that attacks against RSA and DSA as a result of this
6127 defect would be very difficult to perform and are not believed likely.
6128 Attacks against DH1024 are considered just feasible, because most of the
6129 work necessary to deduce information about a private key may be performed
6130 offline. The amount of resources required for such an attack would be
6131 significant. However, for an attack on TLS to be meaningful, the server
6132 would have to share the DH1024 private key among multiple clients, which is
6133 no longer an option since CVE-2016-0701.
6135 This only affects processors that support the AVX2 but not ADX extensions
6136 like Intel Haswell (4th generation).
6138 This issue was reported to OpenSSL by David Benjamin (Google). The issue
6139 was originally found via the OSS-Fuzz project.
6144 ### Changes between 1.0.2l and 1.0.2m [2 Nov 2017]
6146 * bn_sqrx8x_internal carry bug on x86_64
6148 There is a carry propagating bug in the x86_64 Montgomery squaring
6149 procedure. No EC algorithms are affected. Analysis suggests that attacks
6150 against RSA and DSA as a result of this defect would be very difficult to
6151 perform and are not believed likely. Attacks against DH are considered just
6152 feasible (although very difficult) because most of the work necessary to
6153 deduce information about a private key may be performed offline. The amount
6154 of resources required for such an attack would be very significant and
6155 likely only accessible to a limited number of attackers. An attacker would
6156 additionally need online access to an unpatched system using the target
6157 private key in a scenario with persistent DH parameters and a private
6158 key that is shared between multiple clients.
6160 This only affects processors that support the BMI1, BMI2 and ADX extensions
6161 like Intel Broadwell (5th generation) and later or AMD Ryzen.
6163 This issue was reported to OpenSSL by the OSS-Fuzz project.
6168 * Malformed X.509 IPAddressFamily could cause OOB read
6170 If an X.509 certificate has a malformed IPAddressFamily extension,
6171 OpenSSL could do a one-byte buffer overread. The most likely result
6172 would be an erroneous display of the certificate in text format.
6174 This issue was reported to OpenSSL by the OSS-Fuzz project.
6178 ### Changes between 1.0.2k and 1.0.2l [25 May 2017]
6180 * Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target
6181 platform rather than 'mingw'.
6185 ### Changes between 1.0.2j and 1.0.2k [26 Jan 2017]
6187 * Truncated packet could crash via OOB read
6189 If one side of an SSL/TLS path is running on a 32-bit host and a specific
6190 cipher is being used, then a truncated packet can cause that host to
6191 perform an out-of-bounds read, usually resulting in a crash.
6193 This issue was reported to OpenSSL by Robert Święcki of Google.
6198 * BN_mod_exp may produce incorrect results on x86_64
6200 There is a carry propagating bug in the x86_64 Montgomery squaring
6201 procedure. No EC algorithms are affected. Analysis suggests that attacks
6202 against RSA and DSA as a result of this defect would be very difficult to
6203 perform and are not believed likely. Attacks against DH are considered just
6204 feasible (although very difficult) because most of the work necessary to
6205 deduce information about a private key may be performed offline. The amount
6206 of resources required for such an attack would be very significant and
6207 likely only accessible to a limited number of attackers. An attacker would
6208 additionally need online access to an unpatched system using the target
6209 private key in a scenario with persistent DH parameters and a private
6210 key that is shared between multiple clients. For example this can occur by
6211 default in OpenSSL DHE based SSL/TLS ciphersuites. Note: This issue is very
6212 similar to CVE-2015-3193 but must be treated as a separate problem.
6214 This issue was reported to OpenSSL by the OSS-Fuzz project.
6219 * Montgomery multiplication may produce incorrect results
6221 There is a carry propagating bug in the Broadwell-specific Montgomery
6222 multiplication procedure that handles input lengths divisible by, but
6223 longer than 256 bits. Analysis suggests that attacks against RSA, DSA
6224 and DH private keys are impossible. This is because the subroutine in
6225 question is not used in operations with the private key itself and an input
6226 of the attacker's direct choice. Otherwise the bug can manifest itself as
6227 transient authentication and key negotiation failures or reproducible
6228 erroneous outcome of public-key operations with specially crafted input.
6229 Among EC algorithms only Brainpool P-512 curves are affected and one
6230 presumably can attack ECDH key negotiation. Impact was not analyzed in
6231 detail, because pre-requisites for attack are considered unlikely. Namely
6232 multiple clients have to choose the curve in question and the server has to
6233 share the private key among them, neither of which is default behaviour.
6234 Even then only clients that chose the curve will be affected.
6236 This issue was publicly reported as transient failures and was not
6237 initially recognized as a security issue. Thanks to Richard Morgan for
6238 providing reproducible case.
6243 * OpenSSL now fails if it receives an unrecognised record type in TLS1.0
6244 or TLS1.1. Previously this only happened in SSLv3 and TLS1.2. This is to
6245 prevent issues where no progress is being made and the peer continually
6246 sends unrecognised record types, using up resources processing them.
6250 ### Changes between 1.0.2i and 1.0.2j [26 Sep 2016]
6252 * Missing CRL sanity check
6254 A bug fix which included a CRL sanity check was added to OpenSSL 1.1.0
6255 but was omitted from OpenSSL 1.0.2i. As a result any attempt to use
6256 CRLs in OpenSSL 1.0.2i will crash with a null pointer exception.
6258 This issue only affects the OpenSSL 1.0.2i
6263 ### Changes between 1.0.2h and 1.0.2i [22 Sep 2016]
6265 * OCSP Status Request extension unbounded memory growth
6267 A malicious client can send an excessively large OCSP Status Request
6268 extension. If that client continually requests renegotiation, sending a
6269 large OCSP Status Request extension each time, then there will be unbounded
6270 memory growth on the server. This will eventually lead to a Denial Of
6271 Service attack through memory exhaustion. Servers with a default
6272 configuration are vulnerable even if they do not support OCSP. Builds using
6273 the "no-ocsp" build time option are not affected.
6275 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
6280 * In order to mitigate the SWEET32 attack, the DES ciphers were moved from
6283 This issue was reported to OpenSSL Karthikeyan Bhargavan and Gaetan
6289 * OOB write in MDC2_Update()
6291 An overflow can occur in MDC2_Update() either if called directly or
6292 through the EVP_DigestUpdate() function using MDC2. If an attacker
6293 is able to supply very large amounts of input data after a previous
6294 call to EVP_EncryptUpdate() with a partial block then a length check
6295 can overflow resulting in a heap corruption.
6297 The amount of data needed is comparable to SIZE_MAX which is impractical
6300 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
6305 * Malformed SHA512 ticket DoS
6307 If a server uses SHA512 for TLS session ticket HMAC it is vulnerable to a
6308 DoS attack where a malformed ticket will result in an OOB read which will
6311 The use of SHA512 in TLS session tickets is comparatively rare as it requires
6312 a custom server callback and ticket lookup mechanism.
6314 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
6319 * OOB write in BN_bn2dec()
6321 The function BN_bn2dec() does not check the return value of BN_div_word().
6322 This can cause an OOB write if an application uses this function with an
6323 overly large BIGNUM. This could be a problem if an overly large certificate
6324 or CRL is printed out from an untrusted source. TLS is not affected because
6325 record limits will reject an oversized certificate before it is parsed.
6327 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
6332 * OOB read in TS_OBJ_print_bio()
6334 The function TS_OBJ_print_bio() misuses OBJ_obj2txt(): the return value is
6335 the total length the OID text representation would use and not the amount
6336 of data written. This will result in OOB reads when large OIDs are
6339 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
6344 * Pointer arithmetic undefined behaviour
6346 Avoid some undefined pointer arithmetic
6348 A common idiom in the codebase is to check limits in the following manner:
6351 Where "p" points to some malloc'd data of SIZE bytes and
6354 "len" here could be from some externally supplied data (e.g. from a TLS
6357 The rules of C pointer arithmetic are such that "p + len" is only well
6358 defined where len <= SIZE. Therefore the above idiom is actually
6359 undefined behaviour.
6361 For example this could cause problems if some malloc implementation
6362 provides an address for "p" such that "p + len" actually overflows for
6363 values of len that are too big and therefore p + len < limit.
6365 This issue was reported to OpenSSL by Guido Vranken
6370 * Constant time flag not preserved in DSA signing
6372 Operations in the DSA signing algorithm should run in constant time in
6373 order to avoid side channel attacks. A flaw in the OpenSSL DSA
6374 implementation means that a non-constant time codepath is followed for
6375 certain operations. This has been demonstrated through a cache-timing
6376 attack to be sufficient for an attacker to recover the private DSA key.
6378 This issue was reported by César Pereida (Aalto University), Billy Brumley
6379 (Tampere University of Technology), and Yuval Yarom (The University of
6380 Adelaide and NICTA).
6385 * DTLS buffered message DoS
6387 In a DTLS connection where handshake messages are delivered out-of-order
6388 those messages that OpenSSL is not yet ready to process will be buffered
6389 for later use. Under certain circumstances, a flaw in the logic means that
6390 those messages do not get removed from the buffer even though the handshake
6391 has been completed. An attacker could force up to approx. 15 messages to
6392 remain in the buffer when they are no longer required. These messages will
6393 be cleared when the DTLS connection is closed. The default maximum size for
6394 a message is 100k. Therefore, the attacker could force an additional 1500k
6395 to be consumed per connection. By opening many simultaneous connections an
6396 attacker could cause a DoS attack through memory exhaustion.
6398 This issue was reported to OpenSSL by Quan Luo.
6403 * DTLS replay protection DoS
6405 A flaw in the DTLS replay attack protection mechanism means that records
6406 that arrive for future epochs update the replay protection "window" before
6407 the MAC for the record has been validated. This could be exploited by an
6408 attacker by sending a record for the next epoch (which does not have to
6409 decrypt or have a valid MAC), with a very large sequence number. This means
6410 that all subsequent legitimate packets are dropped causing a denial of
6411 service for a specific DTLS connection.
6413 This issue was reported to OpenSSL by the OCAP audit team.
6418 * Certificate message OOB reads
6420 In OpenSSL 1.0.2 and earlier some missing message length checks can result
6421 in OOB reads of up to 2 bytes beyond an allocated buffer. There is a
6422 theoretical DoS risk but this has not been observed in practice on common
6425 The messages affected are client certificate, client certificate request
6426 and server certificate. As a result the attack can only be performed
6427 against a client or a server which enables client authentication.
6429 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
6434 ### Changes between 1.0.2g and 1.0.2h [3 May 2016]
6436 * Prevent padding oracle in AES-NI CBC MAC check
6438 A MITM attacker can use a padding oracle attack to decrypt traffic
6439 when the connection uses an AES CBC cipher and the server support
6442 This issue was introduced as part of the fix for Lucky 13 padding
6443 attack ([CVE-2013-0169]). The padding check was rewritten to be in
6444 constant time by making sure that always the same bytes are read and
6445 compared against either the MAC or padding bytes. But it no longer
6446 checked that there was enough data to have both the MAC and padding
6449 This issue was reported by Juraj Somorovsky using TLS-Attacker.
6453 * Fix EVP_EncodeUpdate overflow
6455 An overflow can occur in the EVP_EncodeUpdate() function which is used for
6456 Base64 encoding of binary data. If an attacker is able to supply very large
6457 amounts of input data then a length check can overflow resulting in a heap
6460 Internally to OpenSSL the EVP_EncodeUpdate() function is primarily used by
6461 the `PEM_write_bio*` family of functions. These are mainly used within the
6462 OpenSSL command line applications, so any application which processes data
6463 from an untrusted source and outputs it as a PEM file should be considered
6464 vulnerable to this issue. User applications that call these APIs directly
6465 with large amounts of untrusted data may also be vulnerable.
6467 This issue was reported by Guido Vranken.
6472 * Fix EVP_EncryptUpdate overflow
6474 An overflow can occur in the EVP_EncryptUpdate() function. If an attacker
6475 is able to supply very large amounts of input data after a previous call to
6476 EVP_EncryptUpdate() with a partial block then a length check can overflow
6477 resulting in a heap corruption. Following an analysis of all OpenSSL
6478 internal usage of the EVP_EncryptUpdate() function all usage is one of two
6479 forms. The first form is where the EVP_EncryptUpdate() call is known to be
6480 the first called function after an EVP_EncryptInit(), and therefore that
6481 specific call must be safe. The second form is where the length passed to
6482 EVP_EncryptUpdate() can be seen from the code to be some small value and
6483 therefore there is no possibility of an overflow. Since all instances are
6484 one of these two forms, it is believed that there can be no overflows in
6485 internal code due to this problem. It should be noted that
6486 EVP_DecryptUpdate() can call EVP_EncryptUpdate() in certain code paths.
6487 Also EVP_CipherUpdate() is a synonym for EVP_EncryptUpdate(). All instances
6488 of these calls have also been analysed too and it is believed there are no
6489 instances in internal usage where an overflow could occur.
6491 This issue was reported by Guido Vranken.
6496 * Prevent ASN.1 BIO excessive memory allocation
6498 When ASN.1 data is read from a BIO using functions such as d2i_CMS_bio()
6499 a short invalid encoding can cause allocation of large amounts of memory
6500 potentially consuming excessive resources or exhausting memory.
6502 Any application parsing untrusted data through d2i BIO functions is
6503 affected. The memory based functions such as d2i_X509() are *not* affected.
6504 Since the memory based functions are used by the TLS library, TLS
6505 applications are not affected.
6507 This issue was reported by Brian Carpenter.
6514 ASN1 Strings that are over 1024 bytes can cause an overread in applications
6515 using the X509_NAME_oneline() function on EBCDIC systems. This could result
6516 in arbitrary stack data being returned in the buffer.
6518 This issue was reported by Guido Vranken.
6523 * Modify behavior of ALPN to invoke callback after SNI/servername
6524 callback, such that updates to the SSL_CTX affect ALPN.
6528 * Remove LOW from the DEFAULT cipher list. This removes singles DES from the
6533 * Only remove the SSLv2 methods with the no-ssl2-method option. When the
6534 methods are enabled and ssl2 is disabled the methods return NULL.
6538 ### Changes between 1.0.2f and 1.0.2g [1 Mar 2016]
6540 * Disable weak ciphers in SSLv3 and up in default builds of OpenSSL.
6541 Builds that are not configured with "enable-weak-ssl-ciphers" will not
6542 provide any "EXPORT" or "LOW" strength ciphers.
6546 * Disable SSLv2 default build, default negotiation and weak ciphers. SSLv2
6547 is by default disabled at build-time. Builds that are not configured with
6548 "enable-ssl2" will not support SSLv2. Even if "enable-ssl2" is used,
6549 users who want to negotiate SSLv2 via the version-flexible SSLv23_method()
6550 will need to explicitly call either of:
6552 SSL_CTX_clear_options(ctx, SSL_OP_NO_SSLv2);
6554 SSL_clear_options(ssl, SSL_OP_NO_SSLv2);
6556 as appropriate. Even if either of those is used, or the application
6557 explicitly uses the version-specific SSLv2_method() or its client and
6558 server variants, SSLv2 ciphers vulnerable to exhaustive search key
6559 recovery have been removed. Specifically, the SSLv2 40-bit EXPORT
6560 ciphers, and SSLv2 56-bit DES are no longer available.
6565 * Fix a double-free in DSA code
6567 A double free bug was discovered when OpenSSL parses malformed DSA private
6568 keys and could lead to a DoS attack or memory corruption for applications
6569 that receive DSA private keys from untrusted sources. This scenario is
6572 This issue was reported to OpenSSL by Adam Langley(Google/BoringSSL) using
6578 * Disable SRP fake user seed to address a server memory leak.
6580 Add a new method SRP_VBASE_get1_by_user that handles the seed properly.
6582 SRP_VBASE_get_by_user had inconsistent memory management behaviour.
6583 In order to fix an unavoidable memory leak, SRP_VBASE_get_by_user
6584 was changed to ignore the "fake user" SRP seed, even if the seed
6587 Users should use SRP_VBASE_get1_by_user instead. Note that in
6588 SRP_VBASE_get1_by_user, caller must free the returned value. Note
6589 also that even though configuring the SRP seed attempts to hide
6590 invalid usernames by continuing the handshake with fake
6591 credentials, this behaviour is not constant time and no strong
6592 guarantees are made that the handshake is indistinguishable from
6593 that of a valid user.
6598 * Fix BN_hex2bn/BN_dec2bn NULL pointer deref/heap corruption
6600 In the BN_hex2bn function the number of hex digits is calculated using an
6601 int value `i`. Later `bn_expand` is called with a value of `i * 4`. For
6602 large values of `i` this can result in `bn_expand` not allocating any
6603 memory because `i * 4` is negative. This can leave the internal BIGNUM data
6604 field as NULL leading to a subsequent NULL ptr deref. For very large values
6605 of `i`, the calculation `i * 4` could be a positive value smaller than `i`.
6606 In this case memory is allocated to the internal BIGNUM data field, but it
6607 is insufficiently sized leading to heap corruption. A similar issue exists
6608 in BN_dec2bn. This could have security consequences if BN_hex2bn/BN_dec2bn
6609 is ever called by user applications with very large untrusted hex/dec data.
6610 This is anticipated to be a rare occurrence.
6612 All OpenSSL internal usage of these functions use data that is not expected
6613 to be untrusted, e.g. config file data or application command line
6614 arguments. If user developed applications generate config file data based
6615 on untrusted data then it is possible that this could also lead to security
6616 consequences. This is also anticipated to be rare.
6618 This issue was reported to OpenSSL by Guido Vranken.
6623 * Fix memory issues in `BIO_*printf` functions
6625 The internal `fmtstr` function used in processing a "%s" format string in
6626 the `BIO_*printf` functions could overflow while calculating the length of a
6627 string and cause an OOB read when printing very long strings.
6629 Additionally the internal `doapr_outch` function can attempt to write to an
6630 OOB memory location (at an offset from the NULL pointer) in the event of a
6631 memory allocation failure. In 1.0.2 and below this could be caused where
6632 the size of a buffer to be allocated is greater than INT_MAX. E.g. this
6633 could be in processing a very long "%s" format string. Memory leaks can
6636 The first issue may mask the second issue dependent on compiler behaviour.
6637 These problems could enable attacks where large amounts of untrusted data
6638 is passed to the `BIO_*printf` functions. If applications use these functions
6639 in this way then they could be vulnerable. OpenSSL itself uses these
6640 functions when printing out human-readable dumps of ASN.1 data. Therefore
6641 applications that print this data could be vulnerable if the data is from
6642 untrusted sources. OpenSSL command line applications could also be
6643 vulnerable where they print out ASN.1 data, or if untrusted data is passed
6644 as command line arguments.
6646 Libssl is not considered directly vulnerable. Additionally certificates etc
6647 received via remote connections via libssl are also unlikely to be able to
6648 trigger these issues because of message size limits enforced within libssl.
6650 This issue was reported to OpenSSL Guido Vranken.
6655 * Side channel attack on modular exponentiation
6657 A side-channel attack was found which makes use of cache-bank conflicts on
6658 the Intel Sandy-Bridge microarchitecture which could lead to the recovery
6659 of RSA keys. The ability to exploit this issue is limited as it relies on
6660 an attacker who has control of code in a thread running on the same
6661 hyper-threaded core as the victim thread which is performing decryptions.
6663 This issue was reported to OpenSSL by Yuval Yarom, The University of
6664 Adelaide and NICTA, Daniel Genkin, Technion and Tel Aviv University, and
6665 Nadia Heninger, University of Pennsylvania with more information at
6666 <http://cachebleed.info>.
6671 * Change the `req` command to generate a 2048-bit RSA/DSA key by default,
6672 if no keysize is specified with default_bits. This fixes an
6673 omission in an earlier change that changed all RSA/DSA key generation
6674 commands to use 2048 bits by default.
6678 ### Changes between 1.0.2e and 1.0.2f [28 Jan 2016]
6680 * DH small subgroups
6682 Historically OpenSSL only ever generated DH parameters based on "safe"
6683 primes. More recently (in version 1.0.2) support was provided for
6684 generating X9.42 style parameter files such as those required for RFC 5114
6685 support. The primes used in such files may not be "safe". Where an
6686 application is using DH configured with parameters based on primes that are
6687 not "safe" then an attacker could use this fact to find a peer's private
6688 DH exponent. This attack requires that the attacker complete multiple
6689 handshakes in which the peer uses the same private DH exponent. For example
6690 this could be used to discover a TLS server's private DH exponent if it's
6691 reusing the private DH exponent or it's using a static DH ciphersuite.
6693 OpenSSL provides the option SSL_OP_SINGLE_DH_USE for ephemeral DH (DHE) in
6694 TLS. It is not on by default. If the option is not set then the server
6695 reuses the same private DH exponent for the life of the server process and
6696 would be vulnerable to this attack. It is believed that many popular
6697 applications do set this option and would therefore not be at risk.
6699 The fix for this issue adds an additional check where a "q" parameter is
6700 available (as is the case in X9.42 based parameters). This detects the
6701 only known attack, and is the only possible defense for static DH
6702 ciphersuites. This could have some performance impact.
6704 Additionally the SSL_OP_SINGLE_DH_USE option has been switched on by
6705 default and cannot be disabled. This could have some performance impact.
6707 This issue was reported to OpenSSL by Antonio Sanso (Adobe).
6712 * SSLv2 doesn't block disabled ciphers
6714 A malicious client can negotiate SSLv2 ciphers that have been disabled on
6715 the server and complete SSLv2 handshakes even if all SSLv2 ciphers have
6716 been disabled, provided that the SSLv2 protocol was not also disabled via
6719 This issue was reported to OpenSSL on 26th December 2015 by Nimrod Aviram
6720 and Sebastian Schinzel.
6725 ### Changes between 1.0.2d and 1.0.2e [3 Dec 2015]
6727 * BN_mod_exp may produce incorrect results on x86_64
6729 There is a carry propagating bug in the x86_64 Montgomery squaring
6730 procedure. No EC algorithms are affected. Analysis suggests that attacks
6731 against RSA and DSA as a result of this defect would be very difficult to
6732 perform and are not believed likely. Attacks against DH are considered just
6733 feasible (although very difficult) because most of the work necessary to
6734 deduce information about a private key may be performed offline. The amount
6735 of resources required for such an attack would be very significant and
6736 likely only accessible to a limited number of attackers. An attacker would
6737 additionally need online access to an unpatched system using the target
6738 private key in a scenario with persistent DH parameters and a private
6739 key that is shared between multiple clients. For example this can occur by
6740 default in OpenSSL DHE based SSL/TLS ciphersuites.
6742 This issue was reported to OpenSSL by Hanno Böck.
6747 * Certificate verify crash with missing PSS parameter
6749 The signature verification routines will crash with a NULL pointer
6750 dereference if presented with an ASN.1 signature using the RSA PSS
6751 algorithm and absent mask generation function parameter. Since these
6752 routines are used to verify certificate signature algorithms this can be
6753 used to crash any certificate verification operation and exploited in a
6754 DoS attack. Any application which performs certificate verification is
6755 vulnerable including OpenSSL clients and servers which enable client
6758 This issue was reported to OpenSSL by Loïc Jonas Etienne (Qnective AG).
6763 * X509_ATTRIBUTE memory leak
6765 When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak
6766 memory. This structure is used by the PKCS#7 and CMS routines so any
6767 application which reads PKCS#7 or CMS data from untrusted sources is
6768 affected. SSL/TLS is not affected.
6770 This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using
6776 * Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs.
6777 This changes the decoding behaviour for some invalid messages,
6778 though the change is mostly in the more lenient direction, and
6779 legacy behaviour is preserved as much as possible.
6783 * In DSA_generate_parameters_ex, if the provided seed is too short,
6786 *Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>*
6788 ### Changes between 1.0.2c and 1.0.2d [9 Jul 2015]
6790 * Alternate chains certificate forgery
6792 During certificate verification, OpenSSL will attempt to find an
6793 alternative certificate chain if the first attempt to build such a chain
6794 fails. An error in the implementation of this logic can mean that an
6795 attacker could cause certain checks on untrusted certificates to be
6796 bypassed, such as the CA flag, enabling them to use a valid leaf
6797 certificate to act as a CA and "issue" an invalid certificate.
6799 This issue was reported to OpenSSL by Adam Langley/David Benjamin
6804 ### Changes between 1.0.2b and 1.0.2c [12 Jun 2015]
6806 * Fix HMAC ABI incompatibility. The previous version introduced an ABI
6807 incompatibility in the handling of HMAC. The previous ABI has now been
6812 ### Changes between 1.0.2a and 1.0.2b [11 Jun 2015]
6814 * Malformed ECParameters causes infinite loop
6816 When processing an ECParameters structure OpenSSL enters an infinite loop
6817 if the curve specified is over a specially malformed binary polynomial
6820 This can be used to perform denial of service against any
6821 system which processes public keys, certificate requests or
6822 certificates. This includes TLS clients and TLS servers with
6823 client authentication enabled.
6825 This issue was reported to OpenSSL by Joseph Barr-Pixton.
6830 * Exploitable out-of-bounds read in X509_cmp_time
6832 X509_cmp_time does not properly check the length of the ASN1_TIME
6833 string and can read a few bytes out of bounds. In addition,
6834 X509_cmp_time accepts an arbitrary number of fractional seconds in the
6837 An attacker can use this to craft malformed certificates and CRLs of
6838 various sizes and potentially cause a segmentation fault, resulting in
6839 a DoS on applications that verify certificates or CRLs. TLS clients
6840 that verify CRLs are affected. TLS clients and servers with client
6841 authentication enabled may be affected if they use custom verification
6844 This issue was reported to OpenSSL by Robert Swiecki (Google), and
6845 independently by Hanno Böck.
6850 * PKCS7 crash with missing EnvelopedContent
6852 The PKCS#7 parsing code does not handle missing inner EncryptedContent
6853 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs
6854 with missing content and trigger a NULL pointer dereference on parsing.
6856 Applications that decrypt PKCS#7 data or otherwise parse PKCS#7
6857 structures from untrusted sources are affected. OpenSSL clients and
6858 servers are not affected.
6860 This issue was reported to OpenSSL by Michal Zalewski (Google).
6865 * CMS verify infinite loop with unknown hash function
6867 When verifying a signedData message the CMS code can enter an infinite loop
6868 if presented with an unknown hash function OID. This can be used to perform
6869 denial of service against any system which verifies signedData messages using
6871 This issue was reported to OpenSSL by Johannes Bauer.
6876 * Race condition handling NewSessionTicket
6878 If a NewSessionTicket is received by a multi-threaded client when attempting to
6879 reuse a previous ticket then a race condition can occur potentially leading to
6880 a double free of the ticket data.
6885 * Only support 256-bit or stronger elliptic curves with the
6886 'ecdh_auto' setting (server) or by default (client). Of supported
6887 curves, prefer P-256 (both).
6891 ### Changes between 1.0.2 and 1.0.2a [19 Mar 2015]
6893 * ClientHello sigalgs DoS fix
6895 If a client connects to an OpenSSL 1.0.2 server and renegotiates with an
6896 invalid signature algorithms extension a NULL pointer dereference will
6897 occur. This can be exploited in a DoS attack against the server.
6899 This issue was was reported to OpenSSL by David Ramos of Stanford
6903 *Stephen Henson and Matt Caswell*
6905 * Multiblock corrupted pointer fix
6907 OpenSSL 1.0.2 introduced the "multiblock" performance improvement. This
6908 feature only applies on 64 bit x86 architecture platforms that support AES
6909 NI instructions. A defect in the implementation of "multiblock" can cause
6910 OpenSSL's internal write buffer to become incorrectly set to NULL when
6911 using non-blocking IO. Typically, when the user application is using a
6912 socket BIO for writing, this will only result in a failed connection.
6913 However if some other BIO is used then it is likely that a segmentation
6914 fault will be triggered, thus enabling a potential DoS attack.
6916 This issue was reported to OpenSSL by Daniel Danner and Rainer Mueller.
6921 * Segmentation fault in DTLSv1_listen fix
6923 The DTLSv1_listen function is intended to be stateless and processes the
6924 initial ClientHello from many peers. It is common for user code to loop
6925 over the call to DTLSv1_listen until a valid ClientHello is received with
6926 an associated cookie. A defect in the implementation of DTLSv1_listen means
6927 that state is preserved in the SSL object from one invocation to the next
6928 that can lead to a segmentation fault. Errors processing the initial
6929 ClientHello can trigger this scenario. An example of such an error could be
6930 that a DTLS1.0 only client is attempting to connect to a DTLS1.2 only
6933 This issue was reported to OpenSSL by Per Allansson.
6938 * Segmentation fault in ASN1_TYPE_cmp fix
6940 The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is
6941 made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check
6942 certificate signature algorithm consistency this can be used to crash any
6943 certificate verification operation and exploited in a DoS attack. Any
6944 application which performs certificate verification is vulnerable including
6945 OpenSSL clients and servers which enable client authentication.
6950 * Segmentation fault for invalid PSS parameters fix
6952 The signature verification routines will crash with a NULL pointer
6953 dereference if presented with an ASN.1 signature using the RSA PSS
6954 algorithm and invalid parameters. Since these routines are used to verify
6955 certificate signature algorithms this can be used to crash any
6956 certificate verification operation and exploited in a DoS attack. Any
6957 application which performs certificate verification is vulnerable including
6958 OpenSSL clients and servers which enable client authentication.
6960 This issue was was reported to OpenSSL by Brian Carpenter.
6965 * ASN.1 structure reuse memory corruption fix
6967 Reusing a structure in ASN.1 parsing may allow an attacker to cause
6968 memory corruption via an invalid write. Such reuse is and has been
6969 strongly discouraged and is believed to be rare.
6971 Applications that parse structures containing CHOICE or ANY DEFINED BY
6972 components may be affected. Certificate parsing (d2i_X509 and related
6973 functions) are however not affected. OpenSSL clients and servers are
6979 * PKCS7 NULL pointer dereferences fix
6981 The PKCS#7 parsing code does not handle missing outer ContentInfo
6982 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with
6983 missing content and trigger a NULL pointer dereference on parsing.
6985 Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or
6986 otherwise parse PKCS#7 structures from untrusted sources are
6987 affected. OpenSSL clients and servers are not affected.
6989 This issue was reported to OpenSSL by Michal Zalewski (Google).
6994 * DoS via reachable assert in SSLv2 servers fix
6996 A malicious client can trigger an OPENSSL_assert (i.e., an abort) in
6997 servers that both support SSLv2 and enable export cipher suites by sending
6998 a specially crafted SSLv2 CLIENT-MASTER-KEY message.
7000 This issue was discovered by Sean Burford (Google) and Emilia Käsper
7001 (OpenSSL development team).
7006 * Empty CKE with client auth and DHE fix
7008 If client auth is used then a server can seg fault in the event of a DHE
7009 ciphersuite being selected and a zero length ClientKeyExchange message
7010 being sent by the client. This could be exploited in a DoS attack.
7015 * Handshake with unseeded PRNG fix
7017 Under certain conditions an OpenSSL 1.0.2 client can complete a handshake
7018 with an unseeded PRNG. The conditions are:
7019 - The client is on a platform where the PRNG has not been seeded
7020 automatically, and the user has not seeded manually
7021 - A protocol specific client method version has been used (i.e. not
7022 SSL_client_methodv23)
7023 - A ciphersuite is used that does not require additional random data from
7024 the PRNG beyond the initial ClientHello client random (e.g. PSK-RC4-SHA).
7026 If the handshake succeeds then the client random that has been used will
7027 have been generated from a PRNG with insufficient entropy and therefore the
7028 output may be predictable.
7030 For example using the following command with an unseeded openssl will
7031 succeed on an unpatched platform:
7033 openssl s_client -psk 1a2b3c4d -tls1_2 -cipher PSK-RC4-SHA
7038 * Use After Free following d2i_ECPrivatekey error fix
7040 A malformed EC private key file consumed via the d2i_ECPrivateKey function
7041 could cause a use after free condition. This, in turn, could cause a double
7042 free in several private key parsing functions (such as d2i_PrivateKey
7043 or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption
7044 for applications that receive EC private keys from untrusted
7045 sources. This scenario is considered rare.
7047 This issue was discovered by the BoringSSL project and fixed in their
7053 * X509_to_X509_REQ NULL pointer deref fix
7055 The function X509_to_X509_REQ will crash with a NULL pointer dereference if
7056 the certificate key is invalid. This function is rarely used in practice.
7058 This issue was discovered by Brian Carpenter.
7063 * Removed the export ciphers from the DEFAULT ciphers
7067 ### Changes between 1.0.1l and 1.0.2 [22 Jan 2015]
7069 * Facilitate "universal" ARM builds targeting range of ARM ISAs, e.g.
7070 ARMv5 through ARMv8, as opposite to "locking" it to single one.
7071 So far those who have to target multiple platforms would compromise
7072 and argue that binary targeting say ARMv5 would still execute on
7073 ARMv8. "Universal" build resolves this compromise by providing
7074 near-optimal performance even on newer platforms.
7078 * Accelerated NIST P-256 elliptic curve implementation for x86_64
7079 (other platforms pending).
7081 *Shay Gueron & Vlad Krasnov (Intel Corp), Andy Polyakov*
7083 * Add support for the SignedCertificateTimestampList certificate and
7084 OCSP response extensions from RFC6962.
7088 * Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.)
7089 for corner cases. (Certain input points at infinity could lead to
7090 bogus results, with non-infinity inputs mapped to infinity too.)
7094 * Initial support for PowerISA 2.0.7, first implemented in POWER8.
7095 This covers AES, SHA256/512 and GHASH. "Initial" means that most
7096 common cases are optimized and there still is room for further
7097 improvements. Vector Permutation AES for Altivec is also added.
7101 * Add support for little-endian ppc64 Linux target.
7103 *Marcelo Cerri (IBM)*
7105 * Initial support for AMRv8 ISA crypto extensions. This covers AES,
7106 SHA1, SHA256 and GHASH. "Initial" means that most common cases
7107 are optimized and there still is room for further improvements.
7108 Both 32- and 64-bit modes are supported.
7110 *Andy Polyakov, Ard Biesheuvel (Linaro)*
7112 * Improved ARMv7 NEON support.
7116 * Support for SPARC Architecture 2011 crypto extensions, first
7117 implemented in SPARC T4. This covers AES, DES, Camellia, SHA1,
7118 SHA256/512, MD5, GHASH and modular exponentiation.
7120 *Andy Polyakov, David Miller*
7122 * Accelerated modular exponentiation for Intel processors, a.k.a.
7125 *Shay Gueron & Vlad Krasnov (Intel Corp)*
7127 * Support for new and upcoming Intel processors, including AVX2,
7128 BMI and SHA ISA extensions. This includes additional "stitched"
7129 implementations, AESNI-SHA256 and GCM, and multi-buffer support
7132 This work was sponsored by Intel Corp.
7136 * Support for DTLS 1.2. This adds two sets of DTLS methods: DTLS_*_method()
7137 supports both DTLS 1.2 and 1.0 and should use whatever version the peer
7138 supports and DTLSv1_2_*_method() which supports DTLS 1.2 only.
7142 * Use algorithm specific chains in SSL_CTX_use_certificate_chain_file():
7143 this fixes a limitation in previous versions of OpenSSL.
7147 * Extended RSA OAEP support via EVP_PKEY API. Options to specify digest,
7148 MGF1 digest and OAEP label.
7152 * Add EVP support for key wrapping algorithms, to avoid problems with
7153 existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in
7154 the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap
7155 algorithms and include tests cases.
7159 * Add functions to allocate and set the fields of an ECDSA_METHOD
7162 *Douglas E. Engert, Steve Henson*
7164 * New functions OPENSSL_gmtime_diff and ASN1_TIME_diff to find the
7165 difference in days and seconds between two tm or ASN1_TIME structures.
7169 * Add -rev test option to s_server to just reverse order of characters
7170 received by client and send back to server. Also prints an abbreviated
7171 summary of the connection parameters.
7175 * New option -brief for s_client and s_server to print out a brief summary
7176 of connection parameters.
7180 * Add callbacks for arbitrary TLS extensions.
7182 *Trevor Perrin <trevp@trevp.net> and Ben Laurie*
7184 * New option -crl_download in several openssl utilities to download CRLs
7185 from CRLDP extension in certificates.
7189 * New options -CRL and -CRLform for s_client and s_server for CRLs.
7193 * New function X509_CRL_diff to generate a delta CRL from the difference
7194 of two full CRLs. Add support to "crl" utility.
7198 * New functions to set lookup_crls function and to retrieve
7199 X509_STORE from X509_STORE_CTX.
7203 * Print out deprecated issuer and subject unique ID fields in
7208 * Extend OCSP I/O functions so they can be used for simple general purpose
7209 HTTP as well as OCSP. New wrapper function which can be used to download
7210 CRLs using the OCSP API.
7214 * Delegate command line handling in s_client/s_server to SSL_CONF APIs.
7218 * `SSL_CONF*` functions. These provide a common framework for application
7219 configuration using configuration files or command lines.
7223 * SSL/TLS tracing code. This parses out SSL/TLS records using the
7224 message callback and prints the results. Needs compile time option
7225 "enable-ssl-trace". New options to s_client and s_server to enable
7230 * New ctrl and macro to retrieve supported points extensions.
7231 Print out extension in s_server and s_client.
7235 * New functions to retrieve certificate signature and signature
7240 * Add functions to retrieve and manipulate the raw cipherlist sent by a
7245 * New Suite B modes for TLS code. These use and enforce the requirements
7246 of RFC6460: restrict ciphersuites, only permit Suite B algorithms and
7247 only use Suite B curves. The Suite B modes can be set by using the
7248 strings "SUITEB128", "SUITEB192" or "SUITEB128ONLY" for the cipherstring.
7252 * New chain verification flags for Suite B levels of security. Check
7253 algorithms are acceptable when flags are set in X509_verify_cert.
7257 * Make tls1_check_chain return a set of flags indicating checks passed
7258 by a certificate chain. Add additional tests to handle client
7259 certificates: checks for matching certificate type and issuer name
7264 * If an attempt is made to use a signature algorithm not in the peer
7265 preference list abort the handshake. If client has no suitable
7266 signature algorithms in response to a certificate request do not
7267 use the certificate.
7271 * If server EC tmp key is not in client preference list abort handshake.
7275 * Add support for certificate stores in CERT structure. This makes it
7276 possible to have different stores per SSL structure or one store in
7277 the parent SSL_CTX. Include distinct stores for certificate chain
7278 verification and chain building. New ctrl SSL_CTRL_BUILD_CERT_CHAIN
7279 to build and store a certificate chain in CERT structure: returning
7280 an error if the chain cannot be built: this will allow applications
7281 to test if a chain is correctly configured.
7283 Note: if the CERT based stores are not set then the parent SSL_CTX
7284 store is used to retain compatibility with existing behaviour.
7288 * New function ssl_set_client_disabled to set a ciphersuite disabled
7289 mask based on the current session, check mask when sending client
7290 hello and checking the requested ciphersuite.
7294 * New ctrls to retrieve and set certificate types in a certificate
7295 request message. Print out received values in s_client. If certificate
7296 types is not set with custom values set sensible values based on
7297 supported signature algorithms.
7301 * Support for distinct client and server supported signature algorithms.
7305 * Add certificate callback. If set this is called whenever a certificate
7306 is required by client or server. An application can decide which
7307 certificate chain to present based on arbitrary criteria: for example
7308 supported signature algorithms. Add very simple example to s_server.
7309 This fixes many of the problems and restrictions of the existing client
7310 certificate callback: for example you can now clear an existing
7311 certificate and specify the whole chain.
7315 * Add new "valid_flags" field to CERT_PKEY structure which determines what
7316 the certificate can be used for (if anything). Set valid_flags field
7317 in new tls1_check_chain function. Simplify ssl_set_cert_masks which used
7318 to have similar checks in it.
7320 Add new "cert_flags" field to CERT structure and include a "strict mode".
7321 This enforces some TLS certificate requirements (such as only permitting
7322 certificate signature algorithms contained in the supported algorithms
7323 extension) which some implementations ignore: this option should be used
7324 with caution as it could cause interoperability issues.
7328 * Update and tidy signature algorithm extension processing. Work out
7329 shared signature algorithms based on preferences and peer algorithms
7330 and print them out in s_client and s_server. Abort handshake if no
7331 shared signature algorithms.
7335 * Add new functions to allow customised supported signature algorithms
7336 for SSL and SSL_CTX structures. Add options to s_client and s_server
7341 * New function SSL_certs_clear() to delete all references to certificates
7342 from an SSL structure. Before this once a certificate had been added
7343 it couldn't be removed.
7347 * Integrate hostname, email address and IP address checking with certificate
7348 verification. New verify options supporting checking in openssl utility.
7352 * Fixes and wildcard matching support to hostname and email checking
7353 functions. Add manual page.
7355 *Florian Weimer (Red Hat Product Security Team)*
7357 * New functions to check a hostname email or IP address against a
7358 certificate. Add options x509 utility to print results of checks against
7363 * Fix OCSP checking.
7365 *Rob Stradling <rob.stradling@comodo.com> and Ben Laurie*
7367 * Initial experimental support for explicitly trusted non-root CAs.
7368 OpenSSL still tries to build a complete chain to a root but if an
7369 intermediate CA has a trust setting included that is used. The first
7370 setting is used: whether to trust (e.g., -addtrust option to the x509
7375 * Add -trusted_first option which attempts to find certificates in the
7376 trusted store even if an untrusted chain is also supplied.
7380 * MIPS assembly pack updates: support for MIPS32r2 and SmartMIPS ASE,
7381 platform support for Linux and Android.
7385 * Support for linux-x32, ILP32 environment in x86_64 framework.
7389 * Experimental multi-implementation support for FIPS capable OpenSSL.
7390 When in FIPS mode the approved implementations are used as normal,
7391 when not in FIPS mode the internal unapproved versions are used instead.
7392 This means that the FIPS capable OpenSSL isn't forced to use the
7393 (often lower performance) FIPS implementations outside FIPS mode.
7397 * Transparently support X9.42 DH parameters when calling
7398 PEM_read_bio_DHparameters. This means existing applications can handle
7399 the new parameter format automatically.
7403 * Initial experimental support for X9.42 DH parameter format: mainly
7404 to support use of 'q' parameter for RFC5114 parameters.
7408 * Add DH parameters from RFC5114 including test data to dhtest.
7412 * Support for automatic EC temporary key parameter selection. If enabled
7413 the most preferred EC parameters are automatically used instead of
7414 hardcoded fixed parameters. Now a server just has to call:
7415 SSL_CTX_set_ecdh_auto(ctx, 1) and the server will automatically
7416 support ECDH and use the most appropriate parameters.
7420 * Enhance and tidy EC curve and point format TLS extension code. Use
7421 static structures instead of allocation if default values are used.
7422 New ctrls to set curves we wish to support and to retrieve shared curves.
7423 Print out shared curves in s_server. New options to s_server and s_client
7424 to set list of supported curves.
7428 * New ctrls to retrieve supported signature algorithms and
7429 supported curve values as an array of NIDs. Extend openssl utility
7430 to print out received values.
7434 * Add new APIs EC_curve_nist2nid and EC_curve_nid2nist which convert
7435 between NIDs and the more common NIST names such as "P-256". Enhance
7436 ecparam utility and ECC method to recognise the NIST names for curves.
7440 * Enhance SSL/TLS certificate chain handling to support different
7441 chains for each certificate instead of one chain in the parent SSL_CTX.
7445 * Support for fixed DH ciphersuite client authentication: where both
7446 server and client use DH certificates with common parameters.
7450 * Support for fixed DH ciphersuites: those requiring DH server
7455 * New function i2d_re_X509_tbs for re-encoding the TBS portion of
7457 Note: Related 1.0.2-beta specific macros X509_get_cert_info,
7458 X509_CINF_set_modified, X509_CINF_get_issuer, X509_CINF_get_extensions and
7459 X509_CINF_get_signature were reverted post internal team review.
7464 ### Changes between 1.0.1t and 1.0.1u [22 Sep 2016]
7466 * OCSP Status Request extension unbounded memory growth
7468 A malicious client can send an excessively large OCSP Status Request
7469 extension. If that client continually requests renegotiation, sending a
7470 large OCSP Status Request extension each time, then there will be unbounded
7471 memory growth on the server. This will eventually lead to a Denial Of
7472 Service attack through memory exhaustion. Servers with a default
7473 configuration are vulnerable even if they do not support OCSP. Builds using
7474 the "no-ocsp" build time option are not affected.
7476 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
7481 * In order to mitigate the SWEET32 attack, the DES ciphers were moved from
7484 This issue was reported to OpenSSL Karthikeyan Bhargavan and Gaetan
7490 * OOB write in MDC2_Update()
7492 An overflow can occur in MDC2_Update() either if called directly or
7493 through the EVP_DigestUpdate() function using MDC2. If an attacker
7494 is able to supply very large amounts of input data after a previous
7495 call to EVP_EncryptUpdate() with a partial block then a length check
7496 can overflow resulting in a heap corruption.
7498 The amount of data needed is comparable to SIZE_MAX which is impractical
7501 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
7506 * Malformed SHA512 ticket DoS
7508 If a server uses SHA512 for TLS session ticket HMAC it is vulnerable to a
7509 DoS attack where a malformed ticket will result in an OOB read which will
7512 The use of SHA512 in TLS session tickets is comparatively rare as it requires
7513 a custom server callback and ticket lookup mechanism.
7515 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
7520 * OOB write in BN_bn2dec()
7522 The function BN_bn2dec() does not check the return value of BN_div_word().
7523 This can cause an OOB write if an application uses this function with an
7524 overly large BIGNUM. This could be a problem if an overly large certificate
7525 or CRL is printed out from an untrusted source. TLS is not affected because
7526 record limits will reject an oversized certificate before it is parsed.
7528 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
7533 * OOB read in TS_OBJ_print_bio()
7535 The function TS_OBJ_print_bio() misuses OBJ_obj2txt(): the return value is
7536 the total length the OID text representation would use and not the amount
7537 of data written. This will result in OOB reads when large OIDs are
7540 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
7545 * Pointer arithmetic undefined behaviour
7547 Avoid some undefined pointer arithmetic
7549 A common idiom in the codebase is to check limits in the following manner:
7552 Where "p" points to some malloc'd data of SIZE bytes and
7555 "len" here could be from some externally supplied data (e.g. from a TLS
7558 The rules of C pointer arithmetic are such that "p + len" is only well
7559 defined where len <= SIZE. Therefore, the above idiom is actually
7560 undefined behaviour.
7562 For example this could cause problems if some malloc implementation
7563 provides an address for "p" such that "p + len" actually overflows for
7564 values of len that are too big and therefore p + len < limit.
7566 This issue was reported to OpenSSL by Guido Vranken
7571 * Constant time flag not preserved in DSA signing
7573 Operations in the DSA signing algorithm should run in constant time in
7574 order to avoid side channel attacks. A flaw in the OpenSSL DSA
7575 implementation means that a non-constant time codepath is followed for
7576 certain operations. This has been demonstrated through a cache-timing
7577 attack to be sufficient for an attacker to recover the private DSA key.
7579 This issue was reported by César Pereida (Aalto University), Billy Brumley
7580 (Tampere University of Technology), and Yuval Yarom (The University of
7581 Adelaide and NICTA).
7586 * DTLS buffered message DoS
7588 In a DTLS connection where handshake messages are delivered out-of-order
7589 those messages that OpenSSL is not yet ready to process will be buffered
7590 for later use. Under certain circumstances, a flaw in the logic means that
7591 those messages do not get removed from the buffer even though the handshake
7592 has been completed. An attacker could force up to approx. 15 messages to
7593 remain in the buffer when they are no longer required. These messages will
7594 be cleared when the DTLS connection is closed. The default maximum size for
7595 a message is 100k. Therefore, the attacker could force an additional 1500k
7596 to be consumed per connection. By opening many simultaneous connections an
7597 attacker could cause a DoS attack through memory exhaustion.
7599 This issue was reported to OpenSSL by Quan Luo.
7604 * DTLS replay protection DoS
7606 A flaw in the DTLS replay attack protection mechanism means that records
7607 that arrive for future epochs update the replay protection "window" before
7608 the MAC for the record has been validated. This could be exploited by an
7609 attacker by sending a record for the next epoch (which does not have to
7610 decrypt or have a valid MAC), with a very large sequence number. This means
7611 that all subsequent legitimate packets are dropped causing a denial of
7612 service for a specific DTLS connection.
7614 This issue was reported to OpenSSL by the OCAP audit team.
7619 * Certificate message OOB reads
7621 In OpenSSL 1.0.2 and earlier some missing message length checks can result
7622 in OOB reads of up to 2 bytes beyond an allocated buffer. There is a
7623 theoretical DoS risk but this has not been observed in practice on common
7626 The messages affected are client certificate, client certificate request
7627 and server certificate. As a result the attack can only be performed
7628 against a client or a server which enables client authentication.
7630 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
7635 ### Changes between 1.0.1s and 1.0.1t [3 May 2016]
7637 * Prevent padding oracle in AES-NI CBC MAC check
7639 A MITM attacker can use a padding oracle attack to decrypt traffic
7640 when the connection uses an AES CBC cipher and the server support
7643 This issue was introduced as part of the fix for Lucky 13 padding
7644 attack ([CVE-2013-0169]). The padding check was rewritten to be in
7645 constant time by making sure that always the same bytes are read and
7646 compared against either the MAC or padding bytes. But it no longer
7647 checked that there was enough data to have both the MAC and padding
7650 This issue was reported by Juraj Somorovsky using TLS-Attacker.
7655 * Fix EVP_EncodeUpdate overflow
7657 An overflow can occur in the EVP_EncodeUpdate() function which is used for
7658 Base64 encoding of binary data. If an attacker is able to supply very large
7659 amounts of input data then a length check can overflow resulting in a heap
7662 Internally to OpenSSL the EVP_EncodeUpdate() function is primarily used by
7663 the `PEM_write_bio*` family of functions. These are mainly used within the
7664 OpenSSL command line applications, so any application which processes data
7665 from an untrusted source and outputs it as a PEM file should be considered
7666 vulnerable to this issue. User applications that call these APIs directly
7667 with large amounts of untrusted data may also be vulnerable.
7669 This issue was reported by Guido Vranken.
7674 * Fix EVP_EncryptUpdate overflow
7676 An overflow can occur in the EVP_EncryptUpdate() function. If an attacker
7677 is able to supply very large amounts of input data after a previous call to
7678 EVP_EncryptUpdate() with a partial block then a length check can overflow
7679 resulting in a heap corruption. Following an analysis of all OpenSSL
7680 internal usage of the EVP_EncryptUpdate() function all usage is one of two
7681 forms. The first form is where the EVP_EncryptUpdate() call is known to be
7682 the first called function after an EVP_EncryptInit(), and therefore that
7683 specific call must be safe. The second form is where the length passed to
7684 EVP_EncryptUpdate() can be seen from the code to be some small value and
7685 therefore there is no possibility of an overflow. Since all instances are
7686 one of these two forms, it is believed that there can be no overflows in
7687 internal code due to this problem. It should be noted that
7688 EVP_DecryptUpdate() can call EVP_EncryptUpdate() in certain code paths.
7689 Also EVP_CipherUpdate() is a synonym for EVP_EncryptUpdate(). All instances
7690 of these calls have also been analysed too and it is believed there are no
7691 instances in internal usage where an overflow could occur.
7693 This issue was reported by Guido Vranken.
7698 * Prevent ASN.1 BIO excessive memory allocation
7700 When ASN.1 data is read from a BIO using functions such as d2i_CMS_bio()
7701 a short invalid encoding can casuse allocation of large amounts of memory
7702 potentially consuming excessive resources or exhausting memory.
7704 Any application parsing untrusted data through d2i BIO functions is
7705 affected. The memory based functions such as d2i_X509() are *not* affected.
7706 Since the memory based functions are used by the TLS library, TLS
7707 applications are not affected.
7709 This issue was reported by Brian Carpenter.
7716 ASN1 Strings that are over 1024 bytes can cause an overread in applications
7717 using the X509_NAME_oneline() function on EBCDIC systems. This could result
7718 in arbitrary stack data being returned in the buffer.
7720 This issue was reported by Guido Vranken.
7725 * Modify behavior of ALPN to invoke callback after SNI/servername
7726 callback, such that updates to the SSL_CTX affect ALPN.
7730 * Remove LOW from the DEFAULT cipher list. This removes singles DES from the
7735 * Only remove the SSLv2 methods with the no-ssl2-method option. When the
7736 methods are enabled and ssl2 is disabled the methods return NULL.
7740 ### Changes between 1.0.1r and 1.0.1s [1 Mar 2016]
7742 * Disable weak ciphers in SSLv3 and up in default builds of OpenSSL.
7743 Builds that are not configured with "enable-weak-ssl-ciphers" will not
7744 provide any "EXPORT" or "LOW" strength ciphers.
7748 * Disable SSLv2 default build, default negotiation and weak ciphers. SSLv2
7749 is by default disabled at build-time. Builds that are not configured with
7750 "enable-ssl2" will not support SSLv2. Even if "enable-ssl2" is used,
7751 users who want to negotiate SSLv2 via the version-flexible SSLv23_method()
7752 will need to explicitly call either of:
7754 SSL_CTX_clear_options(ctx, SSL_OP_NO_SSLv2);
7756 SSL_clear_options(ssl, SSL_OP_NO_SSLv2);
7758 as appropriate. Even if either of those is used, or the application
7759 explicitly uses the version-specific SSLv2_method() or its client and
7760 server variants, SSLv2 ciphers vulnerable to exhaustive search key
7761 recovery have been removed. Specifically, the SSLv2 40-bit EXPORT
7762 ciphers, and SSLv2 56-bit DES are no longer available.
7767 * Fix a double-free in DSA code
7769 A double free bug was discovered when OpenSSL parses malformed DSA private
7770 keys and could lead to a DoS attack or memory corruption for applications
7771 that receive DSA private keys from untrusted sources. This scenario is
7774 This issue was reported to OpenSSL by Adam Langley(Google/BoringSSL) using
7780 * Disable SRP fake user seed to address a server memory leak.
7782 Add a new method SRP_VBASE_get1_by_user that handles the seed properly.
7784 SRP_VBASE_get_by_user had inconsistent memory management behaviour.
7785 In order to fix an unavoidable memory leak, SRP_VBASE_get_by_user
7786 was changed to ignore the "fake user" SRP seed, even if the seed
7789 Users should use SRP_VBASE_get1_by_user instead. Note that in
7790 SRP_VBASE_get1_by_user, caller must free the returned value. Note
7791 also that even though configuring the SRP seed attempts to hide
7792 invalid usernames by continuing the handshake with fake
7793 credentials, this behaviour is not constant time and no strong
7794 guarantees are made that the handshake is indistinguishable from
7795 that of a valid user.
7800 * Fix BN_hex2bn/BN_dec2bn NULL pointer deref/heap corruption
7802 In the BN_hex2bn function the number of hex digits is calculated using an
7803 int value `i`. Later `bn_expand` is called with a value of `i * 4`. For
7804 large values of `i` this can result in `bn_expand` not allocating any
7805 memory because `i * 4` is negative. This can leave the internal BIGNUM data
7806 field as NULL leading to a subsequent NULL ptr deref. For very large values
7807 of `i`, the calculation `i * 4` could be a positive value smaller than `i`.
7808 In this case memory is allocated to the internal BIGNUM data field, but it
7809 is insufficiently sized leading to heap corruption. A similar issue exists
7810 in BN_dec2bn. This could have security consequences if BN_hex2bn/BN_dec2bn
7811 is ever called by user applications with very large untrusted hex/dec data.
7812 This is anticipated to be a rare occurrence.
7814 All OpenSSL internal usage of these functions use data that is not expected
7815 to be untrusted, e.g. config file data or application command line
7816 arguments. If user developed applications generate config file data based
7817 on untrusted data then it is possible that this could also lead to security
7818 consequences. This is also anticipated to be rare.
7820 This issue was reported to OpenSSL by Guido Vranken.
7825 * Fix memory issues in `BIO_*printf` functions
7827 The internal `fmtstr` function used in processing a "%s" format string in
7828 the `BIO_*printf` functions could overflow while calculating the length of a
7829 string and cause an OOB read when printing very long strings.
7831 Additionally the internal `doapr_outch` function can attempt to write to an
7832 OOB memory location (at an offset from the NULL pointer) in the event of a
7833 memory allocation failure. In 1.0.2 and below this could be caused where
7834 the size of a buffer to be allocated is greater than INT_MAX. E.g. this
7835 could be in processing a very long "%s" format string. Memory leaks can
7838 The first issue may mask the second issue dependent on compiler behaviour.
7839 These problems could enable attacks where large amounts of untrusted data
7840 is passed to the `BIO_*printf` functions. If applications use these functions
7841 in this way then they could be vulnerable. OpenSSL itself uses these
7842 functions when printing out human-readable dumps of ASN.1 data. Therefore
7843 applications that print this data could be vulnerable if the data is from
7844 untrusted sources. OpenSSL command line applications could also be
7845 vulnerable where they print out ASN.1 data, or if untrusted data is passed
7846 as command line arguments.
7848 Libssl is not considered directly vulnerable. Additionally certificates etc
7849 received via remote connections via libssl are also unlikely to be able to
7850 trigger these issues because of message size limits enforced within libssl.
7852 This issue was reported to OpenSSL Guido Vranken.
7857 * Side channel attack on modular exponentiation
7859 A side-channel attack was found which makes use of cache-bank conflicts on
7860 the Intel Sandy-Bridge microarchitecture which could lead to the recovery
7861 of RSA keys. The ability to exploit this issue is limited as it relies on
7862 an attacker who has control of code in a thread running on the same
7863 hyper-threaded core as the victim thread which is performing decryptions.
7865 This issue was reported to OpenSSL by Yuval Yarom, The University of
7866 Adelaide and NICTA, Daniel Genkin, Technion and Tel Aviv University, and
7867 Nadia Heninger, University of Pennsylvania with more information at
7868 <http://cachebleed.info>.
7873 * Change the req command to generate a 2048-bit RSA/DSA key by default,
7874 if no keysize is specified with default_bits. This fixes an
7875 omission in an earlier change that changed all RSA/DSA key generation
7876 commands to use 2048 bits by default.
7880 ### Changes between 1.0.1q and 1.0.1r [28 Jan 2016]
7882 * Protection for DH small subgroup attacks
7884 As a precautionary measure the SSL_OP_SINGLE_DH_USE option has been
7885 switched on by default and cannot be disabled. This could have some
7890 * SSLv2 doesn't block disabled ciphers
7892 A malicious client can negotiate SSLv2 ciphers that have been disabled on
7893 the server and complete SSLv2 handshakes even if all SSLv2 ciphers have
7894 been disabled, provided that the SSLv2 protocol was not also disabled via
7897 This issue was reported to OpenSSL on 26th December 2015 by Nimrod Aviram
7898 and Sebastian Schinzel.
7903 * Reject DH handshakes with parameters shorter than 1024 bits.
7907 ### Changes between 1.0.1p and 1.0.1q [3 Dec 2015]
7909 * Certificate verify crash with missing PSS parameter
7911 The signature verification routines will crash with a NULL pointer
7912 dereference if presented with an ASN.1 signature using the RSA PSS
7913 algorithm and absent mask generation function parameter. Since these
7914 routines are used to verify certificate signature algorithms this can be
7915 used to crash any certificate verification operation and exploited in a
7916 DoS attack. Any application which performs certificate verification is
7917 vulnerable including OpenSSL clients and servers which enable client
7920 This issue was reported to OpenSSL by Loïc Jonas Etienne (Qnective AG).
7925 * X509_ATTRIBUTE memory leak
7927 When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak
7928 memory. This structure is used by the PKCS#7 and CMS routines so any
7929 application which reads PKCS#7 or CMS data from untrusted sources is
7930 affected. SSL/TLS is not affected.
7932 This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using
7938 * Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs.
7939 This changes the decoding behaviour for some invalid messages,
7940 though the change is mostly in the more lenient direction, and
7941 legacy behaviour is preserved as much as possible.
7945 * In DSA_generate_parameters_ex, if the provided seed is too short,
7946 use a random seed, as already documented.
7948 *Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>*
7950 ### Changes between 1.0.1o and 1.0.1p [9 Jul 2015]
7952 * Alternate chains certificate forgery
7954 During certificate verification, OpenSSL will attempt to find an
7955 alternative certificate chain if the first attempt to build such a chain
7956 fails. An error in the implementation of this logic can mean that an
7957 attacker could cause certain checks on untrusted certificates to be
7958 bypassed, such as the CA flag, enabling them to use a valid leaf
7959 certificate to act as a CA and "issue" an invalid certificate.
7961 This issue was reported to OpenSSL by Adam Langley/David Benjamin
7967 * Race condition handling PSK identify hint
7969 If PSK identity hints are received by a multi-threaded client then
7970 the values are wrongly updated in the parent SSL_CTX structure. This can
7971 result in a race condition potentially leading to a double free of the
7977 ### Changes between 1.0.1n and 1.0.1o [12 Jun 2015]
7979 * Fix HMAC ABI incompatibility. The previous version introduced an ABI
7980 incompatibility in the handling of HMAC. The previous ABI has now been
7983 ### Changes between 1.0.1m and 1.0.1n [11 Jun 2015]
7985 * Malformed ECParameters causes infinite loop
7987 When processing an ECParameters structure OpenSSL enters an infinite loop
7988 if the curve specified is over a specially malformed binary polynomial
7991 This can be used to perform denial of service against any
7992 system which processes public keys, certificate requests or
7993 certificates. This includes TLS clients and TLS servers with
7994 client authentication enabled.
7996 This issue was reported to OpenSSL by Joseph Barr-Pixton.
8001 * Exploitable out-of-bounds read in X509_cmp_time
8003 X509_cmp_time does not properly check the length of the ASN1_TIME
8004 string and can read a few bytes out of bounds. In addition,
8005 X509_cmp_time accepts an arbitrary number of fractional seconds in the
8008 An attacker can use this to craft malformed certificates and CRLs of
8009 various sizes and potentially cause a segmentation fault, resulting in
8010 a DoS on applications that verify certificates or CRLs. TLS clients
8011 that verify CRLs are affected. TLS clients and servers with client
8012 authentication enabled may be affected if they use custom verification
8015 This issue was reported to OpenSSL by Robert Swiecki (Google), and
8016 independently by Hanno Böck.
8021 * PKCS7 crash with missing EnvelopedContent
8023 The PKCS#7 parsing code does not handle missing inner EncryptedContent
8024 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs
8025 with missing content and trigger a NULL pointer dereference on parsing.
8027 Applications that decrypt PKCS#7 data or otherwise parse PKCS#7
8028 structures from untrusted sources are affected. OpenSSL clients and
8029 servers are not affected.
8031 This issue was reported to OpenSSL by Michal Zalewski (Google).
8036 * CMS verify infinite loop with unknown hash function
8038 When verifying a signedData message the CMS code can enter an infinite loop
8039 if presented with an unknown hash function OID. This can be used to perform
8040 denial of service against any system which verifies signedData messages using
8042 This issue was reported to OpenSSL by Johannes Bauer.
8047 * Race condition handling NewSessionTicket
8049 If a NewSessionTicket is received by a multi-threaded client when attempting to
8050 reuse a previous ticket then a race condition can occur potentially leading to
8051 a double free of the ticket data.
8056 * Reject DH handshakes with parameters shorter than 768 bits.
8058 *Kurt Roeckx and Emilia Kasper*
8060 * dhparam: generate 2048-bit parameters by default.
8062 *Kurt Roeckx and Emilia Kasper*
8064 ### Changes between 1.0.1l and 1.0.1m [19 Mar 2015]
8066 * Segmentation fault in ASN1_TYPE_cmp fix
8068 The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is
8069 made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check
8070 certificate signature algorithm consistency this can be used to crash any
8071 certificate verification operation and exploited in a DoS attack. Any
8072 application which performs certificate verification is vulnerable including
8073 OpenSSL clients and servers which enable client authentication.
8078 * ASN.1 structure reuse memory corruption fix
8080 Reusing a structure in ASN.1 parsing may allow an attacker to cause
8081 memory corruption via an invalid write. Such reuse is and has been
8082 strongly discouraged and is believed to be rare.
8084 Applications that parse structures containing CHOICE or ANY DEFINED BY
8085 components may be affected. Certificate parsing (d2i_X509 and related
8086 functions) are however not affected. OpenSSL clients and servers are
8092 * PKCS7 NULL pointer dereferences fix
8094 The PKCS#7 parsing code does not handle missing outer ContentInfo
8095 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with
8096 missing content and trigger a NULL pointer dereference on parsing.
8098 Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or
8099 otherwise parse PKCS#7 structures from untrusted sources are
8100 affected. OpenSSL clients and servers are not affected.
8102 This issue was reported to OpenSSL by Michal Zalewski (Google).
8107 * DoS via reachable assert in SSLv2 servers fix
8109 A malicious client can trigger an OPENSSL_assert (i.e., an abort) in
8110 servers that both support SSLv2 and enable export cipher suites by sending
8111 a specially crafted SSLv2 CLIENT-MASTER-KEY message.
8113 This issue was discovered by Sean Burford (Google) and Emilia Käsper
8114 (OpenSSL development team).
8119 * Use After Free following d2i_ECPrivatekey error fix
8121 A malformed EC private key file consumed via the d2i_ECPrivateKey function
8122 could cause a use after free condition. This, in turn, could cause a double
8123 free in several private key parsing functions (such as d2i_PrivateKey
8124 or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption
8125 for applications that receive EC private keys from untrusted
8126 sources. This scenario is considered rare.
8128 This issue was discovered by the BoringSSL project and fixed in their
8134 * X509_to_X509_REQ NULL pointer deref fix
8136 The function X509_to_X509_REQ will crash with a NULL pointer dereference if
8137 the certificate key is invalid. This function is rarely used in practice.
8139 This issue was discovered by Brian Carpenter.
8144 * Removed the export ciphers from the DEFAULT ciphers
8148 ### Changes between 1.0.1k and 1.0.1l [15 Jan 2015]
8150 * Build fixes for the Windows and OpenVMS platforms
8152 *Matt Caswell and Richard Levitte*
8154 ### Changes between 1.0.1j and 1.0.1k [8 Jan 2015]
8156 * Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS
8157 message can cause a segmentation fault in OpenSSL due to a NULL pointer
8158 dereference. This could lead to a Denial Of Service attack. Thanks to
8159 Markus Stenberg of Cisco Systems, Inc. for reporting this issue.
8164 * Fix DTLS memory leak in dtls1_buffer_record. A memory leak can occur in the
8165 dtls1_buffer_record function under certain conditions. In particular this
8166 could occur if an attacker sent repeated DTLS records with the same
8167 sequence number but for the next epoch. The memory leak could be exploited
8168 by an attacker in a Denial of Service attack through memory exhaustion.
8169 Thanks to Chris Mueller for reporting this issue.
8174 * Fix issue where no-ssl3 configuration sets method to NULL. When openssl is
8175 built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl
8176 method would be set to NULL which could later result in a NULL pointer
8177 dereference. Thanks to Frank Schmirler for reporting this issue.
8182 * Abort handshake if server key exchange message is omitted for ephemeral
8185 Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for
8186 reporting this issue.
8191 * Remove non-export ephemeral RSA code on client and server. This code
8192 violated the TLS standard by allowing the use of temporary RSA keys in
8193 non-export ciphersuites and could be used by a server to effectively
8194 downgrade the RSA key length used to a value smaller than the server
8195 certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at
8196 INRIA or reporting this issue.
8201 * Fixed issue where DH client certificates are accepted without verification.
8202 An OpenSSL server will accept a DH certificate for client authentication
8203 without the certificate verify message. This effectively allows a client to
8204 authenticate without the use of a private key. This only affects servers
8205 which trust a client certificate authority which issues certificates
8206 containing DH keys: these are extremely rare and hardly ever encountered.
8207 Thanks for Karthikeyan Bhargavan of the PROSECCO team at INRIA or reporting
8213 * Ensure that the session ID context of an SSL is updated when its
8214 SSL_CTX is updated via SSL_set_SSL_CTX.
8216 The session ID context is typically set from the parent SSL_CTX,
8217 and can vary with the CTX.
8221 * Fix various certificate fingerprint issues.
8223 By using non-DER or invalid encodings outside the signed portion of a
8224 certificate the fingerprint can be changed without breaking the signature.
8225 Although no details of the signed portion of the certificate can be changed
8226 this can cause problems with some applications: e.g. those using the
8227 certificate fingerprint for blacklists.
8229 1. Reject signatures with non zero unused bits.
8231 If the BIT STRING containing the signature has non zero unused bits reject
8232 the signature. All current signature algorithms require zero unused bits.
8234 2. Check certificate algorithm consistency.
8236 Check the AlgorithmIdentifier inside TBS matches the one in the
8237 certificate signature. NB: this will result in signature failure
8238 errors for some broken certificates.
8240 Thanks to Konrad Kraszewski from Google for reporting this issue.
8242 3. Check DSA/ECDSA signatures use DER.
8244 Re-encode DSA/ECDSA signatures and compare with the original received
8245 signature. Return an error if there is a mismatch.
8247 This will reject various cases including garbage after signature
8248 (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS
8249 program for discovering this case) and use of BER or invalid ASN.1 INTEGERs
8250 (negative or with leading zeroes).
8252 Further analysis was conducted and fixes were developed by Stephen Henson
8253 of the OpenSSL core team.
8259 * Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect
8260 results on some platforms, including x86_64. This bug occurs at random
8261 with a very low probability, and is not known to be exploitable in any
8262 way, though its exact impact is difficult to determine. Thanks to Pieter
8263 Wuille (Blockstream) who reported this issue and also suggested an initial
8264 fix. Further analysis was conducted by the OpenSSL development team and
8265 Adam Langley of Google. The final fix was developed by Andy Polyakov of
8266 the OpenSSL core team.
8271 * Do not resume sessions on the server if the negotiated protocol
8272 version does not match the session's version. Resuming with a different
8273 version, while not strictly forbidden by the RFC, is of questionable
8274 sanity and breaks all known clients.
8276 *David Benjamin, Emilia Käsper*
8278 * Tighten handling of the ChangeCipherSpec (CCS) message: reject
8279 early CCS messages during renegotiation. (Note that because
8280 renegotiation is encrypted, this early CCS was not exploitable.)
8284 * Tighten client-side session ticket handling during renegotiation:
8285 ensure that the client only accepts a session ticket if the server sends
8286 the extension anew in the ServerHello. Previously, a TLS client would
8287 reuse the old extension state and thus accept a session ticket if one was
8288 announced in the initial ServerHello.
8290 Similarly, ensure that the client requires a session ticket if one
8291 was advertised in the ServerHello. Previously, a TLS client would
8292 ignore a missing NewSessionTicket message.
8296 ### Changes between 1.0.1i and 1.0.1j [15 Oct 2014]
8300 A flaw in the DTLS SRTP extension parsing code allows an attacker, who
8301 sends a carefully crafted handshake message, to cause OpenSSL to fail
8302 to free up to 64k of memory causing a memory leak. This could be
8303 exploited in a Denial Of Service attack. This issue affects OpenSSL
8304 1.0.1 server implementations for both SSL/TLS and DTLS regardless of
8305 whether SRTP is used or configured. Implementations of OpenSSL that
8306 have been compiled with OPENSSL_NO_SRTP defined are not affected.
8308 The fix was developed by the OpenSSL team.
8313 * Session Ticket Memory Leak.
8315 When an OpenSSL SSL/TLS/DTLS server receives a session ticket the
8316 integrity of that ticket is first verified. In the event of a session
8317 ticket integrity check failing, OpenSSL will fail to free memory
8318 causing a memory leak. By sending a large number of invalid session
8319 tickets an attacker could exploit this issue in a Denial Of Service
8325 * Build option no-ssl3 is incomplete.
8327 When OpenSSL is configured with "no-ssl3" as a build option, servers
8328 could accept and complete an SSL 3.0 handshake, and clients could be
8329 configured to send them.
8332 *Akamai and the OpenSSL team*
8334 * Add support for TLS_FALLBACK_SCSV.
8335 Client applications doing fallback retries should call
8336 SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV).
8339 *Adam Langley, Bodo Moeller*
8341 * Add additional DigestInfo checks.
8343 Re-encode DigestInto in DER and check against the original when
8344 verifying RSA signature: this will reject any improperly encoded
8345 DigestInfo structures.
8347 Note: this is a precautionary measure and no attacks are currently known.
8351 ### Changes between 1.0.1h and 1.0.1i [6 Aug 2014]
8353 * Fix SRP buffer overrun vulnerability. Invalid parameters passed to the
8354 SRP code can be overrun an internal buffer. Add sanity check that
8355 g, A, B < N to SRP code.
8357 Thanks to Sean Devlin and Watson Ladd of Cryptography Services, NCC
8358 Group for discovering this issue.
8363 * A flaw in the OpenSSL SSL/TLS server code causes the server to negotiate
8364 TLS 1.0 instead of higher protocol versions when the ClientHello message
8365 is badly fragmented. This allows a man-in-the-middle attacker to force a
8366 downgrade to TLS 1.0 even if both the server and the client support a
8367 higher protocol version, by modifying the client's TLS records.
8369 Thanks to David Benjamin and Adam Langley (Google) for discovering and
8370 researching this issue.
8375 * OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject
8376 to a denial of service attack. A malicious server can crash the client
8377 with a null pointer dereference (read) by specifying an anonymous (EC)DH
8378 ciphersuite and sending carefully crafted handshake messages.
8380 Thanks to Felix Gröbert (Google) for discovering and researching this
8386 * By sending carefully crafted DTLS packets an attacker could cause openssl
8387 to leak memory. This can be exploited through a Denial of Service attack.
8388 Thanks to Adam Langley for discovering and researching this issue.
8393 * An attacker can force openssl to consume large amounts of memory whilst
8394 processing DTLS handshake messages. This can be exploited through a
8395 Denial of Service attack.
8396 Thanks to Adam Langley for discovering and researching this issue.
8401 * An attacker can force an error condition which causes openssl to crash
8402 whilst processing DTLS packets due to memory being freed twice. This
8403 can be exploited through a Denial of Service attack.
8404 Thanks to Adam Langley and Wan-Teh Chang for discovering and researching
8410 * If a multithreaded client connects to a malicious server using a resumed
8411 session and the server sends an ec point format extension it could write
8412 up to 255 bytes to freed memory.
8414 Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this
8420 * A malicious server can crash an OpenSSL client with a null pointer
8421 dereference (read) by specifying an SRP ciphersuite even though it was not
8422 properly negotiated with the client. This can be exploited through a
8423 Denial of Service attack.
8425 Thanks to Joonas Kuorilehto and Riku Hietamäki (Codenomicon) for
8426 discovering and researching this issue.
8431 * A flaw in OBJ_obj2txt may cause pretty printing functions such as
8432 X509_name_oneline, X509_name_print_ex et al. to leak some information
8433 from the stack. Applications may be affected if they echo pretty printing
8434 output to the attacker.
8436 Thanks to Ivan Fratric (Google) for discovering this issue.
8439 *Emilia Käsper, and Steve Henson*
8441 * Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.)
8442 for corner cases. (Certain input points at infinity could lead to
8443 bogus results, with non-infinity inputs mapped to infinity too.)
8447 ### Changes between 1.0.1g and 1.0.1h [5 Jun 2014]
8449 * Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted
8450 handshake can force the use of weak keying material in OpenSSL
8451 SSL/TLS clients and servers.
8453 Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and
8454 researching this issue. ([CVE-2014-0224])
8456 *KIKUCHI Masashi, Steve Henson*
8458 * Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an
8459 OpenSSL DTLS client the code can be made to recurse eventually crashing
8462 Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue.
8465 *Imre Rad, Steve Henson*
8467 * Fix DTLS invalid fragment vulnerability. A buffer overrun attack can
8468 be triggered by sending invalid DTLS fragments to an OpenSSL DTLS
8469 client or server. This is potentially exploitable to run arbitrary
8470 code on a vulnerable client or server.
8472 Thanks to Jüri Aedla for reporting this issue. ([CVE-2014-0195])
8474 *Jüri Aedla, Steve Henson*
8476 * Fix bug in TLS code where clients enable anonymous ECDH ciphersuites
8477 are subject to a denial of service attack.
8479 Thanks to Felix Gröbert and Ivan Fratric at Google for discovering
8480 this issue. ([CVE-2014-3470])
8482 *Felix Gröbert, Ivan Fratric, Steve Henson*
8484 * Harmonize version and its documentation. -f flag is used to display
8487 *mancha <mancha1@zoho.com>*
8489 * Fix eckey_priv_encode so it immediately returns an error upon a failure
8490 in i2d_ECPrivateKey.
8492 *mancha <mancha1@zoho.com>*
8494 * Fix some double frees. These are not thought to be exploitable.
8496 *mancha <mancha1@zoho.com>*
8498 ### Changes between 1.0.1f and 1.0.1g [7 Apr 2014]
8500 * A missing bounds check in the handling of the TLS heartbeat extension
8501 can be used to reveal up to 64k of memory to a connected client or
8504 Thanks for Neel Mehta of Google Security for discovering this bug and to
8505 Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for
8506 preparing the fix ([CVE-2014-0160])
8508 *Adam Langley, Bodo Moeller*
8510 * Fix for the attack described in the paper "Recovering OpenSSL
8511 ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack"
8512 by Yuval Yarom and Naomi Benger. Details can be obtained from:
8513 <http://eprint.iacr.org/2014/140>
8515 Thanks to Yuval Yarom and Naomi Benger for discovering this
8516 flaw and to Yuval Yarom for supplying a fix ([CVE-2014-0076])
8518 *Yuval Yarom and Naomi Benger*
8520 * TLS pad extension: draft-agl-tls-padding-03
8522 Workaround for the "TLS hang bug" (see FAQ and PR#2771): if the
8523 TLS client Hello record length value would otherwise be > 255 and
8524 less that 512 pad with a dummy extension containing zeroes so it
8525 is at least 512 bytes long.
8527 *Adam Langley, Steve Henson*
8529 ### Changes between 1.0.1e and 1.0.1f [6 Jan 2014]
8531 * Fix for TLS record tampering bug. A carefully crafted invalid
8532 handshake could crash OpenSSL with a NULL pointer exception.
8533 Thanks to Anton Johansson for reporting this issues.
8536 * Keep original DTLS digest and encryption contexts in retransmission
8537 structures so we can use the previous session parameters if they need
8538 to be resent. ([CVE-2013-6450])
8542 * Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which
8543 avoids preferring ECDHE-ECDSA ciphers when the client appears to be
8544 Safari on OS X. Safari on OS X 10.8..10.8.3 advertises support for
8545 several ECDHE-ECDSA ciphers, but fails to negotiate them. The bug
8546 is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing
8547 10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer.
8549 *Rob Stradling, Adam Langley*
8551 ### Changes between 1.0.1d and 1.0.1e [11 Feb 2013]
8553 * Correct fix for CVE-2013-0169. The original didn't work on AES-NI
8554 supporting platforms or when small records were transferred.
8556 *Andy Polyakov, Steve Henson*
8558 ### Changes between 1.0.1c and 1.0.1d [5 Feb 2013]
8560 * Make the decoding of SSLv3, TLS and DTLS CBC records constant time.
8562 This addresses the flaw in CBC record processing discovered by
8563 Nadhem Alfardan and Kenny Paterson. Details of this attack can be found
8564 at: <http://www.isg.rhul.ac.uk/tls/>
8566 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
8567 Security Group at Royal Holloway, University of London
8568 (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and
8569 Emilia Käsper for the initial patch.
8572 *Emilia Käsper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson*
8574 * Fix flaw in AESNI handling of TLS 1.2 and 1.1 records for CBC mode
8575 ciphersuites which can be exploited in a denial of service attack.
8576 Thanks go to and to Adam Langley <agl@chromium.org> for discovering
8577 and detecting this bug and to Wolfgang Ettlinger
8578 <wolfgang.ettlinger@gmail.com> for independently discovering this issue.
8583 * Return an error when checking OCSP signatures when key is NULL.
8584 This fixes a DoS attack. ([CVE-2013-0166])
8588 * Make openssl verify return errors.
8590 *Chris Palmer <palmer@google.com> and Ben Laurie*
8592 * Call OCSP Stapling callback after ciphersuite has been chosen, so
8593 the right response is stapled. Also change SSL_get_certificate()
8594 so it returns the certificate actually sent.
8595 See <http://rt.openssl.org/Ticket/Display.html?id=2836>.
8597 *Rob Stradling <rob.stradling@comodo.com>*
8599 * Fix possible deadlock when decoding public keys.
8603 * Don't use TLS 1.0 record version number in initial client hello
8608 ### Changes between 1.0.1b and 1.0.1c [10 May 2012]
8610 * Sanity check record length before skipping explicit IV in TLS
8611 1.2, 1.1 and DTLS to fix DoS attack.
8613 Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic
8614 fuzzing as a service testing platform.
8619 * Initialise tkeylen properly when encrypting CMS messages.
8620 Thanks to Solar Designer of Openwall for reporting this issue.
8624 * In FIPS mode don't try to use composite ciphers as they are not
8629 ### Changes between 1.0.1a and 1.0.1b [26 Apr 2012]
8631 * OpenSSL 1.0.0 sets SSL_OP_ALL to 0x80000FFFL and OpenSSL 1.0.1 and
8632 1.0.1a set SSL_OP_NO_TLSv1_1 to 0x00000400L which would unfortunately
8633 mean any application compiled against OpenSSL 1.0.0 headers setting
8634 SSL_OP_ALL would also set SSL_OP_NO_TLSv1_1, unintentionally disabling
8635 TLS 1.1 also. Fix this by changing the value of SSL_OP_NO_TLSv1_1 to
8636 0x10000000L Any application which was previously compiled against
8637 OpenSSL 1.0.1 or 1.0.1a headers and which cares about SSL_OP_NO_TLSv1_1
8638 will need to be recompiled as a result. Letting be results in
8639 inability to disable specifically TLS 1.1 and in client context,
8640 in unlike event, limit maximum offered version to TLS 1.0 [see below].
8644 * In order to ensure interoperability SSL_OP_NO_protocolX does not
8645 disable just protocol X, but all protocols above X *if* there are
8646 protocols *below* X still enabled. In more practical terms it means
8647 that if application wants to disable TLS1.0 in favor of TLS1.1 and
8648 above, it's not sufficient to pass `SSL_OP_NO_TLSv1`, one has to pass
8649 `SSL_OP_NO_TLSv1|SSL_OP_NO_SSLv3|SSL_OP_NO_SSLv2`. This applies to
8654 ### Changes between 1.0.1 and 1.0.1a [19 Apr 2012]
8656 * Check for potentially exploitable overflows in asn1_d2i_read_bio
8657 BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer
8658 in CRYPTO_realloc_clean.
8660 Thanks to Tavis Ormandy, Google Security Team, for discovering this
8661 issue and to Adam Langley <agl@chromium.org> for fixing it.
8664 *Adam Langley (Google), Tavis Ormandy, Google Security Team*
8666 * Don't allow TLS 1.2 SHA-256 ciphersuites in TLS 1.0, 1.1 connections.
8670 * Workarounds for some broken servers that "hang" if a client hello
8671 record length exceeds 255 bytes.
8673 1. Do not use record version number > TLS 1.0 in initial client
8674 hello: some (but not all) hanging servers will now work.
8675 2. If we set OPENSSL_MAX_TLS1_2_CIPHER_LENGTH this will truncate
8676 the number of ciphers sent in the client hello. This should be
8677 set to an even number, such as 50, for example by passing:
8678 -DOPENSSL_MAX_TLS1_2_CIPHER_LENGTH=50 to config or Configure.
8679 Most broken servers should now work.
8680 3. If all else fails setting OPENSSL_NO_TLS1_2_CLIENT will disable
8681 TLS 1.2 client support entirely.
8685 * Fix SEGV in Vector Permutation AES module observed in OpenSSH.
8689 ### Changes between 1.0.0h and 1.0.1 [14 Mar 2012]
8691 * Add compatibility with old MDC2 signatures which use an ASN1 OCTET
8692 STRING form instead of a DigestInfo.
8696 * The format used for MDC2 RSA signatures is inconsistent between EVP
8697 and the RSA_sign/RSA_verify functions. This was made more apparent when
8698 OpenSSL used RSA_sign/RSA_verify for some RSA signatures in particular
8699 those which went through EVP_PKEY_METHOD in 1.0.0 and later. Detect
8700 the correct format in RSA_verify so both forms transparently work.
8704 * Some servers which support TLS 1.0 can choke if we initially indicate
8705 support for TLS 1.2 and later renegotiate using TLS 1.0 in the RSA
8706 encrypted premaster secret. As a workaround use the maximum permitted
8707 client version in client hello, this should keep such servers happy
8708 and still work with previous versions of OpenSSL.
8712 * Add support for TLS/DTLS heartbeats.
8714 *Robin Seggelmann <seggelmann@fh-muenster.de>*
8716 * Add support for SCTP.
8718 *Robin Seggelmann <seggelmann@fh-muenster.de>*
8720 * Improved PRNG seeding for VOS.
8722 *Paul Green <Paul.Green@stratus.com>*
8724 * Extensive assembler packs updates, most notably:
8726 - x86[_64]: AES-NI, PCLMULQDQ, RDRAND support;
8727 - x86[_64]: SSSE3 support (SHA1, vector-permutation AES);
8728 - x86_64: bit-sliced AES implementation;
8729 - ARM: NEON support, contemporary platforms optimizations;
8730 - s390x: z196 support;
8731 - `*`: GHASH and GF(2^m) multiplication implementations;
8735 * Make TLS-SRP code conformant with RFC 5054 API cleanup
8736 (removal of unnecessary code)
8738 *Peter Sylvester <peter.sylvester@edelweb.fr>*
8740 * Add TLS key material exporter from RFC 5705.
8744 * Add DTLS-SRTP negotiation from RFC 5764.
8748 * Add Next Protocol Negotiation,
8749 <http://tools.ietf.org/html/draft-agl-tls-nextprotoneg-00>. Can be
8750 disabled with a no-npn flag to config or Configure. Code donated
8753 *Adam Langley <agl@google.com> and Ben Laurie*
8755 * Add optional 64-bit optimized implementations of elliptic curves NIST-P224,
8756 NIST-P256, NIST-P521, with constant-time single point multiplication on
8757 typical inputs. Compiler support for the nonstandard type `__uint128_t` is
8758 required to use this (present in gcc 4.4 and later, for 64-bit builds).
8759 Code made available under Apache License version 2.0.
8761 Specify "enable-ec_nistp_64_gcc_128" on the Configure (or config) command
8762 line to include this in your build of OpenSSL, and run "make depend" (or
8763 "make update"). This enables the following EC_METHODs:
8765 EC_GFp_nistp224_method()
8766 EC_GFp_nistp256_method()
8767 EC_GFp_nistp521_method()
8769 EC_GROUP_new_by_curve_name() will automatically use these (while
8770 EC_GROUP_new_curve_GFp() currently prefers the more flexible
8773 *Emilia Käsper, Adam Langley, Bodo Moeller (Google)*
8775 * Use type ossl_ssize_t instead of ssize_t which isn't available on
8776 all platforms. Move ssize_t definition from e_os.h to the public
8777 header file e_os2.h as it now appears in public header file cms.h
8781 * New -sigopt option to the ca, req and x509 utilities. Additional
8782 signature parameters can be passed using this option and in
8787 * Add RSA PSS signing function. This will generate and set the
8788 appropriate AlgorithmIdentifiers for PSS based on those in the
8789 corresponding EVP_MD_CTX structure. No application support yet.
8793 * Support for companion algorithm specific ASN1 signing routines.
8794 New function ASN1_item_sign_ctx() signs a pre-initialised
8795 EVP_MD_CTX structure and sets AlgorithmIdentifiers based on
8796 the appropriate parameters.
8800 * Add new algorithm specific ASN1 verification initialisation function
8801 to EVP_PKEY_ASN1_METHOD: this is not in EVP_PKEY_METHOD since the ASN1
8802 handling will be the same no matter what EVP_PKEY_METHOD is used.
8803 Add a PSS handler to support verification of PSS signatures: checked
8804 against a number of sample certificates.
8808 * Add signature printing for PSS. Add PSS OIDs.
8810 *Steve Henson, Martin Kaiser <lists@kaiser.cx>*
8812 * Add algorithm specific signature printing. An individual ASN1 method
8813 can now print out signatures instead of the standard hex dump.
8815 More complex signatures (e.g. PSS) can print out more meaningful
8816 information. Include DSA version that prints out the signature
8821 * Password based recipient info support for CMS library: implementing
8826 * Split password based encryption into PBES2 and PBKDF2 functions. This
8827 neatly separates the code into cipher and PBE sections and is required
8828 for some algorithms that split PBES2 into separate pieces (such as
8829 password based CMS).
8833 * Session-handling fixes:
8834 - Fix handling of connections that are resuming with a session ID,
8835 but also support Session Tickets.
8836 - Fix a bug that suppressed issuing of a new ticket if the client
8837 presented a ticket with an expired session.
8838 - Try to set the ticket lifetime hint to something reasonable.
8839 - Make tickets shorter by excluding irrelevant information.
8840 - On the client side, don't ignore renewed tickets.
8842 *Adam Langley, Bodo Moeller (Google)*
8844 * Fix PSK session representation.
8848 * Add RC4-MD5 and AESNI-SHA1 "stitched" implementations.
8850 This work was sponsored by Intel.
8854 * Add GCM support to TLS library. Some custom code is needed to split
8855 the IV between the fixed (from PRF) and explicit (from TLS record)
8856 portions. This adds all GCM ciphersuites supported by RFC5288 and
8857 RFC5289. Generalise some `AES*` cipherstrings to include GCM and
8858 add a special AESGCM string for GCM only.
8862 * Expand range of ctrls for AES GCM. Permit setting invocation
8863 field on decrypt and retrieval of invocation field only on encrypt.
8867 * Add HMAC ECC ciphersuites from RFC5289. Include SHA384 PRF support.
8868 As required by RFC5289 these ciphersuites cannot be used if for
8869 versions of TLS earlier than 1.2.
8873 * For FIPS capable OpenSSL interpret a NULL default public key method
8874 as unset and return the appropriate default but do *not* set the default.
8875 This means we can return the appropriate method in applications that
8876 switch between FIPS and non-FIPS modes.
8880 * Redirect HMAC and CMAC operations to FIPS module in FIPS mode. If an
8881 ENGINE is used then we cannot handle that in the FIPS module so we
8882 keep original code iff non-FIPS operations are allowed.
8886 * Add -attime option to openssl utilities.
8888 *Peter Eckersley <pde@eff.org>, Ben Laurie and Steve Henson*
8890 * Redirect DSA and DH operations to FIPS module in FIPS mode.
8894 * Redirect ECDSA and ECDH operations to FIPS module in FIPS mode. Also use
8895 FIPS EC methods unconditionally for now.
8899 * New build option no-ec2m to disable characteristic 2 code.
8903 * Backport libcrypto audit of return value checking from 1.1.0-dev; not
8904 all cases can be covered as some introduce binary incompatibilities.
8908 * Redirect RSA operations to FIPS module including keygen,
8909 encrypt, decrypt, sign and verify. Block use of non FIPS RSA methods.
8913 * Add similar low-level API blocking to ciphers.
8917 * low-level digest APIs are not approved in FIPS mode: any attempt
8918 to use these will cause a fatal error. Applications that *really* want
8919 to use them can use the `private_*` version instead.
8923 * Redirect cipher operations to FIPS module for FIPS builds.
8927 * Redirect digest operations to FIPS module for FIPS builds.
8931 * Update build system to add "fips" flag which will link in fipscanister.o
8932 for static and shared library builds embedding a signature if needed.
8936 * Output TLS supported curves in preference order instead of numerical
8937 order. This is currently hardcoded for the highest order curves first.
8938 This should be configurable so applications can judge speed vs strength.
8942 * Add TLS v1.2 server support for client authentication.
8946 * Add support for FIPS mode in ssl library: disable SSLv3, non-FIPS ciphers
8951 * Functions FIPS_mode_set() and FIPS_mode() which call the underlying
8952 FIPS modules versions.
8956 * Add TLS v1.2 client side support for client authentication. Keep cache
8957 of handshake records longer as we don't know the hash algorithm to use
8958 until after the certificate request message is received.
8962 * Initial TLS v1.2 client support. Add a default signature algorithms
8963 extension including all the algorithms we support. Parse new signature
8964 format in client key exchange. Relax some ECC signing restrictions for
8965 TLS v1.2 as indicated in RFC5246.
8969 * Add server support for TLS v1.2 signature algorithms extension. Switch
8970 to new signature format when needed using client digest preference.
8971 All server ciphersuites should now work correctly in TLS v1.2. No client
8972 support yet and no support for client certificates.
8976 * Initial TLS v1.2 support. Add new SHA256 digest to ssl code, switch
8977 to SHA256 for PRF when using TLS v1.2 and later. Add new SHA256 based
8978 ciphersuites. At present only RSA key exchange ciphersuites work with
8979 TLS v1.2. Add new option for TLS v1.2 replacing the old and obsolete
8980 SSL_OP_PKCS1_CHECK flags with SSL_OP_NO_TLSv1_2. New TLSv1.2 methods
8981 and version checking.
8985 * New option OPENSSL_NO_SSL_INTERN. If an application can be compiled
8986 with this defined it will not be affected by any changes to ssl internal
8987 structures. Add several utility functions to allow openssl application
8988 to work with OPENSSL_NO_SSL_INTERN defined.
8992 * A long standing patch to add support for SRP from EdelWeb (Peter
8993 Sylvester and Christophe Renou) was integrated.
8994 *Christophe Renou <christophe.renou@edelweb.fr>, Peter Sylvester
8995 <peter.sylvester@edelweb.fr>, Tom Wu <tjw@cs.stanford.edu>, and
8998 * Add functions to copy EVP_PKEY_METHOD and retrieve flags and id.
9002 * Permit abbreviated handshakes when renegotiating using the function
9003 SSL_renegotiate_abbreviated().
9005 *Robin Seggelmann <seggelmann@fh-muenster.de>*
9007 * Add call to ENGINE_register_all_complete() to
9008 ENGINE_load_builtin_engines(), so some implementations get used
9009 automatically instead of needing explicit application support.
9013 * Add support for TLS key exporter as described in RFC5705.
9015 *Robin Seggelmann <seggelmann@fh-muenster.de>, Steve Henson*
9017 * Initial TLSv1.1 support. Since TLSv1.1 is very similar to TLS v1.0 only
9018 a few changes are required:
9020 Add SSL_OP_NO_TLSv1_1 flag.
9021 Add TLSv1_1 methods.
9022 Update version checking logic to handle version 1.1.
9023 Add explicit IV handling (ported from DTLS code).
9024 Add command line options to s_client/s_server.
9031 ### Changes between 1.0.0s and 1.0.0t [3 Dec 2015]
9033 * X509_ATTRIBUTE memory leak
9035 When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak
9036 memory. This structure is used by the PKCS#7 and CMS routines so any
9037 application which reads PKCS#7 or CMS data from untrusted sources is
9038 affected. SSL/TLS is not affected.
9040 This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using
9046 * Race condition handling PSK identify hint
9048 If PSK identity hints are received by a multi-threaded client then
9049 the values are wrongly updated in the parent SSL_CTX structure. This can
9050 result in a race condition potentially leading to a double free of the
9056 ### Changes between 1.0.0r and 1.0.0s [11 Jun 2015]
9058 * Malformed ECParameters causes infinite loop
9060 When processing an ECParameters structure OpenSSL enters an infinite loop
9061 if the curve specified is over a specially malformed binary polynomial
9064 This can be used to perform denial of service against any
9065 system which processes public keys, certificate requests or
9066 certificates. This includes TLS clients and TLS servers with
9067 client authentication enabled.
9069 This issue was reported to OpenSSL by Joseph Barr-Pixton.
9074 * Exploitable out-of-bounds read in X509_cmp_time
9076 X509_cmp_time does not properly check the length of the ASN1_TIME
9077 string and can read a few bytes out of bounds. In addition,
9078 X509_cmp_time accepts an arbitrary number of fractional seconds in the
9081 An attacker can use this to craft malformed certificates and CRLs of
9082 various sizes and potentially cause a segmentation fault, resulting in
9083 a DoS on applications that verify certificates or CRLs. TLS clients
9084 that verify CRLs are affected. TLS clients and servers with client
9085 authentication enabled may be affected if they use custom verification
9088 This issue was reported to OpenSSL by Robert Swiecki (Google), and
9089 independently by Hanno Böck.
9094 * PKCS7 crash with missing EnvelopedContent
9096 The PKCS#7 parsing code does not handle missing inner EncryptedContent
9097 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs
9098 with missing content and trigger a NULL pointer dereference on parsing.
9100 Applications that decrypt PKCS#7 data or otherwise parse PKCS#7
9101 structures from untrusted sources are affected. OpenSSL clients and
9102 servers are not affected.
9104 This issue was reported to OpenSSL by Michal Zalewski (Google).
9109 * CMS verify infinite loop with unknown hash function
9111 When verifying a signedData message the CMS code can enter an infinite loop
9112 if presented with an unknown hash function OID. This can be used to perform
9113 denial of service against any system which verifies signedData messages using
9115 This issue was reported to OpenSSL by Johannes Bauer.
9120 * Race condition handling NewSessionTicket
9122 If a NewSessionTicket is received by a multi-threaded client when attempting to
9123 reuse a previous ticket then a race condition can occur potentially leading to
9124 a double free of the ticket data.
9129 ### Changes between 1.0.0q and 1.0.0r [19 Mar 2015]
9131 * Segmentation fault in ASN1_TYPE_cmp fix
9133 The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is
9134 made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check
9135 certificate signature algorithm consistency this can be used to crash any
9136 certificate verification operation and exploited in a DoS attack. Any
9137 application which performs certificate verification is vulnerable including
9138 OpenSSL clients and servers which enable client authentication.
9143 * ASN.1 structure reuse memory corruption fix
9145 Reusing a structure in ASN.1 parsing may allow an attacker to cause
9146 memory corruption via an invalid write. Such reuse is and has been
9147 strongly discouraged and is believed to be rare.
9149 Applications that parse structures containing CHOICE or ANY DEFINED BY
9150 components may be affected. Certificate parsing (d2i_X509 and related
9151 functions) are however not affected. OpenSSL clients and servers are
9157 * PKCS7 NULL pointer dereferences fix
9159 The PKCS#7 parsing code does not handle missing outer ContentInfo
9160 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with
9161 missing content and trigger a NULL pointer dereference on parsing.
9163 Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or
9164 otherwise parse PKCS#7 structures from untrusted sources are
9165 affected. OpenSSL clients and servers are not affected.
9167 This issue was reported to OpenSSL by Michal Zalewski (Google).
9172 * DoS via reachable assert in SSLv2 servers fix
9174 A malicious client can trigger an OPENSSL_assert (i.e., an abort) in
9175 servers that both support SSLv2 and enable export cipher suites by sending
9176 a specially crafted SSLv2 CLIENT-MASTER-KEY message.
9178 This issue was discovered by Sean Burford (Google) and Emilia Käsper
9179 (OpenSSL development team).
9184 * Use After Free following d2i_ECPrivatekey error fix
9186 A malformed EC private key file consumed via the d2i_ECPrivateKey function
9187 could cause a use after free condition. This, in turn, could cause a double
9188 free in several private key parsing functions (such as d2i_PrivateKey
9189 or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption
9190 for applications that receive EC private keys from untrusted
9191 sources. This scenario is considered rare.
9193 This issue was discovered by the BoringSSL project and fixed in their
9199 * X509_to_X509_REQ NULL pointer deref fix
9201 The function X509_to_X509_REQ will crash with a NULL pointer dereference if
9202 the certificate key is invalid. This function is rarely used in practice.
9204 This issue was discovered by Brian Carpenter.
9209 * Removed the export ciphers from the DEFAULT ciphers
9213 ### Changes between 1.0.0p and 1.0.0q [15 Jan 2015]
9215 * Build fixes for the Windows and OpenVMS platforms
9217 *Matt Caswell and Richard Levitte*
9219 ### Changes between 1.0.0o and 1.0.0p [8 Jan 2015]
9221 * Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS
9222 message can cause a segmentation fault in OpenSSL due to a NULL pointer
9223 dereference. This could lead to a Denial Of Service attack. Thanks to
9224 Markus Stenberg of Cisco Systems, Inc. for reporting this issue.
9229 * Fix DTLS memory leak in dtls1_buffer_record. A memory leak can occur in the
9230 dtls1_buffer_record function under certain conditions. In particular this
9231 could occur if an attacker sent repeated DTLS records with the same
9232 sequence number but for the next epoch. The memory leak could be exploited
9233 by an attacker in a Denial of Service attack through memory exhaustion.
9234 Thanks to Chris Mueller for reporting this issue.
9239 * Fix issue where no-ssl3 configuration sets method to NULL. When openssl is
9240 built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl
9241 method would be set to NULL which could later result in a NULL pointer
9242 dereference. Thanks to Frank Schmirler for reporting this issue.
9247 * Abort handshake if server key exchange message is omitted for ephemeral
9250 Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for
9251 reporting this issue.
9256 * Remove non-export ephemeral RSA code on client and server. This code
9257 violated the TLS standard by allowing the use of temporary RSA keys in
9258 non-export ciphersuites and could be used by a server to effectively
9259 downgrade the RSA key length used to a value smaller than the server
9260 certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at
9261 INRIA or reporting this issue.
9266 * Fixed issue where DH client certificates are accepted without verification.
9267 An OpenSSL server will accept a DH certificate for client authentication
9268 without the certificate verify message. This effectively allows a client to
9269 authenticate without the use of a private key. This only affects servers
9270 which trust a client certificate authority which issues certificates
9271 containing DH keys: these are extremely rare and hardly ever encountered.
9272 Thanks for Karthikeyan Bhargavan of the PROSECCO team at INRIA or reporting
9278 * Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect
9279 results on some platforms, including x86_64. This bug occurs at random
9280 with a very low probability, and is not known to be exploitable in any
9281 way, though its exact impact is difficult to determine. Thanks to Pieter
9282 Wuille (Blockstream) who reported this issue and also suggested an initial
9283 fix. Further analysis was conducted by the OpenSSL development team and
9284 Adam Langley of Google. The final fix was developed by Andy Polyakov of
9285 the OpenSSL core team.
9290 * Fix various certificate fingerprint issues.
9292 By using non-DER or invalid encodings outside the signed portion of a
9293 certificate the fingerprint can be changed without breaking the signature.
9294 Although no details of the signed portion of the certificate can be changed
9295 this can cause problems with some applications: e.g. those using the
9296 certificate fingerprint for blacklists.
9298 1. Reject signatures with non zero unused bits.
9300 If the BIT STRING containing the signature has non zero unused bits reject
9301 the signature. All current signature algorithms require zero unused bits.
9303 2. Check certificate algorithm consistency.
9305 Check the AlgorithmIdentifier inside TBS matches the one in the
9306 certificate signature. NB: this will result in signature failure
9307 errors for some broken certificates.
9309 Thanks to Konrad Kraszewski from Google for reporting this issue.
9311 3. Check DSA/ECDSA signatures use DER.
9313 Re-encode DSA/ECDSA signatures and compare with the original received
9314 signature. Return an error if there is a mismatch.
9316 This will reject various cases including garbage after signature
9317 (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS
9318 program for discovering this case) and use of BER or invalid ASN.1 INTEGERs
9319 (negative or with leading zeroes).
9321 Further analysis was conducted and fixes were developed by Stephen Henson
9322 of the OpenSSL core team.
9328 ### Changes between 1.0.0n and 1.0.0o [15 Oct 2014]
9330 * Session Ticket Memory Leak.
9332 When an OpenSSL SSL/TLS/DTLS server receives a session ticket the
9333 integrity of that ticket is first verified. In the event of a session
9334 ticket integrity check failing, OpenSSL will fail to free memory
9335 causing a memory leak. By sending a large number of invalid session
9336 tickets an attacker could exploit this issue in a Denial Of Service
9342 * Build option no-ssl3 is incomplete.
9344 When OpenSSL is configured with "no-ssl3" as a build option, servers
9345 could accept and complete an SSL 3.0 handshake, and clients could be
9346 configured to send them.
9349 *Akamai and the OpenSSL team*
9351 * Add support for TLS_FALLBACK_SCSV.
9352 Client applications doing fallback retries should call
9353 SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV).
9356 *Adam Langley, Bodo Moeller*
9358 * Add additional DigestInfo checks.
9360 Re-encode DigestInto in DER and check against the original when
9361 verifying RSA signature: this will reject any improperly encoded
9362 DigestInfo structures.
9364 Note: this is a precautionary measure and no attacks are currently known.
9368 ### Changes between 1.0.0m and 1.0.0n [6 Aug 2014]
9370 * OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject
9371 to a denial of service attack. A malicious server can crash the client
9372 with a null pointer dereference (read) by specifying an anonymous (EC)DH
9373 ciphersuite and sending carefully crafted handshake messages.
9375 Thanks to Felix Gröbert (Google) for discovering and researching this
9381 * By sending carefully crafted DTLS packets an attacker could cause openssl
9382 to leak memory. This can be exploited through a Denial of Service attack.
9383 Thanks to Adam Langley for discovering and researching this issue.
9388 * An attacker can force openssl to consume large amounts of memory whilst
9389 processing DTLS handshake messages. This can be exploited through a
9390 Denial of Service attack.
9391 Thanks to Adam Langley for discovering and researching this issue.
9396 * An attacker can force an error condition which causes openssl to crash
9397 whilst processing DTLS packets due to memory being freed twice. This
9398 can be exploited through a Denial of Service attack.
9399 Thanks to Adam Langley and Wan-Teh Chang for discovering and researching
9405 * If a multithreaded client connects to a malicious server using a resumed
9406 session and the server sends an ec point format extension it could write
9407 up to 255 bytes to freed memory.
9409 Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this
9415 * A flaw in OBJ_obj2txt may cause pretty printing functions such as
9416 X509_name_oneline, X509_name_print_ex et al. to leak some information
9417 from the stack. Applications may be affected if they echo pretty printing
9418 output to the attacker.
9420 Thanks to Ivan Fratric (Google) for discovering this issue.
9423 *Emilia Käsper, and Steve Henson*
9425 * Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.)
9426 for corner cases. (Certain input points at infinity could lead to
9427 bogus results, with non-infinity inputs mapped to infinity too.)
9431 ### Changes between 1.0.0l and 1.0.0m [5 Jun 2014]
9433 * Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted
9434 handshake can force the use of weak keying material in OpenSSL
9435 SSL/TLS clients and servers.
9437 Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and
9438 researching this issue. ([CVE-2014-0224])
9440 *KIKUCHI Masashi, Steve Henson*
9442 * Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an
9443 OpenSSL DTLS client the code can be made to recurse eventually crashing
9446 Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue.
9449 *Imre Rad, Steve Henson*
9451 * Fix DTLS invalid fragment vulnerability. A buffer overrun attack can
9452 be triggered by sending invalid DTLS fragments to an OpenSSL DTLS
9453 client or server. This is potentially exploitable to run arbitrary
9454 code on a vulnerable client or server.
9456 Thanks to Jüri Aedla for reporting this issue. ([CVE-2014-0195])
9458 *Jüri Aedla, Steve Henson*
9460 * Fix bug in TLS code where clients enable anonymous ECDH ciphersuites
9461 are subject to a denial of service attack.
9463 Thanks to Felix Gröbert and Ivan Fratric at Google for discovering
9464 this issue. ([CVE-2014-3470])
9466 *Felix Gröbert, Ivan Fratric, Steve Henson*
9468 * Harmonize version and its documentation. -f flag is used to display
9471 *mancha <mancha1@zoho.com>*
9473 * Fix eckey_priv_encode so it immediately returns an error upon a failure
9474 in i2d_ECPrivateKey.
9476 *mancha <mancha1@zoho.com>*
9478 * Fix some double frees. These are not thought to be exploitable.
9480 *mancha <mancha1@zoho.com>*
9482 * Fix for the attack described in the paper "Recovering OpenSSL
9483 ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack"
9484 by Yuval Yarom and Naomi Benger. Details can be obtained from:
9485 <http://eprint.iacr.org/2014/140>
9487 Thanks to Yuval Yarom and Naomi Benger for discovering this
9488 flaw and to Yuval Yarom for supplying a fix ([CVE-2014-0076])
9490 *Yuval Yarom and Naomi Benger*
9492 ### Changes between 1.0.0k and 1.0.0l [6 Jan 2014]
9494 * Keep original DTLS digest and encryption contexts in retransmission
9495 structures so we can use the previous session parameters if they need
9496 to be resent. ([CVE-2013-6450])
9500 * Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which
9501 avoids preferring ECDHE-ECDSA ciphers when the client appears to be
9502 Safari on OS X. Safari on OS X 10.8..10.8.3 advertises support for
9503 several ECDHE-ECDSA ciphers, but fails to negotiate them. The bug
9504 is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing
9505 10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer.
9507 *Rob Stradling, Adam Langley*
9509 ### Changes between 1.0.0j and 1.0.0k [5 Feb 2013]
9511 * Make the decoding of SSLv3, TLS and DTLS CBC records constant time.
9513 This addresses the flaw in CBC record processing discovered by
9514 Nadhem Alfardan and Kenny Paterson. Details of this attack can be found
9515 at: <http://www.isg.rhul.ac.uk/tls/>
9517 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
9518 Security Group at Royal Holloway, University of London
9519 (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and
9520 Emilia Käsper for the initial patch.
9523 *Emilia Käsper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson*
9525 * Return an error when checking OCSP signatures when key is NULL.
9526 This fixes a DoS attack. ([CVE-2013-0166])
9530 * Call OCSP Stapling callback after ciphersuite has been chosen, so
9531 the right response is stapled. Also change SSL_get_certificate()
9532 so it returns the certificate actually sent.
9533 See <http://rt.openssl.org/Ticket/Display.html?id=2836>.
9534 (This is a backport)
9536 *Rob Stradling <rob.stradling@comodo.com>*
9538 * Fix possible deadlock when decoding public keys.
9542 ### Changes between 1.0.0i and 1.0.0j [10 May 2012]
9544 [NB: OpenSSL 1.0.0i and later 1.0.0 patch levels were released after
9547 * Sanity check record length before skipping explicit IV in DTLS
9550 Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic
9551 fuzzing as a service testing platform.
9556 * Initialise tkeylen properly when encrypting CMS messages.
9557 Thanks to Solar Designer of Openwall for reporting this issue.
9561 ### Changes between 1.0.0h and 1.0.0i [19 Apr 2012]
9563 * Check for potentially exploitable overflows in asn1_d2i_read_bio
9564 BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer
9565 in CRYPTO_realloc_clean.
9567 Thanks to Tavis Ormandy, Google Security Team, for discovering this
9568 issue and to Adam Langley <agl@chromium.org> for fixing it.
9571 *Adam Langley (Google), Tavis Ormandy, Google Security Team*
9573 ### Changes between 1.0.0g and 1.0.0h [12 Mar 2012]
9575 * Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness
9576 in CMS and PKCS7 code. When RSA decryption fails use a random key for
9577 content decryption and always return the same error. Note: this attack
9578 needs on average 2^20 messages so it only affects automated senders. The
9579 old behaviour can be re-enabled in the CMS code by setting the
9580 CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where
9581 an MMA defence is not necessary.
9582 Thanks to Ivan Nestlerode <inestlerode@us.ibm.com> for discovering
9583 this issue. ([CVE-2012-0884])
9587 * Fix CVE-2011-4619: make sure we really are receiving a
9588 client hello before rejecting multiple SGC restarts. Thanks to
9589 Ivan Nestlerode <inestlerode@us.ibm.com> for discovering this bug.
9593 ### Changes between 1.0.0f and 1.0.0g [18 Jan 2012]
9595 * Fix for DTLS DoS issue introduced by fix for CVE-2011-4109.
9596 Thanks to Antonio Martin, Enterprise Secure Access Research and
9597 Development, Cisco Systems, Inc. for discovering this bug and
9598 preparing a fix. ([CVE-2012-0050])
9602 ### Changes between 1.0.0e and 1.0.0f [4 Jan 2012]
9604 * Nadhem Alfardan and Kenny Paterson have discovered an extension
9605 of the Vaudenay padding oracle attack on CBC mode encryption
9606 which enables an efficient plaintext recovery attack against
9607 the OpenSSL implementation of DTLS. Their attack exploits timing
9608 differences arising during decryption processing. A research
9609 paper describing this attack can be found at:
9610 <http://www.isg.rhul.ac.uk/~kp/dtls.pdf>
9611 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
9612 Security Group at Royal Holloway, University of London
9613 (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann
9614 <seggelmann@fh-muenster.de> and Michael Tuexen <tuexen@fh-muenster.de>
9615 for preparing the fix. ([CVE-2011-4108])
9617 *Robin Seggelmann, Michael Tuexen*
9619 * Clear bytes used for block padding of SSL 3.0 records.
9622 *Adam Langley (Google)*
9624 * Only allow one SGC handshake restart for SSL/TLS. Thanks to George
9625 Kadianakis <desnacked@gmail.com> for discovering this issue and
9626 Adam Langley for preparing the fix. ([CVE-2011-4619])
9628 *Adam Langley (Google)*
9630 * Check parameters are not NULL in GOST ENGINE. ([CVE-2012-0027])
9632 *Andrey Kulikov <amdeich@gmail.com>*
9634 * Prevent malformed RFC3779 data triggering an assertion failure.
9635 Thanks to Andrew Chi, BBN Technologies, for discovering the flaw
9636 and Rob Austein <sra@hactrn.net> for fixing it. ([CVE-2011-4577])
9638 *Rob Austein <sra@hactrn.net>*
9640 * Improved PRNG seeding for VOS.
9642 *Paul Green <Paul.Green@stratus.com>*
9644 * Fix ssl_ciph.c set-up race.
9646 *Adam Langley (Google)*
9648 * Fix spurious failures in ecdsatest.c.
9650 *Emilia Käsper (Google)*
9652 * Fix the BIO_f_buffer() implementation (which was mixing different
9653 interpretations of the `..._len` fields).
9655 *Adam Langley (Google)*
9657 * Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than
9658 BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent
9659 threads won't reuse the same blinding coefficients.
9661 This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING
9662 lock to call BN_BLINDING_invert_ex, and avoids one use of
9663 BN_BLINDING_update for each BN_BLINDING structure (previously,
9664 the last update always remained unused).
9666 *Emilia Käsper (Google)*
9668 * In ssl3_clear, preserve s3->init_extra along with s3->rbuf.
9670 *Bob Buckholz (Google)*
9672 ### Changes between 1.0.0d and 1.0.0e [6 Sep 2011]
9674 * Fix bug where CRLs with nextUpdate in the past are sometimes accepted
9675 by initialising X509_STORE_CTX properly. ([CVE-2011-3207])
9677 *Kaspar Brand <ossl@velox.ch>*
9679 * Fix SSL memory handling for (EC)DH ciphersuites, in particular
9680 for multi-threaded use of ECDH. ([CVE-2011-3210])
9682 *Adam Langley (Google)*
9684 * Fix x509_name_ex_d2i memory leak on bad inputs.
9688 * Remove hard coded ecdsaWithSHA1 signature tests in ssl code and check
9689 signature public key algorithm by using OID xref utilities instead.
9690 Before this you could only use some ECC ciphersuites with SHA1 only.
9694 * Add protection against ECDSA timing attacks as mentioned in the paper
9695 by Billy Bob Brumley and Nicola Tuveri, see:
9696 <http://eprint.iacr.org/2011/232.pdf>
9698 *Billy Bob Brumley and Nicola Tuveri*
9700 ### Changes between 1.0.0c and 1.0.0d [8 Feb 2011]
9702 * Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014
9704 *Neel Mehta, Adam Langley, Bodo Moeller (Google)*
9706 * Fix bug in string printing code: if *any* escaping is enabled we must
9707 escape the escape character (backslash) or the resulting string is
9712 ### Changes between 1.0.0b and 1.0.0c [2 Dec 2010]
9714 * Disable code workaround for ancient and obsolete Netscape browsers
9715 and servers: an attacker can use it in a ciphersuite downgrade attack.
9716 Thanks to Martin Rex for discovering this bug. CVE-2010-4180
9720 * Fixed J-PAKE implementation error, originally discovered by
9721 Sebastien Martini, further info and confirmation from Stefan
9722 Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252
9726 ### Changes between 1.0.0a and 1.0.0b [16 Nov 2010]
9728 * Fix extension code to avoid race conditions which can result in a buffer
9729 overrun vulnerability: resumed sessions must not be modified as they can
9730 be shared by multiple threads. CVE-2010-3864
9734 * Fix WIN32 build system to correctly link an ENGINE directory into
9739 ### Changes between 1.0.0 and 1.0.0a [01 Jun 2010]
9741 * Check return value of int_rsa_verify in pkey_rsa_verifyrecover
9744 *Steve Henson, Peter-Michael Hager <hager@dortmund.net>*
9746 ### Changes between 0.9.8n and 1.0.0 [29 Mar 2010]
9748 * Add "missing" function EVP_CIPHER_CTX_copy(). This copies a cipher
9749 context. The operation can be customised via the ctrl mechanism in
9750 case ENGINEs want to include additional functionality.
9754 * Tolerate yet another broken PKCS#8 key format: private key value negative.
9758 * Add new -subject_hash_old and -issuer_hash_old options to x509 utility to
9759 output hashes compatible with older versions of OpenSSL.
9761 *Willy Weisz <weisz@vcpc.univie.ac.at>*
9763 * Fix compression algorithm handling: if resuming a session use the
9764 compression algorithm of the resumed session instead of determining
9765 it from client hello again. Don't allow server to change algorithm.
9769 * Add load_crls() function to commands tidying load_certs() too. Add option
9770 to verify utility to allow additional CRLs to be included.
9774 * Update OCSP request code to permit adding custom headers to the request:
9775 some responders need this.
9779 * The function EVP_PKEY_sign() returns <=0 on error: check return code
9782 *Julia Lawall <julia@diku.dk>*
9784 * Update verify callback code in `apps/s_cb.c` and `apps/verify.c`, it
9785 needlessly dereferenced structures, used obsolete functions and
9786 didn't handle all updated verify codes correctly.
9790 * Disable MD2 in the default configuration.
9794 * In BIO_pop() and BIO_push() use the ctrl argument (which was NULL) to
9795 indicate the initial BIO being pushed or popped. This makes it possible
9796 to determine whether the BIO is the one explicitly called or as a result
9797 of the ctrl being passed down the chain. Fix BIO_pop() and SSL BIOs so
9798 it handles reference counts correctly and doesn't zero out the I/O bio
9799 when it is not being explicitly popped. WARNING: applications which
9800 included workarounds for the old buggy behaviour will need to be modified
9801 or they could free up already freed BIOs.
9805 * Extend the uni2asc/asc2uni => OPENSSL_uni2asc/OPENSSL_asc2uni
9806 renaming to all platforms (within the 0.9.8 branch, this was
9807 done conditionally on Netware platforms to avoid a name clash).
9809 *Guenter <lists@gknw.net>*
9811 * Add ECDHE and PSK support to DTLS.
9813 *Michael Tuexen <tuexen@fh-muenster.de>*
9815 * Add CHECKED_STACK_OF macro to safestack.h, otherwise safestack can't
9820 * Add "missing" function EVP_MD_flags() (without this the only way to
9821 retrieve a digest flags is by accessing the structure directly. Update
9822 `EVP_MD_do_all*()` and `EVP_CIPHER_do_all*()` to include the name a digest
9823 or cipher is registered as in the "from" argument. Print out all
9824 registered digests in the dgst usage message instead of manually
9825 attempting to work them out.
9829 * If no SSLv2 ciphers are used don't use an SSLv2 compatible client hello:
9830 this allows the use of compression and extensions. Change default cipher
9831 string to remove SSLv2 ciphersuites. This effectively avoids ancient SSLv2
9832 by default unless an application cipher string requests it.
9836 * Alter match criteria in PKCS12_parse(). It used to try to use local
9837 key ids to find matching certificates and keys but some PKCS#12 files
9838 don't follow the (somewhat unwritten) rules and this strategy fails.
9839 Now just gather all certificates together and the first private key
9840 then look for the first certificate that matches the key.
9844 * Support use of registered digest and cipher names for dgst and cipher
9845 commands instead of having to add each one as a special case. So now
9852 openssl dgst -sha256 foo
9854 and this works for ENGINE based algorithms too.
9858 * Update Gost ENGINE to support parameter files.
9860 *Victor B. Wagner <vitus@cryptocom.ru>*
9862 * Support GeneralizedTime in ca utility.
9864 *Oliver Martin <oliver@volatilevoid.net>, Steve Henson*
9866 * Enhance the hash format used for certificate directory links. The new
9867 form uses the canonical encoding (meaning equivalent names will work
9868 even if they aren't identical) and uses SHA1 instead of MD5. This form
9869 is incompatible with the older format and as a result c_rehash should
9870 be used to rebuild symbolic links.
9874 * Make PKCS#8 the default write format for private keys, replacing the
9875 traditional format. This form is standardised, more secure and doesn't
9876 include an implicit MD5 dependency.
9880 * Add a $gcc_devteam_warn option to Configure. The idea is that any code
9881 committed to OpenSSL should pass this lot as a minimum.
9885 * Add session ticket override functionality for use by EAP-FAST.
9887 *Jouni Malinen <j@w1.fi>*
9889 * Modify HMAC functions to return a value. Since these can be implemented
9890 in an ENGINE errors can occur.
9894 * Type-checked OBJ_bsearch_ex.
9898 * Type-checked OBJ_bsearch. Also some constification necessitated
9899 by type-checking. Still to come: TXT_DB, bsearch(?),
9900 OBJ_bsearch_ex, qsort, CRYPTO_EX_DATA, ASN1_VALUE, ASN1_STRING,
9905 * New function OPENSSL_gmtime_adj() to add a specific number of days and
9906 seconds to a tm structure directly, instead of going through OS
9907 specific date routines. This avoids any issues with OS routines such
9908 as the year 2038 bug. New `*_adj()` functions for ASN1 time structures
9909 and X509_time_adj_ex() to cover the extended range. The existing
9910 X509_time_adj() is still usable and will no longer have any date issues.
9914 * Delta CRL support. New use deltas option which will attempt to locate
9915 and search any appropriate delta CRLs available.
9917 This work was sponsored by Google.
9921 * Support for CRLs partitioned by reason code. Reorganise CRL processing
9922 code and add additional score elements. Validate alternate CRL paths
9923 as part of the CRL checking and indicate a new error "CRL path validation
9924 error" in this case. Applications wanting additional details can use
9925 the verify callback and check the new "parent" field. If this is not
9926 NULL CRL path validation is taking place. Existing applications won't
9927 see this because it requires extended CRL support which is off by
9930 This work was sponsored by Google.
9934 * Support for freshest CRL extension.
9936 This work was sponsored by Google.
9940 * Initial indirect CRL support. Currently only supported in the CRLs
9941 passed directly and not via lookup. Process certificate issuer
9942 CRL entry extension and lookup CRL entries by bother issuer name
9943 and serial number. Check and process CRL issuer entry in IDP extension.
9945 This work was sponsored by Google.
9949 * Add support for distinct certificate and CRL paths. The CRL issuer
9950 certificate is validated separately in this case. Only enabled if
9951 an extended CRL support flag is set: this flag will enable additional
9952 CRL functionality in future.
9954 This work was sponsored by Google.
9958 * Add support for policy mappings extension.
9960 This work was sponsored by Google.
9964 * Fixes to pathlength constraint, self issued certificate handling,
9965 policy processing to align with RFC3280 and PKITS tests.
9967 This work was sponsored by Google.
9971 * Support for name constraints certificate extension. DN, email, DNS
9972 and URI types are currently supported.
9974 This work was sponsored by Google.
9978 * To cater for systems that provide a pointer-based thread ID rather
9979 than numeric, deprecate the current numeric thread ID mechanism and
9980 replace it with a structure and associated callback type. This
9981 mechanism allows a numeric "hash" to be extracted from a thread ID in
9982 either case, and on platforms where pointers are larger than 'long',
9983 mixing is done to help ensure the numeric 'hash' is usable even if it
9984 can't be guaranteed unique. The default mechanism is to use "&errno"
9985 as a pointer-based thread ID to distinguish between threads.
9987 Applications that want to provide their own thread IDs should now use
9988 CRYPTO_THREADID_set_callback() to register a callback that will call
9989 either CRYPTO_THREADID_set_numeric() or CRYPTO_THREADID_set_pointer().
9991 Note that ERR_remove_state() is now deprecated, because it is tied
9992 to the assumption that thread IDs are numeric. ERR_remove_state(0)
9993 to free the current thread's error state should be replaced by
9994 ERR_remove_thread_state(NULL).
9996 (This new approach replaces the functions CRYPTO_set_idptr_callback(),
9997 CRYPTO_get_idptr_callback(), and CRYPTO_thread_idptr() that existed in
9998 OpenSSL 0.9.9-dev between June 2006 and August 2008. Also, if an
9999 application was previously providing a numeric thread callback that
10000 was inappropriate for distinguishing threads, then uniqueness might
10001 have been obtained with &errno that happened immediately in the
10002 intermediate development versions of OpenSSL; this is no longer the
10003 case, the numeric thread callback will now override the automatic use
10006 *Geoff Thorpe, with help from Bodo Moeller*
10008 * Initial support for different CRL issuing certificates. This covers a
10009 simple case where the self issued certificates in the chain exist and
10010 the real CRL issuer is higher in the existing chain.
10012 This work was sponsored by Google.
10016 * Removed effectively defunct crypto/store from the build.
10020 * Revamp of STACK to provide stronger type-checking. Still to come:
10021 TXT_DB, bsearch(?), OBJ_bsearch, qsort, CRYPTO_EX_DATA, ASN1_VALUE,
10022 ASN1_STRING, CONF_VALUE.
10026 * Add a new SSL_MODE_RELEASE_BUFFERS mode flag to release unused buffer
10027 RAM on SSL connections. This option can save about 34k per idle SSL.
10031 * Revamp of LHASH to provide stronger type-checking. Still to come:
10032 STACK, TXT_DB, bsearch, qsort.
10036 * Initial support for Cryptographic Message Syntax (aka CMS) based
10037 on RFC3850, RFC3851 and RFC3852. New cms directory and cms utility,
10038 support for data, signedData, compressedData, digestedData and
10039 encryptedData, envelopedData types included. Scripts to check against
10040 RFC4134 examples draft and interop and consistency checks of many
10041 content types and variants.
10045 * Add options to enc utility to support use of zlib compression BIO.
10049 * Extend mk1mf to support importing of options and assembly language
10050 files from Configure script, currently only included in VC-WIN32.
10051 The assembly language rules can now optionally generate the source
10052 files from the associated perl scripts.
10056 * Implement remaining functionality needed to support GOST ciphersuites.
10057 Interop testing has been performed using CryptoPro implementations.
10059 *Victor B. Wagner <vitus@cryptocom.ru>*
10061 * s390x assembler pack.
10065 * ARMv4 assembler pack. ARMv4 refers to v4 and later ISA, not CPU
10070 * Implement Opaque PRF Input TLS extension as specified in
10071 draft-rescorla-tls-opaque-prf-input-00.txt. Since this is not an
10072 official specification yet and no extension type assignment by
10073 IANA exists, this extension (for now) will have to be explicitly
10074 enabled when building OpenSSL by providing the extension number
10075 to use. For example, specify an option
10077 -DTLSEXT_TYPE_opaque_prf_input=0x9527
10079 to the "config" or "Configure" script to enable the extension,
10080 assuming extension number 0x9527 (which is a completely arbitrary
10081 and unofficial assignment based on the MD5 hash of the Internet
10082 Draft). Note that by doing so, you potentially lose
10083 interoperability with other TLS implementations since these might
10084 be using the same extension number for other purposes.
10086 SSL_set_tlsext_opaque_prf_input(ssl, src, len) is used to set the
10087 opaque PRF input value to use in the handshake. This will create
10088 an internal copy of the length-'len' string at 'src', and will
10089 return non-zero for success.
10091 To get more control and flexibility, provide a callback function
10094 SSL_CTX_set_tlsext_opaque_prf_input_callback(ctx, cb)
10095 SSL_CTX_set_tlsext_opaque_prf_input_callback_arg(ctx, arg)
10099 int (*cb)(SSL *, void *peerinput, size_t len, void *arg);
10102 Callback function 'cb' will be called in handshakes, and is
10103 expected to use SSL_set_tlsext_opaque_prf_input() as appropriate.
10104 Argument 'arg' is for application purposes (the value as given to
10105 SSL_CTX_set_tlsext_opaque_prf_input_callback_arg() will directly
10106 be provided to the callback function). The callback function
10107 has to return non-zero to report success: usually 1 to use opaque
10108 PRF input just if possible, or 2 to enforce use of the opaque PRF
10109 input. In the latter case, the library will abort the handshake
10110 if opaque PRF input is not successfully negotiated.
10112 Arguments 'peerinput' and 'len' given to the callback function
10113 will always be NULL and 0 in the case of a client. A server will
10114 see the client's opaque PRF input through these variables if
10115 available (NULL and 0 otherwise). Note that if the server
10116 provides an opaque PRF input, the length must be the same as the
10117 length of the client's opaque PRF input.
10119 Note that the callback function will only be called when creating
10120 a new session (session resumption can resume whatever was
10121 previously negotiated), and will not be called in SSL 2.0
10122 handshakes; thus, SSL_CTX_set_options(ctx, SSL_OP_NO_SSLv2) or
10123 SSL_set_options(ssl, SSL_OP_NO_SSLv2) is especially recommended
10124 for applications that need to enforce opaque PRF input.
10128 * Update ssl code to support digests other than SHA1+MD5 for handshake
10131 *Victor B. Wagner <vitus@cryptocom.ru>*
10133 * Add RFC4507 support to OpenSSL. This includes the corrections in
10134 RFC4507bis. The encrypted ticket format is an encrypted encoded
10135 SSL_SESSION structure, that way new session features are automatically
10138 If a client application caches session in an SSL_SESSION structure
10139 support is transparent because tickets are now stored in the encoded
10142 The SSL_CTX structure automatically generates keys for ticket
10143 protection in servers so again support should be possible
10144 with no application modification.
10146 If a client or server wishes to disable RFC4507 support then the option
10147 SSL_OP_NO_TICKET can be set.
10149 Add a TLS extension debugging callback to allow the contents of any client
10150 or server extensions to be examined.
10152 This work was sponsored by Google.
10156 * Final changes to avoid use of pointer pointer casts in OpenSSL.
10157 OpenSSL should now compile cleanly on gcc 4.2
10159 *Peter Hartley <pdh@utter.chaos.org.uk>, Steve Henson*
10161 * Update SSL library to use new EVP_PKEY MAC API. Include generic MAC
10162 support including streaming MAC support: this is required for GOST
10163 ciphersuite support.
10165 *Victor B. Wagner <vitus@cryptocom.ru>, Steve Henson*
10167 * Add option -stream to use PKCS#7 streaming in smime utility. New
10168 function i2d_PKCS7_bio_stream() and PEM_write_PKCS7_bio_stream()
10169 to output in BER and PEM format.
10173 * Experimental support for use of HMAC via EVP_PKEY interface. This
10174 allows HMAC to be handled via the `EVP_DigestSign*()` interface. The
10175 EVP_PKEY "key" in this case is the HMAC key, potentially allowing
10176 ENGINE support for HMAC keys which are unextractable. New -mac and
10177 -macopt options to dgst utility.
10181 * New option -sigopt to dgst utility. Update dgst to use
10182 `EVP_Digest{Sign,Verify}*`. These two changes make it possible to use
10183 alternative signing parameters such as X9.31 or PSS in the dgst
10188 * Change ssl_cipher_apply_rule(), the internal function that does
10189 the work each time a ciphersuite string requests enabling
10190 ("foo+bar"), moving ("+foo+bar"), disabling ("-foo+bar", or
10191 removing ("!foo+bar") a class of ciphersuites: Now it maintains
10192 the order of disabled ciphersuites such that those ciphersuites
10193 that most recently went from enabled to disabled not only stay
10194 in order with respect to each other, but also have higher priority
10195 than other disabled ciphersuites the next time ciphersuites are
10198 This means that you can now say, e.g., "PSK:-PSK:HIGH" to enable
10199 the same ciphersuites as with "HIGH" alone, but in a specific
10200 order where the PSK ciphersuites come first (since they are the
10201 most recently disabled ciphersuites when "HIGH" is parsed).
10203 Also, change ssl_create_cipher_list() (using this new
10204 functionality) such that between otherwise identical
10205 ciphersuites, ephemeral ECDH is preferred over ephemeral DH in
10210 * Change ssl_create_cipher_list() so that it automatically
10211 arranges the ciphersuites in reasonable order before starting
10212 to process the rule string. Thus, the definition for "DEFAULT"
10213 (SSL_DEFAULT_CIPHER_LIST) now is just "ALL:!aNULL:!eNULL", but
10214 remains equivalent to `"AES:ALL:!aNULL:!eNULL:+aECDH:+kRSA:+RC4:@STRENGTH"`.
10215 This makes it much easier to arrive at a reasonable default order
10216 in applications for which anonymous ciphers are OK (meaning
10217 that you can't actually use DEFAULT).
10219 *Bodo Moeller; suggested by Victor Duchovni*
10221 * Split the SSL/TLS algorithm mask (as used for ciphersuite string
10222 processing) into multiple integers instead of setting
10223 "SSL_MKEY_MASK" bits, "SSL_AUTH_MASK" bits, "SSL_ENC_MASK",
10224 "SSL_MAC_MASK", and "SSL_SSL_MASK" bits all in a single integer.
10225 (These masks as well as the individual bit definitions are hidden
10226 away into the non-exported interface ssl/ssl_locl.h, so this
10227 change to the definition of the SSL_CIPHER structure shouldn't
10228 affect applications.) This give us more bits for each of these
10229 categories, so there is no longer a need to coagulate AES128 and
10230 AES256 into a single algorithm bit, and to coagulate Camellia128
10231 and Camellia256 into a single algorithm bit, which has led to all
10234 Thus, among other things, the kludge introduced in 0.9.7m and
10235 0.9.8e for masking out AES256 independently of AES128 or masking
10236 out Camellia256 independently of AES256 is not needed here in 0.9.9.
10238 With the change, we also introduce new ciphersuite aliases that
10239 so far were missing: "AES128", "AES256", "CAMELLIA128", and
10244 * Add support for dsa-with-SHA224 and dsa-with-SHA256.
10245 Use the leftmost N bytes of the signature input if the input is
10246 larger than the prime q (with N being the size in bytes of q).
10250 * Very *very* experimental PKCS#7 streaming encoder support. Nothing uses
10251 it yet and it is largely untested.
10255 * Add support for the ecdsa-with-SHA224/256/384/512 signature types.
10259 * Initial incomplete changes to avoid need for function casts in OpenSSL
10260 some compilers (gcc 4.2 and later) reject their use. Safestack is
10261 reimplemented. Update ASN1 to avoid use of legacy functions.
10265 * Win32/64 targets are linked with Winsock2.
10269 * Add an X509_CRL_METHOD structure to allow CRL processing to be redirected
10270 to external functions. This can be used to increase CRL handling
10271 efficiency especially when CRLs are very large by (for example) storing
10272 the CRL revoked certificates in a database.
10276 * Overhaul of by_dir code. Add support for dynamic loading of CRLs so
10277 new CRLs added to a directory can be used. New command line option
10278 -verify_return_error to s_client and s_server. This causes real errors
10279 to be returned by the verify callback instead of carrying on no matter
10280 what. This reflects the way a "real world" verify callback would behave.
10284 * GOST engine, supporting several GOST algorithms and public key formats.
10285 Kindly donated by Cryptocom.
10289 * Partial support for Issuing Distribution Point CRL extension. CRLs
10290 partitioned by DP are handled but no indirect CRL or reason partitioning
10291 (yet). Complete overhaul of CRL handling: now the most suitable CRL is
10292 selected via a scoring technique which handles IDP and AKID in CRLs.
10296 * New X509_STORE_CTX callbacks lookup_crls() and lookup_certs() which
10297 will ultimately be used for all verify operations: this will remove the
10298 X509_STORE dependency on certificate verification and allow alternative
10299 lookup methods. X509_STORE based implementations of these two callbacks.
10303 * Allow multiple CRLs to exist in an X509_STORE with matching issuer names.
10304 Modify get_crl() to find a valid (unexpired) CRL if possible.
10308 * New function X509_CRL_match() to check if two CRLs are identical. Normally
10309 this would be called X509_CRL_cmp() but that name is already used by
10310 a function that just compares CRL issuer names. Cache several CRL
10311 extensions in X509_CRL structure and cache CRLDP in X509.
10315 * Store a "canonical" representation of X509_NAME structure (ASN1 Name)
10316 this maps equivalent X509_NAME structures into a consistent structure.
10317 Name comparison can then be performed rapidly using memcmp().
10321 * Non-blocking OCSP request processing. Add -timeout option to ocsp
10326 * Allow digests to supply their own micalg string for S/MIME type using
10327 the ctrl EVP_MD_CTRL_MICALG.
10331 * During PKCS7 signing pass the PKCS7 SignerInfo structure to the
10332 EVP_PKEY_METHOD before and after signing via the EVP_PKEY_CTRL_PKCS7_SIGN
10333 ctrl. It can then customise the structure before and/or after signing
10338 * New function OBJ_add_sigid() to allow application defined signature OIDs
10339 to be added to OpenSSLs internal tables. New function OBJ_sigid_free()
10340 to free up any added signature OIDs.
10344 * New functions EVP_CIPHER_do_all(), EVP_CIPHER_do_all_sorted(),
10345 EVP_MD_do_all() and EVP_MD_do_all_sorted() to enumerate internal
10346 digest and cipher tables. New options added to openssl utility:
10347 list-message-digest-algorithms and list-cipher-algorithms.
10351 * Change the array representation of binary polynomials: the list
10352 of degrees of non-zero coefficients is now terminated with -1.
10353 Previously it was terminated with 0, which was also part of the
10354 value; thus, the array representation was not applicable to
10355 polynomials where t^0 has coefficient zero. This change makes
10356 the array representation useful in a more general context.
10360 * Various modifications and fixes to SSL/TLS cipher string
10361 handling. For ECC, the code now distinguishes between fixed ECDH
10362 with RSA certificates on the one hand and with ECDSA certificates
10363 on the other hand, since these are separate ciphersuites. The
10364 unused code for Fortezza ciphersuites has been removed.
10366 For consistency with EDH, ephemeral ECDH is now called "EECDH"
10367 (not "ECDHE"). For consistency with the code for DH
10368 certificates, use of ECDH certificates is now considered ECDH
10369 authentication, not RSA or ECDSA authentication (the latter is
10370 merely the CA's signing algorithm and not actively used in the
10373 The temporary ciphersuite alias "ECCdraft" is no longer
10374 available, and ECC ciphersuites are no longer excluded from "ALL"
10375 and "DEFAULT". The following aliases now exist for RFC 4492
10376 ciphersuites, most of these by analogy with the DH case:
10378 kECDHr - ECDH cert, signed with RSA
10379 kECDHe - ECDH cert, signed with ECDSA
10380 kECDH - ECDH cert (signed with either RSA or ECDSA)
10381 kEECDH - ephemeral ECDH
10382 ECDH - ECDH cert or ephemeral ECDH
10385 aECDSA - ECDSA cert
10388 AECDH - anonymous ECDH
10389 EECDH - non-anonymous ephemeral ECDH (equivalent to "kEECDH:-AECDH")
10393 * Add additional S/MIME capabilities for AES and GOST ciphers if supported.
10394 Use correct micalg parameters depending on digest(s) in signed message.
10398 * Add engine support for EVP_PKEY_ASN1_METHOD. Add functions to process
10399 an ENGINE asn1 method. Support ENGINE lookups in the ASN1 code.
10403 * Initial engine support for EVP_PKEY_METHOD. New functions to permit
10404 an engine to register a method. Add ENGINE lookups for methods and
10405 functional reference processing.
10409 * New functions `EVP_Digest{Sign,Verify)*`. These are enhanced versions of
10410 `EVP_{Sign,Verify}*` which allow an application to customise the signature
10415 * New -resign option to smime utility. This adds one or more signers
10416 to an existing PKCS#7 signedData structure. Also -md option to use an
10417 alternative message digest algorithm for signing.
10421 * Tidy up PKCS#7 routines and add new functions to make it easier to
10422 create PKCS7 structures containing multiple signers. Update smime
10423 application to support multiple signers.
10427 * New -macalg option to pkcs12 utility to allow setting of an alternative
10432 * Initial support for PKCS#5 v2.0 PRFs other than default SHA1 HMAC.
10433 Reorganize PBE internals to lookup from a static table using NIDs,
10434 add support for HMAC PBE OID translation. Add a EVP_CIPHER ctrl:
10435 EVP_CTRL_PBE_PRF_NID this allows a cipher to specify an alternative
10436 PRF which will be automatically used with PBES2.
10440 * Replace the algorithm specific calls to generate keys in "req" with the
10445 * Update PKCS#7 enveloped data routines to use new API. This is now
10446 supported by any public key method supporting the encrypt operation. A
10447 ctrl is added to allow the public key algorithm to examine or modify
10448 the PKCS#7 RecipientInfo structure if it needs to: for RSA this is
10453 * Add a ctrl to asn1 method to allow a public key algorithm to express
10454 a default digest type to use. In most cases this will be SHA1 but some
10455 algorithms (such as GOST) need to specify an alternative digest. The
10456 return value indicates how strong the preference is 1 means optional and
10457 2 is mandatory (that is it is the only supported type). Modify
10458 ASN1_item_sign() to accept a NULL digest argument to indicate it should
10459 use the default md. Update openssl utilities to use the default digest
10460 type for signing if it is not explicitly indicated.
10464 * Use OID cross reference table in ASN1_sign() and ASN1_verify(). New
10465 EVP_MD flag EVP_MD_FLAG_PKEY_METHOD_SIGNATURE. This uses the relevant
10466 signing method from the key type. This effectively removes the link
10467 between digests and public key types.
10471 * Add an OID cross reference table and utility functions. Its purpose is to
10472 translate between signature OIDs such as SHA1WithrsaEncryption and SHA1,
10473 rsaEncryption. This will allow some of the algorithm specific hackery
10474 needed to use the correct OID to be removed.
10478 * Remove algorithm specific dependencies when setting PKCS7_SIGNER_INFO
10479 structures for PKCS7_sign(). They are now set up by the relevant public
10484 * Add provisional EC pkey method with support for ECDSA and ECDH.
10488 * Add support for key derivation (agreement) in the API, DH method and
10493 * Add DSA pkey method and DH pkey methods, extend DH ASN1 method to support
10494 public and private key formats. As a side effect these add additional
10495 command line functionality not previously available: DSA signatures can be
10496 generated and verified using pkeyutl and DH key support and generation in
10503 *Oliver Tappe <zooey@hirschkaefer.de>*
10505 * New make target "install_html_docs" installs HTML renditions of the
10508 *Oliver Tappe <zooey@hirschkaefer.de>*
10510 * New utility "genpkey" this is analogous to "genrsa" etc except it can
10511 generate keys for any algorithm. Extend and update EVP_PKEY_METHOD to
10512 support key and parameter generation and add initial key generation
10513 functionality for RSA.
10517 * Add functions for main EVP_PKEY_method operations. The undocumented
10518 functions `EVP_PKEY_{encrypt,decrypt}` have been renamed to
10519 `EVP_PKEY_{encrypt,decrypt}_old`.
10523 * Initial definitions for EVP_PKEY_METHOD. This will be a high level public
10524 key API, doesn't do much yet.
10528 * New function EVP_PKEY_asn1_get0_info() to retrieve information about
10529 public key algorithms. New option to openssl utility:
10530 "list-public-key-algorithms" to print out info.
10534 * Implement the Supported Elliptic Curves Extension for
10535 ECC ciphersuites from draft-ietf-tls-ecc-12.txt.
10539 * Don't free up OIDs in OBJ_cleanup() if they are in use by EVP_MD or
10540 EVP_CIPHER structures to avoid later problems in EVP_cleanup().
10544 * New utilities pkey and pkeyparam. These are similar to algorithm specific
10545 utilities such as rsa, dsa, dsaparam etc except they process any key
10550 * Transfer public key printing routines to EVP_PKEY_ASN1_METHOD. New
10551 functions EVP_PKEY_print_public(), EVP_PKEY_print_private(),
10552 EVP_PKEY_print_param() to print public key data from an EVP_PKEY
10557 * Initial support for pluggable public key ASN1.
10558 De-spaghettify the public key ASN1 handling. Move public and private
10559 key ASN1 handling to a new EVP_PKEY_ASN1_METHOD structure. Relocate
10560 algorithm specific handling to a single module within the relevant
10561 algorithm directory. Add functions to allow (near) opaque processing
10562 of public and private key structures.
10566 * Implement the Supported Point Formats Extension for
10567 ECC ciphersuites from draft-ietf-tls-ecc-12.txt.
10571 * Add initial support for RFC 4279 PSK TLS ciphersuites. Add members
10572 for the psk identity [hint] and the psk callback functions to the
10573 SSL_SESSION, SSL and SSL_CTX structure.
10576 PSK-RC4-SHA, PSK-3DES-EDE-CBC-SHA, PSK-AES128-CBC-SHA,
10580 SSL_CTX_use_psk_identity_hint
10581 SSL_get_psk_identity_hint
10582 SSL_get_psk_identity
10583 SSL_use_psk_identity_hint
10585 *Mika Kousa and Pasi Eronen of Nokia Corporation*
10587 * Add RFC 3161 compliant time stamp request creation, response generation
10588 and response verification functionality.
10590 *Zoltán Glózik <zglozik@opentsa.org>, The OpenTSA Project*
10592 * Add initial support for TLS extensions, specifically for the server_name
10593 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now
10594 have new members for a hostname. The SSL data structure has an
10595 additional member `SSL_CTX *initial_ctx` so that new sessions can be
10596 stored in that context to allow for session resumption, even after the
10597 SSL has been switched to a new SSL_CTX in reaction to a client's
10598 server_name extension.
10600 New functions (subject to change):
10602 SSL_get_servername()
10603 SSL_get_servername_type()
10606 New CTRL codes and macros (subject to change):
10608 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB
10609 - SSL_CTX_set_tlsext_servername_callback()
10610 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG
10611 - SSL_CTX_set_tlsext_servername_arg()
10612 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name()
10614 openssl s_client has a new '-servername ...' option.
10616 openssl s_server has new options '-servername_host ...', '-cert2 ...',
10617 '-key2 ...', '-servername_fatal' (subject to change). This allows
10618 testing the HostName extension for a specific single hostname ('-cert'
10619 and '-key' remain fallbacks for handshakes without HostName
10620 negotiation). If the unrecognized_name alert has to be sent, this by
10621 default is a warning; it becomes fatal with the '-servername_fatal'
10624 *Peter Sylvester, Remy Allais, Christophe Renou*
10626 * Whirlpool hash implementation is added.
10630 * BIGNUM code on 64-bit SPARCv9 targets is switched from bn(64,64) to
10631 bn(64,32). Because of instruction set limitations it doesn't have
10632 any negative impact on performance. This was done mostly in order
10633 to make it possible to share assembler modules, such as bn_mul_mont
10634 implementations, between 32- and 64-bit builds without hassle.
10638 * Move code previously exiled into file crypto/ec/ec2_smpt.c
10639 to ec2_smpl.c, and no longer require the OPENSSL_EC_BIN_PT_COMP
10644 * New candidate for BIGNUM assembler implementation, bn_mul_mont,
10645 dedicated Montgomery multiplication procedure, is introduced.
10646 BN_MONT_CTX is modified to allow bn_mul_mont to reach for higher
10647 "64-bit" performance on certain 32-bit targets.
10651 * New option SSL_OP_NO_COMP to disable use of compression selectively
10652 in SSL structures. New SSL ctrl to set maximum send fragment size.
10653 Save memory by setting the I/O buffer sizes dynamically instead of
10654 using the maximum available value.
10658 * New option -V for 'openssl ciphers'. This prints the ciphersuite code
10659 in addition to the text details.
10663 * Very, very preliminary EXPERIMENTAL support for printing of general
10664 ASN1 structures. This currently produces rather ugly output and doesn't
10665 handle several customised structures at all.
10669 * Integrated support for PVK file format and some related formats such
10670 as MS PUBLICKEYBLOB and PRIVATEKEYBLOB. Command line switches to support
10671 these in the 'rsa' and 'dsa' utilities.
10675 * Support for PKCS#1 RSAPublicKey format on rsa utility command line.
10679 * Remove the ancient ASN1_METHOD code. This was only ever used in one
10680 place for the (very old) "NETSCAPE" format certificates which are now
10681 handled using new ASN1 code equivalents.
10685 * Let the TLSv1_method() etc. functions return a 'const' SSL_METHOD
10686 pointer and make the SSL_METHOD parameter in SSL_CTX_new,
10687 SSL_CTX_set_ssl_version and SSL_set_ssl_method 'const'.
10691 * Modify CRL distribution points extension code to print out previously
10692 unsupported fields. Enhance extension setting code to allow setting of
10697 * Add print and set support for Issuing Distribution Point CRL extension.
10701 * Change 'Configure' script to enable Camellia by default.
10708 ### Changes between 0.9.8m and 0.9.8n [24 Mar 2010]
10710 * When rejecting SSL/TLS records due to an incorrect version number, never
10711 update s->server with a new major version number. As of
10712 - OpenSSL 0.9.8m if 'short' is a 16-bit type,
10713 - OpenSSL 0.9.8f if 'short' is longer than 16 bits,
10714 the previous behavior could result in a read attempt at NULL when
10715 receiving specific incorrect SSL/TLS records once record payload
10716 protection is active. ([CVE-2010-0740])
10718 *Bodo Moeller, Adam Langley <agl@chromium.org>*
10720 * Fix for CVE-2010-0433 where some kerberos enabled versions of OpenSSL
10721 could be crashed if the relevant tables were not present (e.g. chrooted).
10723 *Tomas Hoger <thoger@redhat.com>*
10725 ### Changes between 0.9.8l and 0.9.8m [25 Feb 2010]
10727 * Always check bn_wexpand() return values for failure. ([CVE-2009-3245])
10729 *Martin Olsson, Neel Mehta*
10731 * Fix X509_STORE locking: Every 'objs' access requires a lock (to
10732 accommodate for stack sorting, always a write lock!).
10736 * On some versions of WIN32 Heap32Next is very slow. This can cause
10737 excessive delays in the RAND_poll(): over a minute. As a workaround
10738 include a time check in the inner Heap32Next loop too.
10742 * The code that handled flushing of data in SSL/TLS originally used the
10743 BIO_CTRL_INFO ctrl to see if any data was pending first. This caused
10744 the problem outlined in PR#1949. The fix suggested there however can
10745 trigger problems with buggy BIO_CTRL_WPENDING (e.g. some versions
10746 of Apache). So instead simplify the code to flush unconditionally.
10747 This should be fine since flushing with no data to flush is a no op.
10751 * Handle TLS versions 2.0 and later properly and correctly use the
10752 highest version of TLS/SSL supported. Although TLS >= 2.0 is some way
10753 off ancient servers have a habit of sticking around for a while...
10757 * Modify compression code so it frees up structures without using the
10758 ex_data callbacks. This works around a problem where some applications
10759 call CRYPTO_cleanup_all_ex_data() before application exit (e.g. when
10760 restarting) then use compression (e.g. SSL with compression) later.
10761 This results in significant per-connection memory leaks and
10762 has caused some security issues including CVE-2008-1678 and
10767 * Constify crypto/cast (i.e., <openssl/cast.h>): a CAST_KEY doesn't
10768 change when encrypting or decrypting.
10772 * Add option SSL_OP_LEGACY_SERVER_CONNECT which will allow clients to
10773 connect and renegotiate with servers which do not support RI.
10774 Until RI is more widely deployed this option is enabled by default.
10778 * Add "missing" ssl ctrls to clear options and mode.
10782 * If client attempts to renegotiate and doesn't support RI respond with
10783 a no_renegotiation alert as required by RFC5746. Some renegotiating
10784 TLS clients will continue a connection gracefully when they receive
10785 the alert. Unfortunately OpenSSL mishandled this alert and would hang
10786 waiting for a server hello which it will never receive. Now we treat a
10787 received no_renegotiation alert as a fatal error. This is because
10788 applications requesting a renegotiation might well expect it to succeed
10789 and would have no code in place to handle the server denying it so the
10790 only safe thing to do is to terminate the connection.
10794 * Add ctrl macro SSL_get_secure_renegotiation_support() which returns 1 if
10795 peer supports secure renegotiation and 0 otherwise. Print out peer
10796 renegotiation support in s_client/s_server.
10800 * Replace the highly broken and deprecated SPKAC certification method with
10801 the updated NID creation version. This should correctly handle UTF8.
10805 * Implement RFC5746. Re-enable renegotiation but require the extension
10806 as needed. Unfortunately, SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION
10807 turns out to be a bad idea. It has been replaced by
10808 SSL_OP_ALLOW_UNSAFE_LEGACY_RENEGOTIATION which can be set with
10809 SSL_CTX_set_options(). This is really not recommended unless you
10810 know what you are doing.
10812 *Eric Rescorla <ekr@networkresonance.com>, Ben Laurie, Steve Henson*
10814 * Fixes to stateless session resumption handling. Use initial_ctx when
10815 issuing and attempting to decrypt tickets in case it has changed during
10816 servername handling. Use a non-zero length session ID when attempting
10817 stateless session resumption: this makes it possible to determine if
10818 a resumption has occurred immediately after receiving server hello
10819 (several places in OpenSSL subtly assume this) instead of later in
10824 * The functions ENGINE_ctrl(), OPENSSL_isservice(),
10825 CMS_get1_RecipientRequest() and RAND_bytes() can return <=0 on error
10826 fixes for a few places where the return code is not checked
10829 *Julia Lawall <julia@diku.dk>*
10831 * Add --strict-warnings option to Configure script to include devteam
10832 warnings in other configurations.
10836 * Add support for --libdir option and LIBDIR variable in makefiles. This
10837 makes it possible to install openssl libraries in locations which
10838 have names other than "lib", for example "/usr/lib64" which some
10841 *Steve Henson, based on patch from Jeremy Utley*
10843 * Don't allow the use of leading 0x80 in OIDs. This is a violation of
10844 X690 8.9.12 and can produce some misleading textual output of OIDs.
10846 *Steve Henson, reported by Dan Kaminsky*
10848 * Delete MD2 from algorithm tables. This follows the recommendation in
10849 several standards that it is not used in new applications due to
10850 several cryptographic weaknesses. For binary compatibility reasons
10851 the MD2 API is still compiled in by default.
10855 * Add compression id to {d2i,i2d}_SSL_SESSION so it is correctly saved
10860 * Rename uni2asc and asc2uni functions to OPENSSL_uni2asc and
10861 OPENSSL_asc2uni conditionally on Netware platforms to avoid a name
10864 *Guenter <lists@gknw.net>*
10866 * Fix the server certificate chain building code to use X509_verify_cert(),
10867 it used to have an ad-hoc builder which was unable to cope with anything
10868 other than a simple chain.
10870 *David Woodhouse <dwmw2@infradead.org>, Steve Henson*
10872 * Don't check self signed certificate signatures in X509_verify_cert()
10873 by default (a flag can override this): it just wastes time without
10874 adding any security. As a useful side effect self signed root CAs
10875 with non-FIPS digests are now usable in FIPS mode.
10879 * In dtls1_process_out_of_seq_message() the check if the current message
10880 is already buffered was missing. For every new message was memory
10881 allocated, allowing an attacker to perform an denial of service attack
10882 with sending out of seq handshake messages until there is no memory
10883 left. Additionally every future message was buffered, even if the
10884 sequence number made no sense and would be part of another handshake.
10885 So only messages with sequence numbers less than 10 in advance will be
10886 buffered. ([CVE-2009-1378])
10888 *Robin Seggelmann, discovered by Daniel Mentz*
10890 * Records are buffered if they arrive with a future epoch to be
10891 processed after finishing the corresponding handshake. There is
10892 currently no limitation to this buffer allowing an attacker to perform
10893 a DOS attack with sending records with future epochs until there is no
10894 memory left. This patch adds the pqueue_size() function to determine
10895 the size of a buffer and limits the record buffer to 100 entries.
10898 *Robin Seggelmann, discovered by Daniel Mentz*
10900 * Keep a copy of frag->msg_header.frag_len so it can be used after the
10901 parent structure is freed. ([CVE-2009-1379])
10905 * Handle non-blocking I/O properly in SSL_shutdown() call.
10907 *Darryl Miles <darryl-mailinglists@netbauds.net>*
10909 * Add `2.5.4.*` OIDs
10911 *Ilya O. <vrghost@gmail.com>*
10913 ### Changes between 0.9.8k and 0.9.8l [5 Nov 2009]
10915 * Disable renegotiation completely - this fixes a severe security
10916 problem ([CVE-2009-3555]) at the cost of breaking all
10917 renegotiation. Renegotiation can be re-enabled by setting
10918 SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION in s3->flags at
10919 run-time. This is really not recommended unless you know what
10924 ### Changes between 0.9.8j and 0.9.8k [25 Mar 2009]
10926 * Don't set val to NULL when freeing up structures, it is freed up by
10927 underlying code. If `sizeof(void *) > sizeof(long)` this can result in
10928 zeroing past the valid field. ([CVE-2009-0789])
10930 *Paolo Ganci <Paolo.Ganci@AdNovum.CH>*
10932 * Fix bug where return value of CMS_SignerInfo_verify_content() was not
10933 checked correctly. This would allow some invalid signed attributes to
10934 appear to verify correctly. ([CVE-2009-0591])
10936 *Ivan Nestlerode <inestlerode@us.ibm.com>*
10938 * Reject UniversalString and BMPString types with invalid lengths. This
10939 prevents a crash in ASN1_STRING_print_ex() which assumes the strings have
10940 a legal length. ([CVE-2009-0590])
10944 * Set S/MIME signing as the default purpose rather than setting it
10945 unconditionally. This allows applications to override it at the store
10950 * Permit restricted recursion of ASN1 strings. This is needed in practice
10951 to handle some structures.
10955 * Improve efficiency of mem_gets: don't search whole buffer each time
10958 *Jeremy Shapiro <jnshapir@us.ibm.com>*
10960 * New -hex option for openssl rand.
10964 * Print out UTF8String and NumericString when parsing ASN1.
10968 * Support NumericString type for name components.
10972 * Allow CC in the environment to override the automatically chosen
10973 compiler. Note that nothing is done to ensure flags work with the
10978 ### Changes between 0.9.8i and 0.9.8j [07 Jan 2009]
10980 * Properly check EVP_VerifyFinal() and similar return values
10983 *Ben Laurie, Bodo Moeller, Google Security Team*
10985 * Enable TLS extensions by default.
10989 * Allow the CHIL engine to be loaded, whether the application is
10990 multithreaded or not. (This does not release the developer from the
10991 obligation to set up the dynamic locking callbacks.)
10993 *Sander Temme <sander@temme.net>*
10995 * Use correct exit code if there is an error in dgst command.
10997 *Steve Henson; problem pointed out by Roland Dirlewanger*
10999 * Tweak Configure so that you need to say "experimental-jpake" to enable
11000 JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications.
11004 * Add experimental JPAKE support, including demo authentication in
11005 s_client and s_server.
11009 * Set the comparison function in v3_addr_canonize().
11011 *Rob Austein <sra@hactrn.net>*
11013 * Add support for XMPP STARTTLS in s_client.
11015 *Philip Paeps <philip@freebsd.org>*
11017 * Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior
11018 to ensure that even with this option, only ciphersuites in the
11019 server's preference list will be accepted. (Note that the option
11020 applies only when resuming a session, so the earlier behavior was
11021 just about the algorithm choice for symmetric cryptography.)
11025 ### Changes between 0.9.8h and 0.9.8i [15 Sep 2008]
11027 * Fix NULL pointer dereference if a DTLS server received
11028 ChangeCipherSpec as first record ([CVE-2009-1386]).
11032 * Fix a state transition in s3_srvr.c and d1_srvr.c
11033 (was using SSL3_ST_CW_CLNT_HELLO_B, should be `..._ST_SW_SRVR_...`).
11035 *Nagendra Modadugu*
11037 * The fix in 0.9.8c that supposedly got rid of unsafe
11038 double-checked locking was incomplete for RSA blinding,
11039 addressing just one layer of what turns out to have been
11040 doubly unsafe triple-checked locking.
11042 So now fix this for real by retiring the MONT_HELPER macro
11043 in crypto/rsa/rsa_eay.c.
11045 *Bodo Moeller; problem pointed out by Marius Schilder*
11047 * Various precautionary measures:
11049 - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h).
11051 - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c).
11052 (NB: This would require knowledge of the secret session ticket key
11053 to exploit, in which case you'd be SOL either way.)
11055 - Change bn_nist.c so that it will properly handle input BIGNUMs
11056 outside the expected range.
11058 - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG
11061 *Neel Mehta, Bodo Moeller*
11063 * Allow engines to be "soft loaded" - i.e. optionally don't die if
11064 the load fails. Useful for distros.
11066 *Ben Laurie and the FreeBSD team*
11068 * Add support for Local Machine Keyset attribute in PKCS#12 files.
11072 * Fix BN_GF2m_mod_arr() top-bit cleanup code.
11076 * Expand ENGINE to support engine supplied SSL client certificate functions.
11078 This work was sponsored by Logica.
11082 * Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows
11083 keystores. Support for SSL/TLS client authentication too.
11084 Not compiled unless enable-capieng specified to Configure.
11086 This work was sponsored by Logica.
11090 * Fix bug in X509_ATTRIBUTE creation: don't set attribute using
11091 ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain
11092 attribute creation routines such as certificate requests and PKCS#12
11097 ### Changes between 0.9.8g and 0.9.8h [28 May 2008]
11099 * Fix flaw if 'Server Key exchange message' is omitted from a TLS
11100 handshake which could lead to a client crash as found using the
11101 Codenomicon TLS test suite ([CVE-2008-1672])
11103 *Steve Henson, Mark Cox*
11105 * Fix double free in TLS server name extensions which could lead to
11106 a remote crash found by Codenomicon TLS test suite ([CVE-2008-0891])
11110 * Clear error queue in SSL_CTX_use_certificate_chain_file()
11112 Clear the error queue to ensure that error entries left from
11113 older function calls do not interfere with the correct operation.
11115 *Lutz Jaenicke, Erik de Castro Lopo*
11117 * Remove root CA certificates of commercial CAs:
11119 The OpenSSL project does not recommend any specific CA and does not
11120 have any policy with respect to including or excluding any CA.
11121 Therefore, it does not make any sense to ship an arbitrary selection
11122 of root CA certificates with the OpenSSL software.
11126 * RSA OAEP patches to fix two separate invalid memory reads.
11127 The first one involves inputs when 'lzero' is greater than
11128 'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes
11129 before the beginning of from). The second one involves inputs where
11130 the 'db' section contains nothing but zeroes (there is a one-byte
11131 invalid read after the end of 'db').
11133 *Ivan Nestlerode <inestlerode@us.ibm.com>*
11135 * Partial backport from 0.9.9-dev:
11137 Introduce bn_mul_mont (dedicated Montgomery multiplication
11138 procedure) as a candidate for BIGNUM assembler implementation.
11139 While 0.9.9-dev uses assembler for various architectures, only
11140 x86_64 is available by default here in the 0.9.8 branch, and
11141 32-bit x86 is available through a compile-time setting.
11143 To try the 32-bit x86 assembler implementation, use Configure
11144 option "enable-montasm" (which exists only for this backport).
11146 As "enable-montasm" for 32-bit x86 disclaims code stability
11147 anyway, in this constellation we activate additional code
11148 backported from 0.9.9-dev for further performance improvements,
11149 namely BN_from_montgomery_word. (To enable this otherwise,
11150 e.g. x86_64, try `-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD`.)
11152 *Andy Polyakov (backport partially by Bodo Moeller)*
11154 * Add TLS session ticket callback. This allows an application to set
11155 TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed
11156 values. This is useful for key rollover for example where several key
11157 sets may exist with different names.
11161 * Reverse ENGINE-internal logic for caching default ENGINE handles.
11162 This was broken until now in 0.9.8 releases, such that the only way
11163 a registered ENGINE could be used (assuming it initialises
11164 successfully on the host) was to explicitly set it as the default
11165 for the relevant algorithms. This is in contradiction with 0.9.7
11166 behaviour and the documentation. With this fix, when an ENGINE is
11167 registered into a given algorithm's table of implementations, the
11168 'uptodate' flag is reset so that auto-discovery will be used next
11169 time a new context for that algorithm attempts to select an
11172 *Ian Lister (tweaked by Geoff Thorpe)*
11174 * Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9
11175 implementation in the following ways:
11177 Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be
11180 Lack of BER streaming support means one pass streaming processing is
11181 only supported if data is detached: setting the streaming flag is
11182 ignored for embedded content.
11184 CMS support is disabled by default and must be explicitly enabled
11185 with the enable-cms configuration option.
11189 * Update the GMP engine glue to do direct copies between BIGNUM and
11190 mpz_t when openssl and GMP use the same limb size. Otherwise the
11191 existing "conversion via a text string export" trick is still used.
11193 *Paul Sheer <paulsheer@gmail.com>*
11195 * Zlib compression BIO. This is a filter BIO which compressed and
11196 uncompresses any data passed through it.
11200 * Add AES_wrap_key() and AES_unwrap_key() functions to implement
11201 RFC3394 compatible AES key wrapping.
11205 * Add utility functions to handle ASN1 structures. ASN1_STRING_set0():
11206 sets string data without copying. X509_ALGOR_set0() and
11207 X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier)
11208 data. Attribute function X509at_get0_data_by_OBJ(): retrieves data
11209 from an X509_ATTRIBUTE structure optionally checking it occurs only
11210 once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied
11215 * Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set()
11216 to get the expected BN_FLG_CONSTTIME behavior.
11218 *Bodo Moeller (Google)*
11222 - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets
11223 - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT)
11224 - added some more tests to do_tests.pl
11225 - fixed RunningProcess usage so that it works with newer LIBC NDKs too
11226 - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency
11227 - added new Configure targets netware-clib-bsdsock, netware-clib-gcc,
11228 netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc
11229 - various changes to netware.pl to enable gcc-cross builds on Win32
11231 - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD)
11232 - various changes to fix missing prototype warnings
11233 - fixed x86nasm.pl to create correct asm files for NASM COFF output
11234 - added AES, WHIRLPOOL and CPUID assembler code to build files
11235 - added missing AES assembler make rules to mk1mf.pl
11236 - fixed order of includes in `apps/ocsp.c` so that `e_os.h` settings apply
11238 *Guenter Knauf <eflash@gmx.net>*
11240 * Implement certificate status request TLS extension defined in RFC3546.
11241 A client can set the appropriate parameters and receive the encoded
11242 OCSP response via a callback. A server can query the supplied parameters
11243 and set the encoded OCSP response in the callback. Add simplified examples
11244 to s_client and s_server.
11248 ### Changes between 0.9.8f and 0.9.8g [19 Oct 2007]
11250 * Fix various bugs:
11251 + Binary incompatibility of ssl_ctx_st structure
11252 + DTLS interoperation with non-compliant servers
11253 + Don't call get_session_cb() without proposed session
11254 + Fix ia64 assembler code
11256 *Andy Polyakov, Steve Henson*
11258 ### Changes between 0.9.8e and 0.9.8f [11 Oct 2007]
11260 * DTLS Handshake overhaul. There were longstanding issues with
11261 OpenSSL DTLS implementation, which were making it impossible for
11262 RFC 4347 compliant client to communicate with OpenSSL server.
11263 Unfortunately just fixing these incompatibilities would "cut off"
11264 pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e
11265 server keeps tolerating non RFC compliant syntax. The opposite is
11266 not true, 0.9.8f client can not communicate with earlier server.
11267 This update even addresses CVE-2007-4995.
11271 * Changes to avoid need for function casts in OpenSSL: some compilers
11272 (gcc 4.2 and later) reject their use.
11273 *Kurt Roeckx <kurt@roeckx.be>, Peter Hartley <pdh@utter.chaos.org.uk>,
11276 * Add RFC4507 support to OpenSSL. This includes the corrections in
11277 RFC4507bis. The encrypted ticket format is an encrypted encoded
11278 SSL_SESSION structure, that way new session features are automatically
11281 If a client application caches session in an SSL_SESSION structure
11282 support is transparent because tickets are now stored in the encoded
11285 The SSL_CTX structure automatically generates keys for ticket
11286 protection in servers so again support should be possible
11287 with no application modification.
11289 If a client or server wishes to disable RFC4507 support then the option
11290 SSL_OP_NO_TICKET can be set.
11292 Add a TLS extension debugging callback to allow the contents of any client
11293 or server extensions to be examined.
11295 This work was sponsored by Google.
11299 * Add initial support for TLS extensions, specifically for the server_name
11300 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now
11301 have new members for a hostname. The SSL data structure has an
11302 additional member `SSL_CTX *initial_ctx` so that new sessions can be
11303 stored in that context to allow for session resumption, even after the
11304 SSL has been switched to a new SSL_CTX in reaction to a client's
11305 server_name extension.
11307 New functions (subject to change):
11309 SSL_get_servername()
11310 SSL_get_servername_type()
11313 New CTRL codes and macros (subject to change):
11315 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB
11316 - SSL_CTX_set_tlsext_servername_callback()
11317 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG
11318 - SSL_CTX_set_tlsext_servername_arg()
11319 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name()
11321 openssl s_client has a new '-servername ...' option.
11323 openssl s_server has new options '-servername_host ...', '-cert2 ...',
11324 '-key2 ...', '-servername_fatal' (subject to change). This allows
11325 testing the HostName extension for a specific single hostname ('-cert'
11326 and '-key' remain fallbacks for handshakes without HostName
11327 negotiation). If the unrecognized_name alert has to be sent, this by
11328 default is a warning; it becomes fatal with the '-servername_fatal'
11331 *Peter Sylvester, Remy Allais, Christophe Renou, Steve Henson*
11333 * Add AES and SSE2 assembly language support to VC++ build.
11337 * Mitigate attack on final subtraction in Montgomery reduction.
11341 * Fix crypto/ec/ec_mult.c to work properly with scalars of value 0
11342 (which previously caused an internal error).
11346 * Squeeze another 10% out of IGE mode when in != out.
11350 * AES IGE mode speedup.
11352 *Dean Gaudet (Google)*
11354 * Add the Korean symmetric 128-bit cipher SEED (see
11355 <http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp>) and
11356 add SEED ciphersuites from RFC 4162:
11358 TLS_RSA_WITH_SEED_CBC_SHA = "SEED-SHA"
11359 TLS_DHE_DSS_WITH_SEED_CBC_SHA = "DHE-DSS-SEED-SHA"
11360 TLS_DHE_RSA_WITH_SEED_CBC_SHA = "DHE-RSA-SEED-SHA"
11361 TLS_DH_anon_WITH_SEED_CBC_SHA = "ADH-SEED-SHA"
11363 To minimize changes between patchlevels in the OpenSSL 0.9.8
11364 series, SEED remains excluded from compilation unless OpenSSL
11365 is configured with 'enable-seed'.
11367 *KISA, Bodo Moeller*
11369 * Mitigate branch prediction attacks, which can be practical if a
11370 single processor is shared, allowing a spy process to extract
11371 information. For detailed background information, see
11372 <http://eprint.iacr.org/2007/039> (O. Aciicmez, S. Gueron,
11373 J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL
11374 and Necessary Software Countermeasures"). The core of the change
11375 are new versions BN_div_no_branch() and
11376 BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(),
11377 respectively, which are slower, but avoid the security-relevant
11378 conditional branches. These are automatically called by BN_div()
11379 and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one
11380 of the input BIGNUMs. Also, BN_is_bit_set() has been changed to
11381 remove a conditional branch.
11383 BN_FLG_CONSTTIME is the new name for the previous
11384 BN_FLG_EXP_CONSTTIME flag, since it now affects more than just
11385 modular exponentiation. (Since OpenSSL 0.9.7h, setting this flag
11386 in the exponent causes BN_mod_exp_mont() to use the alternative
11387 implementation in BN_mod_exp_mont_consttime().) The old name
11388 remains as a deprecated alias.
11390 Similarly, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general
11391 RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses
11392 constant-time implementations for more than just exponentiation.
11393 Here too the old name is kept as a deprecated alias.
11395 BN_BLINDING_new() will now use BN_dup() for the modulus so that
11396 the BN_BLINDING structure gets an independent copy of the
11397 modulus. This means that the previous `BIGNUM *m` argument to
11398 BN_BLINDING_new() and to BN_BLINDING_create_param() now
11399 essentially becomes `const BIGNUM *m`, although we can't actually
11400 change this in the header file before 0.9.9. It allows
11401 RSA_setup_blinding() to use BN_with_flags() on the modulus to
11402 enable BN_FLG_CONSTTIME.
11404 *Matthew D Wood (Intel Corp)*
11406 * In the SSL/TLS server implementation, be strict about session ID
11407 context matching (which matters if an application uses a single
11408 external cache for different purposes). Previously,
11409 out-of-context reuse was forbidden only if SSL_VERIFY_PEER was
11410 set. This did ensure strict client verification, but meant that,
11411 with applications using a single external cache for quite
11412 different requirements, clients could circumvent ciphersuite
11413 restrictions for a given session ID context by starting a session
11414 in a different context.
11418 * Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that
11419 a ciphersuite string such as "DEFAULT:RSA" cannot enable
11420 authentication-only ciphersuites.
11424 * Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was
11425 not complete and could lead to a possible single byte overflow
11426 ([CVE-2007-5135]) [Ben Laurie]
11428 ### Changes between 0.9.8d and 0.9.8e [23 Feb 2007]
11430 * Since AES128 and AES256 (and similarly Camellia128 and
11431 Camellia256) share a single mask bit in the logic of
11432 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a
11433 kludge to work properly if AES128 is available and AES256 isn't
11434 (or if Camellia128 is available and Camellia256 isn't).
11438 * Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c
11439 (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters):
11440 When a point or a seed is encoded in a BIT STRING, we need to
11441 prevent the removal of trailing zero bits to get the proper DER
11442 encoding. (By default, crypto/asn1/a_bitstr.c assumes the case
11443 of a NamedBitList, for which trailing 0 bits need to be removed.)
11447 * Have SSL/TLS server implementation tolerate "mismatched" record
11448 protocol version while receiving ClientHello even if the
11449 ClientHello is fragmented. (The server can't insist on the
11450 particular protocol version it has chosen before the ServerHello
11451 message has informed the client about his choice.)
11455 * Add RFC 3779 support.
11457 *Rob Austein for ARIN, Ben Laurie*
11459 * Load error codes if they are not already present instead of using a
11460 static variable. This allows them to be cleanly unloaded and reloaded.
11461 Improve header file function name parsing.
11465 * extend SMTP and IMAP protocol emulation in s_client to use EHLO
11466 or CAPABILITY handshake as required by RFCs.
11468 *Goetz Babin-Ebell*
11470 ### Changes between 0.9.8c and 0.9.8d [28 Sep 2006]
11472 * Introduce limits to prevent malicious keys being able to
11473 cause a denial of service. ([CVE-2006-2940])
11475 *Steve Henson, Bodo Moeller*
11477 * Fix ASN.1 parsing of certain invalid structures that can result
11478 in a denial of service. ([CVE-2006-2937]) [Steve Henson]
11480 * Fix buffer overflow in SSL_get_shared_ciphers() function.
11481 ([CVE-2006-3738]) [Tavis Ormandy and Will Drewry, Google Security Team]
11483 * Fix SSL client code which could crash if connecting to a
11484 malicious SSLv2 server. ([CVE-2006-4343])
11486 *Tavis Ormandy and Will Drewry, Google Security Team*
11488 * Since 0.9.8b, ciphersuite strings naming explicit ciphersuites
11489 match only those. Before that, "AES256-SHA" would be interpreted
11490 as a pattern and match "AES128-SHA" too (since AES128-SHA got
11491 the same strength classification in 0.9.7h) as we currently only
11492 have a single AES bit in the ciphersuite description bitmap.
11493 That change, however, also applied to ciphersuite strings such as
11494 "RC4-MD5" that intentionally matched multiple ciphersuites --
11495 namely, SSL 2.0 ciphersuites in addition to the more common ones
11496 from SSL 3.0/TLS 1.0.
11498 So we change the selection algorithm again: Naming an explicit
11499 ciphersuite selects this one ciphersuite, and any other similar
11500 ciphersuite (same bitmap) from *other* protocol versions.
11501 Thus, "RC4-MD5" again will properly select both the SSL 2.0
11502 ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite.
11504 Since SSL 2.0 does not have any ciphersuites for which the
11505 128/256 bit distinction would be relevant, this works for now.
11506 The proper fix will be to use different bits for AES128 and
11507 AES256, which would have avoided the problems from the beginning;
11508 however, bits are scarce, so we can only do this in a new release
11509 (not just a patchlevel) when we can change the SSL_CIPHER
11510 definition to split the single 'unsigned long mask' bitmap into
11511 multiple values to extend the available space.
11515 ### Changes between 0.9.8b and 0.9.8c [05 Sep 2006]
11517 * Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher
11518 ([CVE-2006-4339]) [Ben Laurie and Google Security Team]
11520 * Add AES IGE and biIGE modes.
11524 * Change the Unix randomness entropy gathering to use poll() when
11525 possible instead of select(), since the latter has some
11526 undesirable limitations.
11528 *Darryl Miles via Richard Levitte and Bodo Moeller*
11530 * Disable "ECCdraft" ciphersuites more thoroughly. Now special
11531 treatment in ssl/ssl_ciph.s makes sure that these ciphersuites
11532 cannot be implicitly activated as part of, e.g., the "AES" alias.
11533 However, please upgrade to OpenSSL 0.9.9[-dev] for
11534 non-experimental use of the ECC ciphersuites to get TLS extension
11535 support, which is required for curve and point format negotiation
11536 to avoid potential handshake problems.
11540 * Disable rogue ciphersuites:
11542 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5")
11543 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5")
11544 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5")
11546 The latter two were purportedly from
11547 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really
11550 Also deactivate the remaining ciphersuites from
11551 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as
11552 unofficial, and the ID has long expired.
11556 * Fix RSA blinding Heisenbug (problems sometimes occurred on
11557 dual-core machines) and other potential thread-safety issues.
11561 * Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key
11562 versions), which is now available for royalty-free use
11563 (see <http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html>).
11564 Also, add Camellia TLS ciphersuites from RFC 4132.
11566 To minimize changes between patchlevels in the OpenSSL 0.9.8
11567 series, Camellia remains excluded from compilation unless OpenSSL
11568 is configured with 'enable-camellia'.
11572 * Disable the padding bug check when compression is in use. The padding
11573 bug check assumes the first packet is of even length, this is not
11574 necessarily true if compression is enabled and can result in false
11575 positives causing handshake failure. The actual bug test is ancient
11576 code so it is hoped that implementations will either have fixed it by
11577 now or any which still have the bug do not support compression.
11581 ### Changes between 0.9.8a and 0.9.8b [04 May 2006]
11583 * When applying a cipher rule check to see if string match is an explicit
11584 cipher suite and only match that one cipher suite if it is.
11588 * Link in manifests for VC++ if needed.
11590 *Austin Ziegler <halostatue@gmail.com>*
11592 * Update support for ECC-based TLS ciphersuites according to
11593 draft-ietf-tls-ecc-12.txt with proposed changes (but without
11594 TLS extensions, which are supported starting with the 0.9.9
11595 branch, not in the OpenSSL 0.9.8 branch).
11599 * New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support
11600 opaque EVP_CIPHER_CTX handling.
11604 * Fixes and enhancements to zlib compression code. We now only use
11605 "zlib1.dll" and use the default `__cdecl` calling convention on Win32
11606 to conform with the standards mentioned here:
11607 <http://www.zlib.net/DLL_FAQ.txt>
11608 Static zlib linking now works on Windows and the new --with-zlib-include
11609 --with-zlib-lib options to Configure can be used to supply the location
11610 of the headers and library. Gracefully handle case where zlib library
11615 * Several fixes and enhancements to the OID generation code. The old code
11616 sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't
11617 handle numbers larger than ULONG_MAX, truncated printing and had a
11618 non standard OBJ_obj2txt() behaviour.
11622 * Add support for building of engines under engine/ as shared libraries
11623 under VC++ build system.
11627 * Corrected the numerous bugs in the Win32 path splitter in DSO.
11628 Hopefully, we will not see any false combination of paths any more.
11632 ### Changes between 0.9.8 and 0.9.8a [11 Oct 2005]
11634 * Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
11635 (part of SSL_OP_ALL). This option used to disable the
11636 countermeasure against man-in-the-middle protocol-version
11637 rollback in the SSL 2.0 server implementation, which is a bad
11638 idea. ([CVE-2005-2969])
11640 *Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
11641 for Information Security, National Institute of Advanced Industrial
11642 Science and Technology [AIST], Japan)*
11644 * Add two function to clear and return the verify parameter flags.
11648 * Keep cipherlists sorted in the source instead of sorting them at
11649 runtime, thus removing the need for a lock.
11653 * Avoid some small subgroup attacks in Diffie-Hellman.
11655 *Nick Mathewson and Ben Laurie*
11657 * Add functions for well-known primes.
11661 * Extended Windows CE support.
11663 *Satoshi Nakamura and Andy Polyakov*
11665 * Initialize SSL_METHOD structures at compile time instead of during
11666 runtime, thus removing the need for a lock.
11670 * Make PKCS7_decrypt() work even if no certificate is supplied by
11671 attempting to decrypt each encrypted key in turn. Add support to
11676 ### Changes between 0.9.7h and 0.9.8 [05 Jul 2005]
11678 [NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after
11681 * Add libcrypto.pc and libssl.pc for those who feel they need them.
11685 * Change CA.sh and CA.pl so they don't bundle the CSR and the private
11686 key into the same file any more.
11690 * Add initial support for Win64, both IA64 and AMD64/x64 flavors.
11694 * Add -utf8 command line and config file option to 'ca'.
11696 *Stefan <stf@udoma.org*
11698 * Removed the macro des_crypt(), as it seems to conflict with some
11699 libraries. Use DES_crypt().
11703 * Correct naming of the 'chil' and '4758cca' ENGINEs. This
11704 involves renaming the source and generated shared-libs for
11705 both. The engines will accept the corrected or legacy ids
11706 ('ncipher' and '4758_cca' respectively) when binding. NB,
11707 this only applies when building 'shared'.
11709 *Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe*
11711 * Add attribute functions to EVP_PKEY structure. Modify
11712 PKCS12_create() to recognize a CSP name attribute and
11713 use it. Make -CSP option work again in pkcs12 utility.
11717 * Add new functionality to the bn blinding code:
11718 - automatic re-creation of the BN_BLINDING parameters after
11719 a fixed number of uses (currently 32)
11720 - add new function for parameter creation
11721 - introduce flags to control the update behaviour of the
11722 BN_BLINDING parameters
11723 - hide BN_BLINDING structure
11724 Add a second BN_BLINDING slot to the RSA structure to improve
11725 performance when a single RSA object is shared among several
11730 * Add support for DTLS.
11732 *Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie*
11734 * Add support for DER encoded private keys (SSL_FILETYPE_ASN1)
11735 to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file()
11739 * Remove buggy and incomplete DH cert support from
11740 ssl/ssl_rsa.c and ssl/s3_both.c
11744 * Use SHA-1 instead of MD5 as the default digest algorithm for
11745 the `apps/openssl` commands.
11749 * Compile clean with "-Wall -Wmissing-prototypes
11750 -Wstrict-prototypes -Wmissing-declarations -Werror". Currently
11751 DEBUG_SAFESTACK must also be set.
11755 * Change ./Configure so that certain algorithms can be disabled by default.
11756 The new counterpiece to "no-xxx" is "enable-xxx".
11758 The patented RC5 and MDC2 algorithms will now be disabled unless
11759 "enable-rc5" and "enable-mdc2", respectively, are specified.
11761 (IDEA remains enabled despite being patented. This is because IDEA
11762 is frequently required for interoperability, and there is no license
11763 fee for non-commercial use. As before, "no-idea" can be used to
11764 avoid this algorithm.)
11768 * Add processing of proxy certificates (see RFC 3820). This work was
11769 sponsored by KTH (The Royal Institute of Technology in Stockholm) and
11770 EGEE (Enabling Grids for E-science in Europe).
11774 * RC4 performance overhaul on modern architectures/implementations, such
11775 as Intel P4, IA-64 and AMD64.
11779 * New utility extract-section.pl. This can be used specify an alternative
11780 section number in a pod file instead of having to treat each file as
11781 a separate case in Makefile. This can be done by adding two lines to the
11784 =for comment openssl_section:XXX
11786 The blank line is mandatory.
11790 * New arguments -certform, -keyform and -pass for s_client and s_server
11791 to allow alternative format key and certificate files and passphrase
11796 * New structure X509_VERIFY_PARAM which combines current verify parameters,
11797 update associated structures and add various utility functions.
11799 Add new policy related verify parameters, include policy checking in
11800 standard verify code. Enhance 'smime' application with extra parameters
11801 to support policy checking and print out.
11805 * Add a new engine to support VIA PadLock ACE extensions in the VIA C3
11806 Nehemiah processors. These extensions support AES encryption in hardware
11807 as well as RNG (though RNG support is currently disabled).
11809 *Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov*
11811 * Deprecate `BN_[get|set]_params()` functions (they were ignored internally).
11815 * New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented.
11817 *Andy Polyakov and a number of other people*
11819 * Improved PowerPC platform support. Most notably BIGNUM assembler
11820 implementation contributed by IBM.
11822 *Suresh Chari, Peter Waltenberg, Andy Polyakov*
11824 * The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public
11825 exponent rather than 'unsigned long'. There is a corresponding change to
11826 the new 'rsa_keygen' element of the RSA_METHOD structure.
11828 *Jelte Jansen, Geoff Thorpe*
11830 * Functionality for creating the initial serial number file is now
11831 moved from CA.pl to the 'ca' utility with a new option -create_serial.
11833 (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial
11834 number file to 1, which is bound to cause problems. To avoid
11835 the problems while respecting compatibility between different 0.9.7
11836 patchlevels, 0.9.7e employed 'openssl x509 -next_serial' in
11837 CA.pl for serial number initialization. With the new release 0.9.8,
11838 we can fix the problem directly in the 'ca' utility.)
11842 * Reduced header interdependencies by declaring more opaque objects in
11843 ossl_typ.h. As a consequence, including some headers (eg. engine.h) will
11844 give fewer recursive includes, which could break lazy source code - so
11845 this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always,
11846 developers should define this symbol when building and using openssl to
11847 ensure they track the recommended behaviour, interfaces, [etc], but
11848 backwards-compatible behaviour prevails when this isn't defined.
11852 * New function X509_POLICY_NODE_print() which prints out policy nodes.
11856 * Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality.
11857 This will generate a random key of the appropriate length based on the
11858 cipher context. The EVP_CIPHER can provide its own random key generation
11859 routine to support keys of a specific form. This is used in the des and
11860 3des routines to generate a key of the correct parity. Update S/MIME
11861 code to use new functions and hence generate correct parity DES keys.
11862 Add EVP_CHECK_DES_KEY #define to return an error if the key is not
11863 valid (weak or incorrect parity).
11867 * Add a local set of CRLs that can be used by X509_verify_cert() as well
11868 as looking them up. This is useful when the verified structure may contain
11869 CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs
11870 present unless the new PKCS7_NO_CRL flag is asserted.
11874 * Extend ASN1 oid configuration module. It now additionally accepts the
11877 shortName = some long name, 1.2.3.4
11881 * Reimplemented the BN_CTX implementation. There is now no more static
11882 limitation on the number of variables it can handle nor the depth of the
11883 "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack
11884 information can now expand as required, and rather than having a single
11885 static array of bignums, BN_CTX now uses a linked-list of such arrays
11886 allowing it to expand on demand whilst maintaining the usefulness of
11887 BN_CTX's "bundling".
11891 * Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD
11892 to allow all RSA operations to function using a single BN_CTX.
11896 * Preliminary support for certificate policy evaluation and checking. This
11897 is initially intended to pass the tests outlined in "Conformance Testing
11898 of Relying Party Client Certificate Path Processing Logic" v1.07.
11902 * bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and
11903 remained unused and not that useful. A variety of other little bignum
11904 tweaks and fixes have also been made continuing on from the audit (see
11909 * Constify all or almost all d2i, c2i, s2i and r2i functions, along with
11910 associated ASN1, EVP and SSL functions and old ASN1 macros.
11914 * BN_zero() only needs to set 'top' and 'neg' to zero for correct results,
11915 and this should never fail. So the return value from the use of
11916 BN_set_word() (which can fail due to needless expansion) is now deprecated;
11917 if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro.
11921 * BN_CTX_get() should return zero-valued bignums, providing the same
11922 initialised value as BN_new().
11924 *Geoff Thorpe, suggested by Ulf Möller*
11926 * Support for inhibitAnyPolicy certificate extension.
11930 * An audit of the BIGNUM code is underway, for which debugging code is
11931 enabled when BN_DEBUG is defined. This makes stricter enforcements on what
11932 is considered valid when processing BIGNUMs, and causes execution to
11933 assert() when a problem is discovered. If BN_DEBUG_RAND is defined,
11934 further steps are taken to deliberately pollute unused data in BIGNUM
11935 structures to try and expose faulty code further on. For now, openssl will
11936 (in its default mode of operation) continue to tolerate the inconsistent
11937 forms that it has tolerated in the past, but authors and packagers should
11938 consider trying openssl and their own applications when compiled with
11939 these debugging symbols defined. It will help highlight potential bugs in
11940 their own code, and will improve the test coverage for OpenSSL itself. At
11941 some point, these tighter rules will become openssl's default to improve
11942 maintainability, though the assert()s and other overheads will remain only
11943 in debugging configurations. See bn.h for more details.
11945 *Geoff Thorpe, Nils Larsch, Ulf Möller*
11947 * BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure
11948 that can only be obtained through BN_CTX_new() (which implicitly
11949 initialises it). The presence of this function only made it possible
11950 to overwrite an existing structure (and cause memory leaks).
11954 * Because of the callback-based approach for implementing LHASH as a
11955 template type, lh_insert() adds opaque objects to hash-tables and
11956 lh_doall() or lh_doall_arg() are typically used with a destructor callback
11957 to clean up those corresponding objects before destroying the hash table
11958 (and losing the object pointers). So some over-zealous constifications in
11959 LHASH have been relaxed so that lh_insert() does not take (nor store) the
11960 objects as "const" and the `lh_doall[_arg]` callback wrappers are not
11961 prototyped to have "const" restrictions on the object pointers they are
11962 given (and so aren't required to cast them away any more).
11966 * The tmdiff.h API was so ugly and minimal that our own timing utility
11967 (speed) prefers to use its own implementation. The two implementations
11968 haven't been consolidated as yet (volunteers?) but the tmdiff API has had
11969 its object type properly exposed (MS_TM) instead of casting to/from
11970 `char *`. This may still change yet if someone realises MS_TM and
11972 aren't necessarily the greatest nomenclatures - but this is what was used
11973 internally to the implementation so I've used that for now.
11977 * Ensure that deprecated functions do not get compiled when
11978 OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of
11979 the self-tests were still using deprecated key-generation functions so
11980 these have been updated also.
11984 * Reorganise PKCS#7 code to separate the digest location functionality
11985 into PKCS7_find_digest(), digest addition into PKCS7_bio_add_digest().
11986 New function PKCS7_set_digest() to set the digest type for PKCS#7
11987 digestedData type. Add additional code to correctly generate the
11988 digestedData type and add support for this type in PKCS7 initialization
11993 * New function PKCS7_set0_type_other() this initializes a PKCS7
11994 structure of type "other".
11998 * Fix prime generation loop in crypto/bn/bn_prime.pl by making
11999 sure the loop does correctly stop and breaking ("division by zero")
12000 modulus operations are not performed. The (pre-generated) prime
12001 table crypto/bn/bn_prime.h was already correct, but it could not be
12002 re-generated on some platforms because of the "division by zero"
12003 situation in the script.
12005 *Ralf S. Engelschall*
12007 * Update support for ECC-based TLS ciphersuites according to
12008 draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with
12009 SHA-1 now is only used for "small" curves (where the
12010 representation of a field element takes up to 24 bytes); for
12011 larger curves, the field element resulting from ECDH is directly
12012 used as premaster secret.
12014 *Douglas Stebila (Sun Microsystems Laboratories)*
12016 * Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2
12017 curve secp160r1 to the tests.
12019 *Douglas Stebila (Sun Microsystems Laboratories)*
12021 * Add the possibility to load symbols globally with DSO.
12023 *Götz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte*
12025 * Add the functions ERR_set_mark() and ERR_pop_to_mark() for better
12026 control of the error stack.
12030 * Add support for STORE in ENGINE.
12034 * Add the STORE type. The intention is to provide a common interface
12035 to certificate and key stores, be they simple file-based stores, or
12036 HSM-type store, or LDAP stores, or...
12037 NOTE: The code is currently UNTESTED and isn't really used anywhere.
12041 * Add a generic structure called OPENSSL_ITEM. This can be used to
12042 pass a list of arguments to any function as well as provide a way
12043 for a function to pass data back to the caller.
12047 * Add the functions BUF_strndup() and BUF_memdup(). BUF_strndup()
12048 works like BUF_strdup() but can be used to duplicate a portion of
12049 a string. The copy gets NUL-terminated. BUF_memdup() duplicates
12054 * Add the function sk_find_ex() which works like sk_find(), but will
12055 return an index to an element even if an exact match couldn't be
12056 found. The index is guaranteed to point at the element where the
12057 searched-for key would be inserted to preserve sorting order.
12061 * Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but
12062 takes an extra flags argument for optional functionality. Currently,
12063 the following flags are defined:
12065 OBJ_BSEARCH_VALUE_ON_NOMATCH
12066 This one gets OBJ_bsearch_ex() to return a pointer to the first
12067 element where the comparing function returns a negative or zero
12070 OBJ_BSEARCH_FIRST_VALUE_ON_MATCH
12071 This one gets OBJ_bsearch_ex() to return a pointer to the first
12072 element where the comparing function returns zero. This is useful
12073 if there are more than one element where the comparing function
12078 * Make it possible to create self-signed certificates with 'openssl ca'
12079 in such a way that the self-signed certificate becomes part of the
12080 CA database and uses the same mechanisms for serial number generation
12081 as all other certificate signing. The new flag '-selfsign' enables
12082 this functionality. Adapt CA.sh and CA.pl.in.
12086 * Add functionality to check the public key of a certificate request
12087 against a given private. This is useful to check that a certificate
12088 request can be signed by that key (self-signing).
12092 * Make it possible to have multiple active certificates with the same
12093 subject in the CA index file. This is done only if the keyword
12094 'unique_subject' is set to 'no' in the main CA section (default
12095 if 'CA_default') of the configuration file. The value is saved
12096 with the database itself in a separate index attribute file,
12097 named like the index file with '.attr' appended to the name.
12101 * Generate multi-valued AVAs using '+' notation in config files for
12106 * Support for nameConstraints certificate extension.
12110 * Support for policyConstraints certificate extension.
12114 * Support for policyMappings certificate extension.
12118 * Make sure the default DSA_METHOD implementation only uses its
12119 dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL,
12120 and change its own handlers to be NULL so as to remove unnecessary
12121 indirection. This lets alternative implementations fallback to the
12122 default implementation more easily.
12126 * Support for directoryName in GeneralName related extensions
12131 * Make it possible to link applications using Makefile.shared.
12132 Make that possible even when linking against static libraries!
12136 * Support for single pass processing for S/MIME signing. This now
12137 means that S/MIME signing can be done from a pipe, in addition
12138 cleartext signing (multipart/signed type) is effectively streaming
12139 and the signed data does not need to be all held in memory.
12141 This is done with a new flag PKCS7_STREAM. When this flag is set
12142 PKCS7_sign() only initializes the PKCS7 structure and the actual signing
12143 is done after the data is output (and digests calculated) in
12144 SMIME_write_PKCS7().
12148 * Add full support for -rpath/-R, both in shared libraries and
12149 applications, at least on the platforms where it's known how
12154 * In crypto/ec/ec_mult.c, implement fast point multiplication with
12155 precomputation, based on wNAF splitting: EC_GROUP_precompute_mult()
12156 will now compute a table of multiples of the generator that
12157 makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul()
12158 faster (notably in the case of a single point multiplication,
12159 scalar * generator).
12161 *Nils Larsch, Bodo Moeller*
12163 * IPv6 support for certificate extensions. The various extensions
12164 which use the IP:a.b.c.d can now take IPv6 addresses using the
12165 formats of RFC1884 2.2 . IPv6 addresses are now also displayed
12170 * Added an ENGINE that implements RSA by performing private key
12171 exponentiations with the GMP library. The conversions to and from
12172 GMP's mpz_t format aren't optimised nor are any montgomery forms
12173 cached, and on x86 it appears OpenSSL's own performance has caught up.
12174 However there are likely to be other architectures where GMP could
12175 provide a boost. This ENGINE is not built in by default, but it can be
12176 specified at Configure time and should be accompanied by the necessary
12177 linker additions, eg;
12178 ./config -DOPENSSL_USE_GMP -lgmp
12182 * "openssl engine" will not display ENGINE/DSO load failure errors when
12183 testing availability of engines with "-t" - the old behaviour is
12184 produced by increasing the feature's verbosity with "-tt".
12188 * ECDSA routines: under certain error conditions uninitialized BN objects
12189 could be freed. Solution: make sure initialization is performed early
12190 enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de>
12195 * Key-generation can now be implemented in RSA_METHOD, DSA_METHOD
12196 and DH_METHOD (eg. by ENGINE implementations) to override the normal
12197 software implementations. For DSA and DH, parameter generation can
12198 also be overridden by providing the appropriate method callbacks.
12202 * Change the "progress" mechanism used in key-generation and
12203 primality testing to functions that take a new BN_GENCB pointer in
12204 place of callback/argument pairs. The new API functions have `_ex`
12205 postfixes and the older functions are reimplemented as wrappers for
12206 the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide
12207 declarations of the old functions to help (graceful) attempts to
12208 migrate to the new functions. Also, the new key-generation API
12209 functions operate on a caller-supplied key-structure and return
12210 success/failure rather than returning a key or NULL - this is to
12211 help make "keygen" another member function of RSA_METHOD etc.
12213 Example for using the new callback interface:
12215 int (*my_callback)(int a, int b, BN_GENCB *cb) = ...;
12216 void *my_arg = ...;
12219 BN_GENCB_set(&my_cb, my_callback, my_arg);
12221 return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb);
12222 /* For the meaning of a, b in calls to my_callback(), see the
12223 * documentation of the function that calls the callback.
12224 * cb will point to my_cb; my_arg can be retrieved as cb->arg.
12225 * my_callback should return 1 if it wants BN_is_prime_ex()
12226 * to continue, or 0 to stop.
12231 * Change the ZLIB compression method to be stateful, and make it
12232 available to TLS with the number defined in
12233 draft-ietf-tls-compression-04.txt.
12237 * Add the ASN.1 structures and functions for CertificatePair, which
12238 is defined as follows (according to X.509_4thEditionDraftV6.pdf):
12240 CertificatePair ::= SEQUENCE {
12241 forward [0] Certificate OPTIONAL,
12242 reverse [1] Certificate OPTIONAL,
12243 -- at least one of the pair shall be present -- }
12245 Also implement the PEM functions to read and write certificate
12246 pairs, and defined the PEM tag as "CERTIFICATE PAIR".
12248 This needed to be defined, mostly for the sake of the LDAP
12249 attribute crossCertificatePair, but may prove useful elsewhere as
12254 * Make it possible to inhibit symlinking of shared libraries in
12255 Makefile.shared, for Cygwin's sake.
12259 * Extend the BIGNUM API by creating a function
12260 void BN_set_negative(BIGNUM *a, int neg);
12261 and a macro that behave like
12262 int BN_is_negative(const BIGNUM *a);
12264 to avoid the need to access 'a->neg' directly in applications.
12268 * Implement fast modular reduction for pseudo-Mersenne primes
12269 used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c).
12270 EC_GROUP_new_curve_GFp() will now automatically use this
12273 *Nils Larsch <nla@trustcenter.de>*
12275 * Add new lock type (CRYPTO_LOCK_BN).
12279 * Change the ENGINE framework to automatically load engines
12280 dynamically from specific directories unless they could be
12281 found to already be built in or loaded. Move all the
12282 current engines except for the cryptodev one to a new
12283 directory engines/.
12284 The engines in engines/ are built as shared libraries if
12285 the "shared" options was given to ./Configure or ./config.
12286 Otherwise, they are inserted in libcrypto.a.
12287 /usr/local/ssl/engines is the default directory for dynamic
12288 engines, but that can be overridden at configure time through
12289 the usual use of --prefix and/or --openssldir, and at run
12290 time with the environment variable OPENSSL_ENGINES.
12292 *Geoff Thorpe and Richard Levitte*
12294 * Add Makefile.shared, a helper makefile to build shared
12295 libraries. Adapt Makefile.org.
12299 * Add version info to Win32 DLLs.
12301 *Peter 'Luna' Runestig" <peter@runestig.com>*
12303 * Add new 'medium level' PKCS#12 API. Certificates and keys
12304 can be added using this API to created arbitrary PKCS#12
12305 files while avoiding the low-level API.
12307 New options to PKCS12_create(), key or cert can be NULL and
12308 will then be omitted from the output file. The encryption
12309 algorithm NIDs can be set to -1 for no encryption, the mac
12310 iteration count can be set to 0 to omit the mac.
12312 Enhance pkcs12 utility by making the -nokeys and -nocerts
12313 options work when creating a PKCS#12 file. New option -nomac
12314 to omit the mac, NONE can be set for an encryption algorithm.
12315 New code is modified to use the enhanced PKCS12_create()
12316 instead of the low-level API.
12320 * Extend ASN1 encoder to support indefinite length constructed
12321 encoding. This can output sequences tags and octet strings in
12322 this form. Modify pk7_asn1.c to support indefinite length
12323 encoding. This is experimental and needs additional code to
12324 be useful, such as an ASN1 bio and some enhanced streaming
12327 Extend template encode functionality so that tagging is passed
12328 down to the template encoder.
12332 * Let 'openssl req' fail if an argument to '-newkey' is not
12333 recognized instead of using RSA as a default.
12337 * Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt.
12338 As these are not official, they are not included in "ALL";
12339 the "ECCdraft" ciphersuite group alias can be used to select them.
12341 *Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)*
12343 * Add ECDH engine support.
12345 *Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)*
12347 * Add ECDH in new directory crypto/ecdh/.
12349 *Douglas Stebila (Sun Microsystems Laboratories)*
12351 * Let BN_rand_range() abort with an error after 100 iterations
12352 without success (which indicates a broken PRNG).
12356 * Change BN_mod_sqrt() so that it verifies that the input value
12357 is really the square of the return value. (Previously,
12358 BN_mod_sqrt would show GIGO behaviour.)
12362 * Add named elliptic curves over binary fields from X9.62, SECG,
12363 and WAP/WTLS; add OIDs that were still missing.
12365 *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
12367 * Extend the EC library for elliptic curves over binary fields
12368 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/).
12371 EC_GF2m_simple_method
12375 EC_GROUP_new_curve_GF2m
12376 EC_GROUP_set_curve_GF2m
12377 EC_GROUP_get_curve_GF2m
12378 EC_POINT_set_affine_coordinates_GF2m
12379 EC_POINT_get_affine_coordinates_GF2m
12380 EC_POINT_set_compressed_coordinates_GF2m
12382 Point compression for binary fields is disabled by default for
12383 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to
12386 As binary polynomials are represented as BIGNUMs, various members
12387 of the EC_GROUP and EC_POINT data structures can be shared
12388 between the implementations for prime fields and binary fields;
12389 the above `..._GF2m functions` (except for EX_GROUP_new_curve_GF2m)
12390 are essentially identical to their `..._GFp` counterparts.
12391 (For simplicity, the `..._GFp` prefix has been dropped from
12392 various internal method names.)
12394 An internal 'field_div' method (similar to 'field_mul' and
12395 'field_sqr') has been added; this is used only for binary fields.
12397 *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
12399 * Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult()
12400 through methods ('mul', 'precompute_mult').
12402 The generic implementations (now internally called 'ec_wNAF_mul'
12403 and 'ec_wNAF_precomputed_mult') remain the default if these
12404 methods are undefined.
12406 *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
12408 * New function EC_GROUP_get_degree, which is defined through
12409 EC_METHOD. For curves over prime fields, this returns the bit
12410 length of the modulus.
12412 *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
12414 * New functions EC_GROUP_dup, EC_POINT_dup.
12415 (These simply call ..._new and ..._copy).
12417 *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
12419 * Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c.
12420 Polynomials are represented as BIGNUMs (where the sign bit is not
12421 used) in the following functions [macros]:
12424 BN_GF2m_sub [= BN_GF2m_add]
12425 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr]
12426 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr]
12427 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr]
12429 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr]
12430 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr]
12431 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr]
12432 BN_GF2m_cmp [= BN_ucmp]
12434 (Note that only the 'mod' functions are actually for fields GF(2^m).
12435 BN_GF2m_add() is misnomer, but this is for the sake of consistency.)
12437 For some functions, an the irreducible polynomial defining a
12438 field can be given as an 'unsigned int[]' with strictly
12439 decreasing elements giving the indices of those bits that are set;
12440 i.e., p[] represents the polynomial
12441 f(t) = t^p[0] + t^p[1] + ... + t^p[k]
12443 p[0] > p[1] > ... > p[k] = 0.
12444 This applies to the following functions:
12447 BN_GF2m_mod_mul_arr
12448 BN_GF2m_mod_sqr_arr
12449 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv]
12450 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div]
12451 BN_GF2m_mod_exp_arr
12452 BN_GF2m_mod_sqrt_arr
12453 BN_GF2m_mod_solve_quad_arr
12457 Conversion can be performed by the following functions:
12462 bntest.c has additional tests for binary polynomial arithmetic.
12464 Two implementations for BN_GF2m_mod_div() are available.
12465 The default algorithm simply uses BN_GF2m_mod_inv() and
12466 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only
12467 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the
12468 copyright notice in crypto/bn/bn_gf2m.c before enabling it).
12470 *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
12472 * Add new error code 'ERR_R_DISABLED' that can be used when some
12473 functionality is disabled at compile-time.
12475 *Douglas Stebila <douglas.stebila@sun.com>*
12477 * Change default behaviour of 'openssl asn1parse' so that more
12478 information is visible when viewing, e.g., a certificate:
12480 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump'
12481 mode the content of non-printable OCTET STRINGs is output in a
12482 style similar to INTEGERs, but with '[HEX DUMP]' prepended to
12483 avoid the appearance of a printable string.
12485 *Nils Larsch <nla@trustcenter.de>*
12487 * Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access
12489 EC_GROUP_set_asn1_flag()
12490 EC_GROUP_get_asn1_flag()
12491 EC_GROUP_set_point_conversion_form()
12492 EC_GROUP_get_point_conversion_form()
12493 These control ASN1 encoding details:
12494 - Curves (i.e., groups) are encoded explicitly unless asn1_flag
12495 has been set to OPENSSL_EC_NAMED_CURVE.
12496 - Points are encoded in uncompressed form by default; options for
12497 asn1_for are as for point2oct, namely
12498 POINT_CONVERSION_COMPRESSED
12499 POINT_CONVERSION_UNCOMPRESSED
12500 POINT_CONVERSION_HYBRID
12502 Also add 'seed' and 'seed_len' members to EC_GROUP with access
12504 EC_GROUP_set_seed()
12505 EC_GROUP_get0_seed()
12506 EC_GROUP_get_seed_len()
12507 This is used only for ASN1 purposes (so far).
12509 *Nils Larsch <nla@trustcenter.de>*
12511 * Add 'field_type' member to EC_METHOD, which holds the NID
12512 of the appropriate field type OID. The new function
12513 EC_METHOD_get_field_type() returns this value.
12515 *Nils Larsch <nla@trustcenter.de>*
12518 EC_POINT_point2bn()
12519 EC_POINT_bn2point()
12520 EC_POINT_point2hex()
12521 EC_POINT_hex2point()
12522 providing useful interfaces to EC_POINT_point2oct() and
12523 EC_POINT_oct2point().
12525 *Nils Larsch <nla@trustcenter.de>*
12527 * Change internals of the EC library so that the functions
12528 EC_GROUP_set_generator()
12529 EC_GROUP_get_generator()
12530 EC_GROUP_get_order()
12531 EC_GROUP_get_cofactor()
12532 are implemented directly in crypto/ec/ec_lib.c and not dispatched
12533 to methods, which would lead to unnecessary code duplication when
12534 adding different types of curves.
12536 *Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller*
12538 * Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM
12539 arithmetic, and such that modified wNAFs are generated
12540 (which avoid length expansion in many cases).
12544 * Add a function EC_GROUP_check_discriminant() (defined via
12545 EC_METHOD) that verifies that the curve discriminant is non-zero.
12547 Add a function EC_GROUP_check() that makes some sanity tests
12548 on a EC_GROUP, its generator and order. This includes
12549 EC_GROUP_check_discriminant().
12551 *Nils Larsch <nla@trustcenter.de>*
12553 * Add ECDSA in new directory crypto/ecdsa/.
12555 Add applications 'openssl ecparam' and 'openssl ecdsa'
12556 (these are based on 'openssl dsaparam' and 'openssl dsa').
12558 ECDSA support is also included in various other files across the
12559 library. Most notably,
12560 - 'openssl req' now has a '-newkey ecdsa:file' option;
12561 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA;
12562 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and
12563 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make
12564 them suitable for ECDSA where domain parameters must be
12565 extracted before the specific public key;
12566 - ECDSA engine support has been added.
12568 *Nils Larsch <nla@trustcenter.de>*
12570 * Include some named elliptic curves, and add OIDs from X9.62,
12571 SECG, and WAP/WTLS. Each curve can be obtained from the new
12573 EC_GROUP_new_by_curve_name(),
12574 and the list of available named curves can be obtained with
12575 EC_get_builtin_curves().
12576 Also add a 'curve_name' member to EC_GROUP objects, which can be
12578 EC_GROUP_set_curve_name()
12579 EC_GROUP_get_curve_name()
12581 *Nils Larsch <larsch@trustcenter.de, Bodo Moeller*
12583 * Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
12584 was actually never needed) and in BN_mul(). The removal in BN_mul()
12585 required a small change in bn_mul_part_recursive() and the addition
12586 of the functions bn_cmp_part_words(), bn_sub_part_words() and
12587 bn_add_part_words(), which do the same thing as bn_cmp_words(),
12588 bn_sub_words() and bn_add_words() except they take arrays with
12593 ### Changes between 0.9.7l and 0.9.7m [23 Feb 2007]
12595 * Cleanse PEM buffers before freeing them since they may contain
12598 *Benjamin Bennett <ben@psc.edu>*
12600 * Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that
12601 a ciphersuite string such as "DEFAULT:RSA" cannot enable
12602 authentication-only ciphersuites.
12606 * Since AES128 and AES256 share a single mask bit in the logic of
12607 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a
12608 kludge to work properly if AES128 is available and AES256 isn't.
12612 * Expand security boundary to match 1.1.1 module.
12616 * Remove redundant features: hash file source, editing of test vectors
12617 modify fipsld to use external fips_premain.c signature.
12621 * New perl script mkfipsscr.pl to create shell scripts or batch files to
12622 run algorithm test programs.
12626 * Make algorithm test programs more tolerant of whitespace.
12630 * Have SSL/TLS server implementation tolerate "mismatched" record
12631 protocol version while receiving ClientHello even if the
12632 ClientHello is fragmented. (The server can't insist on the
12633 particular protocol version it has chosen before the ServerHello
12634 message has informed the client about his choice.)
12638 * Load error codes if they are not already present instead of using a
12639 static variable. This allows them to be cleanly unloaded and reloaded.
12643 ### Changes between 0.9.7k and 0.9.7l [28 Sep 2006]
12645 * Introduce limits to prevent malicious keys being able to
12646 cause a denial of service. ([CVE-2006-2940])
12648 *Steve Henson, Bodo Moeller*
12650 * Fix ASN.1 parsing of certain invalid structures that can result
12651 in a denial of service. ([CVE-2006-2937]) [Steve Henson]
12653 * Fix buffer overflow in SSL_get_shared_ciphers() function.
12654 ([CVE-2006-3738]) [Tavis Ormandy and Will Drewry, Google Security Team]
12656 * Fix SSL client code which could crash if connecting to a
12657 malicious SSLv2 server. ([CVE-2006-4343])
12659 *Tavis Ormandy and Will Drewry, Google Security Team*
12661 * Change ciphersuite string processing so that an explicit
12662 ciphersuite selects this one ciphersuite (so that "AES256-SHA"
12663 will no longer include "AES128-SHA"), and any other similar
12664 ciphersuite (same bitmap) from *other* protocol versions (so that
12665 "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the
12666 SSL 3.0/TLS 1.0 ciphersuite). This is a backport combining
12667 changes from 0.9.8b and 0.9.8d.
12671 ### Changes between 0.9.7j and 0.9.7k [05 Sep 2006]
12673 * Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher
12674 ([CVE-2006-4339]) [Ben Laurie and Google Security Team]
12676 * Change the Unix randomness entropy gathering to use poll() when
12677 possible instead of select(), since the latter has some
12678 undesirable limitations.
12680 *Darryl Miles via Richard Levitte and Bodo Moeller*
12682 * Disable rogue ciphersuites:
12684 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5")
12685 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5")
12686 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5")
12688 The latter two were purportedly from
12689 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really
12692 Also deactivate the remaining ciphersuites from
12693 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as
12694 unofficial, and the ID has long expired.
12698 * Fix RSA blinding Heisenbug (problems sometimes occurred on
12699 dual-core machines) and other potential thread-safety issues.
12703 ### Changes between 0.9.7i and 0.9.7j [04 May 2006]
12705 * Adapt fipsld and the build system to link against the validated FIPS
12706 module in FIPS mode.
12710 * Fixes for VC++ 2005 build under Windows.
12714 * Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make
12715 from a Windows bash shell such as MSYS. It is autodetected from the
12716 "config" script when run from a VC++ environment. Modify standard VC++
12717 build to use fipscanister.o from the GNU make build.
12721 ### Changes between 0.9.7h and 0.9.7i [14 Oct 2005]
12723 * Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS.
12724 The value now differs depending on if you build for FIPS or not.
12725 BEWARE! A program linked with a shared FIPSed libcrypto can't be
12726 safely run with a non-FIPSed libcrypto, as it may crash because of
12727 the difference induced by this change.
12731 ### Changes between 0.9.7g and 0.9.7h [11 Oct 2005]
12733 * Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
12734 (part of SSL_OP_ALL). This option used to disable the
12735 countermeasure against man-in-the-middle protocol-version
12736 rollback in the SSL 2.0 server implementation, which is a bad
12737 idea. ([CVE-2005-2969])
12739 *Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
12740 for Information Security, National Institute of Advanced Industrial
12741 Science and Technology [AIST, Japan)]*
12743 * Minimal support for X9.31 signatures and PSS padding modes. This is
12744 mainly for FIPS compliance and not fully integrated at this stage.
12748 * For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform
12749 the exponentiation using a fixed-length exponent. (Otherwise,
12750 the information leaked through timing could expose the secret key
12751 after many signatures; cf. Bleichenbacher's attack on DSA with
12756 * Make a new fixed-window mod_exp implementation the default for
12757 RSA, DSA, and DH private-key operations so that the sequence of
12758 squares and multiplies and the memory access pattern are
12759 independent of the particular secret key. This will mitigate
12760 cache-timing and potential related attacks.
12762 BN_mod_exp_mont_consttime() is the new exponentiation implementation,
12763 and this is automatically used by BN_mod_exp_mont() if the new flag
12764 BN_FLG_EXP_CONSTTIME is set for the exponent. RSA, DSA, and DH
12765 will use this BN flag for private exponents unless the flag
12766 RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or
12767 DH_FLAG_NO_EXP_CONSTTIME, respectively, is set.
12769 *Matthew D Wood (Intel Corp), with some changes by Bodo Moeller*
12771 * Change the client implementation for SSLv23_method() and
12772 SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0
12773 Client Hello message format if the SSL_OP_NO_SSLv2 option is set.
12774 (Previously, the SSL 2.0 backwards compatible Client Hello
12775 message format would be used even with SSL_OP_NO_SSLv2.)
12779 * Add support for smime-type MIME parameter in S/MIME messages which some
12784 * New function BN_MONT_CTX_set_locked() to set montgomery parameters in
12785 a threadsafe manner. Modify rsa code to use new function and add calls
12786 to dsa and dh code (which had race conditions before).
12790 * Include the fixed error library code in the C error file definitions
12791 instead of fixing them up at runtime. This keeps the error code
12792 structures constant.
12796 ### Changes between 0.9.7f and 0.9.7g [11 Apr 2005]
12798 [NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after
12801 * Fixes for newer kerberos headers. NB: the casts are needed because
12802 the 'length' field is signed on one version and unsigned on another
12803 with no (?) obvious way to tell the difference, without these VC++
12804 complains. Also the "definition" of FAR (blank) is no longer included
12805 nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up
12806 some needed definitions.
12810 * Undo Cygwin change.
12814 * Added support for proxy certificates according to RFC 3820.
12815 Because they may be a security thread to unaware applications,
12816 they must be explicitly allowed in run-time. See
12817 docs/HOWTO/proxy_certificates.txt for further information.
12821 ### Changes between 0.9.7e and 0.9.7f [22 Mar 2005]
12823 * Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating
12824 server and client random values. Previously
12825 (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in
12826 less random data when sizeof(time_t) > 4 (some 64 bit platforms).
12828 This change has negligible security impact because:
12830 1. Server and client random values still have 24 bytes of pseudo random
12833 2. Server and client random values are sent in the clear in the initial
12836 3. The master secret is derived using the premaster secret (48 bytes in
12837 size for static RSA ciphersuites) as well as client server and random
12840 The OpenSSL team would like to thank the UK NISCC for bringing this issue
12843 *Stephen Henson, reported by UK NISCC*
12845 * Use Windows randomness collection on Cygwin.
12849 * Fix hang in EGD/PRNGD query when communication socket is closed
12850 prematurely by EGD/PRNGD.
12852 *Darren Tucker <dtucker@zip.com.au> via Lutz Jänicke, resolves #1014*
12854 * Prompt for pass phrases when appropriate for PKCS12 input format.
12858 * Back-port of selected performance improvements from development
12859 branch, as well as improved support for PowerPC platforms.
12863 * Add lots of checks for memory allocation failure, error codes to indicate
12864 failure and freeing up memory if a failure occurs.
12866 *Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson*
12868 * Add new -passin argument to dgst.
12872 * Perform some character comparisons of different types in X509_NAME_cmp:
12873 this is needed for some certificates that re-encode DNs into UTF8Strings
12874 (in violation of RFC3280) and can't or won't issue name rollover
12879 * Make an explicit check during certificate validation to see that
12880 the CA setting in each certificate on the chain is correct. As a
12881 side effect always do the following basic checks on extensions,
12882 not just when there's an associated purpose to the check:
12884 - if there is an unhandled critical extension (unless the user
12885 has chosen to ignore this fault)
12886 - if the path length has been exceeded (if one is set at all)
12887 - that certain extensions fit the associated purpose (if one has
12892 ### Changes between 0.9.7d and 0.9.7e [25 Oct 2004]
12894 * Avoid a race condition when CRLs are checked in a multi threaded
12895 environment. This would happen due to the reordering of the revoked
12896 entries during signature checking and serial number lookup. Now the
12897 encoding is cached and the serial number sort performed under a lock.
12898 Add new STACK function sk_is_sorted().
12902 * Add Delta CRL to the extension code.
12906 * Various fixes to s3_pkt.c so alerts are sent properly.
12908 *David Holmes <d.holmes@f5.com>*
12910 * Reduce the chances of duplicate issuer name and serial numbers (in
12911 violation of RFC3280) using the OpenSSL certificate creation utilities.
12912 This is done by creating a random 64 bit value for the initial serial
12913 number when a serial number file is created or when a self signed
12914 certificate is created using 'openssl req -x509'. The initial serial
12915 number file is created using 'openssl x509 -next_serial' in CA.pl
12916 rather than being initialized to 1.
12920 ### Changes between 0.9.7c and 0.9.7d [17 Mar 2004]
12922 * Fix null-pointer assignment in do_change_cipher_spec() revealed
12923 by using the Codenomicon TLS Test Tool ([CVE-2004-0079])
12925 *Joe Orton, Steve Henson*
12927 * Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites
12930 *Joe Orton, Steve Henson*
12932 * Make it possible to have multiple active certificates with the same
12933 subject in the CA index file. This is done only if the keyword
12934 'unique_subject' is set to 'no' in the main CA section (default
12935 if 'CA_default') of the configuration file. The value is saved
12936 with the database itself in a separate index attribute file,
12937 named like the index file with '.attr' appended to the name.
12941 * X509 verify fixes. Disable broken certificate workarounds when
12942 X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if
12943 keyUsage extension present. Don't accept CRLs with unhandled critical
12944 extensions: since verify currently doesn't process CRL extensions this
12945 rejects a CRL with *any* critical extensions. Add new verify error codes
12950 * When creating an OCSP nonce use an OCTET STRING inside the extnValue.
12951 A clarification of RFC2560 will require the use of OCTET STRINGs and
12952 some implementations cannot handle the current raw format. Since OpenSSL
12953 copies and compares OCSP nonces as opaque blobs without any attempt at
12954 parsing them this should not create any compatibility issues.
12958 * New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when
12959 calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without
12960 this HMAC (and other) operations are several times slower than OpenSSL
12965 * Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex().
12967 *Peter Sylvester <Peter.Sylvester@EdelWeb.fr>*
12969 * Use the correct content when signing type "other".
12973 ### Changes between 0.9.7b and 0.9.7c [30 Sep 2003]
12975 * Fix various bugs revealed by running the NISCC test suite:
12977 Stop out of bounds reads in the ASN1 code when presented with
12978 invalid tags (CVE-2003-0543 and CVE-2003-0544).
12980 Free up ASN1_TYPE correctly if ANY type is invalid ([CVE-2003-0545]).
12982 If verify callback ignores invalid public key errors don't try to check
12983 certificate signature with the NULL public key.
12987 * New -ignore_err option in ocsp application to stop the server
12988 exiting on the first error in a request.
12992 * In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
12993 if the server requested one: as stated in TLS 1.0 and SSL 3.0
12998 * In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
12999 extra data after the compression methods not only for TLS 1.0
13000 but also for SSL 3.0 (as required by the specification).
13002 *Bodo Moeller; problem pointed out by Matthias Loepfe*
13004 * Change X509_certificate_type() to mark the key as exported/exportable
13005 when it's 512 *bits* long, not 512 bytes.
13009 * Change AES_cbc_encrypt() so it outputs exact multiple of
13010 blocks during encryption.
13014 * Various fixes to base64 BIO and non blocking I/O. On write
13015 flushes were not handled properly if the BIO retried. On read
13016 data was not being buffered properly and had various logic bugs.
13017 This also affects blocking I/O when the data being decoded is a
13022 * Various S/MIME bugfixes and compatibility changes:
13023 output correct application/pkcs7 MIME type if
13024 PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures.
13025 Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening
13026 of files as .eml work). Correctly handle very long lines in MIME
13031 ### Changes between 0.9.7a and 0.9.7b [10 Apr 2003]
13033 * Countermeasure against the Klima-Pokorny-Rosa extension of
13034 Bleichbacher's attack on PKCS #1 v1.5 padding: treat
13035 a protocol version number mismatch like a decryption error
13036 in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
13040 * Turn on RSA blinding by default in the default implementation
13041 to avoid a timing attack. Applications that don't want it can call
13042 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
13043 They would be ill-advised to do so in most cases.
13045 *Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller*
13047 * Change RSA blinding code so that it works when the PRNG is not
13048 seeded (in this case, the secret RSA exponent is abused as
13049 an unpredictable seed -- if it is not unpredictable, there
13050 is no point in blinding anyway). Make RSA blinding thread-safe
13051 by remembering the creator's thread ID in rsa->blinding and
13052 having all other threads use local one-time blinding factors
13053 (this requires more computation than sharing rsa->blinding, but
13054 avoids excessive locking; and if an RSA object is not shared
13055 between threads, blinding will still be very fast).
13059 * Fixed a typo bug that would cause ENGINE_set_default() to set an
13060 ENGINE as defaults for all supported algorithms irrespective of
13061 the 'flags' parameter. 'flags' is now honoured, so applications
13062 should make sure they are passing it correctly.
13066 * Target "mingw" now allows native Windows code to be generated in
13067 the Cygwin environment as well as with the MinGW compiler.
13071 ### Changes between 0.9.7 and 0.9.7a [19 Feb 2003]
13073 * In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
13074 via timing by performing a MAC computation even if incorrect
13075 block cipher padding has been found. This is a countermeasure
13076 against active attacks where the attacker has to distinguish
13077 between bad padding and a MAC verification error. ([CVE-2003-0078])
13079 *Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
13080 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
13081 Martin Vuagnoux (EPFL, Ilion)*
13083 * Make the no-err option work as intended. The intention with no-err
13084 is not to have the whole error stack handling routines removed from
13085 libcrypto, it's only intended to remove all the function name and
13086 reason texts, thereby removing some of the footprint that may not
13087 be interesting if those errors aren't displayed anyway.
13089 NOTE: it's still possible for any application or module to have its
13090 own set of error texts inserted. The routines are there, just not
13091 used by default when no-err is given.
13095 * Add support for FreeBSD on IA64.
13097 *dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454*
13099 * Adjust DES_cbc_cksum() so it returns the same value as the MIT
13100 Kerberos function mit_des_cbc_cksum(). Before this change,
13101 the value returned by DES_cbc_cksum() was like the one from
13102 mit_des_cbc_cksum(), except the bytes were swapped.
13104 *Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte*
13106 * Allow an application to disable the automatic SSL chain building.
13107 Before this a rather primitive chain build was always performed in
13108 ssl3_output_cert_chain(): an application had no way to send the
13109 correct chain if the automatic operation produced an incorrect result.
13111 Now the chain builder is disabled if either:
13113 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert().
13115 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set.
13117 The reasoning behind this is that an application would not want the
13118 auto chain building to take place if extra chain certificates are
13119 present and it might also want a means of sending no additional
13120 certificates (for example the chain has two certificates and the
13125 * Add the possibility to build without the ENGINE framework.
13127 *Steven Reddie <smr@essemer.com.au> via Richard Levitte*
13129 * Under Win32 gmtime() can return NULL: check return value in
13130 OPENSSL_gmtime(). Add error code for case where gmtime() fails.
13134 * DSA routines: under certain error conditions uninitialized BN objects
13135 could be freed. Solution: make sure initialization is performed early
13136 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>,
13137 Nils Larsch <nla@trustcenter.de> via PR#459)
13141 * Another fix for SSLv2 session ID handling: the session ID was incorrectly
13142 checked on reconnect on the client side, therefore session resumption
13143 could still fail with a "ssl session id is different" error. This
13144 behaviour is masked when SSL_OP_ALL is used due to
13145 SSL_OP_MICROSOFT_SESS_ID_BUG being set.
13146 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
13147 followup to PR #377.
13151 * IA-32 assembler support enhancements: unified ELF targets, support
13152 for SCO/Caldera platforms, fix for Cygwin shared build.
13156 * Add support for FreeBSD on sparc64. As a consequence, support for
13157 FreeBSD on non-x86 processors is separate from x86 processors on
13158 the config script, much like the NetBSD support.
13160 *Richard Levitte & Kris Kennaway <kris@obsecurity.org>*
13162 ### Changes between 0.9.6h and 0.9.7 [31 Dec 2002]
13164 [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after
13167 * Fix session ID handling in SSLv2 client code: the SERVER FINISHED
13168 code (06) was taken as the first octet of the session ID and the last
13169 octet was ignored consequently. As a result SSLv2 client side session
13170 caching could not have worked due to the session ID mismatch between
13172 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
13177 * Change the declaration of needed Kerberos libraries to use EX_LIBS
13178 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is
13183 * The hw_ncipher.c engine requires dynamic locks. Unfortunately, it
13184 seems that in spite of existing for more than a year, many application
13185 author have done nothing to provide the necessary callbacks, which
13186 means that this particular engine will not work properly anywhere.
13187 This is a very unfortunate situation which forces us, in the name
13188 of usability, to give the hw_ncipher.c a static lock, which is part
13190 NOTE: This is for the 0.9.7 series ONLY. This hack will never
13191 appear in 0.9.8 or later. We EXPECT application authors to have
13192 dealt properly with this when 0.9.8 is released (unless we actually
13193 make such changes in the libcrypto locking code that changes will
13194 have to be made anyway).
13198 * In asn1_d2i_read_bio() repeatedly call BIO_read() until all content
13199 octets have been read, EOF or an error occurs. Without this change
13200 some truncated ASN1 structures will not produce an error.
13204 * Disable Heimdal support, since it hasn't been fully implemented.
13205 Still give the possibility to force the use of Heimdal, but with
13206 warnings and a request that patches get sent to openssl-dev.
13210 * Add the VC-CE target, introduce the WINCE sysname, and add
13211 INSTALL.WCE and appropriate conditionals to make it build.
13213 *Steven Reddie <smr@essemer.com.au> via Richard Levitte*
13215 * Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and
13216 cygssl-x.y.z.dll, where x, y and z are the major, minor and
13217 edit numbers of the version.
13219 *Corinna Vinschen <vinschen@redhat.com> and Richard Levitte*
13221 * Introduce safe string copy and catenation functions
13222 (BUF_strlcpy() and BUF_strlcat()).
13224 *Ben Laurie (CHATS) and Richard Levitte*
13226 * Avoid using fixed-size buffers for one-line DNs.
13228 *Ben Laurie (CHATS)*
13230 * Add BUF_MEM_grow_clean() to avoid information leakage when
13231 resizing buffers containing secrets, and use where appropriate.
13233 *Ben Laurie (CHATS)*
13235 * Avoid using fixed size buffers for configuration file location.
13237 *Ben Laurie (CHATS)*
13239 * Avoid filename truncation for various CA files.
13241 *Ben Laurie (CHATS)*
13243 * Use sizeof in preference to magic numbers.
13245 *Ben Laurie (CHATS)*
13247 * Avoid filename truncation in cert requests.
13249 *Ben Laurie (CHATS)*
13251 * Add assertions to check for (supposedly impossible) buffer
13254 *Ben Laurie (CHATS)*
13256 * Don't cache truncated DNS entries in the local cache (this could
13257 potentially lead to a spoofing attack).
13259 *Ben Laurie (CHATS)*
13261 * Fix various buffers to be large enough for hex/decimal
13262 representations in a platform independent manner.
13264 *Ben Laurie (CHATS)*
13266 * Add CRYPTO_realloc_clean() to avoid information leakage when
13267 resizing buffers containing secrets, and use where appropriate.
13269 *Ben Laurie (CHATS)*
13271 * Add BIO_indent() to avoid much slightly worrying code to do
13274 *Ben Laurie (CHATS)*
13276 * Convert sprintf()/BIO_puts() to BIO_printf().
13278 *Ben Laurie (CHATS)*
13280 * buffer_gets() could terminate with the buffer only half
13283 *Ben Laurie (CHATS)*
13285 * Add assertions to prevent user-supplied crypto functions from
13286 overflowing internal buffers by having large block sizes, etc.
13288 *Ben Laurie (CHATS)*
13290 * New OPENSSL_assert() macro (similar to assert(), but enabled
13293 *Ben Laurie (CHATS)*
13295 * Eliminate unused copy of key in RC4.
13297 *Ben Laurie (CHATS)*
13299 * Eliminate unused and incorrectly sized buffers for IV in pem.h.
13301 *Ben Laurie (CHATS)*
13303 * Fix off-by-one error in EGD path.
13305 *Ben Laurie (CHATS)*
13307 * If RANDFILE path is too long, ignore instead of truncating.
13309 *Ben Laurie (CHATS)*
13311 * Eliminate unused and incorrectly sized X.509 structure
13314 *Ben Laurie (CHATS)*
13316 * Eliminate unused and dangerous function knumber().
13318 *Ben Laurie (CHATS)*
13320 * Eliminate unused and dangerous structure, KSSL_ERR.
13322 *Ben Laurie (CHATS)*
13324 * Protect against overlong session ID context length in an encoded
13325 session object. Since these are local, this does not appear to be
13328 *Ben Laurie (CHATS)*
13330 * Change from security patch (see 0.9.6e below) that did not affect
13331 the 0.9.6 release series:
13333 Remote buffer overflow in SSL3 protocol - an attacker could
13334 supply an oversized master key in Kerberos-enabled versions.
13337 *Ben Laurie (CHATS)*
13339 * Change the SSL kerb5 codes to match RFC 2712.
13343 * Make -nameopt work fully for req and add -reqopt switch.
13345 *Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson*
13347 * The "block size" for block ciphers in CFB and OFB mode should be 1.
13349 *Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>*
13351 * Make sure tests can be performed even if the corresponding algorithms
13352 have been removed entirely. This was also the last step to make
13353 OpenSSL compilable with DJGPP under all reasonable conditions.
13355 *Richard Levitte, Doug Kaufman <dkaufman@rahul.net>*
13357 * Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT
13358 to allow version independent disabling of normally unselected ciphers,
13359 which may be activated as a side-effect of selecting a single cipher.
13361 (E.g., cipher list string "RSA" enables ciphersuites that are left
13362 out of "ALL" because they do not provide symmetric encryption.
13363 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.)
13365 *Lutz Jaenicke, Bodo Moeller*
13367 * Add appropriate support for separate platform-dependent build
13368 directories. The recommended way to make a platform-dependent
13369 build directory is the following (tested on Linux), maybe with
13372 # Place yourself outside of the OpenSSL source tree. In
13373 # this example, the environment variable OPENSSL_SOURCE
13374 # is assumed to contain the absolute OpenSSL source directory.
13375 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`"
13376 cd objtree/"`uname -s`-`uname -r`-`uname -m`"
13377 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do
13378 mkdir -p `dirname $F`
13379 ln -s $OPENSSL_SOURCE/$F $F
13382 To be absolutely sure not to disturb the source tree, a "make clean"
13383 is a good thing. If it isn't successful, don't worry about it,
13384 it probably means the source directory is very clean.
13388 * Make sure any ENGINE control commands make local copies of string
13389 pointers passed to them whenever necessary. Otherwise it is possible
13390 the caller may have overwritten (or deallocated) the original string
13391 data when a later ENGINE operation tries to use the stored values.
13393 *Götz Babin-Ebell <babinebell@trustcenter.de>*
13395 * Improve diagnostics in file reading and command-line digests.
13397 *Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>*
13399 * Add AES modes CFB and OFB to the object database. Correct an
13400 error in AES-CFB decryption.
13404 * Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this
13405 allows existing EVP_CIPHER_CTX structures to be reused after
13406 calling `EVP_*Final()`. This behaviour is used by encryption
13407 BIOs and some applications. This has the side effect that
13408 applications must explicitly clean up cipher contexts with
13409 EVP_CIPHER_CTX_cleanup() or they will leak memory.
13413 * Check the values of dna and dnb in bn_mul_recursive before calling
13414 bn_mul_comba (a non zero value means the a or b arrays do not contain
13415 n2 elements) and fallback to bn_mul_normal if either is not zero.
13419 * Fix escaping of non-ASCII characters when using the -subj option
13420 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>)
13424 * Make object definitions compliant to LDAP (RFC2256): SN is the short
13425 form for "surname", serialNumber has no short form.
13426 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798;
13427 therefore remove "mail" short name for "internet 7".
13428 The OID for unique identifiers in X509 certificates is
13429 x500UniqueIdentifier, not uniqueIdentifier.
13430 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>)
13434 * Add an "init" command to the ENGINE config module and auto initialize
13435 ENGINEs. Without any "init" command the ENGINE will be initialized
13436 after all ctrl commands have been executed on it. If init=1 the
13437 ENGINE is initialized at that point (ctrls before that point are run
13438 on the uninitialized ENGINE and after on the initialized one). If
13439 init=0 then the ENGINE will not be initialized at all.
13443 * Fix the 'app_verify_callback' interface so that the user-defined
13444 argument is actually passed to the callback: In the
13445 SSL_CTX_set_cert_verify_callback() prototype, the callback
13446 declaration has been changed from
13449 int (*cb)(X509_STORE_CTX *,void *);
13450 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call
13451 i=s->ctx->app_verify_callback(&ctx)
13452 has been changed into
13453 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg).
13455 To update applications using SSL_CTX_set_cert_verify_callback(),
13456 a dummy argument can be added to their callback functions.
13458 *D. K. Smetters <smetters@parc.xerox.com>*
13460 * Added the '4758cca' ENGINE to support IBM 4758 cards.
13462 *Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe*
13464 * Add and OPENSSL_LOAD_CONF define which will cause
13465 OpenSSL_add_all_algorithms() to load the openssl.cnf config file.
13466 This allows older applications to transparently support certain
13467 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading.
13468 Two new functions OPENSSL_add_all_algorithms_noconf() which will never
13469 load the config file and OPENSSL_add_all_algorithms_conf() which will
13470 always load it have also been added.
13474 * Add the OFB, CFB and CTR (all with 128 bit feedback) to AES.
13475 Adjust NIDs and EVP layer.
13477 *Stephen Sprunk <stephen@sprunk.org> and Richard Levitte*
13479 * Config modules support in openssl utility.
13481 Most commands now load modules from the config file,
13482 though in a few (such as version) this isn't done
13483 because it couldn't be used for anything.
13485 In the case of ca and req the config file used is
13486 the same as the utility itself: that is the -config
13487 command line option can be used to specify an
13492 * Move default behaviour from OPENSSL_config(). If appname is NULL
13493 use "openssl_conf" if filename is NULL use default openssl config file.
13497 * Add an argument to OPENSSL_config() to allow the use of an alternative
13498 config section name. Add a new flag to tolerate a missing config file
13499 and move code to CONF_modules_load_file().
13503 * Support for crypto accelerator cards from Accelerated Encryption
13504 Processing, www.aep.ie. (Use engine 'aep')
13505 The support was copied from 0.9.6c [engine] and adapted/corrected
13506 to work with the new engine framework.
13508 *AEP Inc. and Richard Levitte*
13510 * Support for SureWare crypto accelerator cards from Baltimore
13511 Technologies. (Use engine 'sureware')
13512 The support was copied from 0.9.6c [engine] and adapted
13513 to work with the new engine framework.
13517 * Have the CHIL engine fork-safe (as defined by nCipher) and actually
13518 make the newer ENGINE framework commands for the CHIL engine work.
13520 *Toomas Kiisk <vix@cyber.ee> and Richard Levitte*
13522 * Make it possible to produce shared libraries on ReliantUNIX.
13524 *Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte*
13526 * Add the configuration target debug-linux-ppro.
13527 Make 'openssl rsa' use the general key loading routines
13528 implemented in `apps.c`, and make those routines able to
13529 handle the key format FORMAT_NETSCAPE and the variant
13532 *Toomas Kiisk <vix@cyber.ee> via Richard Levitte*
13534 * Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
13536 *Toomas Kiisk <vix@cyber.ee> via Richard Levitte*
13538 * Add -keyform to rsautl, and document -engine.
13540 *Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>*
13542 * Change BIO_new_file (crypto/bio/bss_file.c) to use new
13543 BIO_R_NO_SUCH_FILE error code rather than the generic
13544 ERR_R_SYS_LIB error code if fopen() fails with ENOENT.
13548 * Add new functions
13549 ERR_peek_last_error
13550 ERR_peek_last_error_line
13551 ERR_peek_last_error_line_data.
13552 These are similar to
13554 ERR_peek_error_line
13555 ERR_peek_error_line_data,
13556 but report on the latest error recorded rather than the first one
13557 still in the error queue.
13559 *Ben Laurie, Bodo Moeller*
13561 * default_algorithms option in ENGINE config module. This allows things
13563 default_algorithms = ALL
13564 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS
13568 * Preliminary ENGINE config module.
13572 * New experimental application configuration code.
13576 * Change the AES code to follow the same name structure as all other
13577 symmetric ciphers, and behave the same way. Move everything to
13578 the directory crypto/aes, thereby obsoleting crypto/rijndael.
13580 *Stephen Sprunk <stephen@sprunk.org> and Richard Levitte*
13582 * SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c.
13584 *Ben Laurie and Theo de Raadt*
13586 * Add option to output public keys in req command.
13588 *Massimiliano Pala madwolf@openca.org*
13590 * Use wNAFs in EC_POINTs_mul() for improved efficiency
13591 (up to about 10% better than before for P-192 and P-224).
13595 * New functions/macros
13597 SSL_CTX_set_msg_callback(ctx, cb)
13598 SSL_CTX_set_msg_callback_arg(ctx, arg)
13599 SSL_set_msg_callback(ssl, cb)
13600 SSL_set_msg_callback_arg(ssl, arg)
13602 to request calling a callback function
13604 void cb(int write_p, int version, int content_type,
13605 const void *buf, size_t len, SSL *ssl, void *arg)
13607 whenever a protocol message has been completely received
13608 (write_p == 0) or sent (write_p == 1). Here 'version' is the
13609 protocol version according to which the SSL library interprets
13610 the current protocol message (SSL2_VERSION, SSL3_VERSION, or
13611 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or
13612 the content type as defined in the SSL 3.0/TLS 1.0 protocol
13613 specification (change_cipher_spec(20), alert(21), handshake(22)).
13614 'buf' and 'len' point to the actual message, 'ssl' to the
13615 SSL object, and 'arg' is the application-defined value set by
13616 SSL[_CTX]_set_msg_callback_arg().
13618 'openssl s_client' and 'openssl s_server' have new '-msg' options
13619 to enable a callback that displays all protocol messages.
13623 * Change the shared library support so shared libraries are built as
13624 soon as the corresponding static library is finished, and thereby get
13625 openssl and the test programs linked against the shared library.
13626 This still only happens when the keyword "shard" has been given to
13627 the configuration scripts.
13629 NOTE: shared library support is still an experimental thing, and
13630 backward binary compatibility is still not guaranteed.
13632 *"Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte*
13634 * Add support for Subject Information Access extension.
13636 *Peter Sylvester <Peter.Sylvester@EdelWeb.fr>*
13638 * Make BUF_MEM_grow() behaviour more consistent: Initialise to zero
13639 additional bytes when new memory had to be allocated, not just
13640 when reusing an existing buffer.
13644 * New command line and configuration option 'utf8' for the req command.
13645 This allows field values to be specified as UTF8 strings.
13649 * Add -multi and -mr options to "openssl speed" - giving multiple parallel
13650 runs for the former and machine-readable output for the latter.
13654 * Add '-noemailDN' option to 'openssl ca'. This prevents inclusion
13655 of the e-mail address in the DN (i.e., it will go into a certificate
13656 extension only). The new configuration file option 'email_in_dn = no'
13657 has the same effect.
13659 *Massimiliano Pala madwolf@openca.org*
13661 * Change all functions with names starting with `des_` to be starting
13662 with `DES_` instead. Add wrappers that are compatible with libdes,
13663 but are named `_ossl_old_des_*`. Finally, add macros that map the
13664 `des_*` symbols to the corresponding `_ossl_old_des_*` if libdes
13665 compatibility is desired. If OpenSSL 0.9.6c compatibility is
13666 desired, the `des_*` symbols will be mapped to `DES_*`, with one
13669 Since we provide two compatibility mappings, the user needs to
13670 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes
13671 compatibility is desired. The default (i.e., when that macro
13672 isn't defined) is OpenSSL 0.9.6c compatibility.
13674 There are also macros that enable and disable the support of old
13675 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT
13676 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those
13677 are defined, the default will apply: to support the old des routines.
13679 In either case, one must include openssl/des.h to get the correct
13680 definitions. Do not try to just include openssl/des_old.h, that
13683 NOTE: This is a major break of an old API into a new one. Software
13684 authors are encouraged to switch to the `DES_` style functions. Some
13685 time in the future, des_old.h and the libdes compatibility functions
13686 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the
13687 default), and then completely removed.
13691 * Test for certificates which contain unsupported critical extensions.
13692 If such a certificate is found during a verify operation it is
13693 rejected by default: this behaviour can be overridden by either
13694 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or
13695 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function
13696 X509_supported_extension() has also been added which returns 1 if a
13697 particular extension is supported.
13701 * Modify the behaviour of EVP cipher functions in similar way to digests
13702 to retain compatibility with existing code.
13706 * Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain
13707 compatibility with existing code. In particular the 'ctx' parameter does
13708 not have to be to be initialized before the call to EVP_DigestInit() and
13709 it is tidied up after a call to EVP_DigestFinal(). New function
13710 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function
13711 EVP_MD_CTX_copy() changed to not require the destination to be
13712 initialized valid and new function EVP_MD_CTX_copy_ex() added which
13713 requires the destination to be valid.
13715 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(),
13716 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex().
13720 * Change ssl3_get_message (ssl/s3_both.c) and the functions using it
13721 so that complete 'Handshake' protocol structures are kept in memory
13722 instead of overwriting 'msg_type' and 'length' with 'body' data.
13726 * Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32.
13728 *Massimo Santin via Richard Levitte*
13730 * Major restructuring to the underlying ENGINE code. This includes
13731 reduction of linker bloat, separation of pure "ENGINE" manipulation
13732 (initialisation, etc) from functionality dealing with implementations
13733 of specific crypto interfaces. This change also introduces integrated
13734 support for symmetric ciphers and digest implementations - so ENGINEs
13735 can now accelerate these by providing EVP_CIPHER and EVP_MD
13736 implementations of their own. This is detailed in
13737 [crypto/engine/README.md](crypto/engine/README.md)
13738 as it couldn't be adequately described here. However, there are a few
13739 API changes worth noting - some RSA, DSA, DH, and RAND functions that
13740 were changed in the original introduction of ENGINE code have now
13741 reverted back - the hooking from this code to ENGINE is now a good
13742 deal more passive and at run-time, operations deal directly with
13743 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than
13744 dereferencing through an ENGINE pointer any more. Also, the ENGINE
13745 functions dealing with `BN_MOD_EXP[_CRT]` handlers have been removed -
13746 they were not being used by the framework as there is no concept of a
13747 BIGNUM_METHOD and they could not be generalised to the new
13748 'ENGINE_TABLE' mechanism that underlies the new code. Similarly,
13749 ENGINE_cpy() has been removed as it cannot be consistently defined in
13754 * Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds.
13758 * Change mkdef.pl to sort symbols that get the same entry number,
13759 and make sure the automatically generated functions `ERR_load_*`
13760 become part of libeay.num as well.
13764 * New function SSL_renegotiate_pending(). This returns true once
13765 renegotiation has been requested (either SSL_renegotiate() call
13766 or HelloRequest/ClientHello received from the peer) and becomes
13767 false once a handshake has been completed.
13768 (For servers, SSL_renegotiate() followed by SSL_do_handshake()
13769 sends a HelloRequest, but does not ensure that a handshake takes
13770 place. SSL_renegotiate_pending() is useful for checking if the
13771 client has followed the request.)
13775 * New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION.
13776 By default, clients may request session resumption even during
13777 renegotiation (if session ID contexts permit); with this option,
13778 session resumption is possible only in the first handshake.
13780 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes
13781 more bits available for options that should not be part of
13782 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION).
13786 * Add some demos for certificate and certificate request creation.
13790 * Make maximum certificate chain size accepted from the peer application
13791 settable (`SSL*_get/set_max_cert_list()`), as proposed by
13792 "Douglas E. Engert" <deengert@anl.gov>.
13796 * Add support for shared libraries for Unixware-7
13797 (Boyd Lynn Gerber <gerberb@zenez.com>).
13801 * Add a "destroy" handler to ENGINEs that allows structural cleanup to
13802 be done prior to destruction. Use this to unload error strings from
13803 ENGINEs that load their own error strings. NB: This adds two new API
13804 functions to "get" and "set" this destroy handler in an ENGINE.
13808 * Alter all existing ENGINE implementations (except "openssl" and
13809 "openbsd") to dynamically instantiate their own error strings. This
13810 makes them more flexible to be built both as statically-linked ENGINEs
13811 and self-contained shared-libraries loadable via the "dynamic" ENGINE.
13812 Also, add stub code to each that makes building them as self-contained
13813 shared-libraries easier (see [README-Engine.md](README-Engine.md)).
13817 * Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE
13818 implementations into applications that are completely implemented in
13819 self-contained shared-libraries. The "dynamic" ENGINE exposes control
13820 commands that can be used to configure what shared-library to load and
13821 to control aspects of the way it is handled. Also, made an update to
13822 the [README-Engine.md](README-Engine.md) file
13823 that brings its information up-to-date and
13824 provides some information and instructions on the "dynamic" ENGINE
13825 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc).
13829 * Make it possible to unload ranges of ERR strings with a new
13830 "ERR_unload_strings" function.
13834 * Add a copy() function to EVP_MD.
13838 * Make EVP_MD routines take a context pointer instead of just the
13839 md_data void pointer.
13843 * Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
13844 that the digest can only process a single chunk of data
13845 (typically because it is provided by a piece of
13846 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
13847 is only going to provide a single chunk of data, and hence the
13848 framework needn't accumulate the data for oneshot drivers.
13852 * As with "ERR", make it possible to replace the underlying "ex_data"
13853 functions. This change also alters the storage and management of global
13854 ex_data state - it's now all inside ex_data.c and all "class" code (eg.
13855 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
13856 index counters. The API functions that use this state have been changed
13857 to take a "class_index" rather than pointers to the class's local STACK
13858 and counter, and there is now an API function to dynamically create new
13859 classes. This centralisation allows us to (a) plug a lot of the
13860 thread-safety problems that existed, and (b) makes it possible to clean
13861 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
13862 such data would previously have always leaked in application code and
13863 workarounds were in place to make the memory debugging turn a blind eye
13864 to it. Application code that doesn't use this new function will still
13865 leak as before, but their memory debugging output will announce it now
13866 rather than letting it slide.
13868 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
13869 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
13870 has a return value to indicate success or failure.
13874 * Make it possible to replace the underlying "ERR" functions such that the
13875 global state (2 LHASH tables and 2 locks) is only used by the "default"
13876 implementation. This change also adds two functions to "get" and "set"
13877 the implementation prior to it being automatically set the first time
13878 any other ERR function takes place. Ie. an application can call "get",
13879 pass the return value to a module it has just loaded, and that module
13880 can call its own "set" function using that value. This means the
13881 module's "ERR" operations will use (and modify) the error state in the
13882 application and not in its own statically linked copy of OpenSSL code.
13886 * Give DH, DSA, and RSA types their own `*_up_ref()` function to increment
13887 reference counts. This performs normal REF_PRINT/REF_CHECK macros on
13888 the operation, and provides a more encapsulated way for external code
13889 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
13890 to use these functions rather than manually incrementing the counts.
13892 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()".
13896 * Add EVP test program.
13900 * Add symmetric cipher support to ENGINE. Expect the API to change!
13904 * New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
13905 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
13906 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
13907 These allow a CRL to be built without having to access X509_CRL fields
13908 directly. Modify 'ca' application to use new functions.
13912 * Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
13913 bug workarounds. Rollback attack detection is a security feature.
13914 The problem will only arise on OpenSSL servers when TLSv1 is not
13915 available (sslv3_server_method() or SSL_OP_NO_TLSv1).
13916 Software authors not wanting to support TLSv1 will have special reasons
13917 for their choice and can explicitly enable this option.
13919 *Bodo Moeller, Lutz Jaenicke*
13921 * Rationalise EVP so it can be extended: don't include a union of
13922 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX
13923 (similar to those existing for EVP_CIPHER_CTX).
13928 EVP_MD_CTX_init(&md); /* new function call */
13929 EVP_DigestInit(&md, EVP_sha1());
13930 EVP_DigestUpdate(&md, in, len);
13931 EVP_DigestFinal(&md, out, NULL);
13932 EVP_MD_CTX_cleanup(&md); /* new function call */
13936 * Make DES key schedule conform to the usual scheme, as well as
13937 correcting its structure. This means that calls to DES functions
13938 now have to pass a pointer to a des_key_schedule instead of a
13939 plain des_key_schedule (which was actually always a pointer
13942 des_key_schedule ks;
13944 des_set_key_checked(..., &ks);
13945 des_ncbc_encrypt(..., &ks, ...);
13947 (Note that a later change renames 'des_...' into 'DES_...'.)
13951 * Initial reduction of linker bloat: the use of some functions, such as
13952 PEM causes large amounts of unused functions to be linked in due to
13953 poor organisation. For example pem_all.c contains every PEM function
13954 which has a knock on effect of linking in large amounts of (unused)
13955 ASN1 code. Grouping together similar functions and splitting unrelated
13956 functions prevents this.
13960 * Cleanup of EVP macros.
13964 * Change historical references to `{NID,SN,LN}_des_ede` and ede3 to add the
13965 correct `_ecb suffix`.
13969 * Add initial OCSP responder support to ocsp application. The
13970 revocation information is handled using the text based index
13971 use by the ca application. The responder can either handle
13972 requests generated internally, supplied in files (for example
13973 via a CGI script) or using an internal minimal server.
13977 * Add configuration choices to get zlib compression for TLS.
13981 * Changes to Kerberos SSL for RFC 2712 compliance:
13982 1. Implemented real KerberosWrapper, instead of just using
13983 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
13984 2. Implemented optional authenticator field of KerberosWrapper.
13986 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
13987 and authenticator structs; see crypto/krb5/.
13989 Generalized Kerberos calls to support multiple Kerberos libraries.
13990 *Vern Staats <staatsvr@asc.hpc.mil>, Jeffrey Altman <jaltman@columbia.edu>
13991 via Richard Levitte*
13993 * Cause 'openssl speed' to use fully hard-coded DSA keys as it
13994 already does with RSA. testdsa.h now has 'priv_key/pub_key'
13995 values for each of the key sizes rather than having just
13996 parameters (and 'speed' generating keys each time).
14000 * Speed up EVP routines.
14003 pe 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes
14004 s-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k
14005 s-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k
14006 s-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k
14008 s-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k
14009 s-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k
14010 s-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k
14013 s-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k
14015 s-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k
14019 * Added the OS2-EMX target.
14021 *"Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte*
14023 * Rewrite commands to use `NCONF` routines instead of the old `CONF`.
14024 New functions to support `NCONF` routines in extension code.
14025 New function `CONF_set_nconf()`
14026 to allow functions which take an `NCONF` to also handle the old `LHASH`
14027 structure: this means that the old `CONF` compatible routines can be
14028 retained (in particular w.rt. extensions) without having to duplicate the
14029 code. New function `X509V3_add_ext_nconf_sk()` to add extensions to a stack.
14033 * Enhance the general user interface with mechanisms for inner control
14034 and with possibilities to have yes/no kind of prompts.
14038 * Change all calls to low-level digest routines in the library and
14039 applications to use EVP. Add missing calls to HMAC_cleanup() and
14040 don't assume HMAC_CTX can be copied using memcpy().
14042 *Verdon Walker <VWalker@novell.com>, Steve Henson*
14044 * Add the possibility to control engines through control names but with
14045 arbitrary arguments instead of just a string.
14046 Change the key loaders to take a UI_METHOD instead of a callback
14047 function pointer. NOTE: this breaks binary compatibility with earlier
14048 versions of OpenSSL [engine].
14049 Adapt the nCipher code for these new conditions and add a card insertion
14054 * Enhance the general user interface with mechanisms to better support
14055 dialog box interfaces, application-defined prompts, the possibility
14056 to use defaults (for example default passwords from somewhere else)
14057 and interrupts/cancellations.
14061 * Tidy up PKCS#12 attribute handling. Add support for the CSP name
14062 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
14066 * Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
14067 tidy up some unnecessarily weird code in 'sk_new()').
14069 *Geoff, reported by Diego Tartara <dtartara@novamens.com>*
14071 * Change the key loading routines for ENGINEs to use the same kind
14072 callback (pem_password_cb) as all other routines that need this
14077 * Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
14078 256 bit (=32 byte) keys. Of course seeding with more entropy bytes
14079 than this minimum value is recommended.
14083 * New random seeder for OpenVMS, using the system process statistics
14084 that are easily reachable.
14088 * Windows apparently can't transparently handle global
14089 variables defined in DLLs. Initialisations such as:
14091 const ASN1_ITEM *it = &ASN1_INTEGER_it;
14093 won't compile. This is used by the any applications that need to
14094 declare their own ASN1 modules. This was fixed by adding the option
14095 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
14096 needed for static libraries under Win32.
14100 * New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
14101 setting of purpose and trust fields. New X509_STORE trust and
14102 purpose functions and tidy up setting in other SSL functions.
14106 * Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
14107 structure. These are inherited by X509_STORE_CTX when it is
14108 initialised. This allows various defaults to be set in the
14109 X509_STORE structure (such as flags for CRL checking and custom
14110 purpose or trust settings) for functions which only use X509_STORE_CTX
14111 internally such as S/MIME.
14113 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
14114 trust settings if they are not set in X509_STORE. This allows X509_STORE
14115 purposes and trust (in S/MIME for example) to override any set by default.
14117 Add command line options for CRL checking to smime, s_client and s_server
14122 * Initial CRL based revocation checking. If the CRL checking flag(s)
14123 are set then the CRL is looked up in the X509_STORE structure and
14124 its validity and signature checked, then if the certificate is found
14125 in the CRL the verify fails with a revoked error.
14127 Various new CRL related callbacks added to X509_STORE_CTX structure.
14129 Command line options added to 'verify' application to support this.
14131 This needs some additional work, such as being able to handle multiple
14132 CRLs with different times, extension based lookup (rather than just
14133 by subject name) and ultimately more complete V2 CRL extension
14138 * Add a general user interface API (crypto/ui/). This is designed
14139 to replace things like des_read_password and friends (backward
14140 compatibility functions using this new API are provided).
14141 The purpose is to remove prompting functions from the DES code
14142 section as well as provide for prompting through dialog boxes in
14143 a window system and the like.
14147 * Add "ex_data" support to ENGINE so implementations can add state at a
14148 per-structure level rather than having to store it globally.
14152 * Make it possible for ENGINE structures to be copied when retrieved by
14153 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
14154 This causes the "original" ENGINE structure to act like a template,
14155 analogous to the RSA vs. RSA_METHOD type of separation. Because of this
14156 operational state can be localised to each ENGINE structure, despite the
14157 fact they all share the same "methods". New ENGINE structures returned in
14158 this case have no functional references and the return value is the single
14159 structural reference. This matches the single structural reference returned
14160 by ENGINE_by_id() normally, when it is incremented on the pre-existing
14165 * Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
14166 needs to match any other type at all we need to manually clear the
14171 * Changes to the "openssl engine" utility to include;
14172 - verbosity levels ('-v', '-vv', and '-vvv') that provide information
14173 about an ENGINE's available control commands.
14174 - executing control commands from command line arguments using the
14175 '-pre' and '-post' switches. '-post' is only used if '-t' is
14176 specified and the ENGINE is successfully initialised. The syntax for
14177 the individual commands are colon-separated, for example;
14178 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
14182 * New dynamic control command support for ENGINEs. ENGINEs can now
14183 declare their own commands (numbers), names (strings), descriptions,
14184 and input types for run-time discovery by calling applications. A
14185 subset of these commands are implicitly classed as "executable"
14186 depending on their input type, and only these can be invoked through
14187 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
14188 can be based on user input, config files, etc). The distinction is
14189 that "executable" commands cannot return anything other than a boolean
14190 result and can only support numeric or string input, whereas some
14191 discoverable commands may only be for direct use through
14192 ENGINE_ctrl(), eg. supporting the exchange of binary data, function
14193 pointers, or other custom uses. The "executable" commands are to
14194 support parameterisations of ENGINE behaviour that can be
14195 unambiguously defined by ENGINEs and used consistently across any
14196 OpenSSL-based application. Commands have been added to all the
14197 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
14198 control over shared-library paths without source code alterations.
14202 * Changed all ENGINE implementations to dynamically allocate their
14203 ENGINEs rather than declaring them statically. Apart from this being
14204 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
14205 this also allows the implementations to compile without using the
14206 internal engine_int.h header.
14210 * Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
14211 'const' value. Any code that should be able to modify a RAND_METHOD
14212 should already have non-const pointers to it (ie. they should only
14213 modify their own ones).
14217 * Made a variety of little tweaks to the ENGINE code.
14218 - "atalla" and "ubsec" string definitions were moved from header files
14219 to C code. "nuron" string definitions were placed in variables
14220 rather than hard-coded - allowing parameterisation of these values
14221 later on via ctrl() commands.
14222 - Removed unused "#if 0"'d code.
14223 - Fixed engine list iteration code so it uses ENGINE_free() to release
14224 structural references.
14225 - Constified the RAND_METHOD element of ENGINE structures.
14226 - Constified various get/set functions as appropriate and added
14227 missing functions (including a catch-all ENGINE_cpy that duplicates
14228 all ENGINE values onto a new ENGINE except reference counts/state).
14229 - Removed NULL parameter checks in get/set functions. Setting a method
14230 or function to NULL is a way of cancelling out a previously set
14231 value. Passing a NULL ENGINE parameter is just plain stupid anyway
14232 and doesn't justify the extra error symbols and code.
14233 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
14234 flags from engine_int.h to engine.h.
14235 - Changed prototypes for ENGINE handler functions (init(), finish(),
14236 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
14240 * Implement binary inversion algorithm for BN_mod_inverse in addition
14241 to the algorithm using long division. The binary algorithm can be
14242 used only if the modulus is odd. On 32-bit systems, it is faster
14243 only for relatively small moduli (roughly 20-30% for 128-bit moduli,
14244 roughly 5-15% for 256-bit moduli), so we use it only for moduli
14245 up to 450 bits. In 64-bit environments, the binary algorithm
14246 appears to be advantageous for much longer moduli; here we use it
14247 for moduli up to 2048 bits.
14251 * Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
14252 could not support the combine flag in choice fields.
14256 * Add a 'copy_extensions' option to the 'ca' utility. This copies
14257 extensions from a certificate request to the certificate.
14261 * Allow multiple 'certopt' and 'nameopt' options to be separated
14262 by commas. Add 'namopt' and 'certopt' options to the 'ca' config
14263 file: this allows the display of the certificate about to be
14264 signed to be customised, to allow certain fields to be included
14265 or excluded and extension details. The old system didn't display
14266 multicharacter strings properly, omitted fields not in the policy
14267 and couldn't display additional details such as extensions.
14271 * Function EC_POINTs_mul for multiple scalar multiplication
14272 of an arbitrary number of elliptic curve points
14273 \sum scalars[i]*points[i],
14274 optionally including the generator defined for the EC_GROUP:
14275 scalar*generator + \sum scalars[i]*points[i].
14277 EC_POINT_mul is a simple wrapper function for the typical case
14278 that the point list has just one item (besides the optional
14283 * First EC_METHODs for curves over GF(p):
14285 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
14286 operations and provides various method functions that can also
14287 operate with faster implementations of modular arithmetic.
14289 EC_GFp_mont_method() reuses most functions that are part of
14290 EC_GFp_simple_method, but uses Montgomery arithmetic.
14292 *Bodo Moeller; point addition and point doubling
14293 implementation directly derived from source code provided by
14294 Lenka Fibikova <fibikova@exp-math.uni-essen.de>*
14296 * Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
14297 crypto/ec/ec_lib.c):
14299 Curves are EC_GROUP objects (with an optional group generator)
14300 based on EC_METHODs that are built into the library.
14302 Points are EC_POINT objects based on EC_GROUP objects.
14304 Most of the framework would be able to handle curves over arbitrary
14305 finite fields, but as there are no obvious types for fields other
14306 than GF(p), some functions are limited to that for now.
14310 * Add the -HTTP option to s_server. It is similar to -WWW, but requires
14311 that the file contains a complete HTTP response.
14315 * Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
14316 change the def and num file printf format specifier from "%-40sXXX"
14317 to "%-39s XXX". The latter will always guarantee a space after the
14318 field while the former will cause them to run together if the field
14319 is 40 of more characters long.
14323 * Constify the cipher and digest 'method' functions and structures
14324 and modify related functions to take constant EVP_MD and EVP_CIPHER
14329 * Hide BN_CTX structure details in bn_lcl.h instead of publishing them
14330 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32.
14334 * Modify `EVP_Digest*()` routines so they now return values. Although the
14335 internal software routines can never fail additional hardware versions
14340 * Clean up crypto/err/err.h and change some error codes to avoid conflicts:
14342 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
14343 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
14346 ERR_R_NESTED_ASN1_ERROR
14348 ERR_R_MISSING_ASN1_EOS
14349 were 4 .. 9, conflicting with
14350 ERR_LIB_RSA (= ERR_R_RSA_LIB)
14352 ERR_LIB_PEM (= ERR_R_PEM_LIB).
14353 They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
14355 Add new error code 'ERR_R_INTERNAL_ERROR'.
14359 * Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
14364 * New option '-subj arg' for 'openssl req' and 'openssl ca'. This
14365 sets the subject name for a new request or supersedes the
14366 subject name in a given request. Formats that can be parsed are
14367 'CN=Some Name, OU=myOU, C=IT'
14369 'CN=Some Name/OU=myOU/C=IT'.
14371 Add options '-batch' and '-verbose' to 'openssl req'.
14373 *Massimiliano Pala <madwolf@hackmasters.net>*
14375 * Introduce the possibility to access global variables through
14376 functions on platform were that's the best way to handle exporting
14377 global variables in shared libraries. To enable this functionality,
14378 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
14379 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
14380 is normally done by Configure or something similar).
14382 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
14383 in the source file (foo.c) like this:
14385 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
14386 OPENSSL_IMPLEMENT_GLOBAL(double,bar);
14388 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
14389 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
14391 OPENSSL_DECLARE_GLOBAL(int,foo);
14392 #define foo OPENSSL_GLOBAL_REF(foo)
14393 OPENSSL_DECLARE_GLOBAL(double,bar);
14394 #define bar OPENSSL_GLOBAL_REF(bar)
14396 The #defines are very important, and therefore so is including the
14397 header file everywhere where the defined globals are used.
14399 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
14400 of ASN.1 items, but that structure is a bit different.
14402 The largest change is in util/mkdef.pl which has been enhanced with
14403 better and easier to understand logic to choose which symbols should
14404 go into the Windows .def files as well as a number of fixes and code
14405 cleanup (among others, algorithm keywords are now sorted
14406 lexicographically to avoid constant rewrites).
14410 * In BN_div() keep a copy of the sign of 'num' before writing the
14411 result to 'rm' because if rm==num the value will be overwritten
14412 and produce the wrong result if 'num' is negative: this caused
14413 problems with BN_mod() and BN_nnmod().
14417 * Function OCSP_request_verify(). This checks the signature on an
14418 OCSP request and verifies the signer certificate. The signer
14419 certificate is just checked for a generic purpose and OCSP request
14424 * Add OCSP_check_validity() function to check the validity of OCSP
14425 responses. OCSP responses are prepared in real time and may only
14426 be a few seconds old. Simply checking that the current time lies
14427 between thisUpdate and nextUpdate max reject otherwise valid responses
14428 caused by either OCSP responder or client clock inaccuracy. Instead
14429 we allow thisUpdate and nextUpdate to fall within a certain period of
14430 the current time. The age of the response can also optionally be
14431 checked. Two new options -validity_period and -status_age added to
14436 * If signature or public key algorithm is unrecognized print out its
14437 OID rather that just UNKNOWN.
14441 * Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
14442 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
14443 ID to be generated from the issuer certificate alone which can then be
14444 passed to OCSP_id_issuer_cmp().
14448 * New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
14449 ASN1 modules to export functions returning ASN1_ITEM pointers
14450 instead of the ASN1_ITEM structures themselves. This adds several
14451 new macros which allow the underlying ASN1 function/structure to
14452 be accessed transparently. As a result code should not use ASN1_ITEM
14453 references directly (such as &X509_it) but instead use the relevant
14454 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
14455 use of the new ASN1 code on platforms where exporting structures
14456 is problematical (for example in shared libraries) but exporting
14457 functions returning pointers to structures is not.
14461 * Add support for overriding the generation of SSL/TLS session IDs.
14462 These callbacks can be registered either in an SSL_CTX or per SSL.
14463 The purpose of this is to allow applications to control, if they wish,
14464 the arbitrary values chosen for use as session IDs, particularly as it
14465 can be useful for session caching in multiple-server environments. A
14466 command-line switch for testing this (and any client code that wishes
14467 to use such a feature) has been added to "s_server".
14469 *Geoff Thorpe, Lutz Jaenicke*
14471 * Modify mkdef.pl to recognise and parse preprocessor conditionals
14472 of the form `#if defined(...) || defined(...) || ...` and
14473 `#if !defined(...) && !defined(...) && ...`. This also avoids
14474 the growing number of special cases it was previously handling.
14478 * Make all configuration macros available for application by making
14479 sure they are available in opensslconf.h, by giving them names starting
14480 with `OPENSSL_` to avoid conflicts with other packages and by making
14481 sure e_os2.h will cover all platform-specific cases together with
14483 Additionally, it is now possible to define configuration/platform-
14484 specific names (called "system identities"). In the C code, these
14485 are prefixed with `OPENSSL_SYSNAME_`. e_os2.h will create another
14486 macro with the name beginning with `OPENSSL_SYS_`, which is determined
14487 from `OPENSSL_SYSNAME_*` or compiler-specific macros depending on
14492 * New option -set_serial to 'req' and 'x509' this allows the serial
14493 number to use to be specified on the command line. Previously self
14494 signed certificates were hard coded with serial number 0 and the
14495 CA options of 'x509' had to use a serial number in a file which was
14500 * New options to 'ca' utility to support V2 CRL entry extensions.
14501 Currently CRL reason, invalidity date and hold instruction are
14502 supported. Add new CRL extensions to V3 code and some new objects.
14506 * New function EVP_CIPHER_CTX_set_padding() this is used to
14507 disable standard block padding (aka PKCS#5 padding) in the EVP
14508 API, which was previously mandatory. This means that the data is
14509 not padded in any way and so the total length much be a multiple
14510 of the block size, otherwise an error occurs.
14514 * Initial (incomplete) OCSP SSL support.
14518 * New function OCSP_parse_url(). This splits up a URL into its host,
14519 port and path components: primarily to parse OCSP URLs. New -url
14520 option to ocsp utility.
14524 * New nonce behavior. The return value of OCSP_check_nonce() now
14525 reflects the various checks performed. Applications can decide
14526 whether to tolerate certain situations such as an absent nonce
14527 in a response when one was present in a request: the ocsp application
14528 just prints out a warning. New function OCSP_add1_basic_nonce()
14529 this is to allow responders to include a nonce in a response even if
14530 the request is nonce-less.
14534 * Disable stdin buffering in `load_cert()` (`apps/apps.c`) so that no certs are
14535 skipped when using openssl x509 multiple times on a single input file,
14536 e.g. `(openssl x509 -out cert1; openssl x509 -out cert2) <certs`.
14540 * Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
14541 set string type: to handle setting ASN1_TIME structures. Fix ca
14542 utility to correctly initialize revocation date of CRLs.
14546 * New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
14547 the clients preferred ciphersuites and rather use its own preferences.
14548 Should help to work around M$ SGC (Server Gated Cryptography) bug in
14549 Internet Explorer by ensuring unchanged hash method during stepup.
14550 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
14554 * Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
14555 to aes and add a new 'exist' option to print out symbols that don't
14560 * Additional options to ocsp utility to allow flags to be set and
14561 additional certificates supplied.
14565 * Add the option -VAfile to 'openssl ocsp', so the user can give the
14566 OCSP client a number of certificate to only verify the response
14571 * Update Rijndael code to version 3.0 and change EVP AES ciphers to
14572 handle the new API. Currently only ECB, CBC modes supported. Add new
14575 Add TLS AES ciphersuites as described in RFC3268, "Advanced
14576 Encryption Standard (AES) Ciphersuites for Transport Layer
14577 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were
14578 not enabled by default and were not part of the "ALL" ciphersuite
14579 alias because they were not yet official; they could be
14580 explicitly requested by specifying the "AESdraft" ciphersuite
14581 group alias. In the final release of OpenSSL 0.9.7, the group
14582 alias is called "AES" and is part of "ALL".)
14584 *Ben Laurie, Steve Henson, Bodo Moeller*
14586 * New function OCSP_copy_nonce() to copy nonce value (if present) from
14587 request to response.
14591 * Functions for OCSP responders. OCSP_request_onereq_count(),
14592 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
14593 extract information from a certificate request. OCSP_response_create()
14594 creates a response and optionally adds a basic response structure.
14595 OCSP_basic_add1_status() adds a complete single response to a basic
14596 response and returns the OCSP_SINGLERESP structure just added (to allow
14597 extensions to be included for example). OCSP_basic_add1_cert() adds a
14598 certificate to a basic response and OCSP_basic_sign() signs a basic
14599 response with various flags. New helper functions ASN1_TIME_check()
14600 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
14601 (converts ASN1_TIME to GeneralizedTime).
14605 * Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
14606 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
14607 structure from a certificate. X509_pubkey_digest() digests the public_key
14608 contents: this is used in various key identifiers.
14612 * Make sk_sort() tolerate a NULL argument.
14614 *Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>*
14616 * New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
14617 passed by the function are trusted implicitly. If any of them signed the
14618 response then it is assumed to be valid and is not verified.
14622 * In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
14623 to data. This was previously part of the PKCS7 ASN1 code. This
14624 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
14625 *Steve Henson, reported by Kenneth R. Robinette
14626 <support@securenetterm.com>*
14628 * Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
14629 routines: without these tracing memory leaks is very painful.
14630 Fix leaks in PKCS12 and PKCS7 routines.
14634 * Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
14635 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
14636 effectively meant GeneralizedTime would never be used. Now it
14637 is initialised to -1 but X509_time_adj() now has to check the value
14638 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
14639 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
14640 *Steve Henson, reported by Kenneth R. Robinette
14641 <support@securenetterm.com>*
14643 * Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
14644 result in a zero length in the ASN1_INTEGER structure which was
14645 not consistent with the structure when d2i_ASN1_INTEGER() was used
14646 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
14647 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
14648 where it did not print out a minus for negative ASN1_INTEGER.
14652 * Add summary printout to ocsp utility. The various functions which
14653 convert status values to strings have been renamed to:
14654 OCSP_response_status_str(), OCSP_cert_status_str() and
14655 OCSP_crl_reason_str() and are no longer static. New options
14656 to verify nonce values and to disable verification. OCSP response
14657 printout format cleaned up.
14661 * Add additional OCSP certificate checks. These are those specified
14662 in RFC2560. This consists of two separate checks: the CA of the
14663 certificate being checked must either be the OCSP signer certificate
14664 or the issuer of the OCSP signer certificate. In the latter case the
14665 OCSP signer certificate must contain the OCSP signing extended key
14666 usage. This check is performed by attempting to match the OCSP
14667 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
14668 in the OCSP_CERTID structures of the response.
14672 * Initial OCSP certificate verification added to OCSP_basic_verify()
14673 and related routines. This uses the standard OpenSSL certificate
14674 verify routines to perform initial checks (just CA validity) and
14675 to obtain the certificate chain. Then additional checks will be
14676 performed on the chain. Currently the root CA is checked to see
14677 if it is explicitly trusted for OCSP signing. This is used to set
14678 a root CA as a global signing root: that is any certificate that
14679 chains to that CA is an acceptable OCSP signing certificate.
14683 * New '-extfile ...' option to 'openssl ca' for reading X.509v3
14684 extensions from a separate configuration file.
14685 As when reading extensions from the main configuration file,
14686 the '-extensions ...' option may be used for specifying the
14689 *Massimiliano Pala <madwolf@comune.modena.it>*
14691 * New OCSP utility. Allows OCSP requests to be generated or
14692 read. The request can be sent to a responder and the output
14693 parsed, outputted or printed in text form. Not complete yet:
14694 still needs to check the OCSP response validity.
14698 * New subcommands for 'openssl ca':
14699 `openssl ca -status <serial>` prints the status of the cert with
14700 the given serial number (according to the index file).
14701 `openssl ca -updatedb` updates the expiry status of certificates
14704 *Massimiliano Pala <madwolf@comune.modena.it>*
14706 * New '-newreq-nodes' command option to CA.pl. This is like
14707 '-newreq', but calls 'openssl req' with the '-nodes' option
14708 so that the resulting key is not encrypted.
14710 *Damien Miller <djm@mindrot.org>*
14712 * New configuration for the GNU Hurd.
14714 *Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte*
14716 * Initial code to implement OCSP basic response verify. This
14717 is currently incomplete. Currently just finds the signer's
14718 certificate and verifies the signature on the response.
14722 * New SSLeay_version code SSLEAY_DIR to determine the compiled-in
14723 value of OPENSSLDIR. This is available via the new '-d' option
14724 to 'openssl version', and is also included in 'openssl version -a'.
14728 * Allowing defining memory allocation callbacks that will be given
14729 file name and line number information in additional arguments
14730 (a `const char*` and an int). The basic functionality remains, as
14731 well as the original possibility to just replace malloc(),
14732 realloc() and free() by functions that do not know about these
14733 additional arguments. To register and find out the current
14734 settings for extended allocation functions, the following
14735 functions are provided:
14737 CRYPTO_set_mem_ex_functions
14738 CRYPTO_set_locked_mem_ex_functions
14739 CRYPTO_get_mem_ex_functions
14740 CRYPTO_get_locked_mem_ex_functions
14742 These work the same way as CRYPTO_set_mem_functions and friends.
14743 `CRYPTO_get_[locked_]mem_functions` now writes 0 where such an
14744 extended allocation function is enabled.
14745 Similarly, `CRYPTO_get_[locked_]mem_ex_functions` writes 0 where
14746 a conventional allocation function is enabled.
14748 *Richard Levitte, Bodo Moeller*
14750 * Finish off removing the remaining LHASH function pointer casts.
14751 There should no longer be any prototype-casting required when using
14752 the LHASH abstraction, and any casts that remain are "bugs". See
14753 the callback types and macros at the head of lhash.h for details
14754 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
14758 * Add automatic query of EGD sockets in RAND_poll() for the unix variant.
14759 If /dev/[u]random devices are not available or do not return enough
14760 entropy, EGD style sockets (served by EGD or PRNGD) will automatically
14762 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
14763 /etc/entropy will be queried once each in this sequence, querying stops
14764 when enough entropy was collected without querying more sockets.
14768 * Change the Unix RAND_poll() variant to be able to poll several
14769 random devices, as specified by DEVRANDOM, until a sufficient amount
14770 of data has been collected. We spend at most 10 ms on each file
14771 (select timeout) and read in non-blocking mode. DEVRANDOM now
14772 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
14773 (previously it was just the string "/dev/urandom"), so on typical
14774 platforms the 10 ms delay will never occur.
14775 Also separate out the Unix variant to its own file, rand_unix.c.
14776 For VMS, there's a currently-empty rand_vms.c.
14780 * Move OCSP client related routines to ocsp_cl.c. These
14781 provide utility functions which an application needing
14782 to issue a request to an OCSP responder and analyse the
14783 response will typically need: as opposed to those which an
14784 OCSP responder itself would need which will be added later.
14786 OCSP_request_sign() signs an OCSP request with an API similar
14787 to PKCS7_sign(). OCSP_response_status() returns status of OCSP
14788 response. OCSP_response_get1_basic() extracts basic response
14789 from response. OCSP_resp_find_status(): finds and extracts status
14790 information from an OCSP_CERTID structure (which will be created
14791 when the request structure is built). These are built from lower
14792 level functions which work on OCSP_SINGLERESP structures but
14793 won't normally be used unless the application wishes to examine
14794 extensions in the OCSP response for example.
14796 Replace nonce routines with a pair of functions.
14797 OCSP_request_add1_nonce() adds a nonce value and optionally
14798 generates a random value. OCSP_check_nonce() checks the
14799 validity of the nonce in an OCSP response.
14803 * Change function OCSP_request_add() to OCSP_request_add0_id().
14804 This doesn't copy the supplied OCSP_CERTID and avoids the
14805 need to free up the newly created id. Change return type
14806 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
14807 This can then be used to add extensions to the request.
14808 Deleted OCSP_request_new(), since most of its functionality
14809 is now in OCSP_REQUEST_new() (and the case insensitive name
14810 clash) apart from the ability to set the request name which
14811 will be added elsewhere.
14815 * Update OCSP API. Remove obsolete extensions argument from
14816 various functions. Extensions are now handled using the new
14817 OCSP extension code. New simple OCSP HTTP function which
14818 can be used to send requests and parse the response.
14822 * Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
14823 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
14824 uses the special reorder version of SET OF to sort the attributes
14825 and reorder them to match the encoded order. This resolves a long
14826 standing problem: a verify on a PKCS7 structure just after signing
14827 it used to fail because the attribute order did not match the
14828 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
14829 it uses the received order. This is necessary to tolerate some broken
14830 software that does not order SET OF. This is handled by encoding
14831 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
14832 to produce the required SET OF.
14836 * Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
14837 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
14838 files to get correct declarations of the ASN.1 item variables.
14842 * Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
14843 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
14844 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
14845 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
14846 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
14847 ASN1_ITEM and no wrapper functions.
14851 * New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
14852 replace the old function pointer based I/O routines. Change most of
14853 the `*_d2i_bio()` and `*_d2i_fp()` functions to use these.
14857 * Enhance mkdef.pl to be more accepting about spacing in C preprocessor
14858 lines, recognize more "algorithms" that can be deselected, and make
14859 it complain about algorithm deselection that isn't recognised.
14863 * New ASN1 functions to handle dup, sign, verify, digest, pack and
14864 unpack operations in terms of ASN1_ITEM. Modify existing wrappers
14865 to use new functions. Add NO_ASN1_OLD which can be set to remove
14866 some old style ASN1 functions: this can be used to determine if old
14867 code will still work when these eventually go away.
14871 * New extension functions for OCSP structures, these follow the
14872 same conventions as certificates and CRLs.
14876 * New function X509V3_add1_i2d(). This automatically encodes and
14877 adds an extension. Its behaviour can be customised with various
14878 flags to append, replace or delete. Various wrappers added for
14879 certificates and CRLs.
14883 * Fix to avoid calling the underlying ASN1 print routine when
14884 an extension cannot be parsed. Correct a typo in the
14885 OCSP_SERVICELOC extension. Tidy up print OCSP format.
14889 * Make mkdef.pl parse some of the ASN1 macros and add appropriate
14890 entries for variables.
14894 * Add functionality to `apps/openssl.c` for detecting locking
14895 problems: As the program is single-threaded, all we have
14896 to do is register a locking callback using an array for
14897 storing which locks are currently held by the program.
14901 * Use a lock around the call to CRYPTO_get_ex_new_index() in
14902 SSL_get_ex_data_X509_STORE_idx(), which is used in
14903 ssl_verify_cert_chain() and thus can be called at any time
14904 during TLS/SSL handshakes so that thread-safety is essential.
14905 Unfortunately, the ex_data design is not at all suited
14906 for multi-threaded use, so it probably should be abolished.
14910 * Added Broadcom "ubsec" ENGINE to OpenSSL.
14912 *Broadcom, tweaked and integrated by Geoff Thorpe*
14914 * Move common extension printing code to new function
14915 X509V3_print_extensions(). Reorganise OCSP print routines and
14916 implement some needed OCSP ASN1 functions. Add OCSP extensions.
14920 * New function X509_signature_print() to remove duplication in some
14925 * Add a special meaning when SET OF and SEQUENCE OF flags are both
14926 set (this was treated exactly the same as SET OF previously). This
14927 is used to reorder the STACK representing the structure to match the
14928 encoding. This will be used to get round a problem where a PKCS7
14929 structure which was signed could not be verified because the STACK
14930 order did not reflect the encoded order.
14934 * Reimplement the OCSP ASN1 module using the new code.
14938 * Update the X509V3 code to permit the use of an ASN1_ITEM structure
14939 for its ASN1 operations. The old style function pointers still exist
14940 for now but they will eventually go away.
14944 * Merge in replacement ASN1 code from the ASN1 branch. This almost
14945 completely replaces the old ASN1 functionality with a table driven
14946 encoder and decoder which interprets an ASN1_ITEM structure describing
14947 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
14948 largely maintained. Almost all of the old asn1_mac.h macro based ASN1
14949 has also been converted to the new form.
14953 * Change BN_mod_exp_recp so that negative moduli are tolerated
14954 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set
14955 so that BN_mod_exp_mont and BN_mod_exp_mont_word work
14956 for negative moduli.
14960 * Fix BN_uadd and BN_usub: Always return non-negative results instead
14961 of not touching the result's sign bit.
14965 * BN_div bugfix: If the result is 0, the sign (res->neg) must not be
14970 * Changed the LHASH code to use prototypes for callbacks, and created
14971 macros to declare and implement thin (optionally static) functions
14972 that provide type-safety and avoid function pointer casting for the
14973 type-specific callbacks.
14977 * Added Kerberos Cipher Suites to be used with TLS, as written in
14979 *Veers Staats <staatsvr@asc.hpc.mil>,
14980 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte*
14982 * Reformat the FAQ so the different questions and answers can be divided
14983 in sections depending on the subject.
14987 * Have the zlib compression code load ZLIB.DLL dynamically under
14992 * New function BN_mod_sqrt for computing square roots modulo a prime
14993 (using the probabilistic Tonelli-Shanks algorithm unless
14994 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can
14995 be handled deterministically).
14997 *Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller*
14999 * Make BN_mod_inverse faster by explicitly handling small quotients
15000 in the Euclid loop. (Speed gain about 20% for small moduli [256 or
15001 512 bits], about 30% for larger ones [1024 or 2048 bits].)
15005 * New function BN_kronecker.
15009 * Fix BN_gcd so that it works on negative inputs; the result is
15010 positive unless both parameters are zero.
15011 Previously something reasonably close to an infinite loop was
15012 possible because numbers could be growing instead of shrinking
15013 in the implementation of Euclid's algorithm.
15017 * Fix BN_is_word() and BN_is_one() macros to take into account the
15018 sign of the number in question.
15020 Fix BN_is_word(a,w) to work correctly for w == 0.
15022 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
15023 because its test if the absolute value of 'a' equals 'w'.
15024 Note that BN_abs_is_word does *not* handle w == 0 reliably;
15025 it exists mostly for use in the implementations of BN_is_zero(),
15026 BN_is_one(), and BN_is_word().
15030 * New function BN_swap.
15034 * Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
15035 the exponentiation functions are more likely to produce reasonable
15036 results on negative inputs.
15040 * Change BN_mod_mul so that the result is always non-negative.
15041 Previously, it could be negative if one of the factors was negative;
15042 I don't think anyone really wanted that behaviour.
15046 * Move `BN_mod_...` functions into new file `crypto/bn/bn_mod.c`
15047 (except for exponentiation, which stays in `crypto/bn/bn_exp.c`,
15048 and `BN_mod_mul_reciprocal`, which stays in `crypto/bn/bn_recp.c`)
15049 and add new functions:
15058 BN_mod_lshift1_quick
15060 BN_mod_lshift_quick
15062 These functions always generate non-negative results.
15064 `BN_nnmod` otherwise is `like BN_mod` (if `BN_mod` computes a remainder `r`
15065 such that `|m| < r < 0`, `BN_nnmod` will output `rem + |m|` instead).
15067 `BN_mod_XXX_quick(r, a, [b,] m)` generates the same result as
15068 `BN_mod_XXX(r, a, [b,] m, ctx)`, but requires that `a` [and `b`]
15069 be reduced modulo `m`.
15071 *Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller*
15074 The following entry accidentally appeared in the CHANGES file
15075 distributed with OpenSSL 0.9.7. The modifications described in
15076 it do *not* apply to OpenSSL 0.9.7.
15078 * Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
15079 was actually never needed) and in BN_mul(). The removal in BN_mul()
15080 required a small change in bn_mul_part_recursive() and the addition
15081 of the functions bn_cmp_part_words(), bn_sub_part_words() and
15082 bn_add_part_words(), which do the same thing as bn_cmp_words(),
15083 bn_sub_words() and bn_add_words() except they take arrays with
15089 * In 'openssl passwd', verify passwords read from the terminal
15090 unless the '-salt' option is used (which usually means that
15091 verification would just waste user's time since the resulting
15092 hash is going to be compared with some given password hash)
15093 or the new '-noverify' option is used.
15095 This is an incompatible change, but it does not affect
15096 non-interactive use of 'openssl passwd' (passwords on the command
15097 line, '-stdin' option, '-in ...' option) and thus should not
15098 cause any problems.
15102 * Remove all references to RSAref, since there's no more need for it.
15106 * Make DSO load along a path given through an environment variable
15107 (SHLIB_PATH) with shl_load().
15111 * Constify the ENGINE code as a result of BIGNUM constification.
15112 Also constify the RSA code and most things related to it. In a
15113 few places, most notable in the depth of the ASN.1 code, ugly
15114 casts back to non-const were required (to be solved at a later
15119 * Make it so the openssl application has all engines loaded by default.
15123 * Constify the BIGNUM routines a little more.
15127 * Add the following functions:
15129 ENGINE_load_cswift()
15131 ENGINE_load_atalla()
15132 ENGINE_load_nuron()
15133 ENGINE_load_builtin_engines()
15135 That way, an application can itself choose if external engines that
15136 are built-in in OpenSSL shall ever be used or not. The benefit is
15137 that applications won't have to be linked with libdl or other dso
15138 libraries unless it's really needed.
15140 Changed 'openssl engine' to load all engines on demand.
15141 Changed the engine header files to avoid the duplication of some
15142 declarations (they differed!).
15146 * 'openssl engine' can now list capabilities.
15150 * Better error reporting in 'openssl engine'.
15154 * Never call load_dh_param(NULL) in s_server.
15158 * Add engine application. It can currently list engines by name and
15159 identity, and test if they are actually available.
15163 * Improve RPM specification file by forcing symbolic linking and making
15164 sure the installed documentation is also owned by root.root.
15166 *Damien Miller <djm@mindrot.org>*
15168 * Give the OpenSSL applications more possibilities to make use of
15169 keys (public as well as private) handled by engines.
15173 * Add OCSP code that comes from CertCo.
15177 * Add VMS support for the Rijndael code.
15181 * Added untested support for Nuron crypto accelerator.
15185 * Add support for external cryptographic devices. This code was
15186 previously distributed separately as the "engine" branch.
15188 *Geoff Thorpe, Richard Levitte*
15190 * Rework the filename-translation in the DSO code. It is now possible to
15191 have far greater control over how a "name" is turned into a filename
15192 depending on the operating environment and any oddities about the
15193 different shared library filenames on each system.
15197 * Support threads on FreeBSD-elf in Configure.
15201 * Fix for SHA1 assembly problem with MASM: it produces
15202 warnings about corrupt line number information when assembling
15203 with debugging information. This is caused by the overlapping
15206 *Bernd Matthes <mainbug@celocom.de>, Steve Henson*
15209 NCONF_get_number() has no error checking at all. As a replacement,
15210 NCONF_get_number_e() is defined (`_e` for "error checking") and is
15211 promoted strongly. The old NCONF_get_number is kept around for
15212 binary backward compatibility.
15213 Make it possible for methods to load from something other than a BIO,
15214 by providing a function pointer that is given a name instead of a BIO.
15215 For example, this could be used to load configuration data from an
15220 * Fix for non blocking accept BIOs. Added new I/O special reason
15221 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
15222 with non blocking I/O was not possible because no retry code was
15223 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
15228 * Added the beginnings of Rijndael support.
15232 * Fix for bug in DirectoryString mask setting. Add support for
15233 X509_NAME_print_ex() in 'req' and X509_print_ex() function
15234 to allow certificate printing to more controllable, additional
15235 'certopt' option to 'x509' to allow new printing options to be
15240 * Clean old EAY MD5 hack from e_os.h.
15244 ### Changes between 0.9.6l and 0.9.6m [17 Mar 2004]
15246 * Fix null-pointer assignment in do_change_cipher_spec() revealed
15247 by using the Codenomicon TLS Test Tool ([CVE-2004-0079])
15249 *Joe Orton, Steve Henson*
15251 ### Changes between 0.9.6k and 0.9.6l [04 Nov 2003]
15253 * Fix additional bug revealed by the NISCC test suite:
15255 Stop bug triggering large recursion when presented with
15256 certain ASN.1 tags ([CVE-2003-0851])
15260 ### Changes between 0.9.6j and 0.9.6k [30 Sep 2003]
15262 * Fix various bugs revealed by running the NISCC test suite:
15264 Stop out of bounds reads in the ASN1 code when presented with
15265 invalid tags (CVE-2003-0543 and CVE-2003-0544).
15267 If verify callback ignores invalid public key errors don't try to check
15268 certificate signature with the NULL public key.
15272 * In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
15273 if the server requested one: as stated in TLS 1.0 and SSL 3.0
15278 * In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
15279 extra data after the compression methods not only for TLS 1.0
15280 but also for SSL 3.0 (as required by the specification).
15282 *Bodo Moeller; problem pointed out by Matthias Loepfe*
15284 * Change X509_certificate_type() to mark the key as exported/exportable
15285 when it's 512 *bits* long, not 512 bytes.
15289 ### Changes between 0.9.6i and 0.9.6j [10 Apr 2003]
15291 * Countermeasure against the Klima-Pokorny-Rosa extension of
15292 Bleichbacher's attack on PKCS #1 v1.5 padding: treat
15293 a protocol version number mismatch like a decryption error
15294 in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
15298 * Turn on RSA blinding by default in the default implementation
15299 to avoid a timing attack. Applications that don't want it can call
15300 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
15301 They would be ill-advised to do so in most cases.
15303 *Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller*
15305 * Change RSA blinding code so that it works when the PRNG is not
15306 seeded (in this case, the secret RSA exponent is abused as
15307 an unpredictable seed -- if it is not unpredictable, there
15308 is no point in blinding anyway). Make RSA blinding thread-safe
15309 by remembering the creator's thread ID in rsa->blinding and
15310 having all other threads use local one-time blinding factors
15311 (this requires more computation than sharing rsa->blinding, but
15312 avoids excessive locking; and if an RSA object is not shared
15313 between threads, blinding will still be very fast).
15317 ### Changes between 0.9.6h and 0.9.6i [19 Feb 2003]
15319 * In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
15320 via timing by performing a MAC computation even if incorrect
15321 block cipher padding has been found. This is a countermeasure
15322 against active attacks where the attacker has to distinguish
15323 between bad padding and a MAC verification error. ([CVE-2003-0078])
15325 *Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
15326 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
15327 Martin Vuagnoux (EPFL, Ilion)*
15329 ### Changes between 0.9.6g and 0.9.6h [5 Dec 2002]
15331 * New function OPENSSL_cleanse(), which is used to cleanse a section of
15332 memory from its contents. This is done with a counter that will
15333 place alternating values in each byte. This can be used to solve
15334 two issues: 1) the removal of calls to memset() by highly optimizing
15335 compilers, and 2) cleansing with other values than 0, since those can
15336 be read through on certain media, for example a swap space on disk.
15340 * Bugfix: client side session caching did not work with external caching,
15341 because the session->cipher setting was not restored when reloading
15342 from the external cache. This problem was masked, when
15343 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set.
15344 (Found by Steve Haslam <steve@araqnid.ddts.net>.)
15348 * Fix client_certificate (ssl/s2_clnt.c): The permissible total
15349 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33.
15351 *Zeev Lieber <zeev-l@yahoo.com>*
15353 * Undo an undocumented change introduced in 0.9.6e which caused
15354 repeated calls to OpenSSL_add_all_ciphers() and
15355 OpenSSL_add_all_digests() to be ignored, even after calling
15360 * Change the default configuration reader to deal with last line not
15361 being properly terminated.
15365 * Change X509_NAME_cmp() so it applies the special rules on handling
15366 DN values that are of type PrintableString, as well as RDNs of type
15367 emailAddress where the value has the type ia5String.
15369 *stefank@valicert.com via Richard Levitte*
15371 * Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half
15372 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently
15373 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be
15374 the bitwise-OR of the two for use by the majority of applications
15375 wanting this behaviour, and update the docs. The documented
15376 behaviour and actual behaviour were inconsistent and had been
15377 changing anyway, so this is more a bug-fix than a behavioural
15380 *Geoff Thorpe, diagnosed by Nadav Har'El*
15382 * Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c
15383 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes).
15387 * Fix initialization code race conditions in
15388 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(),
15389 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(),
15390 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(),
15391 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(),
15392 ssl2_get_cipher_by_char(),
15393 ssl3_get_cipher_by_char().
15395 *Patrick McCormick <patrick@tellme.com>, Bodo Moeller*
15397 * Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after
15398 the cached sessions are flushed, as the remove_cb() might use ex_data
15399 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com>
15400 (see [openssl.org #212]).
15402 *Geoff Thorpe, Lutz Jaenicke*
15404 * Fix typo in OBJ_txt2obj which incorrectly passed the content
15405 length, instead of the encoding length to d2i_ASN1_OBJECT.
15409 ### Changes between 0.9.6f and 0.9.6g [9 Aug 2002]
15411 * [In 0.9.6g-engine release:]
15412 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use `_stdcall`).
15414 *Lynn Gazis <lgazis@rainbow.com>*
15416 ### Changes between 0.9.6e and 0.9.6f [8 Aug 2002]
15418 * Fix ASN1 checks. Check for overflow by comparing with LONG_MAX
15419 and get fix the header length calculation.
15420 *Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>,
15421 Alon Kantor <alonk@checkpoint.com> (and others), Steve Henson*
15423 * Use proper error handling instead of 'assertions' in buffer
15424 overflow checks added in 0.9.6e. This prevents DoS (the
15425 assertions could call abort()).
15427 *Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller*
15429 ### Changes between 0.9.6d and 0.9.6e [30 Jul 2002]
15431 * Add various sanity checks to asn1_get_length() to reject
15432 the ASN1 length bytes if they exceed sizeof(long), will appear
15433 negative or the content length exceeds the length of the
15436 *Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>*
15438 * Fix cipher selection routines: ciphers without encryption had no flags
15439 for the cipher strength set and where therefore not handled correctly
15440 by the selection routines (PR #130).
15444 * Fix EVP_dsa_sha macro.
15449 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS
15450 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure
15451 that was added in OpenSSL 0.9.6d.
15453 As the countermeasure turned out to be incompatible with some
15454 broken SSL implementations, the new option is part of SSL_OP_ALL.
15455 SSL_OP_ALL is usually employed when compatibility with weird SSL
15456 implementations is desired (e.g. '-bugs' option to 's_client' and
15457 's_server'), so the new option is automatically set in many
15462 * Changes in security patch:
15464 Changes marked "(CHATS)" were sponsored by the Defense Advanced
15465 Research Projects Agency (DARPA) and Air Force Research Laboratory,
15466 Air Force Materiel Command, USAF, under agreement number
15469 * Add various sanity checks to asn1_get_length() to reject
15470 the ASN1 length bytes if they exceed sizeof(long), will appear
15471 negative or the content length exceeds the length of the
15472 supplied buffer. ([CVE-2002-0659])
15474 *Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>*
15476 * Assertions for various potential buffer overflows, not known to
15477 happen in practice.
15479 *Ben Laurie (CHATS)*
15481 * Various temporary buffers to hold ASCII versions of integers were
15482 too small for 64 bit platforms. ([CVE-2002-0655])
15483 *Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)>*
15485 * Remote buffer overflow in SSL3 protocol - an attacker could
15486 supply an oversized session ID to a client. ([CVE-2002-0656])
15488 *Ben Laurie (CHATS)*
15490 * Remote buffer overflow in SSL2 protocol - an attacker could
15491 supply an oversized client master key. ([CVE-2002-0656])
15493 *Ben Laurie (CHATS)*
15495 ### Changes between 0.9.6c and 0.9.6d [9 May 2002]
15497 * Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not
15498 encoded as NULL) with id-dsa-with-sha1.
15500 *Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller*
15502 * Check various `X509_...()` return values in `apps/req.c`.
15504 *Nils Larsch <nla@trustcenter.de>*
15506 * Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines:
15507 an end-of-file condition would erroneously be flagged, when the CRLF
15508 was just at the end of a processed block. The bug was discovered when
15509 processing data through a buffering memory BIO handing the data to a
15510 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov
15511 <ptsekov@syntrex.com> and Nedelcho Stanev.
15515 * Implement a countermeasure against a vulnerability recently found
15516 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment
15517 before application data chunks to avoid the use of known IVs
15518 with data potentially chosen by the attacker.
15522 * Fix length checks in ssl3_get_client_hello().
15526 * TLS/SSL library bugfix: use s->s3->in_read_app_data differently
15527 to prevent ssl3_read_internal() from incorrectly assuming that
15528 ssl3_read_bytes() found application data while handshake
15529 processing was enabled when in fact s->s3->in_read_app_data was
15530 merely automatically cleared during the initial handshake.
15532 *Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>*
15534 * Fix object definitions for Private and Enterprise: they were not
15535 recognized in their shortname (=lowercase) representation. Extend
15536 obj_dat.pl to issue an error when using undefined keywords instead
15537 of silently ignoring the problem (Svenning Sorensen
15538 <sss@sss.dnsalias.net>).
15542 * Fix DH_generate_parameters() so that it works for 'non-standard'
15543 generators, i.e. generators other than 2 and 5. (Previously, the
15544 code did not properly initialise the 'add' and 'rem' values to
15545 BN_generate_prime().)
15547 In the new general case, we do not insist that 'generator' is
15548 actually a primitive root: This requirement is rather pointless;
15549 a generator of the order-q subgroup is just as good, if not
15554 * Map new X509 verification errors to alerts. Discovered and submitted by
15555 Tom Wu <tom@arcot.com>.
15559 * Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from
15560 returning non-zero before the data has been completely received
15561 when using non-blocking I/O.
15563 *Bodo Moeller; problem pointed out by John Hughes*
15565 * Some of the ciphers missed the strength entry (SSL_LOW etc).
15567 *Ben Laurie, Lutz Jaenicke*
15569 * Fix bug in SSL_clear(): bad sessions were not removed (found by
15570 Yoram Zahavi <YoramZ@gilian.com>).
15574 * Add information about CygWin 1.3 and on, and preserve proper
15575 configuration for the versions before that.
15577 *Corinna Vinschen <vinschen@redhat.com> and Richard Levitte*
15579 * Make removal from session cache (SSL_CTX_remove_session()) more robust:
15580 check whether we deal with a copy of a session and do not delete from
15581 the cache in this case. Problem reported by "Izhar Shoshani Levi"
15582 <izhar@checkpoint.com>.
15586 * Do not store session data into the internal session cache, if it
15587 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP
15588 flag is set). Proposed by Aslam <aslam@funk.com>.
15592 * Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested
15597 * [In 0.9.6d-engine release:]
15598 Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
15600 *Toomas Kiisk <vix@cyber.ee> via Richard Levitte*
15602 * Add the configuration target linux-s390x.
15604 *Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte*
15606 * The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of
15607 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag
15608 variable as an indication that a ClientHello message has been
15609 received. As the flag value will be lost between multiple
15610 invocations of ssl3_accept when using non-blocking I/O, the
15611 function may not be aware that a handshake has actually taken
15612 place, thus preventing a new session from being added to the
15615 To avoid this problem, we now set s->new_session to 2 instead of
15616 using a local variable.
15618 *Lutz Jaenicke, Bodo Moeller*
15620 * Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c)
15621 if the SSL_R_LENGTH_MISMATCH error is detected.
15623 *Geoff Thorpe, Bodo Moeller*
15625 * New 'shared_ldflag' column in Configure platform table.
15629 * Fix EVP_CIPHER_mode macro.
15631 *"Dan S. Camper" <dan@bti.net>*
15633 * Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown
15634 type, we must throw them away by setting rr->length to 0.
15636 *D P Chang <dpc@qualys.com>*
15638 ### Changes between 0.9.6b and 0.9.6c [21 dec 2001]
15640 * Fix BN_rand_range bug pointed out by Dominikus Scherkl
15641 <Dominikus.Scherkl@biodata.com>. (The previous implementation
15642 worked incorrectly for those cases where range = `10..._2` and
15643 `3*range` is two bits longer than range.)
15647 * Only add signing time to PKCS7 structures if it is not already
15652 * Fix crypto/objects/objects.h: "ld-ce" should be "id-ce",
15653 OBJ_ld_ce should be OBJ_id_ce.
15654 Also some ip-pda OIDs in crypto/objects/objects.txt were
15655 incorrect (cf. RFC 3039).
15657 *Matt Cooper, Frederic Giudicelli, Bodo Moeller*
15659 * Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid()
15660 returns early because it has nothing to do.
15662 *Andy Schneider <andy.schneider@bjss.co.uk>*
15664 * [In 0.9.6c-engine release:]
15665 Fix mutex callback return values in crypto/engine/hw_ncipher.c.
15667 *Andy Schneider <andy.schneider@bjss.co.uk>*
15669 * [In 0.9.6c-engine release:]
15670 Add support for Cryptographic Appliance's keyserver technology.
15671 (Use engine 'keyclient')
15673 *Cryptographic Appliances and Geoff Thorpe*
15675 * Add a configuration entry for OS/390 Unix. The C compiler 'c89'
15676 is called via tools/c89.sh because arguments have to be
15677 rearranged (all '-L' options must appear before the first object
15680 *Richard Shapiro <rshapiro@abinitio.com>*
15682 * [In 0.9.6c-engine release:]
15683 Add support for Broadcom crypto accelerator cards, backported
15686 *Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox*
15688 * [In 0.9.6c-engine release:]
15689 Add support for SureWare crypto accelerator cards from
15690 Baltimore Technologies. (Use engine 'sureware')
15692 *Baltimore Technologies and Mark Cox*
15694 * [In 0.9.6c-engine release:]
15695 Add support for crypto accelerator cards from Accelerated
15696 Encryption Processing, www.aep.ie. (Use engine 'aep')
15698 *AEP Inc. and Mark Cox*
15700 * Add a configuration entry for gcc on UnixWare.
15702 *Gary Benson <gbenson@redhat.com>*
15704 * Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake
15705 messages are stored in a single piece (fixed-length part and
15706 variable-length part combined) and fix various bugs found on the way.
15710 * Disable caching in BIO_gethostbyname(), directly use gethostbyname()
15711 instead. BIO_gethostbyname() does not know what timeouts are
15712 appropriate, so entries would stay in cache even when they have
15714 *Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com>*
15716 * Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when
15717 faced with a pathologically small ClientHello fragment that does
15718 not contain client_version: Instead of aborting with an error,
15719 simply choose the highest available protocol version (i.e.,
15720 TLS 1.0 unless it is disabled). In practice, ClientHello
15721 messages are never sent like this, but this change gives us
15722 strictly correct behaviour at least for TLS.
15726 * Fix SSL handshake functions and SSL_clear() such that SSL_clear()
15727 never resets s->method to s->ctx->method when called from within
15728 one of the SSL handshake functions.
15730 *Bodo Moeller; problem pointed out by Niko Baric*
15732 * In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert
15733 (sent using the client's version number) if client_version is
15734 smaller than the protocol version in use. Also change
15735 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if
15736 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then
15737 the client will at least see that alert.
15741 * Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation
15746 * Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a
15747 client receives HelloRequest while in a handshake.
15749 *Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>*
15751 * Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C
15752 should end in 'break', not 'goto end' which circumvents various
15753 cleanups done in state SSL_ST_OK. But session related stuff
15754 must be disabled for SSL_ST_OK in the case that we just sent a
15757 Also avoid some overhead by not calling ssl_init_wbio_buffer()
15758 before just sending a HelloRequest.
15760 *Bodo Moeller, Eric Rescorla <ekr@rtfm.com>*
15762 * Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't
15763 reveal whether illegal block cipher padding was found or a MAC
15764 verification error occurred. (Neither SSLerr() codes nor alerts
15765 are directly visible to potential attackers, but the information
15766 may leak via logfiles.)
15768 Similar changes are not required for the SSL 2.0 implementation
15769 because the number of padding bytes is sent in clear for SSL 2.0,
15770 and the extra bytes are just ignored. However ssl/s2_pkt.c
15771 failed to verify that the purported number of padding bytes is in
15776 * Add OpenUNIX-8 support including shared libraries
15777 (Boyd Lynn Gerber <gerberb@zenez.com>).
15781 * Improve RSA_padding_check_PKCS1_OAEP() check again to avoid
15782 'wristwatch attack' using huge encoding parameters (cf.
15783 James H. Manger's CRYPTO 2001 paper). Note that the
15784 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use
15785 encoding parameters and hence was not vulnerable.
15789 * BN_sqr() bug fix.
15791 *Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>*
15793 * Rabin-Miller test analyses assume uniformly distributed witnesses,
15794 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
15795 followed by modular reduction.
15797 *Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>*
15799 * Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
15800 equivalent based on BN_pseudo_rand() instead of BN_rand().
15804 * s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
15805 This function was broken, as the check for a new client hello message
15806 to handle SGC did not allow these large messages.
15807 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
15811 * Add alert descriptions for TLSv1 to `SSL_alert_desc_string[_long]()`.
15815 * Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
15816 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
15820 * Rework the configuration and shared library support for Tru64 Unix.
15821 The configuration part makes use of modern compiler features and
15822 still retains old compiler behavior for those that run older versions
15823 of the OS. The shared library support part includes a variant that
15824 uses the RPATH feature, and is available through the special
15825 configuration target "alpha-cc-rpath", which will never be selected
15828 *Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte*
15830 * In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
15831 with the same message size as in ssl3_get_certificate_request().
15832 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
15833 messages might inadvertently be reject as too long.
15835 *Petr Lampa <lampa@fee.vutbr.cz>*
15837 * Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
15841 * Modified SSL library such that the verify_callback that has been set
15842 specifically for an SSL object with SSL_set_verify() is actually being
15843 used. Before the change, a verify_callback set with this function was
15844 ignored and the verify_callback() set in the SSL_CTX at the time of
15845 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
15846 to allow the necessary settings.
15850 * Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
15851 explicitly to NULL, as at least on Solaris 8 this seems not always to be
15852 done automatically (in contradiction to the requirements of the C
15853 standard). This made problems when used from OpenSSH.
15857 * In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
15858 dh->length and always used
15860 BN_rand_range(priv_key, dh->p).
15862 BN_rand_range() is not necessary for Diffie-Hellman, and this
15863 specific range makes Diffie-Hellman unnecessarily inefficient if
15864 dh->length (recommended exponent length) is much smaller than the
15865 length of dh->p. We could use BN_rand_range() if the order of
15866 the subgroup was stored in the DH structure, but we only have
15871 BN_rand(priv_key, l, ...)
15873 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
15880 RSA_eay_public_encrypt
15881 RSA_eay_private_decrypt
15882 RSA_eay_private_encrypt (signing)
15883 RSA_eay_public_decrypt (signature verification)
15885 (default implementations for RSA_public_encrypt,
15886 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
15887 always reject numbers >= n.
15891 * In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
15892 to synchronize access to 'locking_thread'. This is necessary on
15893 systems where access to 'locking_thread' (an 'unsigned long'
15894 variable) is not atomic.
15898 * In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
15899 *before* setting the 'crypto_lock_rand' flag. The previous code had
15900 a race condition if 0 is a valid thread ID.
15902 *Travis Vitek <vitek@roguewave.com>*
15904 * Add support for shared libraries under Irix.
15906 *Albert Chin-A-Young <china@thewrittenword.com>*
15908 * Add configuration option to build on Linux on both big-endian and
15909 little-endian MIPS.
15911 *Ralf Baechle <ralf@uni-koblenz.de>*
15913 * Add the possibility to create shared libraries on HP-UX.
15917 ### Changes between 0.9.6a and 0.9.6b [9 Jul 2001]
15919 * Change ssleay_rand_bytes (crypto/rand/md_rand.c)
15920 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
15921 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
15922 PRNG state recovery was possible based on the output of
15923 one PRNG request appropriately sized to gain knowledge on
15924 'md' followed by enough consecutive 1-byte PRNG requests
15925 to traverse all of 'state'.
15927 1. When updating 'md_local' (the current thread's copy of 'md')
15928 during PRNG output generation, hash all of the previous
15929 'md_local' value, not just the half used for PRNG output.
15931 2. Make the number of bytes from 'state' included into the hash
15932 independent from the number of PRNG bytes requested.
15934 The first measure alone would be sufficient to avoid
15935 Markku-Juhani's attack. (Actually it had never occurred
15936 to me that the half of 'md_local' used for chaining was the
15937 half from which PRNG output bytes were taken -- I had always
15938 assumed that the secret half would be used.) The second
15939 measure makes sure that additional data from 'state' is never
15940 mixed into 'md_local' in small portions; this heuristically
15941 further strengthens the PRNG.
15945 * Fix crypto/bn/asm/mips3.s.
15949 * When only the key is given to "enc", the IV is undefined. Print out
15950 an error message in this case.
15954 * Handle special case when X509_NAME is empty in X509 printing routines.
15958 * In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
15959 positive and less than q.
15963 * Don't change `*pointer` in CRYPTO_add_lock() is add_lock_callback is
15964 used: it isn't thread safe and the add_lock_callback should handle
15967 *Paul Rose <Paul.Rose@bridge.com>*
15969 * Verify that incoming data obeys the block size in
15970 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
15976 *Ulf Möller, Bodo Möller*
15978 * The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
15979 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5
15980 when fixing the server behaviour for backwards-compatible 'client
15981 hello' messages. (Note that the attack is impractical against
15982 SSL 3.0 and TLS 1.0 anyway because length and version checking
15983 means that the probability of guessing a valid ciphertext is
15984 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
15987 Before 0.9.5, the countermeasure (hide the error by generating a
15988 random 'decryption result') did not work properly because
15989 ERR_clear_error() was missing, meaning that SSL_get_error() would
15990 detect the supposedly ignored error.
15992 Both problems are now fixed.
15996 * In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
15997 (previously it was 1024).
16001 * Fix for compatibility mode trust settings: ignore trust settings
16002 unless some valid trust or reject settings are present.
16006 * Fix for blowfish EVP: its a variable length cipher.
16010 * Fix various bugs related to DSA S/MIME verification. Handle missing
16011 parameters in DSA public key structures and return an error in the
16012 DSA routines if parameters are absent.
16016 * In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
16017 in the current directory if neither $RANDFILE nor $HOME was set.
16018 RAND_file_name() in 0.9.6a returned NULL in this case. This has
16019 caused some confusion to Windows users who haven't defined $HOME.
16020 Thus RAND_file_name() is changed again: e_os.h can define a
16021 DEFAULT_HOME, which will be used if $HOME is not set.
16022 For Windows, we use "C:"; on other platforms, we still require
16023 environment variables.
16025 * Move 'if (!initialized) RAND_poll()' into regions protected by
16026 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids
16027 having multiple threads call RAND_poll() concurrently.
16031 * In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
16032 combination of a flag and a thread ID variable.
16033 Otherwise while one thread is in ssleay_rand_bytes (which sets the
16034 flag), *other* threads can enter ssleay_add_bytes without obeying
16035 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock
16036 that they do not hold after the first thread unsets add_do_not_lock).
16040 * Change bctest again: '-x' expressions are not available in all
16041 versions of 'test'.
16045 ### Changes between 0.9.6 and 0.9.6a [5 Apr 2001]
16047 * Fix a couple of memory leaks in PKCS7_dataDecode()
16049 *Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>*
16051 * Change Configure and Makefiles to provide EXE_EXT, which will contain
16052 the default extension for executables, if any. Also, make the perl
16053 scripts that use symlink() to test if it really exists and use "cp"
16054 if it doesn't. All this made OpenSSL compilable and installable in
16059 * Fix for asn1_GetSequence() for indefinite length constructed data.
16060 If SEQUENCE is length is indefinite just set c->slen to the total
16061 amount of data available.
16063 *Steve Henson, reported by shige@FreeBSD.org*
16065 *This change does not apply to 0.9.7.*
16067 * Change bctest to avoid here-documents inside command substitution
16068 (workaround for FreeBSD /bin/sh bug).
16069 For compatibility with Ultrix, avoid shell functions (introduced
16070 in the bctest version that searches along $PATH).
16074 * Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes
16075 with des_encrypt() defined on some operating systems, like Solaris
16080 * Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
16081 On the Importance of Eliminating Errors in Cryptographic
16082 Computations, J. Cryptology 14 (2001) 2, 101-119,
16083 <http://theory.stanford.edu/~dabo/papers/faults.ps.gz>).
16087 * MIPS assembler BIGNUM division bug fix.
16091 * Disabled incorrect Alpha assembler code.
16095 * Fix PKCS#7 decode routines so they correctly update the length
16096 after reading an EOC for the EXPLICIT tag.
16100 *This change does not apply to 0.9.7.*
16102 * Fix bug in PKCS#12 key generation routines. This was triggered
16103 if a 3DES key was generated with a 0 initial byte. Include
16104 PKCS12_BROKEN_KEYGEN compilation option to retain the old
16105 (but broken) behaviour.
16109 * Enhance bctest to search for a working bc along $PATH and print
16112 *Tim Rice <tim@multitalents.net> via Richard Levitte*
16114 * Fix memory leaks in err.c: free err_data string if necessary;
16115 don't write to the wrong index in ERR_set_error_data.
16119 * Implement ssl23_peek (analogous to ssl23_read), which previously
16124 * Replace rdtsc with `_emit` statements for VC++ version 5.
16126 *Jeremy Cooper <jeremy@baymoo.org>*
16128 * Make it possible to reuse SSLv2 sessions.
16132 * In copy_email() check for >= 0 as a return value for
16133 X509_NAME_get_index_by_NID() since 0 is a valid index.
16135 *Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>*
16137 * Avoid coredump with unsupported or invalid public keys by checking if
16138 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
16139 PKCS7_verify() fails with non detached data.
16143 * Don't use getenv in library functions when run as setuid/setgid.
16144 New function OPENSSL_issetugid().
16148 * Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
16149 due to incorrect handling of multi-threading:
16151 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
16153 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
16155 3. Count how many times MemCheck_off() has been called so that
16156 nested use can be treated correctly. This also avoids
16157 inband-signalling in the previous code (which relied on the
16158 assumption that thread ID 0 is impossible).
16162 * Add "-rand" option also to s_client and s_server.
16166 * Fix CPU detection on Irix 6.x.
16167 *Kurt Hockenbury <khockenb@stevens-tech.edu> and
16168 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>*
16170 * Fix X509_NAME bug which produced incorrect encoding if X509_NAME
16175 *This change does not apply to 0.9.7.*
16177 * Use the cached encoding of an X509_NAME structure rather than
16178 copying it. This is apparently the reason for the libsafe "errors"
16179 but the code is actually correct.
16183 * Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
16184 Bleichenbacher's DSA attack.
16185 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
16186 to be set and top=0 forces the highest bit to be set; top=-1 is new
16187 and leaves the highest bit random.
16189 *Ulf Moeller, Bodo Moeller*
16191 * In the `NCONF_...`-based implementations for `CONF_...` queries
16192 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
16193 a temporary CONF structure with the data component set to NULL
16194 (which gives segmentation faults in lh_retrieve).
16195 Instead, use NULL for the CONF pointer in CONF_get_string and
16196 CONF_get_number (which may use environment variables) and directly
16197 return NULL from CONF_get_section.
16201 * Fix potential buffer overrun for EBCDIC.
16205 * Tolerate nonRepudiation as being valid for S/MIME signing and certSign
16206 keyUsage if basicConstraints absent for a CA.
16210 * Make SMIME_write_PKCS7() write mail header values with a format that
16211 is more generally accepted (no spaces before the semicolon), since
16212 some programs can't parse those values properly otherwise. Also make
16213 sure BIO's that break lines after each write do not create invalid
16218 * Make the CRL encoding routines work with empty SEQUENCE OF. The
16219 macros previously used would not encode an empty SEQUENCE OF
16220 and break the signature.
16224 *This change does not apply to 0.9.7.*
16226 * Zero the premaster secret after deriving the master secret in
16231 * Add some EVP_add_digest_alias registrations (as found in
16232 OpenSSL_add_all_digests()) to SSL_library_init()
16233 aka OpenSSL_add_ssl_algorithms(). This provides improved
16234 compatibility with peers using X.509 certificates
16235 with unconventional AlgorithmIdentifier OIDs.
16239 * Fix for Irix with NO_ASM.
16241 *"Bruce W. Forsberg" <bruce.forsberg@baesystems.com>*
16243 * ./config script fixes.
16245 *Ulf Moeller, Richard Levitte*
16247 * Fix 'openssl passwd -1'.
16251 * Change PKCS12_key_gen_asc() so it can cope with non null
16252 terminated strings whose length is passed in the passlen
16253 parameter, for example from PEM callbacks. This was done
16254 by adding an extra length parameter to asc2uni().
16256 *Steve Henson, reported by <oddissey@samsung.co.kr>*
16258 * Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
16259 call failed, free the DSA structure.
16263 * Fix to uni2asc() to cope with zero length Unicode strings.
16264 These are present in some PKCS#12 files.
16268 * Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
16269 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
16270 when writing a 32767 byte record.
16272 *Bodo Moeller; problem reported by Eric Day <eday@concentric.net>*
16274 * In `RSA_eay_public_{en,ed}crypt` and RSA_eay_mod_exp (rsa_eay.c),
16275 obtain lock CRYPTO_LOCK_RSA before setting `rsa->_method_mod_{n,p,q}`.
16277 (RSA objects have a reference count access to which is protected
16278 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
16279 so they are meant to be shared between threads.)
16280 *Bodo Moeller, Geoff Thorpe; original patch submitted by
16281 "Reddie, Steven" <Steven.Reddie@ca.com>*
16283 * Fix a deadlock in CRYPTO_mem_leaks().
16287 * Use better test patterns in bntest.
16291 * rand_win.c fix for Borland C.
16295 * BN_rshift bugfix for n == 0.
16299 * Add a 'bctest' script that checks for some known 'bc' bugs
16300 so that 'make test' does not abort just because 'bc' is broken.
16304 * Store verify_result within SSL_SESSION also for client side to
16305 avoid potential security hole. (Re-used sessions on the client side
16306 always resulted in verify_result==X509_V_OK, not using the original
16307 result of the server certificate verification.)
16311 * Fix ssl3_pending: If the record in s->s3->rrec is not of type
16312 SSL3_RT_APPLICATION_DATA, return 0.
16313 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
16318 Both ssl2_peek and ssl3_peek, which were totally broken in earlier
16319 releases, have been re-implemented by renaming the previous
16320 implementations of ssl2_read and ssl3_read to ssl2_read_internal
16321 and ssl3_read_internal, respectively, and adding 'peek' parameters
16322 to them. The new ssl[23]_{read,peek} functions are calls to
16323 ssl[23]_read_internal with the 'peek' flag set appropriately.
16324 A 'peek' parameter has also been added to ssl3_read_bytes, which
16325 does the actual work for ssl3_read_internal.
16329 * Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
16330 the method-specific "init()" handler. Also clean up ex_data after
16331 calling the method-specific "finish()" handler. Previously, this was
16332 happening the other way round.
16336 * Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
16337 The previous value, 12, was not always sufficient for BN_mod_exp().
16341 * Make sure that shared libraries get the internal name engine with
16342 the full version number and not just 0. This should mark the
16343 shared libraries as not backward compatible. Of course, this should
16344 be changed again when we can guarantee backward binary compatibility.
16348 * Fix typo in get_cert_by_subject() in by_dir.c
16350 *Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>*
16352 * Rework the system to generate shared libraries:
16354 - Make note of the expected extension for the shared libraries and
16355 if there is a need for symbolic links from for example libcrypto.so.0
16356 to libcrypto.so.0.9.7. There is extended info in Configure for
16359 - Make as few rebuilds of the shared libraries as possible.
16361 - Still avoid linking the OpenSSL programs with the shared libraries.
16363 - When installing, install the shared libraries separately from the
16368 * Fix SSL_CTX_set_read_ahead macro to actually use its argument.
16370 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
16371 and not in SSL_clear because the latter is also used by the
16372 accept/connect functions; previously, the settings made by
16373 SSL_set_read_ahead would be lost during the handshake.
16375 *Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>*
16377 * Correct util/mkdef.pl to be selective about disabled algorithms.
16378 Previously, it would create entries for disabled algorithms no
16383 * Added several new manual pages for SSL_* function.
16387 ### Changes between 0.9.5a and 0.9.6 [24 Sep 2000]
16389 * In ssl23_get_client_hello, generate an error message when faced
16390 with an initial SSL 3.0/TLS record that is too small to contain the
16391 first two bytes of the ClientHello message, i.e. client_version.
16392 (Note that this is a pathologic case that probably has never happened
16393 in real life.) The previous approach was to use the version number
16394 from the record header as a substitute; but our protocol choice
16395 should not depend on that one because it is not authenticated
16396 by the Finished messages.
16400 * More robust randomness gathering functions for Windows.
16402 *Jeffrey Altman <jaltman@columbia.edu>*
16404 * For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
16405 not set then we don't setup the error code for issuer check errors
16406 to avoid possibly overwriting other errors which the callback does
16407 handle. If an application does set the flag then we assume it knows
16408 what it is doing and can handle the new informational codes
16413 * Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
16414 a general "ANY" type, as such it should be able to decode anything
16415 including tagged types. However it didn't check the class so it would
16416 wrongly interpret tagged types in the same way as their universal
16417 counterpart and unknown types were just rejected. Changed so that the
16418 tagged and unknown types are handled in the same way as a SEQUENCE:
16419 that is the encoding is stored intact. There is also a new type
16420 "V_ASN1_OTHER" which is used when the class is not universal, in this
16421 case we have no idea what the actual type is so we just lump them all
16426 * On VMS, stdout may very well lead to a file that is written to
16427 in a record-oriented fashion. That means that every write() will
16428 write a separate record, which will be read separately by the
16429 programs trying to read from it. This can be very confusing.
16431 The solution is to put a BIO filter in the way that will buffer
16432 text until a linefeed is reached, and then write everything a
16433 line at a time, so every record written will be an actual line,
16434 not chunks of lines and not (usually doesn't happen, but I've
16435 seen it once) several lines in one record. BIO_f_linebuffer() is
16438 Currently, it's a VMS-only method, because that's where it has
16439 been tested well enough.
16443 * Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
16444 it can return incorrect results.
16445 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
16446 but it was in 0.9.6-beta[12].)
16450 * Disable the check for content being present when verifying detached
16451 signatures in pk7_smime.c. Some versions of Netscape (wrongly)
16452 include zero length content when signing messages.
16456 * New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
16457 BIO_ctrl (for BIO pairs).
16461 * Add DSO method for VMS.
16465 * Bug fix: Montgomery multiplication could produce results with the
16470 * Add RPM specification openssl.spec and modify it to build three
16471 packages. The default package contains applications, application
16472 documentation and run-time libraries. The devel package contains
16473 include files, static libraries and function documentation. The
16474 doc package contains the contents of the doc directory. The original
16475 openssl.spec was provided by Damien Miller <djm@mindrot.org>.
16479 * Add a large number of documentation files for many SSL routines.
16481 *Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>*
16483 * Add a configuration entry for Sony News 4.
16485 *NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>*
16487 * Don't set the two most significant bits to one when generating a
16488 random number < q in the DSA library.
16492 * New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default
16493 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
16494 the underlying transport is blocking) if a handshake took place.
16495 (The default behaviour is needed by applications such as s_client
16496 and s_server that use select() to determine when to use SSL_read;
16497 but for applications that know in advance when to expect data, it
16498 just makes things more complicated.)
16502 * Add RAND_egd_bytes(), which gives control over the number of bytes read
16507 * Add a few more EBCDIC conditionals that make `req` and `x509`
16508 work better on such systems.
16510 *Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>*
16512 * Add two demo programs for PKCS12_parse() and PKCS12_create().
16513 Update PKCS12_parse() so it copies the friendlyName and the
16514 keyid to the certificates aux info.
16518 * Fix bug in PKCS7_verify() which caused an infinite loop
16519 if there was more than one signature.
16521 *Sven Uszpelkat <su@celocom.de>*
16523 * Major change in util/mkdef.pl to include extra information
16524 about each symbol, as well as presenting variables as well
16525 as functions. This change means that there's n more need
16526 to rebuild the .num files when some algorithms are excluded.
16530 * Allow the verify time to be set by an application,
16531 rather than always using the current time.
16535 * Phase 2 verify code reorganisation. The certificate
16536 verify code now looks up an issuer certificate by a
16537 number of criteria: subject name, authority key id
16538 and key usage. It also verifies self signed certificates
16539 by the same criteria. The main comparison function is
16540 X509_check_issued() which performs these checks.
16542 Lot of changes were necessary in order to support this
16543 without completely rewriting the lookup code.
16545 Authority and subject key identifier are now cached.
16547 The LHASH 'certs' is X509_STORE has now been replaced
16548 by a STACK_OF(X509_OBJECT). This is mainly because an
16549 LHASH can't store or retrieve multiple objects with
16550 the same hash value.
16552 As a result various functions (which were all internal
16553 use only) have changed to handle the new X509_STORE
16554 structure. This will break anything that messed round
16555 with X509_STORE internally.
16557 The functions X509_STORE_add_cert() now checks for an
16558 exact match, rather than just subject name.
16560 The X509_STORE API doesn't directly support the retrieval
16561 of multiple certificates matching a given criteria, however
16562 this can be worked round by performing a lookup first
16563 (which will fill the cache with candidate certificates)
16564 and then examining the cache for matches. This is probably
16565 the best we can do without throwing out X509_LOOKUP
16566 entirely (maybe later...).
16568 The X509_VERIFY_CTX structure has been enhanced considerably.
16570 All certificate lookup operations now go via a get_issuer()
16571 callback. Although this currently uses an X509_STORE it
16572 can be replaced by custom lookups. This is a simple way
16573 to bypass the X509_STORE hackery necessary to make this
16574 work and makes it possible to use more efficient techniques
16575 in future. A very simple version which uses a simple
16576 STACK for its trusted certificate store is also provided
16577 using X509_STORE_CTX_trusted_stack().
16579 The verify_cb() and verify() callbacks now have equivalents
16580 in the X509_STORE_CTX structure.
16582 X509_STORE_CTX also has a 'flags' field which can be used
16583 to customise the verify behaviour.
16587 * Add new PKCS#7 signing option PKCS7_NOSMIMECAP which
16588 excludes S/MIME capabilities.
16592 * When a certificate request is read in keep a copy of the
16593 original encoding of the signed data and use it when outputting
16594 again. Signatures then use the original encoding rather than
16595 a decoded, encoded version which may cause problems if the
16596 request is improperly encoded.
16600 * For consistency with other BIO_puts implementations, call
16601 buffer_write(b, ...) directly in buffer_puts instead of calling
16604 In BIO_puts, increment b->num_write as in BIO_write.
16606 *Peter.Sylvester@EdelWeb.fr*
16608 * Fix BN_mul_word for the case where the word is 0. (We have to use
16609 BN_zero, we may not return a BIGNUM with an array consisting of
16610 words set to zero.)
16614 * Avoid calling abort() from within the library when problems are
16615 detected, except if preprocessor symbols have been defined
16616 (such as REF_CHECK, BN_DEBUG etc.).
16620 * New openssl application 'rsautl'. This utility can be
16621 used for low-level RSA operations. DER public key
16622 BIO/fp routines also added.
16626 * New Configure entry and patches for compiling on QNX 4.
16628 *Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>*
16630 * A demo state-machine implementation was sponsored by
16631 Nuron (<http://www.nuron.com/>) and is now available in
16632 demos/state_machine.
16636 * New options added to the 'dgst' utility for signature
16637 generation and verification.
16641 * Unrecognized PKCS#7 content types are now handled via a
16642 catch all ASN1_TYPE structure. This allows unsupported
16643 types to be stored as a "blob" and an application can
16644 encode and decode it manually.
16648 * Fix various signed/unsigned issues to make a_strex.c
16649 compile under VC++.
16651 *Oscar Jacobsson <oscar.jacobsson@celocom.com>*
16653 * ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
16654 length if passed a buffer. ASN1_INTEGER_to_BN failed
16655 if passed a NULL BN and its argument was negative.
16657 *Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>*
16659 * Modification to PKCS#7 encoding routines to output definite
16660 length encoding. Since currently the whole structures are in
16661 memory there's not real point in using indefinite length
16662 constructed encoding. However if OpenSSL is compiled with
16663 the flag PKCS7_INDEFINITE_ENCODING the old form is used.
16667 * Added BIO_vprintf() and BIO_vsnprintf().
16671 * Added more prefixes to parse for in the strings written
16672 through a logging bio, to cover all the levels that are available
16673 through syslog. The prefixes are now:
16675 PANIC, EMERG, EMR => LOG_EMERG
16676 ALERT, ALR => LOG_ALERT
16677 CRIT, CRI => LOG_CRIT
16678 ERROR, ERR => LOG_ERR
16679 WARNING, WARN, WAR => LOG_WARNING
16680 NOTICE, NOTE, NOT => LOG_NOTICE
16681 INFO, INF => LOG_INFO
16682 DEBUG, DBG => LOG_DEBUG
16684 and as before, if none of those prefixes are present at the
16685 beginning of the string, LOG_ERR is chosen.
16687 On Win32, the `LOG_*` levels are mapped according to this:
16689 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE
16690 LOG_WARNING => EVENTLOG_WARNING_TYPE
16691 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE
16695 * Made it possible to reconfigure with just the configuration
16696 argument "reconf" or "reconfigure". The command line arguments
16697 are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
16698 and are retrieved from there when reconfiguring.
16704 *Assar Westerlund <assar@sics.se>, Richard Levitte*
16706 * Add the arguments -CAfile and -CApath to the pkcs12 utility.
16710 * The obj_dat.pl script was messing up the sorting of object
16711 names. The reason was that it compared the quoted version
16712 of strings as a result "OCSP" > "OCSP Signing" because
16713 " > SPACE. Changed script to store unquoted versions of
16714 names and add quotes on output. It was also omitting some
16715 names from the lookup table if they were given a default
16716 value (that is if SN is missing it is given the same
16717 value as LN and vice versa), these are now added on the
16718 grounds that if an object has a name we should be able to
16719 look it up. Finally added warning output when duplicate
16720 short or long names are found.
16724 * Changes needed for Tandem NSK.
16726 *Scott Uroff <scott@xypro.com>*
16728 * Fix SSL 2.0 rollback checking: Due to an off-by-one error in
16729 RSA_padding_check_SSLv23(), special padding was never detected
16730 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
16731 version rollback attacks was not effective.
16733 In s23_clnt.c, don't use special rollback-attack detection padding
16734 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
16735 client; similarly, in s23_srvr.c, don't do the rollback check if
16736 SSL 2.0 is the only protocol enabled in the server.
16740 * Make it possible to get hexdumps of unprintable data with 'openssl
16741 asn1parse'. By implication, the functions ASN1_parse_dump() and
16742 BIO_dump_indent() are added.
16746 * New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
16747 these print out strings and name structures based on various
16748 flags including RFC2253 support and proper handling of
16749 multibyte characters. Added options to the 'x509' utility
16750 to allow the various flags to be set.
16754 * Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
16755 Also change the functions X509_cmp_current_time() and
16756 X509_gmtime_adj() work with an ASN1_TIME structure,
16757 this will enable certificates using GeneralizedTime in validity
16758 dates to be checked.
16762 * Make the NEG_PUBKEY_BUG code (which tolerates invalid
16763 negative public key encodings) on by default,
16764 NO_NEG_PUBKEY_BUG can be set to disable it.
16768 * New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
16769 content octets. An i2c_ASN1_OBJECT is unnecessary because
16770 the encoding can be trivially obtained from the structure.
16774 * crypto/err.c locking bugfix: Use write locks (`CRYPTO_w_[un]lock`),
16775 not read locks (`CRYPTO_r_[un]lock`).
16779 * A first attempt at creating official support for shared
16780 libraries through configuration. I've kept it so the
16781 default is static libraries only, and the OpenSSL programs
16782 are always statically linked for now, but there are
16783 preparations for dynamic linking in place.
16784 This has been tested on Linux and Tru64.
16788 * Randomness polling function for Win9x, as described in:
16789 Peter Gutmann, Software Generation of Practically Strong
16794 * Fix so PRNG is seeded in req if using an already existing
16799 * New options to smime application. -inform and -outform
16800 allow alternative formats for the S/MIME message including
16801 PEM and DER. The -content option allows the content to be
16802 specified separately. This should allow things like Netscape
16803 form signing output easier to verify.
16807 * Fix the ASN1 encoding of tags using the 'long form'.
16811 * New ASN1 functions, `i2c_*` and `c2i_*` for INTEGER and BIT
16812 STRING types. These convert content octets to and from the
16813 underlying type. The actual tag and length octets are
16814 already assumed to have been read in and checked. These
16815 are needed because all other string types have virtually
16816 identical handling apart from the tag. By having versions
16817 of the ASN1 functions that just operate on content octets
16818 IMPLICIT tagging can be handled properly. It also allows
16819 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
16820 and ASN1_INTEGER are identical apart from the tag.
16824 * Change the handling of OID objects as follows:
16826 - New object identifiers are inserted in objects.txt, following
16827 the syntax given in [crypto/objects/README.md](crypto/objects/README.md).
16828 - objects.pl is used to process obj_mac.num and create a new
16830 - obj_dat.pl is used to create a new obj_dat.h, using the data in
16833 This is currently kind of a hack, and the perl code in objects.pl
16834 isn't very elegant, but it works as I intended. The simplest way
16835 to check that it worked correctly is to look in obj_dat.h and
16836 check the array nid_objs and make sure the objects haven't moved
16837 around (this is important!). Additions are OK, as well as
16838 consistent name changes.
16842 * Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
16846 * Addition of the command line parameter '-rand file' to 'openssl req'.
16847 The given file adds to whatever has already been seeded into the
16848 random pool through the RANDFILE configuration file option or
16849 environment variable, or the default random state file.
16853 * mkstack.pl now sorts each macro group into lexical order.
16854 Previously the output order depended on the order the files
16855 appeared in the directory, resulting in needless rewriting
16860 * Patches to make OpenSSL compile under Win32 again. Mostly
16861 work arounds for the VC++ problem that it treats func() as
16862 func(void). Also stripped out the parts of mkdef.pl that
16863 added extra typesafe functions: these no longer exist.
16867 * Reorganisation of the stack code. The macros are now all
16868 collected in safestack.h . Each macro is defined in terms of
16869 a "stack macro" of the form `SKM_<name>(type, a, b)`. The
16870 DEBUG_SAFESTACK is now handled in terms of function casts,
16871 this has the advantage of retaining type safety without the
16872 use of additional functions. If DEBUG_SAFESTACK is not defined
16873 then the non typesafe macros are used instead. Also modified the
16874 mkstack.pl script to handle the new form. Needs testing to see
16875 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
16876 the default if no major problems. Similar behaviour for ASN1_SET_OF
16877 and PKCS12_STACK_OF.
16881 * When some versions of IIS use the 'NET' form of private key the
16882 key derivation algorithm is different. Normally MD5(password) is
16883 used as a 128 bit RC4 key. In the modified case
16884 MD5(MD5(password) + "SGCKEYSALT") is used instead. Added some
16885 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
16886 as the old Netscape_RSA functions except they have an additional
16887 'sgckey' parameter which uses the modified algorithm. Also added
16888 an -sgckey command line option to the rsa utility. Thanks to
16889 Adrian Peck <bertie@ncipher.com> for posting details of the modified
16890 algorithm to openssl-dev.
16894 * The evp_local.h macros were using 'c.##kname' which resulted in
16895 invalid expansion on some systems (SCO 5.0.5 for example).
16896 Corrected to 'c.kname'.
16898 *Phillip Porch <root@theporch.com>*
16900 * New X509_get1_email() and X509_REQ_get1_email() functions that return
16901 a STACK of email addresses from a certificate or request, these look
16902 in the subject name and the subject alternative name extensions and
16903 omit any duplicate addresses.
16907 * Re-implement BN_mod_exp2_mont using independent (and larger) windows.
16908 This makes DSA verification about 2 % faster.
16912 * Increase maximum window size in `BN_mod_exp_...` to 6 bits instead of 5
16913 (meaning that now 2^5 values will be precomputed, which is only 4 KB
16914 plus overhead for 1024 bit moduli).
16915 This makes exponentiations about 0.5 % faster for 1024 bit
16916 exponents (as measured by "openssl speed rsa2048").
16920 * Rename memory handling macros to avoid conflicts with other
16922 Malloc => OPENSSL_malloc
16923 Malloc_locked => OPENSSL_malloc_locked
16924 Realloc => OPENSSL_realloc
16925 Free => OPENSSL_free
16929 * New function BN_mod_exp_mont_word for small bases (roughly 15%
16930 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
16934 * CygWin32 support.
16936 *John Jarvie <jjarvie@newsguy.com>*
16938 * The type-safe stack code has been rejigged. It is now only compiled
16939 in when OpenSSL is configured with the DEBUG_SAFESTACK option and
16940 by default all type-specific stack functions are "#define"d back to
16941 standard stack functions. This results in more streamlined output
16942 but retains the type-safety checking possibilities of the original
16947 * The STACK code has been cleaned up, and certain type declarations
16948 that didn't make a lot of sense have been brought in line. This has
16949 also involved a cleanup of sorts in safestack.h to more correctly
16950 map type-safe stack functions onto their plain stack counterparts.
16951 This work has also resulted in a variety of "const"ifications of
16952 lots of the code, especially `_cmp` operations which should normally
16953 be prototyped with "const" parameters anyway.
16957 * When generating bytes for the first time in md_rand.c, 'stir the pool'
16958 by seeding with STATE_SIZE dummy bytes (with zero entropy count).
16959 (The PRNG state consists of two parts, the large pool 'state' and 'md',
16960 where all of 'md' is used each time the PRNG is used, but 'state'
16961 is used only indexed by a cyclic counter. As entropy may not be
16962 well distributed from the beginning, 'md' is important as a
16963 chaining variable. However, the output function chains only half
16964 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains
16965 all of 'md', and seeding with STATE_SIZE dummy bytes will result
16966 in all of 'state' being rewritten, with the new values depending
16967 on virtually all of 'md'. This overcomes the 80 bit limitation.)
16971 * In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when
16972 the handshake is continued after ssl_verify_cert_chain();
16973 otherwise, if SSL_VERIFY_NONE is set, remaining error codes
16974 can lead to 'unexplainable' connection aborts later.
16976 *Bodo Moeller; problem tracked down by Lutz Jaenicke*
16978 * Major EVP API cipher revision.
16979 Add hooks for extra EVP features. This allows various cipher
16980 parameters to be set in the EVP interface. Support added for variable
16981 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and
16982 setting of RC2 and RC5 parameters.
16984 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length
16987 Remove lots of duplicated code from the EVP library. For example *every*
16988 cipher init() function handles the 'iv' in the same way according to the
16989 cipher mode. They also all do nothing if the 'key' parameter is NULL and
16990 for CFB and OFB modes they zero ctx->num.
16992 New functionality allows removal of S/MIME code RC2 hack.
16994 Most of the routines have the same form and so can be declared in terms
16997 By shifting this to the top level EVP_CipherInit() it can be removed from
16998 all individual ciphers. If the cipher wants to handle IVs or keys
16999 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT
17002 Change lots of functions like EVP_EncryptUpdate() to now return a
17003 value: although software versions of the algorithms cannot fail
17004 any installed hardware versions can.
17008 * Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if
17009 this option is set, tolerate broken clients that send the negotiated
17010 protocol version number instead of the requested protocol version
17015 * Call dh_tmp_cb (set by `..._TMP_DH_CB`) with correct 'is_export' flag;
17016 i.e. non-zero for export ciphersuites, zero otherwise.
17017 Previous versions had this flag inverted, inconsistent with
17018 rsa_tmp_cb (..._TMP_RSA_CB).
17020 *Bodo Moeller; problem reported by Amit Chopra*
17022 * Add missing DSA library text string. Work around for some IIS
17023 key files with invalid SEQUENCE encoding.
17027 * Add a document (doc/standards.txt) that list all kinds of standards
17028 and so on that are implemented in OpenSSL.
17032 * Enhance c_rehash script. Old version would mishandle certificates
17033 with the same subject name hash and wouldn't handle CRLs at all.
17034 Added -fingerprint option to crl utility, to support new c_rehash
17039 * Eliminate non-ANSI declarations in crypto.h and stack.h.
17043 * Fix for SSL server purpose checking. Server checking was
17044 rejecting certificates which had extended key usage present
17045 but no ssl client purpose.
17047 *Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>*
17049 * Make PKCS#12 code work with no password. The PKCS#12 spec
17050 is a little unclear about how a blank password is handled.
17051 Since the password in encoded as a BMPString with terminating
17052 double NULL a zero length password would end up as just the
17053 double NULL. However no password at all is different and is
17054 handled differently in the PKCS#12 key generation code. NS
17055 treats a blank password as zero length. MSIE treats it as no
17056 password on export: but it will try both on import. We now do
17057 the same: PKCS12_parse() tries zero length and no password if
17058 the password is set to "" or NULL (NULL is now a valid password:
17059 it wasn't before) as does the pkcs12 application.
17063 * Bugfixes in `apps/x509.c`: Avoid a memory leak; and don't use
17064 perror when PEM_read_bio_X509_REQ fails, the error message must
17065 be obtained from the error queue.
17069 * Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing
17070 it in ERR_remove_state if appropriate, and change ERR_get_state
17071 accordingly to avoid race conditions (this is necessary because
17072 thread_hash is no longer constant once set).
17076 * Bugfix for linux-elf makefile.one.
17080 * RSA_get_default_method() will now cause a default
17081 RSA_METHOD to be chosen if one doesn't exist already.
17082 Previously this was only set during a call to RSA_new()
17083 or RSA_new_method(NULL) meaning it was possible for
17084 RSA_get_default_method() to return NULL.
17088 * Added native name translation to the existing DSO code
17089 that will convert (if the flag to do so is set) filenames
17090 that are sufficiently small and have no path information
17091 into a canonical native form. Eg. "blah" converted to
17092 "libblah.so" or "blah.dll" etc.
17096 * New function ERR_error_string_n(e, buf, len) which is like
17097 ERR_error_string(e, buf), but writes at most 'len' bytes
17098 including the 0 terminator. For ERR_error_string_n, 'buf'
17101 *Damien Miller <djm@mindrot.org>, Bodo Moeller*
17103 * CONF library reworked to become more general. A new CONF
17104 configuration file reader "class" is implemented as well as a
17105 new functions (`NCONF_*`, for "New CONF") to handle it. The now
17106 old `CONF_*` functions are still there, but are reimplemented to
17107 work in terms of the new functions. Also, a set of functions
17108 to handle the internal storage of the configuration data is
17109 provided to make it easier to write new configuration file
17110 reader "classes" (I can definitely see something reading a
17111 configuration file in XML format, for example), called `_CONF_*`,
17112 or "the configuration storage API"...
17114 The new configuration file reading functions are:
17116 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio,
17117 NCONF_get_section, NCONF_get_string, NCONF_get_numbre
17119 NCONF_default, NCONF_WIN32
17121 NCONF_dump_fp, NCONF_dump_bio
17123 NCONF_default and NCONF_WIN32 are method (or "class") choosers,
17124 NCONF_new creates a new CONF object. This works in the same way
17125 as other interfaces in OpenSSL, like the BIO interface.
17126 `NCONF_dump_*` dump the internal storage of the configuration file,
17127 which is useful for debugging. All other functions take the same
17128 arguments as the old `CONF_*` functions with the exception of the
17129 first that must be a `CONF *` instead of a `LHASH *`.
17131 To make it easier to use the new classes with the old `CONF_*` functions,
17132 the function CONF_set_default_method is provided.
17136 * Add '-tls1' option to 'openssl ciphers', which was already
17137 mentioned in the documentation but had not been implemented.
17138 (This option is not yet really useful because even the additional
17139 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.)
17143 * Initial DSO code added into libcrypto for letting OpenSSL (and
17144 OpenSSL-based applications) load shared libraries and bind to
17145 them in a portable way.
17147 *Geoff Thorpe, with contributions from Richard Levitte*
17149 ### Changes between 0.9.5 and 0.9.5a [1 Apr 2000]
17151 * Make sure _lrotl and _lrotr are only used with MSVC.
17153 * Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status
17154 (the default implementation of RAND_status).
17156 * Rename openssl x509 option '-crlext', which was added in 0.9.5,
17157 to '-clrext' (= clear extensions), as intended and documented.
17158 *Bodo Moeller; inconsistency pointed out by Michael Attili
17159 <attili@amaxo.com>*
17161 * Fix for HMAC. It wasn't zeroing the rest of the block if the key length
17162 was larger than the MD block size.
17164 *Steve Henson, pointed out by Yost William <YostW@tce.com>*
17166 * Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument
17167 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set()
17168 using the passed key: if the passed key was a private key the result
17169 of X509_print(), for example, would be to print out all the private key
17174 * des_quad_cksum() byte order bug fix.
17175 *Ulf Möller, using the problem description in krb4-0.9.7, where
17176 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>*
17178 * Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly
17181 *Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>*
17183 * For easily testing in shell scripts whether some command
17184 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX'
17185 returns with exit code 0 iff no command of the given name is available.
17186 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases,
17187 the output goes to stdout and nothing is printed to stderr.
17188 Additional arguments are always ignored.
17190 Since for each cipher there is a command of the same name,
17191 the 'no-cipher' compilation switches can be tested this way.
17193 ('openssl no-XXX' is not able to detect pseudo-commands such
17194 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.)
17198 * Update test suite so that 'make test' succeeds in 'no-rsa' configuration.
17202 * For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE
17203 is set; it will be thrown away anyway because each handshake creates
17205 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition
17206 to parameters -- in previous versions (since OpenSSL 0.9.3) the
17207 'default key' from SSL_CTX_set_tmp_dh would always be lost, meaning
17208 you effectively got SSL_OP_SINGLE_DH_USE when using this macro.
17212 * New s_client option -ign_eof: EOF at stdin is ignored, and
17213 'Q' and 'R' lose their special meanings (quit/renegotiate).
17214 This is part of what -quiet does; unlike -quiet, -ign_eof
17215 does not suppress any output.
17219 * Add compatibility options to the purpose and trust code. The
17220 purpose X509_PURPOSE_ANY is "any purpose" which automatically
17221 accepts a certificate or CA, this was the previous behaviour,
17222 with all the associated security issues.
17224 X509_TRUST_COMPAT is the old trust behaviour: only and
17225 automatically trust self signed roots in certificate store. A
17226 new trust setting X509_TRUST_DEFAULT is used to specify that
17227 a purpose has no associated trust setting and it should instead
17228 use the value in the default purpose.
17232 * Fix the PKCS#8 DSA private key code so it decodes keys again
17233 and fix a memory leak.
17237 * In util/mkerr.pl (which implements 'make errors'), preserve
17238 reason strings from the previous version of the .c file, as
17239 the default to have only downcase letters (and digits) in
17240 automatically generated reasons codes is not always appropriate.
17244 * In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table
17245 using strerror. Previously, ERR_reason_error_string() returned
17246 library names as reason strings for SYSerr; but SYSerr is a special
17247 case where small numbers are errno values, not library numbers.
17251 * Add '-dsaparam' option to 'openssl dhparam' application. This
17252 converts DSA parameters into DH parameters. (When creating parameters,
17253 DSA_generate_parameters is used.)
17257 * Include 'length' (recommended exponent length) in C code generated
17258 by 'openssl dhparam -C'.
17262 * The second argument to set_label in perlasm was already being used
17263 so couldn't be used as a "file scope" flag. Moved to third argument
17268 * In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes
17269 instead of RAND_bytes for encryption IVs and salts.
17273 * Include RAND_status() into RAND_METHOD instead of implementing
17274 it only for md_rand.c Otherwise replacing the PRNG by calling
17275 RAND_set_rand_method would be impossible.
17279 * Don't let DSA_generate_key() enter an infinite loop if the random
17280 number generation fails.
17284 * New 'rand' application for creating pseudo-random output.
17288 * Added configuration support for Linux/IA64
17290 *Rolf Haberrecker <rolf@suse.de>*
17292 * Assembler module support for Mingw32.
17296 * Shared library support for HPUX (in shlib/).
17298 *Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous*
17300 * Shared library support for Solaris gcc.
17302 *Lutz Behnke <behnke@trustcenter.de>*
17304 ### Changes between 0.9.4 and 0.9.5 [28 Feb 2000]
17306 * PKCS7_encrypt() was adding text MIME headers twice because they
17307 were added manually and by SMIME_crlf_copy().
17311 * In bntest.c don't call BN_rand with zero bits argument.
17313 *Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>*
17315 * BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n]
17316 case was implemented. This caused BN_div_recp() to fail occasionally.
17320 * Add an optional second argument to the set_label() in the perl
17321 assembly language builder. If this argument exists and is set
17322 to 1 it signals that the assembler should use a symbol whose
17323 scope is the entire file, not just the current function. This
17324 is needed with MASM which uses the format label:: for this scope.
17326 *Steve Henson, pointed out by Peter Runestig <peter@runestig.com>*
17328 * Change the ASN1 types so they are typedefs by default. Before
17329 almost all types were #define'd to ASN1_STRING which was causing
17330 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING)
17335 * Change names of new functions to the new get1/get0 naming
17336 convention: After 'get1', the caller owns a reference count
17337 and has to call `..._free`; 'get0' returns a pointer to some
17338 data structure without incrementing reference counters.
17339 (Some of the existing 'get' functions increment a reference
17340 counter, some don't.)
17341 Similarly, 'set1' and 'add1' functions increase reference
17342 counters or duplicate objects.
17346 * Allow for the possibility of temp RSA key generation failure:
17347 the code used to assume it always worked and crashed on failure.
17351 * Fix potential buffer overrun problem in BIO_printf().
17352 *Ulf Möller, using public domain code by Patrick Powell; problem
17353 pointed out by David Sacerdote <das33@cornell.edu>*
17355 * Support EGD <http://www.lothar.com/tech/crypto/>. New functions
17356 RAND_egd() and RAND_status(). In the command line application,
17357 the EGD socket can be specified like a seed file using RANDFILE
17362 * Allow the string CERTIFICATE to be tolerated in PKCS#7 structures.
17363 Some CAs (e.g. Verisign) distribute certificates in this form.
17367 * Remove the SSL_ALLOW_ADH compile option and set the default cipher
17368 list to exclude them. This means that no special compilation option
17369 is needed to use anonymous DH: it just needs to be included in the
17374 * Change the EVP_MD_CTX_type macro so its meaning consistent with
17375 EVP_MD_type. The old functionality is available in a new macro called
17376 EVP_MD_md(). Change code that uses it and update docs.
17380 * `..._ctrl` functions now have corresponding `..._callback_ctrl` functions
17381 where the `void *` argument is replaced by a function pointer argument.
17382 Previously `void *` was abused to point to functions, which works on
17383 many platforms, but is not correct. As these functions are usually
17384 called by macros defined in OpenSSL header files, most source code
17385 should work without changes.
17389 * `<openssl/opensslconf.h>` (which is created by Configure) now contains
17390 sections with information on -D... compiler switches used for
17391 compiling the library so that applications can see them. To enable
17392 one of these sections, a pre-processor symbol `OPENSSL_..._DEFINES`
17393 must be defined. E.g.,
17394 #define OPENSSL_ALGORITHM_DEFINES
17395 #include <openssl/opensslconf.h>
17396 defines all pertinent `NO_<algo>` symbols, such as NO_IDEA, NO_RSA, etc.
17398 *Richard Levitte, Ulf and Bodo Möller*
17400 * Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS
17405 * Change the 'other' type in certificate aux info to a STACK_OF
17406 X509_ALGOR. Although not an AlgorithmIdentifier as such it has
17407 the required ASN1 format: arbitrary types determined by an OID.
17411 * Add some PEM_write_X509_REQ_NEW() functions and a command line
17412 argument to 'req'. This is not because the function is newer or
17413 better than others it just uses the work 'NEW' in the certificate
17414 request header lines. Some software needs this.
17418 * Reorganise password command line arguments: now passwords can be
17419 obtained from various sources. Delete the PEM_cb function and make
17420 it the default behaviour: i.e. if the callback is NULL and the
17421 usrdata argument is not NULL interpret it as a null terminated pass
17422 phrase. If usrdata and the callback are NULL then the pass phrase
17423 is prompted for as usual.
17427 * Add support for the Compaq Atalla crypto accelerator. If it is installed,
17428 the support is automatically enabled. The resulting binaries will
17429 autodetect the card and use it if present.
17431 *Ben Laurie and Compaq Inc.*
17433 * Work around for Netscape hang bug. This sends certificate request
17434 and server done in one record. Since this is perfectly legal in the
17435 SSL/TLS protocol it isn't a "bug" option and is on by default. See
17436 the bugs/SSLv3 entry for more info.
17440 * HP-UX tune-up: new unified configs, HP C compiler bug workaround.
17444 * Add -rand argument to smime and pkcs12 applications and read/write
17449 * New 'passwd' tool for crypt(3) and apr1 password hashes.
17453 * Add command line password options to the remaining applications.
17457 * Bug fix for BN_div_recp() for numerators with an even number of
17462 * More tests in bntest.c, and changed test_bn output.
17466 * ./config recognizes MacOS X now.
17470 * Bug fix for BN_div() when the first words of num and divisor are
17471 equal (it gave wrong results if `(rem=(n1-q*d0)&BN_MASK2) < d0)`.
17475 * Add support for various broken PKCS#8 formats, and command line
17476 options to produce them.
17480 * New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to
17481 get temporary BIGNUMs from a BN_CTX.
17485 * Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont()
17490 * Change the `SSLeay_add_all_*()` functions to `OpenSSL_add_all_*()` and
17491 include a #define from the old name to the new. The original intent
17492 was that statically linked binaries could for example just call
17493 SSLeay_add_all_ciphers() to just add ciphers to the table and not
17494 link with digests. This never worked because SSLeay_add_all_digests()
17495 and SSLeay_add_all_ciphers() were in the same source file so calling
17496 one would link with the other. They are now in separate source files.
17500 * Add a new -notext option to 'ca' and a -pubkey option to 'spkac'.
17504 * Use a less unusual form of the Miller-Rabin primality test (it used
17505 a binary algorithm for exponentiation integrated into the Miller-Rabin
17506 loop, our standard modexp algorithms are faster).
17510 * Support for the EBCDIC character set completed.
17512 *Martin Kraemer <Martin.Kraemer@Mch.SNI.De>*
17514 * Source code cleanups: use const where appropriate, eliminate casts,
17515 use `void *` instead of `char *` in lhash.
17519 * Bugfix: ssl3_send_server_key_exchange was not restartable
17520 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of
17521 this the server could overwrite ephemeral keys that the client
17526 * Turn DSA_is_prime into a macro that calls BN_is_prime,
17527 using 50 iterations of the Rabin-Miller test.
17529 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50
17530 iterations of the Rabin-Miller test as required by the appendix
17531 to FIPS PUB 186[-1]) instead of DSA_is_prime.
17532 As BN_is_prime_fasttest includes trial division, DSA parameter
17533 generation becomes much faster.
17535 This implies a change for the callback functions in DSA_is_prime
17536 and DSA_generate_parameters: The callback function is called once
17537 for each positive witness in the Rabin-Miller test, not just
17538 occasionally in the inner loop; and the parameters to the
17539 callback function now provide an iteration count for the outer
17540 loop rather than for the current invocation of the inner loop.
17541 DSA_generate_parameters additionally can call the callback
17542 function with an 'iteration count' of -1, meaning that a
17543 candidate has passed the trial division test (when q is generated
17544 from an application-provided seed, trial division is skipped).
17548 * New function BN_is_prime_fasttest that optionally does trial
17549 division before starting the Rabin-Miller test and has
17550 an additional BN_CTX * argument (whereas BN_is_prime always
17551 has to allocate at least one BN_CTX).
17552 'callback(1, -1, cb_arg)' is called when a number has passed the
17553 trial division stage.
17557 * Fix for bug in CRL encoding. The validity dates weren't being handled
17562 * New -pkcs12 option to CA.pl script to write out a PKCS#12 file.
17566 * New function BN_pseudo_rand().
17570 * Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable)
17571 bignum version of BN_from_montgomery() with the working code from
17572 SSLeay 0.9.0 (the word based version is faster anyway), and clean up
17577 * Avoid a race condition in s2_clnt.c (function get_server_hello) that
17578 made it impossible to use the same SSL_SESSION data structure in
17579 SSL2 clients in multiple threads.
17583 * The return value of RAND_load_file() no longer counts bytes obtained
17584 by stat(). RAND_load_file(..., -1) is new and uses the complete file
17585 to seed the PRNG (previously an explicit byte count was required).
17587 *Ulf Möller, Bodo Möller*
17589 * Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes
17590 used `char *` instead of `void *` and had casts all over the place.
17594 * Make BN_generate_prime() return NULL on error if ret!=NULL.
17598 * Retain source code compatibility for BN_prime_checks macro:
17599 BN_is_prime(..., BN_prime_checks, ...) now uses
17600 BN_prime_checks_for_size to determine the appropriate number of
17601 Rabin-Miller iterations.
17605 * Diffie-Hellman uses "safe" primes: DH_check() return code renamed to
17606 DH_CHECK_P_NOT_SAFE_PRIME.
17607 (Check if this is true? OpenPGP calls them "strong".)
17611 * Merge the functionality of "dh" and "gendh" programs into a new program
17612 "dhparam". The old programs are retained for now but will handle DH keys
17613 (instead of parameters) in future.
17617 * Make the ciphers, s_server and s_client programs check the return values
17618 when a new cipher list is set.
17622 * Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit
17623 ciphers. Before when the 56bit ciphers were enabled the sorting was
17626 The syntax for the cipher sorting has been extended to support sorting by
17627 cipher-strength (using the strength_bits hard coded in the tables).
17628 The new command is `@STRENGTH` (see also `doc/apps/ciphers.pod`).
17630 Fix a bug in the cipher-command parser: when supplying a cipher command
17631 string with an "undefined" symbol (neither command nor alphanumeric
17632 *A-Za-z0-9*, ssl_set_cipher_list used to hang in an endless loop. Now
17633 an error is flagged.
17635 Due to the strength-sorting extension, the code of the
17636 ssl_create_cipher_list() function was completely rearranged. I hope that
17637 the readability was also increased :-)
17639 *Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>*
17641 * Minor change to 'x509' utility. The -CAcreateserial option now uses 1
17642 for the first serial number and places 2 in the serial number file. This
17643 avoids problems when the root CA is created with serial number zero and
17644 the first user certificate has the same issuer name and serial number
17649 * Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses
17650 the new code. Add documentation for this stuff.
17654 * Changes to X509_ATTRIBUTE utilities. These have been renamed from
17655 `X509_*()` to `X509at_*()` on the grounds that they don't handle X509
17656 structures and behave in an analogous way to the X509v3 functions:
17657 they shouldn't be called directly but wrapper functions should be used
17660 So we also now have some wrapper functions that call the X509at functions
17661 when passed certificate requests. (TO DO: similar things can be done with
17662 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other
17663 things. Some of these need some d2i or i2d and print functionality
17664 because they handle more complex structures.)
17668 * Add missing #ifndefs that caused missing symbols when building libssl
17669 as a shared library without RSA. Use #ifndef NO_SSL2 instead of
17670 NO_RSA in `ssl/s2*.c`.
17672 *Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf Möller*
17674 * Precautions against using the PRNG uninitialized: RAND_bytes() now
17675 has a return value which indicates the quality of the random data
17676 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's
17677 error queue. New function RAND_pseudo_bytes() generates output that is
17678 guaranteed to be unique but not unpredictable. RAND_add is like
17679 RAND_seed, but takes an extra argument for an entropy estimate
17680 (RAND_seed always assumes full entropy).
17684 * Do more iterations of Rabin-Miller probable prime test (specifically,
17685 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes
17686 instead of only 2 for all lengths; see BN_prime_checks_for_size definition
17687 in crypto/bn/bn_prime.c for the complete table). This guarantees a
17688 false-positive rate of at most 2^-80 for random input.
17692 * Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs.
17696 * New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain
17697 in the 0.9.5 release), this returns the chain
17698 from an X509_CTX structure with a dup of the stack and all
17699 the X509 reference counts upped: so the stack will exist
17700 after X509_CTX_cleanup() has been called. Modify pkcs12.c
17703 Also make SSL_SESSION_print() print out the verify return
17708 * Add manpage for the pkcs12 command. Also change the default
17709 behaviour so MAC iteration counts are used unless the new
17710 -nomaciter option is used. This improves file security and
17711 only older versions of MSIE (4.0 for example) need it.
17715 * Honor the no-xxx Configure options when creating .DEF files.
17719 * Add PKCS#10 attributes to field table: challengePassword,
17720 unstructuredName and unstructuredAddress. These are taken from
17721 draft PKCS#9 v2.0 but are compatible with v1.2 provided no
17722 international characters are used.
17724 More changes to X509_ATTRIBUTE code: allow the setting of types
17725 based on strings. Remove the 'loc' parameter when adding
17726 attributes because these will be a SET OF encoding which is sorted
17731 * Initial changes to the 'req' utility to allow request generation
17732 automation. This will allow an application to just generate a template
17733 file containing all the field values and have req construct the
17736 Initial support for X509_ATTRIBUTE handling. Stacks of these are
17737 used all over the place including certificate requests and PKCS#7
17738 structures. They are currently handled manually where necessary with
17739 some primitive wrappers for PKCS#7. The new functions behave in a
17740 manner analogous to the X509 extension functions: they allow
17741 attributes to be looked up by NID and added.
17743 Later something similar to the X509V3 code would be desirable to
17744 automatically handle the encoding, decoding and printing of the
17745 more complex types. The string types like challengePassword can
17746 be handled by the string table functions.
17748 Also modified the multi byte string table handling. Now there is
17749 a 'global mask' which masks out certain types. The table itself
17750 can use the flag STABLE_NO_MASK to ignore the mask setting: this
17751 is useful when for example there is only one permissible type
17752 (as in countryName) and using the mask might result in no valid
17757 * Clean up 'Finished' handling, and add functions SSL_get_finished and
17758 SSL_get_peer_finished to allow applications to obtain the latest
17759 Finished messages sent to the peer or expected from the peer,
17760 respectively. (SSL_get_peer_finished is usually the Finished message
17761 actually received from the peer, otherwise the protocol will be aborted.)
17763 As the Finished message are message digests of the complete handshake
17764 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can
17765 be used for external authentication procedures when the authentication
17766 provided by SSL/TLS is not desired or is not enough.
17770 * Enhanced support for Alpha Linux is added. Now ./config checks if
17771 the host supports BWX extension and if Compaq C is present on the
17772 $PATH. Just exploiting of the BWX extension results in 20-30%
17773 performance kick for some algorithms, e.g. DES and RC4 to mention
17774 a couple. Compaq C in turn generates ~20% faster code for MD5 and
17779 * Add support for MS "fast SGC". This is arguably a violation of the
17780 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with
17781 weak crypto and after checking the certificate is SGC a second one
17782 with strong crypto. MS SGC stops the first handshake after receiving
17783 the server certificate message and sends a second client hello. Since
17784 a server will typically do all the time consuming operations before
17785 expecting any further messages from the client (server key exchange
17786 is the most expensive) there is little difference between the two.
17788 To get OpenSSL to support MS SGC we have to permit a second client
17789 hello message after we have sent server done. In addition we have to
17790 reset the MAC if we do get this second client hello.
17794 * Add a function 'd2i_AutoPrivateKey()' this will automatically decide
17795 if a DER encoded private key is RSA or DSA traditional format. Changed
17796 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional"
17797 format DER encoded private key. Newer code should use PKCS#8 format which
17798 has the key type encoded in the ASN1 structure. Added DER private key
17799 support to pkcs8 application.
17803 * SSL 3/TLS 1 servers now don't request certificates when an anonymous
17804 ciphersuites has been selected (as required by the SSL 3/TLS 1
17805 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT
17806 is set, we interpret this as a request to violate the specification
17807 (the worst that can happen is a handshake failure, and 'correct'
17808 behaviour would result in a handshake failure anyway).
17812 * In SSL_CTX_add_session, take into account that there might be multiple
17813 SSL_SESSION structures with the same session ID (e.g. when two threads
17814 concurrently obtain them from an external cache).
17815 The internal cache can handle only one SSL_SESSION with a given ID,
17816 so if there's a conflict, we now throw out the old one to achieve
17821 * Add OIDs for idea and blowfish in CBC mode. This will allow both
17822 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to
17823 some routines that use cipher OIDs: some ciphers do not have OIDs
17824 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for
17829 * Simplify the trust setting structure and code. Now we just have
17830 two sequences of OIDs for trusted and rejected settings. These will
17831 typically have values the same as the extended key usage extension
17832 and any application specific purposes.
17834 The trust checking code now has a default behaviour: it will just
17835 check for an object with the same NID as the passed id. Functions can
17836 be provided to override either the default behaviour or the behaviour
17837 for a given id. SSL client, server and email already have functions
17838 in place for compatibility: they check the NID and also return "trusted"
17839 if the certificate is self signed.
17843 * Add d2i,i2d bio/fp functions for PrivateKey: these convert the
17844 traditional format into an EVP_PKEY structure.
17848 * Add a password callback function PEM_cb() which either prompts for
17849 a password if usr_data is NULL or otherwise assumes it is a null
17850 terminated password. Allow passwords to be passed on command line
17851 environment or config files in a few more utilities.
17855 * Add a bunch of DER and PEM functions to handle PKCS#8 format private
17856 keys. Add some short names for PKCS#8 PBE algorithms and allow them
17857 to be specified on the command line for the pkcs8 and pkcs12 utilities.
17858 Update documentation.
17862 * Support for ASN1 "NULL" type. This could be handled before by using
17863 ASN1_TYPE but there wasn't any function that would try to read a NULL
17864 and produce an error if it couldn't. For compatibility we also have
17865 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and
17866 don't allocate anything because they don't need to.
17870 * Initial support for MacOS is now provided. Examine INSTALL.MacOS
17873 *Andy Polyakov, Roy Woods <roy@centicsystems.ca>*
17875 * Rebuild of the memory allocation routines used by OpenSSL code and
17876 possibly others as well. The purpose is to make an interface that
17877 provide hooks so anyone can build a separate set of allocation and
17878 deallocation routines to be used by OpenSSL, for example memory
17879 pool implementations, or something else, which was previously hard
17880 since Malloc(), Realloc() and Free() were defined as macros having
17881 the values malloc, realloc and free, respectively (except for Win32
17882 compilations). The same is provided for memory debugging code.
17883 OpenSSL already comes with functionality to find memory leaks, but
17884 this gives people a chance to debug other memory problems.
17886 With these changes, a new set of functions and macros have appeared:
17888 CRYPTO_set_mem_debug_functions() [F]
17889 CRYPTO_get_mem_debug_functions() [F]
17890 CRYPTO_dbg_set_options() [F]
17891 CRYPTO_dbg_get_options() [F]
17892 CRYPTO_malloc_debug_init() [M]
17894 The memory debug functions are NULL by default, unless the library
17895 is compiled with CRYPTO_MDEBUG or friends is defined. If someone
17896 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which
17897 gives the standard debugging functions that come with OpenSSL) or
17898 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions
17899 provided by the library user) must be used. When the standard
17900 debugging functions are used, CRYPTO_dbg_set_options can be used to
17901 request additional information:
17902 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting
17903 the CRYPTO_MDEBUG_xxx macro when compiling the library.
17905 Also, things like CRYPTO_set_mem_functions will always give the
17906 expected result (the new set of functions is used for allocation
17907 and deallocation) at all times, regardless of platform and compiler
17910 To finish it up, some functions that were never use in any other
17911 way than through macros have a new API and new semantic:
17913 CRYPTO_dbg_malloc()
17914 CRYPTO_dbg_realloc()
17917 All macros of value have retained their old syntax.
17919 *Richard Levitte and Bodo Moeller*
17921 * Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the
17922 ordering of SMIMECapabilities wasn't in "strength order" and there
17923 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature
17928 * Some ASN1 types with illegal zero length encoding (INTEGER,
17929 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines.
17931 *Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson*
17933 * Merge in my S/MIME library for OpenSSL. This provides a simple
17934 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough
17935 functionality to handle multipart/signed properly) and a utility
17936 called 'smime' to call all this stuff. This is based on code I
17937 originally wrote for Celo who have kindly allowed it to be
17938 included in OpenSSL.
17942 * Add variants des_set_key_checked and des_set_key_unchecked of
17943 des_set_key (aka des_key_sched). Global variable des_check_key
17944 decides which of these is called by des_set_key; this way
17945 des_check_key behaves as it always did, but applications and
17946 the library itself, which was buggy for des_check_key == 1,
17947 have a cleaner way to pick the version they need.
17951 * New function PKCS12_newpass() which changes the password of a
17956 * Modify X509_TRUST and X509_PURPOSE so it also uses a static and
17957 dynamic mix. In both cases the ids can be used as an index into the
17958 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add()
17959 functions so they accept a list of the field values and the
17960 application doesn't need to directly manipulate the X509_TRUST
17965 * Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't
17970 * Modify the way the V3 extension code looks up extensions. This now
17971 works in a similar way to the object code: we have some "standard"
17972 extensions in a static table which is searched with OBJ_bsearch()
17973 and the application can add dynamic ones if needed. The file
17974 crypto/x509v3/ext_dat.h now has the info: this file needs to be
17975 updated whenever a new extension is added to the core code and kept
17976 in ext_nid order. There is a simple program 'tabtest.c' which checks
17977 this. New extensions are not added too often so this file can readily
17978 be maintained manually.
17980 There are two big advantages in doing things this way. The extensions
17981 can be looked up immediately and no longer need to be "added" using
17982 X509V3_add_standard_extensions(): this function now does nothing.
17983 Side note: I get *lots* of email saying the extension code doesn't
17984 work because people forget to call this function.
17985 Also no dynamic allocation is done unless new extensions are added:
17986 so if we don't add custom extensions there is no need to call
17987 X509V3_EXT_cleanup().
17991 * Modify enc utility's salting as follows: make salting the default. Add a
17992 magic header, so unsalted files fail gracefully instead of just decrypting
17993 to garbage. This is because not salting is a big security hole, so people
17994 should be discouraged from doing it.
17998 * Fixes and enhancements to the 'x509' utility. It allowed a message
17999 digest to be passed on the command line but it only used this
18000 parameter when signing a certificate. Modified so all relevant
18001 operations are affected by the digest parameter including the
18002 -fingerprint and -x509toreq options. Also -x509toreq choked if a
18003 DSA key was used because it didn't fix the digest.
18007 * Initial certificate chain verify code. Currently tests the untrusted
18008 certificates for consistency with the verify purpose (which is set
18009 when the X509_STORE_CTX structure is set up) and checks the pathlength.
18011 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour:
18012 this is because it will reject chains with invalid extensions whereas
18013 every previous version of OpenSSL and SSLeay made no checks at all.
18015 Trust code: checks the root CA for the relevant trust settings. Trust
18016 settings have an initial value consistent with the verify purpose: e.g.
18017 if the verify purpose is for SSL client use it expects the CA to be
18018 trusted for SSL client use. However the default value can be changed to
18019 permit custom trust settings: one example of this would be to only trust
18020 certificates from a specific "secure" set of CAs.
18022 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions
18023 which should be used for version portability: especially since the
18024 verify structure is likely to change more often now.
18026 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions
18027 to set them. If not set then assume SSL clients will verify SSL servers
18030 Two new options to the verify program: -untrusted allows a set of
18031 untrusted certificates to be passed in and -purpose which sets the
18032 intended purpose of the certificate. If a purpose is set then the
18033 new chain verify code is used to check extension consistency.
18037 * Support for the authority information access extension.
18041 * Modify RSA and DSA PEM read routines to transparently handle
18042 PKCS#8 format private keys. New *_PUBKEY_* functions that handle
18043 public keys in a format compatible with certificate
18044 SubjectPublicKeyInfo structures. Unfortunately there were already
18045 functions called *_PublicKey_* which used various odd formats so
18046 these are retained for compatibility: however the DSA variants were
18047 never in a public release so they have been deleted. Changed dsa/rsa
18048 utilities to handle the new format: note no releases ever handled public
18049 keys so we should be OK.
18051 The primary motivation for this change is to avoid the same fiasco
18052 that dogs private keys: there are several incompatible private key
18053 formats some of which are standard and some OpenSSL specific and
18054 require various evil hacks to allow partial transparent handling and
18055 even then it doesn't work with DER formats. Given the option anything
18056 other than PKCS#8 should be dumped: but the other formats have to
18057 stay in the name of compatibility.
18059 With public keys and the benefit of hindsight one standard format
18060 is used which works with EVP_PKEY, RSA or DSA structures: though
18061 it clearly returns an error if you try to read the wrong kind of key.
18063 Added a -pubkey option to the 'x509' utility to output the public key.
18064 Also rename the `EVP_PKEY_get_*()` to `EVP_PKEY_rget_*()`
18065 (renamed to `EVP_PKEY_get1_*()` in the OpenSSL 0.9.5 release) and add
18066 `EVP_PKEY_rset_*()` functions (renamed to `EVP_PKEY_set1_*()`)
18067 that do the same as the `EVP_PKEY_assign_*()` except they up the
18068 reference count of the added key (they don't "swallow" the
18073 * Fixes to crypto/x509/by_file.c the code to read in certificates and
18074 CRLs would fail if the file contained no certificates or no CRLs:
18075 added a new function to read in both types and return the number
18076 read: this means that if none are read it will be an error. The
18077 DER versions of the certificate and CRL reader would always fail
18078 because it isn't possible to mix certificates and CRLs in DER format
18079 without choking one or the other routine. Changed this to just read
18080 a certificate: this is the best we can do. Also modified the code
18081 in `apps/verify.c` to take notice of return codes: it was previously
18082 attempting to read in certificates from NULL pointers and ignoring
18083 any errors: this is one reason why the cert and CRL reader seemed
18084 to work. It doesn't check return codes from the default certificate
18085 routines: these may well fail if the certificates aren't installed.
18089 * Code to support otherName option in GeneralName.
18093 * First update to verify code. Change the verify utility
18094 so it warns if it is passed a self signed certificate:
18095 for consistency with the normal behaviour. X509_verify
18096 has been modified to it will now verify a self signed
18097 certificate if *exactly* the same certificate appears
18098 in the store: it was previously impossible to trust a
18099 single self signed certificate. This means that:
18100 openssl verify ss.pem
18101 now gives a warning about a self signed certificate but
18102 openssl verify -CAfile ss.pem ss.pem
18107 * For servers, store verify_result in SSL_SESSION data structure
18108 (and add it to external session representation).
18109 This is needed when client certificate verifications fails,
18110 but an application-provided verification callback (set by
18111 SSL_CTX_set_cert_verify_callback) allows accepting the session
18112 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK
18113 but returns 1): When the session is reused, we have to set
18114 ssl->verify_result to the appropriate error code to avoid
18117 *Bodo Moeller, problem pointed out by Lutz Jaenicke*
18119 * Fix a bug in the new PKCS#7 code: it didn't consider the
18120 case in PKCS7_dataInit() where the signed PKCS7 structure
18121 didn't contain any existing data because it was being created.
18123 *Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson*
18125 * Add a salt to the key derivation routines in enc.c. This
18126 forms the first 8 bytes of the encrypted file. Also add a
18127 -S option to allow a salt to be input on the command line.
18131 * New function X509_cmp(). Oddly enough there wasn't a function
18132 to compare two certificates. We do this by working out the SHA1
18133 hash and comparing that. X509_cmp() will be needed by the trust
18138 * SSL_get1_session() is like SSL_get_session(), but increments
18139 the reference count in the SSL_SESSION returned.
18141 *Geoff Thorpe <geoff@eu.c2.net>*
18143 * Fix for 'req': it was adding a null to request attributes.
18144 Also change the X509_LOOKUP and X509_INFO code to handle
18145 certificate auxiliary information.
18149 * Add support for 40 and 64 bit RC2 and RC4 algorithms: document
18154 * Add the possibility to add extra information to the memory leak
18155 detecting output, to form tracebacks, showing from where each
18156 allocation was originated: CRYPTO_push_info("constant string") adds
18157 the string plus current file name and line number to a per-thread
18158 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info()
18159 is like calling CYRPTO_pop_info() until the stack is empty.
18160 Also updated memory leak detection code to be multi-thread-safe.
18164 * Add options -text and -noout to pkcs7 utility and delete the
18165 encryption options which never did anything. Update docs.
18169 * Add options to some of the utilities to allow the pass phrase
18170 to be included on either the command line (not recommended on
18171 OSes like Unix) or read from the environment. Update the
18172 manpages and fix a few bugs.
18176 * Add a few manpages for some of the openssl commands.
18180 * Fix the -revoke option in ca. It was freeing up memory twice,
18181 leaking and not finding already revoked certificates.
18185 * Extensive changes to support certificate auxiliary information.
18186 This involves the use of X509_CERT_AUX structure and X509_AUX
18187 functions. An X509_AUX function such as PEM_read_X509_AUX()
18188 can still read in a certificate file in the usual way but it
18189 will also read in any additional "auxiliary information". By
18190 doing things this way a fair degree of compatibility can be
18191 retained: existing certificates can have this information added
18192 using the new 'x509' options.
18194 Current auxiliary information includes an "alias" and some trust
18195 settings. The trust settings will ultimately be used in enhanced
18196 certificate chain verification routines: currently a certificate
18197 can only be trusted if it is self signed and then it is trusted
18202 * Fix assembler for Alpha (tested only on DEC OSF not Linux or `*BSD`).
18203 The problem was that one of the replacement routines had not been working
18204 since SSLeay releases. For now the offending routine has been replaced
18205 with non-optimised assembler. Even so, this now gives around 95%
18206 performance improvement for 1024 bit RSA signs.
18210 * Hack to fix PKCS#7 decryption when used with some unorthodox RC2
18211 handling. Most clients have the effective key size in bits equal to
18212 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key.
18213 A few however don't do this and instead use the size of the decrypted key
18214 to determine the RC2 key length and the AlgorithmIdentifier to determine
18215 the effective key length. In this case the effective key length can still
18216 be 40 bits but the key length can be 168 bits for example. This is fixed
18217 by manually forcing an RC2 key into the EVP_PKEY structure because the
18218 EVP code can't currently handle unusual RC2 key sizes: it always assumes
18219 the key length and effective key length are equal.
18223 * Add a bunch of functions that should simplify the creation of
18224 X509_NAME structures. Now you should be able to do:
18225 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0);
18226 and have it automatically work out the correct field type and fill in
18227 the structures. The more adventurous can try:
18228 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0);
18229 and it will (hopefully) work out the correct multibyte encoding.
18233 * Change the 'req' utility to use the new field handling and multibyte
18234 copy routines. Before the DN field creation was handled in an ad hoc
18235 way in req, ca, and x509 which was rather broken and didn't support
18236 BMPStrings or UTF8Strings. Since some software doesn't implement
18237 BMPStrings or UTF8Strings yet, they can be enabled using the config file
18238 using the dirstring_type option. See the new comment in the default
18239 openssl.cnf for more info.
18243 * Make crypto/rand/md_rand.c more robust:
18244 - Assure unique random numbers after fork().
18245 - Make sure that concurrent threads access the global counter and
18246 md serializably so that we never lose entropy in them
18247 or use exactly the same state in multiple threads.
18248 Access to the large state is not always serializable because
18249 the additional locking could be a performance killer, and
18250 md should be large enough anyway.
18254 * New file `apps/app_rand.c` with commonly needed functionality
18255 for handling the random seed file.
18257 Use the random seed file in some applications that previously did not:
18259 dsaparam -genkey (which also ignored its '-rand' option),
18262 x509 (when signing).
18263 Except on systems with /dev/urandom, it is crucial to have a random
18264 seed file at least for key creation, DSA signing, and for DH exchanges;
18265 for RSA signatures we could do without one.
18267 gendh and gendsa (unlike genrsa) used to read only the first byte
18268 of each file listed in the '-rand' option. The function as previously
18269 found in genrsa is now in app_rand.c and is used by all programs
18270 that support '-rand'.
18274 * In RAND_write_file, use mode 0600 for creating files;
18275 don't just chmod when it may be too late.
18279 * Report an error from X509_STORE_load_locations
18280 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed.
18284 * New function ASN1_mbstring_copy() this copies a string in either
18285 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format
18286 into an ASN1_STRING type. A mask of permissible types is passed
18287 and it chooses the "minimal" type to use or an error if not type
18292 * Add function equivalents to the various macros in asn1.h. The old
18293 macros are retained with an `M_` prefix. Code inside the library can
18294 use the `M_` macros. External code (including the openssl utility)
18295 should *NOT* in order to be "shared library friendly".
18299 * Add various functions that can check a certificate's extensions
18300 to see if it usable for various purposes such as SSL client,
18301 server or S/MIME and CAs of these types. This is currently
18302 VERY EXPERIMENTAL but will ultimately be used for certificate chain
18303 verification. Also added a -purpose flag to x509 utility to
18304 print out all the purposes.
18308 * Add a CRYPTO_EX_DATA to X509 certificate structure and associated
18313 * New `X509V3_{X509,CRL,REVOKED}_get_d2i()` functions. These will search
18314 for, obtain and decode and extension and obtain its critical flag.
18315 This allows all the necessary extension code to be handled in a
18316 single function call.
18320 * RC4 tune-up featuring 30-40% performance improvement on most RISC
18321 platforms. See crypto/rc4/rc4_enc.c for further details.
18325 * New -noout option to asn1parse. This causes no output to be produced
18326 its main use is when combined with -strparse and -out to extract data
18327 from a file (which may not be in ASN.1 format).
18331 * Fix for pkcs12 program. It was hashing an invalid certificate pointer
18332 when producing the local key id.
18334 *Richard Levitte <levitte@stacken.kth.se>*
18336 * New option -dhparam in s_server. This allows a DH parameter file to be
18337 stated explicitly. If it is not stated then it tries the first server
18338 certificate file. The previous behaviour hard coded the filename
18343 * Add -pubin and -pubout options to the rsa and dsa commands. These allow
18344 a public key to be input or output. For example:
18345 openssl rsa -in key.pem -pubout -out pubkey.pem
18346 Also added necessary DSA public key functions to handle this.
18350 * Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained
18351 in the message. This was handled by allowing
18352 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it.
18354 *Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>*
18356 * Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null
18357 to the end of the strings whereas this didn't. This would cause problems
18358 if strings read with d2i_ASN1_bytes() were later modified.
18360 *Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>*
18362 * Fix for base64 decode bug. When a base64 bio reads only one line of
18363 data and it contains EOF it will end up returning an error. This is
18364 caused by input 46 bytes long. The cause is due to the way base64
18365 BIOs find the start of base64 encoded data. They do this by trying a
18366 trial decode on each line until they find one that works. When they
18367 do a flag is set and it starts again knowing it can pass all the
18368 data directly through the decoder. Unfortunately it doesn't reset
18369 the context it uses. This means that if EOF is reached an attempt
18370 is made to pass two EOFs through the context and this causes the
18371 resulting error. This can also cause other problems as well. As is
18372 usual with these problems it takes *ages* to find and the fix is
18373 trivial: move one line.
18375 *Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer)*
18377 * Ugly workaround to get s_client and s_server working under Windows. The
18378 old code wouldn't work because it needed to select() on sockets and the
18379 tty (for keypresses and to see if data could be written). Win32 only
18380 supports select() on sockets so we select() with a 1s timeout on the
18381 sockets and then see if any characters are waiting to be read, if none
18382 are present then we retry, we also assume we can always write data to
18383 the tty. This isn't nice because the code then blocks until we've
18384 received a complete line of data and it is effectively polling the
18385 keyboard at 1s intervals: however it's quite a bit better than not
18386 working at all :-) A dedicated Windows application might handle this
18387 with an event loop for example.
18391 * Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign
18392 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions
18393 will be called when RSA_sign() and RSA_verify() are used. This is useful
18394 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available.
18395 For this to work properly RSA_public_decrypt() and RSA_private_encrypt()
18396 should *not* be used: RSA_sign() and RSA_verify() must be used instead.
18397 This necessitated the support of an extra signature type NID_md5_sha1
18398 for SSL signatures and modifications to the SSL library to use it instead
18399 of calling RSA_public_decrypt() and RSA_private_encrypt().
18403 * Add new -verify -CAfile and -CApath options to the crl program, these
18404 will lookup a CRL issuers certificate and verify the signature in a
18405 similar way to the verify program. Tidy up the crl program so it
18406 no longer accesses structures directly. Make the ASN1 CRL parsing a bit
18407 less strict. It will now permit CRL extensions even if it is not
18408 a V2 CRL: this will allow it to tolerate some broken CRLs.
18412 * Initialize all non-automatic variables each time one of the openssl
18413 sub-programs is started (this is necessary as they may be started
18414 multiple times from the "OpenSSL>" prompt).
18416 *Lennart Bang, Bodo Moeller*
18418 * Preliminary compilation option RSA_NULL which disables RSA crypto without
18419 removing all other RSA functionality (this is what NO_RSA does). This
18420 is so (for example) those in the US can disable those operations covered
18421 by the RSA patent while allowing storage and parsing of RSA keys and RSA
18426 * Non-copying interface to BIO pairs.
18427 (still largely untested)
18431 * New function ASN1_tag2str() to convert an ASN1 tag to a descriptive
18432 ASCII string. This was handled independently in various places before.
18436 * New functions UTF8_getc() and UTF8_putc() that parse and generate
18437 UTF8 strings a character at a time.
18441 * Use client_version from client hello to select the protocol
18442 (s23_srvr.c) and for RSA client key exchange verification
18443 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications.
18447 * Add various utility functions to handle SPKACs, these were previously
18448 handled by poking round in the structure internals. Added new function
18449 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to
18450 print, verify and generate SPKACs. Based on an original idea from
18451 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified.
18455 * RIPEMD160 is operational on all platforms and is back in 'make test'.
18459 * Allow the config file extension section to be overwritten on the
18460 command line. Based on an original idea from Massimiliano Pala
18461 <madwolf@comune.modena.it>. The new option is called -extensions
18462 and can be applied to ca, req and x509. Also -reqexts to override
18463 the request extensions in req and -crlexts to override the crl extensions
18468 * Add new feature to the SPKAC handling in ca. Now you can include
18469 the same field multiple times by preceding it by "XXXX." for example:
18472 this is the same syntax as used in the req config file.
18476 * Allow certificate extensions to be added to certificate requests. These
18477 are specified in a 'req_extensions' option of the req section of the
18478 config file. They can be printed out with the -text option to req but
18479 are otherwise ignored at present.
18483 * Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first
18484 data read consists of only the final block it would not decrypted because
18485 EVP_CipherUpdate() would correctly report zero bytes had been decrypted.
18486 A misplaced 'break' also meant the decrypted final block might not be
18487 copied until the next read.
18491 * Initial support for DH_METHOD. Again based on RSA_METHOD. Also added
18492 a few extra parameters to the DH structure: these will be useful if
18493 for example we want the value of 'q' or implement X9.42 DH.
18497 * Initial support for DSA_METHOD. This is based on the RSA_METHOD and
18498 provides hooks that allow the default DSA functions or functions on a
18499 "per key" basis to be replaced. This allows hardware acceleration and
18500 hardware key storage to be handled without major modification to the
18501 library. Also added low-level modexp hooks and CRYPTO_EX structure and
18502 associated functions.
18506 * Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO
18507 as "read only": it can't be written to and the buffer it points to will
18508 not be freed. Reading from a read only BIO is much more efficient than
18509 a normal memory BIO. This was added because there are several times when
18510 an area of memory needs to be read from a BIO. The previous method was
18511 to create a memory BIO and write the data to it, this results in two
18512 copies of the data and an O(n^2) reading algorithm. There is a new
18513 function BIO_new_mem_buf() which creates a read only memory BIO from
18514 an area of memory. Also modified the PKCS#7 routines to use read only
18519 * Bugfix: ssl23_get_client_hello did not work properly when called in
18520 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of
18521 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read,
18522 but a retry condition occurred while trying to read the rest.
18526 * The PKCS7_ENC_CONTENT_new() function was setting the content type as
18527 NID_pkcs7_encrypted by default: this was wrong since this should almost
18528 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle
18529 the encrypted data type: this is a more sensible place to put it and it
18530 allows the PKCS#12 code to be tidied up that duplicated this
18535 * Changed obj_dat.pl script so it takes its input and output files on
18536 the command line. This should avoid shell escape redirection problems
18541 * Initial support for certificate extension requests, these are included
18542 in things like Xenroll certificate requests. Included functions to allow
18543 extensions to be obtained and added.
18547 * -crlf option to s_client and s_server for sending newlines as
18548 CRLF (as required by many protocols).
18552 ### Changes between 0.9.3a and 0.9.4 [09 Aug 1999]
18554 * Install libRSAglue.a when OpenSSL is built with RSAref.
18556 *Ralf S. Engelschall*
18558 * A few more `#ifndef NO_FP_API / #endif` pairs for consistency.
18560 *Andrija Antonijevic <TheAntony2@bigfoot.com>*
18562 * Fix -startdate and -enddate (which was missing) arguments to 'ca'
18567 * New function DSA_dup_DH, which duplicates DSA parameters/keys as
18568 DH parameters/keys (q is lost during that conversion, but the resulting
18569 DH parameters contain its length).
18571 For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is
18572 much faster than DH_generate_parameters (which creates parameters
18573 where `p = 2*q + 1`), and also the smaller q makes DH computations
18574 much more efficient (160-bit exponentiation instead of 1024-bit
18575 exponentiation); so this provides a convenient way to support DHE
18576 ciphersuites in SSL/TLS servers (see ssl/ssltest.c). It is of
18577 utter importance to use
18578 SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
18580 SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
18581 when such DH parameters are used, because otherwise small subgroup
18582 attacks may become possible!
18586 * Avoid memory leak in i2d_DHparams.
18590 * Allow the -k option to be used more than once in the enc program:
18591 this allows the same encrypted message to be read by multiple recipients.
18595 * New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts
18596 an ASN1_OBJECT to a text string. If the "no_name" parameter is set then
18597 it will always use the numerical form of the OID, even if it has a short
18602 * Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp
18603 method only got called if p,q,dmp1,dmq1,iqmp components were present,
18604 otherwise bn_mod_exp was called. In the case of hardware keys for example
18605 no private key components need be present and it might store extra data
18606 in the RSA structure, which cannot be accessed from bn_mod_exp.
18607 By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for
18608 private key operations.
18612 * Added support for SPARC Linux.
18616 * pem_password_cb function type incompatibly changed from
18617 typedef int pem_password_cb(char *buf, int size, int rwflag);
18619 ....(char *buf, int size, int rwflag, void *userdata);
18620 so that applications can pass data to their callbacks:
18621 The `PEM[_ASN1]_{read,write}...` functions and macros now take an
18622 additional void * argument, which is just handed through whenever
18623 the password callback is called.
18625 *Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller*
18627 New function SSL_CTX_set_default_passwd_cb_userdata.
18629 Compatibility note: As many C implementations push function arguments
18630 onto the stack in reverse order, the new library version is likely to
18631 interoperate with programs that have been compiled with the old
18632 pem_password_cb definition (PEM_whatever takes some data that
18633 happens to be on the stack as its last argument, and the callback
18634 just ignores this garbage); but there is no guarantee whatsoever that
18637 * The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=...
18638 (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused
18639 problems not only on Windows, but also on some Unix platforms.
18640 To avoid problematic command lines, these definitions are now in an
18641 auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl
18642 for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds).
18646 * MIPS III/IV assembler module is reimplemented.
18650 * More DES library cleanups: remove references to srand/rand and
18651 delete an unused file.
18655 * Add support for the free Netwide assembler (NASM) under Win32,
18656 since not many people have MASM (ml) and it can be hard to obtain.
18657 This is currently experimental but it seems to work OK and pass all
18658 the tests. Check out INSTALL.W32 for info.
18662 * Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections
18663 without temporary keys kept an extra copy of the server key,
18664 and connections with temporary keys did not free everything in case
18669 * New function RSA_check_key and new openssl rsa option -check
18670 for verifying the consistency of RSA keys.
18672 *Ulf Moeller, Bodo Moeller*
18674 * Various changes to make Win32 compile work:
18675 1. Casts to avoid "loss of data" warnings in p5_crpt2.c
18676 2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned
18677 comparison" warnings.
18678 3. Add `sk_<TYPE>_sort` to DEF file generator and do make update.
18682 * Add a debugging option to PKCS#5 v2 key generation function: when
18683 you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and
18684 derived keys are printed to stderr.
18688 * Copy the flags in ASN1_STRING_dup().
18690 *Roman E. Pavlov <pre@mo.msk.ru>*
18692 * The x509 application mishandled signing requests containing DSA
18693 keys when the signing key was also DSA and the parameters didn't match.
18695 It was supposed to omit the parameters when they matched the signing key:
18696 the verifying software was then supposed to automatically use the CA's
18697 parameters if they were absent from the end user certificate.
18699 Omitting parameters is no longer recommended. The test was also
18700 the wrong way round! This was probably due to unusual behaviour in
18701 EVP_cmp_parameters() which returns 1 if the parameters match.
18702 This meant that parameters were omitted when they *didn't* match and
18703 the certificate was useless. Certificates signed with 'ca' didn't have
18706 *Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>*
18708 * Memory leak checking (-DCRYPTO_MDEBUG) had some problems.
18709 The interface is as follows:
18710 Applications can use
18711 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(),
18712 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop();
18713 "off" is now the default.
18714 The library internally uses
18715 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(),
18716 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on()
18717 to disable memory-checking temporarily.
18719 Some inconsistent states that previously were possible (and were
18720 even the default) are now avoided.
18722 -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time
18723 with each memory chunk allocated; this is occasionally more helpful
18724 than just having a counter.
18726 -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID.
18728 -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future
18733 * Introduce "mode" for SSL structures (with defaults in SSL_CTX),
18734 which largely parallels "options", but is for changing API behaviour,
18735 whereas "options" are about protocol behaviour.
18736 Initial "mode" flags are:
18738 SSL_MODE_ENABLE_PARTIAL_WRITE Allow SSL_write to report success when
18739 a single record has been written.
18740 SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER Don't insist that SSL_write
18741 retries use the same buffer location.
18742 (But all of the contents must be
18747 * Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options
18750 * Fix problems with no-hmac etc.
18752 *Ulf Möller, pointed out by Brian Wellington <bwelling@tislabs.com>*
18754 * New functions RSA_get_default_method(), RSA_set_method() and
18755 RSA_get_method(). These allows replacement of RSA_METHODs without having
18756 to mess around with the internals of an RSA structure.
18760 * Fix memory leaks in DSA_do_sign and DSA_is_prime.
18761 Also really enable memory leak checks in openssl.c and in some
18764 *Chad C. Mulligan, Bodo Moeller*
18766 * Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess
18767 up the length of negative integers. This has now been simplified to just
18768 store the length when it is first determined and use it later, rather
18769 than trying to keep track of where data is copied and updating it to
18771 *Steve Henson, reported by Brien Wheeler <bwheeler@authentica-security.com>*
18773 * Add a new function PKCS7_signatureVerify. This allows the verification
18774 of a PKCS#7 signature but with the signing certificate passed to the
18775 function itself. This contrasts with PKCS7_dataVerify which assumes the
18776 certificate is present in the PKCS#7 structure. This isn't always the
18777 case: certificates can be omitted from a PKCS#7 structure and be
18778 distributed by "out of band" means (such as a certificate database).
18782 * Complete the `PEM_*` macros with DECLARE_PEM versions to replace the
18783 function prototypes in pem.h, also change util/mkdef.pl to add the
18784 necessary function names.
18788 * mk1mf.pl (used by Windows builds) did not properly read the
18789 options set by Configure in the top level Makefile, and Configure
18790 was not even able to write more than one option correctly.
18791 Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended.
18795 * New functions CONF_load_bio() and CONF_load_fp() to allow a config
18796 file to be loaded from a BIO or FILE pointer. The BIO version will
18797 for example allow memory BIOs to contain config info.
18801 * New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS.
18802 Whoever hopes to achieve shared-library compatibility across versions
18803 must use this, not the compile-time macro.
18804 (Exercise 0.9.4: Which is the minimum library version required by
18806 Note: All this applies only to multi-threaded programs, others don't
18811 * Add missing case to s3_clnt.c state machine -- one of the new SSL tests
18812 through a BIO pair triggered the default case, i.e.
18813 SSLerr(...,SSL_R_UNKNOWN_STATE).
18817 * New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications
18818 can use the SSL library even if none of the specific BIOs is
18823 * Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value
18824 for the encoded length.
18826 *Jeon KyoungHo <khjeon@sds.samsung.co.kr>*
18828 * Add initial documentation of the X509V3 functions.
18832 * Add a new pair of functions PEM_write_PKCS8PrivateKey() and
18833 PEM_write_bio_PKCS8PrivateKey() that are equivalent to
18834 PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more
18835 secure PKCS#8 private key format with a high iteration count.
18839 * Fix determination of Perl interpreter: A perl or perl5
18840 *directory* in $PATH was also accepted as the interpreter.
18842 *Ralf S. Engelschall*
18844 * Fix demos/sign/sign.c: well there wasn't anything strictly speaking
18845 wrong with it but it was very old and did things like calling
18846 PEM_ASN1_read() directly and used MD5 for the hash not to mention some
18847 unusual formatting.
18851 * Fix demos/selfsign.c: it used obsolete and deleted functions, changed
18852 to use the new extension code.
18856 * Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c
18857 with macros. This should make it easier to change their form, add extra
18858 arguments etc. Fix a few PEM prototypes which didn't have cipher as a
18863 * Add to configuration table a new entry that can specify an alternative
18864 name for unistd.h (for pre-POSIX systems); we need this for NeXTstep,
18865 according to Mark Crispin <MRC@Panda.COM>.
18869 * DES CBC did not update the IV. Weird.
18873 des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does.
18874 Changing the behaviour of the former might break existing programs --
18875 where IV updating is needed, des_ncbc_encrypt can be used.
18878 * When bntest is run from "make test" it drives bc to check its
18879 calculations, as well as internally checking them. If an internal check
18880 fails, it needs to cause bc to give a non-zero result or make test carries
18881 on without noticing the failure. Fixed.
18885 * DES library cleanups.
18889 * Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be
18890 used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit
18891 ciphers. NOTE: although the key derivation function has been verified
18892 against some published test vectors it has not been extensively tested
18893 yet. Added a -v2 "cipher" option to pkcs8 application to allow the use
18898 * Instead of "mkdir -p", which is not fully portable, use new
18899 Perl script "util/mkdir-p.pl".
18903 * Rewrite the way password based encryption (PBE) is handled. It used to
18904 assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter
18905 structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms
18906 but doesn't apply to PKCS#5 v2.0 where it can be something else. Now
18907 the 'parameter' field of the AlgorithmIdentifier is passed to the
18908 underlying key generation function so it must do its own ASN1 parsing.
18909 This has also changed the EVP_PBE_CipherInit() function which now has a
18910 'parameter' argument instead of literal salt and iteration count values
18911 and the function EVP_PBE_ALGOR_CipherInit() has been deleted.
18915 * Support for PKCS#5 v1.5 compatible password based encryption algorithms
18916 and PKCS#8 functionality. New 'pkcs8' application linked to openssl.
18917 Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE
18918 KEY" because this clashed with PKCS#8 unencrypted string. Since this
18919 value was just used as a "magic string" and not used directly its
18920 value doesn't matter.
18924 * Introduce some semblance of const correctness to BN. Shame C doesn't
18929 * "linux-sparc64" configuration (ultrapenguin).
18931 *Ray Miller <ray.miller@oucs.ox.ac.uk>*
18932 "linux-sparc" configuration.
18934 *Christian Forster <fo@hawo.stw.uni-erlangen.de>*
18936 * config now generates no-xxx options for missing ciphers.
18940 * Support the EBCDIC character set (work in progress).
18941 File ebcdic.c not yet included because it has a different license.
18943 *Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>*
18945 * Support BS2000/OSD-POSIX.
18947 *Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>*
18949 * Make callbacks for key generation use `void *` instead of `char *`.
18953 * Make S/MIME samples compile (not yet tested).
18957 * Additional typesafe stacks.
18961 * New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x).
18965 ### Changes between 0.9.3 and 0.9.3a [29 May 1999]
18967 * New configuration variant "sco5-gcc".
18969 * Updated some demos.
18971 *Sean O Riordain, Wade Scholine*
18973 * Add missing BIO_free at exit of pkcs12 application.
18977 * Fix memory leak in conf.c.
18981 * Updates for Win32 to assembler version of MD5.
18985 * Set #! path to perl in `apps/der_chop` to where we found it
18986 instead of using a fixed path.
18990 * SHA library changes for irix64-mips4-cc.
18994 * Improvements for VMS support.
18998 ### Changes between 0.9.2b and 0.9.3 [24 May 1999]
19000 * Bignum library bug fix. IRIX 6 passes "make test" now!
19001 This also avoids the problems with SC4.2 and unpatched SC5.
19003 *Andy Polyakov <appro@fy.chalmers.se>*
19005 * New functions sk_num, sk_value and sk_set to replace the previous macros.
19006 These are required because of the typesafe stack would otherwise break
19007 existing code. If old code used a structure member which used to be STACK
19008 and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with
19009 sk_num or sk_value it would produce an error because the num, data members
19010 are not present in STACK_OF. Now it just produces a warning. sk_set
19011 replaces the old method of assigning a value to sk_value
19012 (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code
19013 that does this will no longer work (and should use sk_set instead) but
19014 this could be regarded as a "questionable" behaviour anyway.
19018 * Fix most of the other PKCS#7 bugs. The "experimental" code can now
19019 correctly handle encrypted S/MIME data.
19023 * Change type of various DES function arguments from des_cblock
19024 (which means, in function argument declarations, pointer to char)
19025 to des_cblock * (meaning pointer to array with 8 char elements),
19026 which allows the compiler to do more typechecking; it was like
19027 that back in SSLeay, but with lots of ugly casts.
19029 Introduce new type const_des_cblock.
19033 * Reorganise the PKCS#7 library and get rid of some of the more obvious
19034 problems: find RecipientInfo structure that matches recipient certificate
19035 and initialise the ASN1 structures properly based on passed cipher.
19039 * Belatedly make the BN tests actually check the results.
19043 * Fix the encoding and decoding of negative ASN1 INTEGERS and conversion
19044 to and from BNs: it was completely broken. New compilation option
19045 NEG_PUBKEY_BUG to allow for some broken certificates that encode public
19046 key elements as negative integers.
19050 * Reorganize and speed up MD5.
19052 *Andy Polyakov <appro@fy.chalmers.se>*
19056 *Richard Levitte <richard@levitte.org>*
19058 * New option -out to asn1parse to allow the parsed structure to be
19059 output to a file. This is most useful when combined with the -strparse
19060 option to examine the output of things like OCTET STRINGS.
19064 * Make SSL library a little more fool-proof by not requiring any longer
19065 that `SSL_set_{accept,connect}_state` be called before
19066 `SSL_{accept,connect}` may be used (`SSL_set_..._state` is omitted
19067 in many applications because usually everything *appeared* to work as
19068 intended anyway -- now it really works as intended).
19072 * Move openssl.cnf out of lib/.
19076 * Fix various things to let OpenSSL even pass "egcc -pipe -O2 -Wall
19077 -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes
19078 -Wmissing-declarations -Wnested-externs -Winline" with EGCS 1.1.2+
19080 *Ralf S. Engelschall*
19082 * Various fixes to the EVP and PKCS#7 code. It may now be able to
19083 handle PKCS#7 enveloped data properly.
19085 *Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve*
19087 * Create a duplicate of the SSL_CTX's CERT in SSL_new instead of
19088 copying pointers. The cert_st handling is changed by this in
19089 various ways (and thus what used to be known as ctx->default_cert
19090 is now called ctx->cert, since we don't resort to `s->ctx->[default_]cert`
19091 any longer when s->cert does not give us what we need).
19092 ssl_cert_instantiate becomes obsolete by this change.
19093 As soon as we've got the new code right (possibly it already is?),
19094 we have solved a couple of bugs of the earlier code where s->cert
19095 was used as if it could not have been shared with other SSL structures.
19097 Note that using the SSL API in certain dirty ways now will result
19098 in different behaviour than observed with earlier library versions:
19099 Changing settings for an `SSL_CTX *ctx` after having done s = SSL_new(ctx)
19100 does not influence s as it used to.
19102 In order to clean up things more thoroughly, inside SSL_SESSION
19103 we don't use CERT any longer, but a new structure SESS_CERT
19104 that holds per-session data (if available); currently, this is
19105 the peer's certificate chain and, for clients, the server's certificate
19106 and temporary key. CERT holds only those values that can have
19107 meaningful defaults in an SSL_CTX.
19111 * New function X509V3_EXT_i2d() to create an X509_EXTENSION structure
19112 from the internal representation. Various PKCS#7 fixes: remove some
19113 evil casts and set the enc_dig_alg field properly based on the signing
19118 * Allow PKCS#12 password to be set from the command line or the
19119 environment. Let 'ca' get its config file name from the environment
19120 variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req'
19125 * Allow certificate policies extension to use an IA5STRING for the
19126 organization field. This is contrary to the PKIX definition but
19127 VeriSign uses it and IE5 only recognises this form. Document 'x509'
19132 * Add PEDANTIC compiler flag to allow compilation with gcc -pedantic,
19133 without disallowing inline assembler and the like for non-pedantic builds.
19137 * Support Borland C++ builder.
19139 *Janez Jere <jj@void.si>, modified by Ulf Möller*
19145 * SHA-1 cleanups and performance enhancements.
19147 *Andy Polyakov <appro@fy.chalmers.se>*
19149 * Sparc v8plus assembler for the bignum library.
19151 *Andy Polyakov <appro@fy.chalmers.se>*
19153 * Accept any -xxx and +xxx compiler options in Configure.
19157 * Update HPUX configuration.
19161 * Add missing `sk_<type>_unshift()` function to safestack.h
19163 *Ralf S. Engelschall*
19165 * New function SSL_CTX_use_certificate_chain_file that sets the
19166 "extra_cert"s in addition to the certificate. (This makes sense
19167 only for "PEM" format files, as chains as a whole are not
19172 * Support verify_depth from the SSL API.
19173 x509_vfy.c had what can be considered an off-by-one-error:
19174 Its depth (which was not part of the external interface)
19175 was actually counting the number of certificates in a chain;
19176 now it really counts the depth.
19180 * Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used
19181 instead of X509err, which often resulted in confusing error
19182 messages since the error codes are not globally unique
19183 (e.g. an alleged error in ssl3_accept when a certificate
19184 didn't match the private key).
19186 * New function SSL_CTX_set_session_id_context that allows to set a default
19187 value (so that you don't need SSL_set_session_id_context for each
19188 connection using the SSL_CTX).
19192 * OAEP decoding bug fix.
19196 * Support INSTALL_PREFIX for package builders, as proposed by
19201 * New Configure options "threads" and "no-threads". For systems
19202 where the proper compiler options are known (currently Solaris
19203 and Linux), "threads" is the default.
19207 * New script util/mklink.pl as a faster substitute for util/mklink.sh.
19211 * Install various scripts to $(OPENSSLDIR)/misc, not to
19212 $(INSTALLTOP)/bin -- they shouldn't clutter directories
19213 such as /usr/local/bin.
19217 * "make linux-shared" to build shared libraries.
19219 *Niels Poppe <niels@netbox.org>*
19221 * New Configure option `no-<cipher>` (rsa, idea, rc5, ...).
19225 * Add the PKCS#12 API documentation to openssl.txt. Preliminary support for
19226 extension adding in x509 utility.
19230 * Remove NOPROTO sections and error code comments.
19234 * Partial rewrite of the DEF file generator to now parse the ANSI
19239 * New Configure options --prefix=DIR and --openssldir=DIR.
19243 * Complete rewrite of the error code script(s). It is all now handled
19244 by one script at the top level which handles error code gathering,
19245 header rewriting and C source file generation. It should be much better
19246 than the old method: it now uses a modified version of Ulf's parser to
19247 read the ANSI prototypes in all header files (thus the old K&R definitions
19248 aren't needed for error creation any more) and do a better job of
19249 translating function codes into names. The old 'ASN1 error code embedded
19250 in a comment' is no longer necessary and it doesn't use .err files which
19251 have now been deleted. Also the error code call doesn't have to appear all
19252 on one line (which resulted in some large lines...).
19256 * Change #include filenames from `<foo.h>` to `<openssl/foo.h>`.
19260 * Change behaviour of ssl2_read when facing length-0 packets: Don't return
19261 0 (which usually indicates a closed connection), but continue reading.
19265 * Fix some race conditions.
19269 * Add support for CRL distribution points extension. Add Certificate
19270 Policies and CRL distribution points documentation.
19274 * Move the autogenerated header file parts to crypto/opensslconf.h.
19278 * Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of
19279 8 of keying material. Merlin has also confirmed interop with this fix
19280 between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0.
19282 *Merlin Hughes <merlin@baltimore.ie>*
19284 * Fix lots of warnings.
19286 *Richard Levitte <levitte@stacken.kth.se>*
19288 * In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if
19289 the directory spec didn't end with a LIST_SEPARATOR_CHAR.
19291 *Richard Levitte <levitte@stacken.kth.se>*
19293 * Fix problems with sizeof(long) == 8.
19295 *Andy Polyakov <appro@fy.chalmers.se>*
19297 * Change functions to ANSI C.
19301 * Fix typos in error codes.
19303 *Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf Möller*
19305 * Remove defunct assembler files from Configure.
19309 * SPARC v8 assembler BIGNUM implementation.
19311 *Andy Polyakov <appro@fy.chalmers.se>*
19313 * Support for Certificate Policies extension: both print and set.
19314 Various additions to support the r2i method this uses.
19318 * A lot of constification, and fix a bug in X509_NAME_oneline() that could
19319 return a const string when you are expecting an allocated buffer.
19323 * Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE
19324 types DirectoryString and DisplayText.
19328 * Add code to allow r2i extensions to access the configuration database,
19329 add an LHASH database driver and add several ctx helper functions.
19333 * Fix an evil bug in bn_expand2() which caused various BN functions to
19334 fail when they extended the size of a BIGNUM.
19338 * Various utility functions to handle SXNet extension. Modify mkdef.pl to
19339 support typesafe stack.
19343 * Fix typo in SSL_[gs]et_options().
19345 *Nils Frostberg <nils@medcom.se>*
19347 * Delete various functions and files that belonged to the (now obsolete)
19348 old X509V3 handling code.
19352 * New Configure option "rsaref".
19356 * Don't auto-generate pem.h.
19360 * Introduce type-safe ASN.1 SETs.
19364 * Convert various additional casted stacks to type-safe STACK_OF() variants.
19366 *Ben Laurie, Ralf S. Engelschall, Steve Henson*
19368 * Introduce type-safe STACKs. This will almost certainly break lots of code
19369 that links with OpenSSL (well at least cause lots of warnings), but fear
19370 not: the conversion is trivial, and it eliminates loads of evil casts. A
19371 few STACKed things have been converted already. Feel free to convert more.
19372 In the fullness of time, I'll do away with the STACK type altogether.
19376 * Add `openssl ca -revoke <certfile>` facility which revokes a certificate
19377 specified in `<certfile>` by updating the entry in the index.txt file.
19378 This way one no longer has to edit the index.txt file manually for
19379 revoking a certificate. The -revoke option does the gory details now.
19381 *Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall*
19383 * Fix `openssl crl -noout -text` combination where `-noout` killed the
19384 `-text` option at all and this way the `-noout -text` combination was
19385 inconsistent in `openssl crl` with the friends in `openssl x509|rsa|dsa`.
19387 *Ralf S. Engelschall*
19389 * Make sure a corresponding plain text error message exists for the
19390 X509_V_ERR_CERT_REVOKED/23 error number which can occur when a
19391 verify callback function determined that a certificate was revoked.
19393 *Ralf S. Engelschall*
19395 * Bugfix: In test/testenc, don't test `openssl <cipher>` for
19396 ciphers that were excluded, e.g. by -DNO_IDEA. Also, test
19397 all available ciphers including rc5, which was forgotten until now.
19398 In order to let the testing shell script know which algorithms
19399 are available, a new (up to now undocumented) command
19400 `openssl list-cipher-commands` is used.
19404 * Bugfix: s_client occasionally would sleep in select() when
19405 it should have checked SSL_pending() first.
19409 * New functions DSA_do_sign and DSA_do_verify to provide access to
19410 the raw DSA values prior to ASN.1 encoding.
19414 * Tweaks to Configure
19416 *Niels Poppe <niels@netbox.org>*
19418 * Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support,
19423 * New variables $(RANLIB) and $(PERL) in the Makefiles.
19427 * New config option to avoid instructions that are illegal on the 80386.
19428 The default code is faster, but requires at least a 486.
19432 * Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and
19433 SSL2_SERVER_VERSION (not used at all) macros, which are now the
19434 same as SSL2_VERSION anyway.
19438 * New "-showcerts" option for s_client.
19442 * Still more PKCS#12 integration. Add pkcs12 application to openssl
19443 application. Various cleanups and fixes.
19447 * More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and
19448 modify error routines to work internally. Add error codes and PBE init
19449 to library startup routines.
19453 * Further PKCS#12 integration. Added password based encryption, PKCS#8 and
19454 packing functions to asn1 and evp. Changed function names and error
19455 codes along the way.
19459 * PKCS12 integration: and so it begins... First of several patches to
19460 slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12
19461 objects to objects.h
19465 * Add a new 'indent' option to some X509V3 extension code. Initial ASN1
19466 and display support for Thawte strong extranet extension.
19470 * Add LinuxPPC support.
19472 *Jeff Dubrule <igor@pobox.org>*
19474 * Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to
19475 bn_div_words in alpha.s.
19477 *Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie*
19479 * Make sure the RSA OAEP test is skipped under -DRSAref because
19480 OAEP isn't supported when OpenSSL is built with RSAref.
19482 *Ulf Moeller <ulf@fitug.de>*
19484 * Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h
19485 so they no longer are missing under -DNOPROTO.
19487 *Soren S. Jorvang <soren@t.dk>*
19489 ### Changes between 0.9.1c and 0.9.2b [22 Mar 1999]
19491 * Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still
19492 doesn't work when the session is reused. Coming soon!
19496 * Fix a security hole, that allows sessions to be reused in the wrong
19497 context thus bypassing client cert protection! All software that uses
19498 client certs and session caches in multiple contexts NEEDS PATCHING to
19499 allow session reuse! A fuller solution is in the works.
19501 *Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)*
19503 * Some more source tree cleanups (removed obsolete files
19504 crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed
19505 permission on "config" script to be executable) and a fix for the INSTALL
19508 *Ulf Moeller <ulf@fitug.de>*
19510 * Remove some legacy and erroneous uses of malloc, free instead of
19513 *Lennart Bang <lob@netstream.se>, with minor changes by Steve*
19515 * Make rsa_oaep_test return non-zero on error.
19517 *Ulf Moeller <ulf@fitug.de>*
19519 * Add support for native Solaris shared libraries. Configure
19520 solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice
19521 if someone would make that last step automatic.
19523 *Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>*
19525 * ctx_size was not built with the right compiler during "make links". Fixed.
19529 * Change the meaning of 'ALL' in the cipher list. It now means "everything
19530 except NULL ciphers". This means the default cipher list will no longer
19531 enable NULL ciphers. They need to be specifically enabled e.g. with
19532 the string "DEFAULT:eNULL".
19536 * Fix to RSA private encryption routines: if p < q then it would
19537 occasionally produce an invalid result. This will only happen with
19538 externally generated keys because OpenSSL (and SSLeay) ensure p > q.
19542 * Be less restrictive and allow also `perl util/perlpath.pl
19543 /path/to/bin/perl` in addition to `perl util/perlpath.pl /path/to/bin`,
19544 because this way one can also use an interpreter named `perl5` (which is
19545 usually the name of Perl 5.xxx on platforms where an Perl 4.x is still
19546 installed as `perl`).
19548 *Matthias Loepfe <Matthias.Loepfe@adnovum.ch>*
19550 * Let util/clean-depend.pl work also with older Perl 5.00x versions.
19552 *Matthias Loepfe <Matthias.Loepfe@adnovum.ch>*
19554 * Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add
19555 advapi32.lib to Win32 build and change the pem test comparison
19556 to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the
19557 suggestion). Fix misplaced ASNI prototypes and declarations in evp.h
19558 and crypto/des/ede_cbcm_enc.c.
19562 * DES quad checksum was broken on big-endian architectures. Fixed.
19566 * Comment out two functions in bio.h that aren't implemented. Fix up the
19567 Win32 test batch file so it (might) work again. The Win32 test batch file
19568 is horrible: I feel ill....
19572 * Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected
19573 in e_os.h. Audit of header files to check ANSI and non ANSI
19574 sections: 10 functions were absent from non ANSI section and not exported
19575 from Windows DLLs. Fixed up libeay.num for new functions.
19579 * Make `openssl version` output lines consistent.
19581 *Ralf S. Engelschall*
19583 * Fix Win32 symbol export lists for BIO functions: Added
19584 BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data
19585 to ms/libeay{16,32}.def.
19587 *Ralf S. Engelschall*
19589 * Second round of fixing the OpenSSL perl/ stuff. It now at least compiled
19590 fine under Unix and passes some trivial tests I've now added. But the
19591 whole stuff is horribly incomplete, so a README.1ST with a disclaimer was
19592 added to make sure no one expects that this stuff really works in the
19593 OpenSSL 0.9.2 release. Additionally I've started to clean the XS sources
19594 up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and
19597 *Ralf S. Engelschall*
19599 * Fix the generation of two part addresses in perl.
19601 *Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie*
19603 * Add config entry for Linux on MIPS.
19605 *John Tobey <jtobey@channel1.com>*
19607 * Make links whenever Configure is run, unless we are on Windoze.
19611 * Permit extensions to be added to CRLs using crl_section in openssl.cnf.
19612 Currently only issuerAltName and AuthorityKeyIdentifier make any sense
19617 * Add a useful kludge to allow package maintainers to specify compiler and
19618 other platforms details on the command line without having to patch the
19619 Configure script every time: One now can use
19620 `perl Configure <id>:<details>`,
19621 i.e. platform ids are allowed to have details appended
19622 to them (separated by colons). This is treated as there would be a static
19623 pre-configured entry in Configure's %table under key `<id>` with value
19624 `<details>` and `perl Configure <id>` is called. So, when you want to
19625 perform a quick test-compile under FreeBSD 3.1 with pgcc and without
19626 assembler stuff you can use `perl Configure "FreeBSD-elf:pgcc:-O6:::"`
19627 now, which overrides the FreeBSD-elf entry on-the-fly.
19629 *Ralf S. Engelschall*
19631 * Disable new TLS1 ciphersuites by default: they aren't official yet.
19635 * Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified
19636 on the `perl Configure ...` command line. This way one can compile
19637 OpenSSL libraries with Position Independent Code (PIC) which is needed
19638 for linking it into DSOs.
19640 *Ralf S. Engelschall*
19642 * Remarkably, export ciphers were totally broken and no-one had noticed!
19647 * Cleaned up the LICENSE document: The official contact for any license
19648 questions now is the OpenSSL core team under openssl-core@openssl.org.
19649 And add a paragraph about the dual-license situation to make sure people
19650 recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply
19651 to the OpenSSL toolkit.
19653 *Ralf S. Engelschall*
19655 * General source tree makefile cleanups: Made `making xxx in yyy...`
19656 display consistent in the source tree and replaced `/bin/rm` by `rm`.
19657 Additionally cleaned up the `make links` target: Remove unnecessary
19658 semicolons, subsequent redundant removes, inline point.sh into mklink.sh
19659 to speed processing and no longer clutter the display with confusing
19660 stuff. Instead only the actually done links are displayed.
19662 *Ralf S. Engelschall*
19664 * Permit null encryption ciphersuites, used for authentication only. It used
19665 to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this.
19666 It is now necessary to set SSL_FORBID_ENULL to prevent the use of null
19671 * Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder
19672 signed attributes when verifying signatures (this would break them),
19673 the detached data encoding was wrong and public keys obtained using
19674 X509_get_pubkey() weren't freed.
19678 * Add text documentation for the BUFFER functions. Also added a work around
19679 to a Win95 console bug. This was triggered by the password read stuff: the
19680 last character typed gets carried over to the next fread(). If you were
19681 generating a new cert request using 'req' for example then the last
19682 character of the passphrase would be CR which would then enter the first
19687 * Added the new 'Includes OpenSSL Cryptography Software' button as
19688 doc/openssl_button.{gif,html} which is similar in style to the old SSLeay
19689 button and can be used by applications based on OpenSSL to show the
19690 relationship to the OpenSSL project.
19692 *Ralf S. Engelschall*
19694 * Remove confusing variables in function signatures in files
19695 ssl/ssl_lib.c and ssl/ssl.h.
19697 *Lennart Bong <lob@kulthea.stacken.kth.se>*
19699 * Don't install bss_file.c under PREFIX/include/
19701 *Lennart Bong <lob@kulthea.stacken.kth.se>*
19703 * Get the Win32 compile working again. Modify mkdef.pl so it can handle
19704 functions that return function pointers and has support for NT specific
19705 stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various
19706 #ifdef WIN32 and WINNTs sprinkled about the place and some changes from
19707 unsigned to signed types: this was killing the Win32 compile.
19711 * Add new certificate file to stack functions,
19712 SSL_add_dir_cert_subjects_to_stack() and
19713 SSL_add_file_cert_subjects_to_stack(). These largely supplant
19714 SSL_load_client_CA_file(), and can be used to add multiple certs easily
19715 to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()).
19716 This means that Apache-SSL and similar packages don't have to mess around
19717 to add as many CAs as they want to the preferred list.
19721 * Experiment with doxygen documentation. Currently only partially applied to
19723 See <http://www.stack.nl/~dimitri/doxygen/index.html>, and run doxygen with
19724 openssl.doxy as the configuration file.
19728 * Get rid of remaining C++-style comments which strict C compilers hate.
19730 *Ralf S. Engelschall, pointed out by Carlos Amengual*
19732 * Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not
19733 compiled in by default: it has problems with large keys.
19737 * Add a bunch of SSL_xxx() functions for configuring the temporary RSA and
19738 DH private keys and/or callback functions which directly correspond to
19739 their SSL_CTX_xxx() counterparts but work on a per-connection basis. This
19740 is needed for applications which have to configure certificates on a
19741 per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis
19743 For the RSA certificate situation is makes no difference, but
19744 for the DSA certificate situation this fixes the "no shared cipher"
19745 problem where the OpenSSL cipher selection procedure failed because the
19746 temporary keys were not overtaken from the context and the API provided
19747 no way to reconfigure them.
19748 The new functions now let applications reconfigure the stuff and they
19749 are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh,
19750 SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback. Additionally a new
19751 non-public-API function ssl_cert_instantiate() is used as a helper
19752 function and also to reduce code redundancy inside ssl_rsa.c.
19754 *Ralf S. Engelschall*
19756 * Move s_server -dcert and -dkey options out of the undocumented feature
19757 area because they are useful for the DSA situation and should be
19758 recognized by the users.
19760 *Ralf S. Engelschall*
19762 * Fix the cipher decision scheme for export ciphers: the export bits are
19763 *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within
19764 SSL_EXP_MASK. So, the original variable has to be used instead of the
19765 already masked variable.
19767 *Richard Levitte <levitte@stacken.kth.se>*
19769 * Fix `port` variable from `int` to `unsigned int` in crypto/bio/b_sock.c
19771 *Richard Levitte <levitte@stacken.kth.se>*
19773 * Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal()
19774 from `int` to `unsigned int` because it is a length and initialized by
19775 EVP_DigestFinal() which expects an `unsigned int *`.
19777 *Richard Levitte <levitte@stacken.kth.se>*
19779 * Don't hard-code path to Perl interpreter on shebang line of Configure
19780 script. Instead use the usual Shell->Perl transition trick.
19782 *Ralf S. Engelschall*
19784 * Make `openssl x509 -noout -modulus`' functional also for DSA certificates
19785 (in addition to RSA certificates) to match the behaviour of `openssl dsa
19786 -noout -modulus` as it's already the case for `openssl rsa -noout
19787 -modulus`. For RSA the -modulus is the real "modulus" while for DSA
19788 currently the public key is printed (a decision which was already done by
19789 `openssl dsa -modulus` in the past) which serves a similar purpose.
19790 Additionally the NO_RSA no longer completely removes the whole -modulus
19791 option; it now only avoids using the RSA stuff. Same applies to NO_DSA
19794 *Ralf S. Engelschall*
19796 * Add Arne Ansper's reliable BIO - this is an encrypted, block-digested
19797 BIO. See the source (crypto/evp/bio_ok.c) for more info.
19799 *Arne Ansper <arne@ats.cyber.ee>*
19801 * Dump the old yucky req code that tried (and failed) to allow raw OIDs
19802 to be added. Now both 'req' and 'ca' can use new objects defined in the
19807 * Add cool BIO that does syslog (or event log on NT).
19809 *Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie*
19811 * Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5,
19812 TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and
19813 TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher
19814 Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt.
19818 * Add preliminary config info for new extension code.
19822 * Make RSA_NO_PADDING really use no padding.
19824 *Ulf Moeller <ulf@fitug.de>*
19826 * Generate errors when private/public key check is done.
19830 * Overhaul for 'crl' utility. New function X509_CRL_print. Partial support
19831 for some CRL extensions and new objects added.
19835 * Really fix the ASN1 IMPLICIT bug this time... Partial support for private
19836 key usage extension and fuller support for authority key id.
19840 * Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved
19841 padding method for RSA, which is recommended for new applications in PKCS
19842 #1 v2.0 (RFC 2437, October 1998).
19843 OAEP (Optimal Asymmetric Encryption Padding) has better theoretical
19844 foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure
19845 against Bleichbacher's attack on RSA.
19846 *Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by
19849 * Updates to the new SSL compression code
19851 *Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)*
19853 * Fix so that the version number in the master secret, when passed
19854 via RSA, checks that if TLS was proposed, but we roll back to SSLv3
19855 (because the server will not accept higher), that the version number
19856 is 0x03,0x01, not 0x03,0x00
19858 *Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)*
19860 * Run extensive memory leak checks on SSL commands. Fixed *lots* of memory
19861 leaks in `ssl/` relating to new `X509_get_pubkey()` behaviour. Also fixes
19862 in `apps/` and an unrelated leak in `crypto/dsa/dsa_vrf.c`.
19866 * Support for RAW extensions where an arbitrary extension can be
19867 created by including its DER encoding. See `apps/openssl.cnf` for
19872 * Make sure latest Perl versions don't interpret some generated C array
19873 code as Perl array code in the crypto/err/err_genc.pl script.
19875 *Lars Weber <3weber@informatik.uni-hamburg.de>*
19877 * Modify ms/do_ms.bat to not generate assembly language makefiles since
19878 not many people have the assembler. Various Win32 compilation fixes and
19879 update to the INSTALL.W32 file with (hopefully) more accurate Win32
19880 build instructions.
19884 * Modify configure script 'Configure' to automatically create crypto/date.h
19885 file under Win32 and also build pem.h from pem.org. New script
19886 util/mkfiles.pl to create the MINFO file on environments that can't do a
19887 'make files': perl util/mkfiles.pl >MINFO should work.
19891 * Major rework of DES function declarations, in the pursuit of correctness
19892 and purity. As a result, many evil casts evaporated, and some weirdness,
19893 too. You may find this causes warnings in your code. Zapping your evil
19894 casts will probably fix them. Mostly.
19898 * Fix for a typo in asn1.h. Bug fix to object creation script
19899 obj_dat.pl. It considered a zero in an object definition to mean
19900 "end of object": none of the objects in objects.h have any zeros
19901 so it wasn't spotted.
19903 *Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>*
19905 * Add support for Triple DES Cipher Block Chaining with Output Feedback
19906 Masking (CBCM). In the absence of test vectors, the best I have been able
19907 to do is check that the decrypt undoes the encrypt, so far. Send me test
19908 vectors if you have them.
19912 * Correct calculation of key length for export ciphers (too much space was
19913 allocated for null ciphers). This has not been tested!
19917 * Modifications to the mkdef.pl for Win32 DEF file creation. The usage
19918 message is now correct (it understands "crypto" and "ssl" on its
19919 command line). There is also now an "update" option. This will update
19920 the util/ssleay.num and util/libeay.num files with any new functions.
19922 perl util/mkdef.pl crypto ssl update
19923 it will update them.
19927 * Overhauled the Perl interface:
19928 - ported BN stuff to OpenSSL's different BN library
19929 - made the perl/ source tree CVS-aware
19930 - renamed the package from SSLeay to OpenSSL (the files still contain
19931 their history because I've copied them in the repository)
19932 - removed obsolete files (the test scripts will be replaced
19933 by better Test::Harness variants in the future)
19935 *Ralf S. Engelschall*
19937 * First cut for a very conservative source tree cleanup:
19938 1. merge various obsolete readme texts into doc/ssleay.txt
19939 where we collect the old documents and readme texts.
19940 2. remove the first part of files where I'm already sure that we no
19941 longer need them because of three reasons: either they are just temporary
19942 files which were left by Eric or they are preserved original files where
19943 I've verified that the diff is also available in the CVS via "cvs diff
19944 -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for
19945 the crypto/md/ stuff).
19947 *Ralf S. Engelschall*
19949 * More extension code. Incomplete support for subject and issuer alt
19950 name, issuer and authority key id. Change the i2v function parameters
19951 and add an extra 'crl' parameter in the X509V3_CTX structure: guess
19952 what that's for :-) Fix to ASN1 macro which messed up
19953 IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED.
19957 * Preliminary support for ENUMERATED type. This is largely copied from the
19962 * Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy.
19964 *Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)*
19966 * Make sure `make rehash` target really finds the `openssl` program.
19968 *Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>*
19970 * Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd
19971 like to hear about it if this slows down other processors.
19975 * Add CygWin32 platform information to Configure script.
19977 *Alan Batie <batie@aahz.jf.intel.com>*
19979 * Fixed ms/32all.bat script: `no_asm` -> `no-asm`
19981 *Rainer W. Gerling <gerling@mpg-gv.mpg.de>*
19983 * New program nseq to manipulate netscape certificate sequences
19987 * Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a
19992 * Fixes to BN code. Previously the default was to define BN_RECURSION
19993 but the BN code had some problems that would cause failures when
19994 doing certificate verification and some other functions.
19996 *Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)*
19998 * Add ASN1 and PEM code to support netscape certificate sequences.
20002 * Add ASN1 and PEM code to support netscape certificate sequences.
20006 * Add several PKIX and private extended key usage OIDs.
20010 * Modify the 'ca' program to handle the new extension code. Modify
20011 openssl.cnf for new extension format, add comments.
20015 * More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req'
20016 and add a sample to openssl.cnf so req -x509 now adds appropriate
20021 * Continued X509 V3 changes. Add to other makefiles, integrate with the
20022 error code, add initial support to X509_print() and x509 application.
20026 * Takes a deep breath and start adding X509 V3 extension support code. Add
20027 files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this
20028 stuff is currently isolated and isn't even compiled yet.
20032 * Continuing patches for GeneralizedTime. Fix up certificate and CRL
20033 ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print.
20034 Removed the versions check from X509 routines when loading extensions:
20035 this allows certain broken certificates that don't set the version
20036 properly to be processed.
20040 * Deal with irritating shit to do with dependencies, in YAAHW (Yet Another
20041 Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which
20042 can still be regenerated with "make depend".
20046 * Spelling mistake in C version of CAST-128.
20048 *Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>*
20050 * Changes to the error generation code. The perl script err-code.pl
20051 now reads in the old error codes and retains the old numbers, only
20052 adding new ones if necessary. It also only changes the .err files if new
20053 codes are added. The makefiles have been modified to only insert errors
20054 when needed (to avoid needlessly modifying header files). This is done
20055 by only inserting errors if the .err file is newer than the auto generated
20056 C file. To rebuild all the error codes from scratch (the old behaviour)
20057 either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl
20058 or delete all the .err files.
20062 * CAST-128 was incorrectly implemented for short keys. The C version has
20063 been fixed, but is untested. The assembler versions are also fixed, but
20064 new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing
20065 to regenerate it if needed.
20066 *Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun
20067 Hagino <itojun@kame.net>*
20069 * File was opened incorrectly in randfile.c.
20071 *Ulf Möller <ulf@fitug.de>*
20073 * Beginning of support for GeneralizedTime. d2i, i2d, check and print
20074 functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or
20075 GeneralizedTime. ASN1_TIME is the proper type used in certificates et
20076 al: it's just almost always a UTCTime. Note this patch adds new error
20077 codes so do a "make errors" if there are problems.
20081 * Correct Linux 1 recognition in config.
20083 *Ulf Möller <ulf@fitug.de>*
20085 * Remove pointless MD5 hash when using DSA keys in ca.
20087 *Anonymous <nobody@replay.com>*
20089 * Generate an error if given an empty string as a cert directory. Also
20090 generate an error if handed NULL (previously returned 0 to indicate an
20091 error, but didn't set one).
20093 *Ben Laurie, reported by Anonymous <nobody@replay.com>*
20095 * Add prototypes to SSL methods. Make SSL_write's buffer const, at last.
20099 * Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct
20100 parameters. This was causing a warning which killed off the Win32 compile.
20104 * Remove C++ style comments from crypto/bn/bn_local.h.
20106 *Neil Costigan <neil.costigan@celocom.com>*
20108 * The function OBJ_txt2nid was broken. It was supposed to return a nid
20109 based on a text string, looking up short and long names and finally
20110 "dot" format. The "dot" format stuff didn't work. Added new function
20111 OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote
20112 OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the
20113 OID is not part of the table.
20117 * Add prototypes to X509 lookup/verify methods, fixing a bug in
20118 X509_LOOKUP_by_alias().
20122 * Sort openssl functions by name.
20126 * Get the `gendsa` command working and add it to the `list` command. Remove
20127 encryption from sample DSA keys (in case anyone is interested the password
20132 * Make *all* `*_free` functions accept a NULL pointer.
20134 *Frans Heymans <fheymans@isaserver.be>*
20136 * If a DH key is generated in s3_srvr.c, don't blow it by trying to use
20139 *Anonymous <nobody@replay.com>*
20141 * s_server should send the CAfile as acceptable CAs, not its own cert.
20143 *Bodo Moeller <3moeller@informatik.uni-hamburg.de>*
20145 * Don't blow it for numeric `-newkey` arguments to `apps/req`.
20147 *Bodo Moeller <3moeller@informatik.uni-hamburg.de>*
20149 * Temp key "for export" tests were wrong in s3_srvr.c.
20151 *Anonymous <nobody@replay.com>*
20153 * Add prototype for temp key callback functions
20154 SSL_CTX_set_tmp_{rsa,dh}_callback().
20158 * Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and
20159 DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey().
20163 * X509_name_add_entry() freed the wrong thing after an error.
20165 *Arne Ansper <arne@ats.cyber.ee>*
20167 * rsa_eay.c would attempt to free a NULL context.
20169 *Arne Ansper <arne@ats.cyber.ee>*
20171 * BIO_s_socket() had a broken should_retry() on Windoze.
20173 *Arne Ansper <arne@ats.cyber.ee>*
20175 * BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH.
20177 *Arne Ansper <arne@ats.cyber.ee>*
20179 * Make sure the already existing X509_STORE->depth variable is initialized
20180 in X509_STORE_new(), but document the fact that this variable is still
20181 unused in the certificate verification process.
20183 *Ralf S. Engelschall*
20185 * Fix the various library and `apps/` files to free up pkeys obtained from
20186 X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions.
20190 * Fix reference counting in X509_PUBKEY_get(). This makes
20191 demos/maurice/example2.c work, amongst others, probably.
20193 *Steve Henson and Ben Laurie*
20195 * First cut of a cleanup for `apps/`. First the `ssleay` program is now named
20196 `openssl` and second, the shortcut symlinks for the `openssl <command>`
20197 are no longer created. This way we have a single and consistent command
20198 line interface `openssl <command>`, similar to `cvs <command>`.
20200 *Ralf S. Engelschall, Paul Sutton and Ben Laurie*
20202 * ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey
20203 BIT STRING wrapper always have zero unused bits.
20207 * Add CA.pl, perl version of CA.sh, add extended key usage OID.
20211 * Make the top-level INSTALL documentation easier to understand.
20215 * Makefiles updated to exit if an error occurs in a sub-directory
20216 make (including if user presses ^C) [Paul Sutton]
20218 * Make Montgomery context stuff explicit in RSA data structure.
20222 * Fix build order of pem and err to allow for generated pem.h.
20226 * Fix renumbering bug in X509_NAME_delete_entry().
20230 * Enhanced the err-ins.pl script so it makes the error library number
20231 global and can add a library name. This is needed for external ASN1 and
20232 other error libraries.
20236 * Fixed sk_insert which never worked properly.
20240 * Fix ASN1 macros so they can handle indefinite length constructed
20241 EXPLICIT tags. Some non standard certificates use these: they can now
20246 * Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc)
20247 into a single doc/ssleay.txt bundle. This way the information is still
20248 preserved but no longer messes up this directory. Now it's new room for
20249 the new set of documentation files.
20251 *Ralf S. Engelschall*
20253 * SETs were incorrectly DER encoded. This was a major pain, because they
20254 shared code with SEQUENCEs, which aren't coded the same. This means that
20255 almost everything to do with SETs or SEQUENCEs has either changed name or
20256 number of arguments.
20258 *Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>*
20260 * Fix test data to work with the above.
20264 * Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but
20265 was already fixed by Eric for 0.9.1 it seems.
20267 *Ben Laurie - pointed out by Ulf Möller <ulf@fitug.de>*
20269 * Autodetect FreeBSD3.
20273 * Fix various bugs in Configure. This affects the following platforms:
20277 unixware-2.0-pentium
20282 * Eliminate generated files from CVS. Reorder tests to regenerate files
20283 before they are needed.
20287 * Generate Makefile.ssl from Makefile.org (to keep CVS happy).
20291 ### Changes between 0.9.1b and 0.9.1c [23-Dec-1998]
20293 * Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and
20294 changed SSLeay to OpenSSL in version strings.
20296 *Ralf S. Engelschall*
20298 * Some fixups to the top-level documents.
20302 * Fixed the nasty bug where rsaref.h was not found under compile-time
20303 because the symlink to include/ was missing.
20305 *Ralf S. Engelschall*
20307 * Incorporated the popular no-RSA/DSA-only patches
20308 which allow to compile an RSA-free SSLeay.
20310 *Andrew Cooke / Interrader Ldt., Ralf S. Engelschall*
20312 * Fixed nasty rehash problem under `make -f Makefile.ssl links`
20313 when "ssleay" is still not found.
20315 *Ralf S. Engelschall*
20317 * Added more platforms to Configure: Cray T3E, HPUX 11,
20319 *Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>*
20321 * Updated the README file.
20323 *Ralf S. Engelschall*
20325 * Added various .cvsignore files in the CVS repository subdirs
20326 to make a "cvs update" really silent.
20328 *Ralf S. Engelschall*
20330 * Recompiled the error-definition header files and added
20331 missing symbols to the Win32 linker tables.
20333 *Ralf S. Engelschall*
20335 * Cleaned up the top-level documents;
20336 o new files: CHANGES and LICENSE
20337 o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay
20338 o merged COPYRIGHT into LICENSE
20339 o removed obsolete TODO file
20340 o renamed MICROSOFT to INSTALL.W32
20342 *Ralf S. Engelschall*
20344 * Removed dummy files from the 0.9.1b source tree:
20345 crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi
20346 crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f
20347 crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f
20348 crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f
20349 util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f
20351 *Ralf S. Engelschall*
20353 * Added various platform portability fixes.
20357 * The Genesis of the OpenSSL rpject:
20358 We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A.
20359 Young and Tim J. Hudson created while they were working for C2Net until
20362 *The OpenSSL Project*
20364 ### Changes between 0.9.0b and 0.9.1b [not released]
20366 * Updated a few CA certificates under certs/
20370 * Changed some BIGNUM api stuff.
20374 * Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD,
20375 DGUX x86, Linux Alpha, etc.
20379 * New COMP library [crypto/comp/] for SSL Record Layer Compression:
20380 RLE (dummy implemented) and ZLIB (really implemented when ZLIB is
20385 * Add -strparse option to asn1pars program which parses nested
20388 *Dr Stephen Henson <shenson@bigfoot.com>*
20390 * Added "oid_file" to ssleay.cnf for "ca" and "req" programs.
20394 * DSA fix for "ca" program.
20398 * Added "-genkey" option to "dsaparam" program.
20402 * Added RIPE MD160 (rmd160) message digest.
20406 * Added -a (all) option to "ssleay version" command.
20410 * Added PLATFORM define which is the id given to Configure.
20414 * Added MemCheck_XXXX functions to crypto/mem.c for memory checking.
20418 * Extended the ASN.1 parser routines.
20422 * Extended BIO routines to support REUSEADDR, seek, tell, etc.
20426 * Added a BN_CTX to the BN library.
20430 * Fixed the weak key values in DES library
20434 * Changed API in EVP library for cipher aliases.
20438 * Added support for RC2/64bit cipher.
20442 * Converted the lhash library to the crypto/mem.c functions.
20446 * Added more recognized ASN.1 object ids.
20450 * Added more RSA padding checks for SSL/TLS.
20454 * Added BIO proxy/filter functionality.
20458 * Added extra_certs to SSL_CTX which can be used
20459 send extra CA certificates to the client in the CA cert chain sending
20460 process. It can be configured with SSL_CTX_add_extra_chain_cert().
20464 * Now Fortezza is denied in the authentication phase because
20465 this is key exchange mechanism is not supported by SSLeay at all.
20469 * Additional PKCS1 checks.
20473 * Support the string "TLSv1" for all TLS v1 ciphers.
20477 * Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the
20478 ex_data index of the SSL context in the X509_STORE_CTX ex_data.
20482 * Fixed a few memory leaks.
20486 * Fixed various code and comment typos.
20490 * A minor bug in ssl/s3_clnt.c where there would always be 4 0
20491 bytes sent in the client random.
20493 *Edward Bishop <ebishop@spyglass.com>*
20497 [CVE-2024-0727]: https://www.openssl.org/news/vulnerabilities.html#CVE-2024-0727
20498 [CVE-2023-6237]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-6237
20499 [CVE-2023-6129]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-6129
20500 [CVE-2023-5678]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-5678
20501 [CVE-2023-5363]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-5363
20502 [CVE-2023-4807]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-4807
20503 [CVE-2023-3817]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-3817
20504 [CVE-2023-3446]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-3446
20505 [CVE-2023-2975]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-2975
20506 [RFC 2578 (STD 58), section 3.5]: https://datatracker.ietf.org/doc/html/rfc2578#section-3.5
20507 [CVE-2023-2650]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-2650
20508 [CVE-2023-1255]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-1255
20509 [CVE-2023-0466]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0466
20510 [CVE-2023-0465]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0465
20511 [CVE-2023-0464]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0464
20512 [CVE-2023-0401]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0401
20513 [CVE-2023-0286]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0286
20514 [CVE-2023-0217]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0217
20515 [CVE-2023-0216]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0216
20516 [CVE-2023-0215]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0215
20517 [CVE-2022-4450]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-4450
20518 [CVE-2022-4304]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-4304
20519 [CVE-2022-4203]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-4203
20520 [CVE-2022-3996]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-3996
20521 [CVE-2022-2274]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-2274
20522 [CVE-2022-2097]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-2097
20523 [CVE-2020-1971]: https://www.openssl.org/news/vulnerabilities.html#CVE-2020-1971
20524 [CVE-2020-1967]: https://www.openssl.org/news/vulnerabilities.html#CVE-2020-1967
20525 [CVE-2019-1563]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1563
20526 [CVE-2019-1559]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1559
20527 [CVE-2019-1552]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1552
20528 [CVE-2019-1551]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1551
20529 [CVE-2019-1549]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1549
20530 [CVE-2019-1547]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1547
20531 [CVE-2019-1543]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1543
20532 [CVE-2018-5407]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-5407
20533 [CVE-2018-0739]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0739
20534 [CVE-2018-0737]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0737
20535 [CVE-2018-0735]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0735
20536 [CVE-2018-0734]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0734
20537 [CVE-2018-0733]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0733
20538 [CVE-2018-0732]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0732
20539 [CVE-2017-3738]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3738
20540 [CVE-2017-3737]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3737
20541 [CVE-2017-3736]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3736
20542 [CVE-2017-3735]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3735
20543 [CVE-2017-3733]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3733
20544 [CVE-2017-3732]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3732
20545 [CVE-2017-3731]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3731
20546 [CVE-2017-3730]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3730
20547 [CVE-2016-7055]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-7055
20548 [CVE-2016-7054]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-7054
20549 [CVE-2016-7053]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-7053
20550 [CVE-2016-7052]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-7052
20551 [CVE-2016-6309]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6309
20552 [CVE-2016-6308]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6308
20553 [CVE-2016-6307]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6307
20554 [CVE-2016-6306]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6306
20555 [CVE-2016-6305]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6305
20556 [CVE-2016-6304]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6304
20557 [CVE-2016-6303]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6303
20558 [CVE-2016-6302]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6302
20559 [CVE-2016-2183]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2183
20560 [CVE-2016-2182]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2182
20561 [CVE-2016-2181]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2181
20562 [CVE-2016-2180]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2180
20563 [CVE-2016-2179]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2179
20564 [CVE-2016-2178]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2178
20565 [CVE-2016-2177]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2177
20566 [CVE-2016-2176]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2176
20567 [CVE-2016-2109]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2109
20568 [CVE-2016-2107]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2107
20569 [CVE-2016-2106]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2106
20570 [CVE-2016-2105]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2105
20571 [CVE-2016-0800]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0800
20572 [CVE-2016-0799]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0799
20573 [CVE-2016-0798]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0798
20574 [CVE-2016-0797]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0797
20575 [CVE-2016-0705]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0705
20576 [CVE-2016-0702]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0702
20577 [CVE-2016-0701]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0701
20578 [CVE-2015-3197]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3197
20579 [CVE-2015-3196]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3196
20580 [CVE-2015-3195]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3195
20581 [CVE-2015-3194]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3194
20582 [CVE-2015-3193]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3193
20583 [CVE-2015-1793]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1793
20584 [CVE-2015-1792]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1792
20585 [CVE-2015-1791]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1791
20586 [CVE-2015-1790]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1790
20587 [CVE-2015-1789]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1789
20588 [CVE-2015-1788]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1788
20589 [CVE-2015-1787]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1787
20590 [CVE-2015-0293]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0293
20591 [CVE-2015-0291]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0291
20592 [CVE-2015-0290]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0290
20593 [CVE-2015-0289]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0289
20594 [CVE-2015-0288]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0288
20595 [CVE-2015-0287]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0287
20596 [CVE-2015-0286]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0286
20597 [CVE-2015-0285]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0285
20598 [CVE-2015-0209]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0209
20599 [CVE-2015-0208]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0208
20600 [CVE-2015-0207]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0207
20601 [CVE-2015-0206]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0206
20602 [CVE-2015-0205]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0205
20603 [CVE-2015-0204]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0204
20604 [CVE-2014-8275]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-8275
20605 [CVE-2014-5139]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-5139
20606 [CVE-2014-3572]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3572
20607 [CVE-2014-3571]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3571
20608 [CVE-2014-3570]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3570
20609 [CVE-2014-3569]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3569
20610 [CVE-2014-3568]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3568
20611 [CVE-2014-3567]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3567
20612 [CVE-2014-3566]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3566
20613 [CVE-2014-3513]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3513
20614 [CVE-2014-3512]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3512
20615 [CVE-2014-3511]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3511
20616 [CVE-2014-3510]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3510
20617 [CVE-2014-3509]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3509
20618 [CVE-2014-3508]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3508
20619 [CVE-2014-3507]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3507
20620 [CVE-2014-3506]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3506
20621 [CVE-2014-3505]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3505
20622 [CVE-2014-3470]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3470
20623 [CVE-2014-0224]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0224
20624 [CVE-2014-0221]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0221
20625 [CVE-2014-0195]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0195
20626 [CVE-2014-0160]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0160
20627 [CVE-2014-0076]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0076
20628 [CVE-2013-6450]: https://www.openssl.org/news/vulnerabilities.html#CVE-2013-6450
20629 [CVE-2013-4353]: https://www.openssl.org/news/vulnerabilities.html#CVE-2013-4353
20630 [CVE-2013-0169]: https://www.openssl.org/news/vulnerabilities.html#CVE-2013-0169
20631 [CVE-2013-0166]: https://www.openssl.org/news/vulnerabilities.html#CVE-2013-0166
20632 [CVE-2012-2686]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-2686
20633 [CVE-2012-2333]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-2333
20634 [CVE-2012-2110]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-2110
20635 [CVE-2012-0884]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-0884
20636 [CVE-2012-0050]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-0050
20637 [CVE-2012-0027]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-0027
20638 [CVE-2011-4619]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4619
20639 [CVE-2011-4577]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4577
20640 [CVE-2011-4576]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4576
20641 [CVE-2011-4109]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4109
20642 [CVE-2011-4108]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4108
20643 [CVE-2011-3210]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-3210
20644 [CVE-2011-3207]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-3207
20645 [CVE-2011-0014]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-0014
20646 [CVE-2010-4252]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-4252
20647 [CVE-2010-4180]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-4180
20648 [CVE-2010-3864]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-3864
20649 [CVE-2010-1633]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-1633
20650 [CVE-2010-0740]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-0740
20651 [CVE-2010-0433]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-0433
20652 [CVE-2009-4355]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-4355
20653 [CVE-2009-3555]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-3555
20654 [CVE-2009-3245]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-3245
20655 [CVE-2009-1386]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-1386
20656 [CVE-2009-1379]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-1379
20657 [CVE-2009-1378]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-1378
20658 [CVE-2009-1377]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-1377
20659 [CVE-2009-0789]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-0789
20660 [CVE-2009-0591]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-0591
20661 [CVE-2009-0590]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-0590
20662 [CVE-2008-5077]: https://www.openssl.org/news/vulnerabilities.html#CVE-2008-5077
20663 [CVE-2008-1678]: https://www.openssl.org/news/vulnerabilities.html#CVE-2008-1678
20664 [CVE-2008-1672]: https://www.openssl.org/news/vulnerabilities.html#CVE-2008-1672
20665 [CVE-2008-0891]: https://www.openssl.org/news/vulnerabilities.html#CVE-2008-0891
20666 [CVE-2007-5135]: https://www.openssl.org/news/vulnerabilities.html#CVE-2007-5135
20667 [CVE-2007-4995]: https://www.openssl.org/news/vulnerabilities.html#CVE-2007-4995
20668 [CVE-2006-4343]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-4343
20669 [CVE-2006-4339]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-4339
20670 [CVE-2006-3738]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-3738
20671 [CVE-2006-2940]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-2940
20672 [CVE-2006-2937]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-2937
20673 [CVE-2005-2969]: https://www.openssl.org/news/vulnerabilities.html#CVE-2005-2969
20674 [CVE-2004-0112]: https://www.openssl.org/news/vulnerabilities.html#CVE-2004-0112
20675 [CVE-2004-0079]: https://www.openssl.org/news/vulnerabilities.html#CVE-2004-0079
20676 [CVE-2003-0851]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0851
20677 [CVE-2003-0545]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0545
20678 [CVE-2003-0544]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0544
20679 [CVE-2003-0543]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0543
20680 [CVE-2003-0078]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0078
20681 [CVE-2002-0659]: https://www.openssl.org/news/vulnerabilities.html#CVE-2002-0659
20682 [CVE-2002-0657]: https://www.openssl.org/news/vulnerabilities.html#CVE-2002-0657
20683 [CVE-2002-0656]: https://www.openssl.org/news/vulnerabilities.html#CVE-2002-0656
20684 [CVE-2002-0655]: https://www.openssl.org/news/vulnerabilities.html#CVE-2002-0655