1OpenSSL CHANGES 2=============== 3 4This is a high-level summary of the most important changes. 5For a full list of changes, see the [git commit log][log] and 6pick the appropriate release branch. 7 8 [log]: https://github.com/openssl/openssl/commits/ 9 10OpenSSL Releases 11---------------- 12 13 - [OpenSSL 3.0](#openssl-30) 14 - [OpenSSL 1.1.1](#openssl-111) 15 - [OpenSSL 1.1.0](#openssl-110) 16 - [OpenSSL 1.0.2](#openssl-102) 17 - [OpenSSL 1.0.1](#openssl-101) 18 - [OpenSSL 1.0.0](#openssl-100) 19 - [OpenSSL 0.9.x](#openssl-09x) 20 21OpenSSL 3.0 22----------- 23 24For OpenSSL 3.0 a [Migration guide][] has been added, so the CHANGES entries 25listed here are only a brief description. 26The migration guide contains more detailed information related to new features, 27breaking changes, and mappings for the large list of deprecated functions. 28 29[Migration guide]: https://github.com/openssl/openssl/tree/master/doc/man7/migration_guide.pod 30 31### Changes between 3.0.2 and 3.0.2+quic [15 Mar 2022] 32 33 * Add QUIC API support from BoringSSL. 34 35 *Todd Short* 36 37### Changes between 3.0.1 and 3.0.2 [15 Mar 2022] 38 39 * Fixed a bug in the BN_mod_sqrt() function that can cause it to loop forever 40 for non-prime moduli. 41 42 Internally this function is used when parsing certificates that contain 43 elliptic curve public keys in compressed form or explicit elliptic curve 44 parameters with a base point encoded in compressed form. 45 46 It is possible to trigger the infinite loop by crafting a certificate that 47 has invalid explicit curve parameters. 48 49 Since certificate parsing happens prior to verification of the certificate 50 signature, any process that parses an externally supplied certificate may thus 51 be subject to a denial of service attack. The infinite loop can also be 52 reached when parsing crafted private keys as they can contain explicit 53 elliptic curve parameters. 54 55 Thus vulnerable situations include: 56 57 - TLS clients consuming server certificates 58 - TLS servers consuming client certificates 59 - Hosting providers taking certificates or private keys from customers 60 - Certificate authorities parsing certification requests from subscribers 61 - Anything else which parses ASN.1 elliptic curve parameters 62 63 Also any other applications that use the BN_mod_sqrt() where the attacker 64 can control the parameter values are vulnerable to this DoS issue. 65 ([CVE-2022-0778]) 66 67 *Tomáš Mráz* 68 69 * Add ciphersuites based on DHE_PSK (RFC 4279) and ECDHE_PSK (RFC 5489) 70 to the list of ciphersuites providing Perfect Forward Secrecy as 71 required by SECLEVEL >= 3. 72 73 *Dmitry Belyavskiy, Nicola Tuveri* 74 75 * Made the AES constant time code for no-asm configurations 76 optional due to the resulting 95% performance degradation. 77 The AES constant time code can be enabled, for no assembly 78 builds, with: ./config no-asm -DOPENSSL_AES_CONST_TIME 79 80 *Paul Dale* 81 82 * Fixed PEM_write_bio_PKCS8PrivateKey() to make it possible to use empty 83 passphrase strings. 84 85 *Darshan Sen* 86 87 * The negative return value handling of the certificate verification callback 88 was reverted. The replacement is to set the verification retry state with 89 the SSL_set_retry_verify() function. 90 91 *Tomáš Mráz* 92 93### Changes between 3.0.0 and 3.0.1 [14 Dec 2021] 94 95 * Fixed invalid handling of X509_verify_cert() internal errors in libssl 96 Internally libssl in OpenSSL calls X509_verify_cert() on the client side to 97 verify a certificate supplied by a server. That function may return a 98 negative return value to indicate an internal error (for example out of 99 memory). Such a negative return value is mishandled by OpenSSL and will cause 100 an IO function (such as SSL_connect() or SSL_do_handshake()) to not indicate 101 success and a subsequent call to SSL_get_error() to return the value 102 SSL_ERROR_WANT_RETRY_VERIFY. This return value is only supposed to be 103 returned by OpenSSL if the application has previously called 104 SSL_CTX_set_cert_verify_callback(). Since most applications do not do this 105 the SSL_ERROR_WANT_RETRY_VERIFY return value from SSL_get_error() will be 106 totally unexpected and applications may not behave correctly as a result. The 107 exact behaviour will depend on the application but it could result in 108 crashes, infinite loops or other similar incorrect responses. 109 110 This issue is made more serious in combination with a separate bug in OpenSSL 111 3.0 that will cause X509_verify_cert() to indicate an internal error when 112 processing a certificate chain. This will occur where a certificate does not 113 include the Subject Alternative Name extension but where a Certificate 114 Authority has enforced name constraints. This issue can occur even with valid 115 chains. 116 ([CVE-2021-4044]) 117 118 *Matt Caswell* 119 120 * Corrected a few file name and file reference bugs in the build, 121 installation and setup scripts, which lead to installation verification 122 failures. Slightly enhanced the installation verification script. 123 124 *Richard Levitte* 125 126 * Fixed EVP_PKEY_eq() to make it possible to use it with strictly private 127 keys. 128 129 *Richard Levitte* 130 131 * Fixed PVK encoder to properly query for the passphrase. 132 133 *Tomáš Mráz* 134 135 * Multiple fixes in the OSSL_HTTP API functions. 136 137 *David von Oheimb* 138 139 * Allow sign extension in OSSL_PARAM_allocate_from_text() for the 140 OSSL_PARAM_INTEGER data type and return error on negative numbers 141 used with the OSSL_PARAM_UNSIGNED_INTEGER data type. Make 142 OSSL_PARAM_BLD_push_BN{,_pad}() return an error on negative numbers. 143 144 *Richard Levitte* 145 146 * Allow copying uninitialized digest contexts with EVP_MD_CTX_copy_ex. 147 148 *Tomáš Mráz* 149 150 * Fixed detection of ARMv7 and ARM64 CPU features on FreeBSD. 151 152 *Allan Jude* 153 154 * Multiple threading fixes. 155 156 *Matt Caswell* 157 158 * Added NULL digest implementation to keep compatibility with 1.1.1 version. 159 160 *Tomáš Mráz* 161 162 * Allow fetching an operation from the provider that owns an unexportable key 163 as a fallback if that is still allowed by the property query. 164 165 *Richard Levitte* 166 167### Changes between 1.1.1 and 3.0.0 [7 sep 2021] 168 169 * TLS_MAX_VERSION, DTLS_MAX_VERSION and DTLS_MIN_VERSION constants are now 170 deprecated. 171 172 *Matt Caswell* 173 174 * The `OPENSSL_s390xcap` environment variable can be used to set bits in the 175 S390X capability vector to zero. This simplifies testing of different code 176 paths on S390X architecture. 177 178 *Patrick Steuer* 179 180 * Encrypting more than 2^64 TLS records with AES-GCM is disallowed 181 as per FIPS 140-2 IG A.5 "Key/IV Pair Uniqueness Requirements from 182 SP 800-38D". The communication will fail at this point. 183 184 *Paul Dale* 185 186 * The EC_GROUP_clear_free() function is deprecated as there is nothing 187 confidential in EC_GROUP data. 188 189 *Nicola Tuveri* 190 191 * The byte order mark (BOM) character is ignored if encountered at the 192 beginning of a PEM-formatted file. 193 194 *Dmitry Belyavskiy* 195 196 * Added CMS support for the Russian GOST algorithms. 197 198 *Dmitry Belyavskiy* 199 200 * Due to move of the implementation of cryptographic operations 201 to the providers, validation of various operation parameters can 202 be postponed until the actual operation is executed where previously 203 it happened immediately when an operation parameter was set. 204 205 For example when setting an unsupported curve with 206 EVP_PKEY_CTX_set_ec_paramgen_curve_nid() this function call will not 207 fail but later keygen operations with the EVP_PKEY_CTX will fail. 208 209 *OpenSSL team members and many third party contributors* 210 211 * The EVP_get_cipherbyname() function will return NULL for algorithms such as 212 "AES-128-SIV", "AES-128-CBC-CTS" and "CAMELLIA-128-CBC-CTS" which were 213 previously only accessible via low level interfaces. Use EVP_CIPHER_fetch() 214 instead to retrieve these algorithms from a provider. 215 216 *Shane Lontis* 217 218 * On build targets where the multilib postfix is set in the build 219 configuration the libdir directory was changing based on whether 220 the lib directory with the multilib postfix exists on the system 221 or not. This unpredictable behavior was removed and eventual 222 multilib postfix is now always added to the default libdir. Use 223 `--libdir=lib` to override the libdir if adding the postfix is 224 undesirable. 225 226 *Jan Lána* 227 228 * The triple DES key wrap functionality now conforms to RFC 3217 but is 229 no longer interoperable with OpenSSL 1.1.1. 230 231 *Paul Dale* 232 233 * The ERR_GET_FUNC() function was removed. With the loss of meaningful 234 function codes, this function can only cause problems for calling 235 applications. 236 237 *Paul Dale* 238 239 * Add a configurable flag to output date formats as ISO 8601. Does not 240 change the default date format. 241 242 *William Edmisten* 243 244 * Version of MSVC earlier than 1300 could get link warnings, which could 245 be suppressed if the undocumented -DI_CAN_LIVE_WITH_LNK4049 was set. 246 Support for this flag has been removed. 247 248 *Rich Salz* 249 250 * Rework and make DEBUG macros consistent. Remove unused -DCONF_DEBUG, 251 -DBN_CTX_DEBUG, and REF_PRINT. Add a new tracing category and use it for 252 printing reference counts. Rename -DDEBUG_UNUSED to -DUNUSED_RESULT_DEBUG 253 Fix BN_DEBUG_RAND so it compiles and, when set, force DEBUG_RAND to be set 254 also. Rename engine_debug_ref to be ENGINE_REF_PRINT also for consistency. 255 256 *Rich Salz* 257 258 * The signatures of the functions to get and set options on SSL and 259 SSL_CTX objects changed from "unsigned long" to "uint64_t" type. 260 Some source code changes may be required. 261 262 *Rich Salz* 263 264 * The public definitions of conf_method_st and conf_st have been 265 deprecated. They will be made opaque in a future release. 266 267 *Rich Salz and Tomáš Mráz* 268 269 * Client-initiated renegotiation is disabled by default. To allow it, use 270 the -client_renegotiation option, the SSL_OP_ALLOW_CLIENT_RENEGOTIATION 271 flag, or the "ClientRenegotiation" config parameter as appropriate. 272 273 *Rich Salz* 274 275 * Add "abspath" and "includedir" pragma's to config files, to prevent, 276 or modify relative pathname inclusion. 277 278 *Rich Salz* 279 280 * OpenSSL includes a cryptographic module that is intended to be FIPS 140-2 281 validated. Please consult the README-FIPS and 282 README-PROVIDERS files, as well as the migration guide. 283 284 *OpenSSL team members and many third party contributors* 285 286 * For the key types DH and DHX the allowed settable parameters are now different. 287 288 *Shane Lontis* 289 290 * The openssl commands that read keys, certificates, and CRLs now 291 automatically detect the PEM or DER format of the input files. 292 293 *David von Oheimb, Richard Levitte, and Tomáš Mráz* 294 295 * Added enhanced PKCS#12 APIs which accept a library context. 296 297 *Jon Spillett* 298 299 * The default manual page suffix ($MANSUFFIX) has been changed to "ossl" 300 301 *Matt Caswell* 302 303 * Added support for Kernel TLS (KTLS). 304 305 *Boris Pismenny, John Baldwin and Andrew Gallatin* 306 307 * Support for RFC 5746 secure renegotiation is now required by default for 308 SSL or TLS connections to succeed. 309 310 *Benjamin Kaduk* 311 312 * The signature of the `copy` functional parameter of the 313 EVP_PKEY_meth_set_copy() function has changed so its `src` argument is 314 now `const EVP_PKEY_CTX *` instead of `EVP_PKEY_CTX *`. Similarly 315 the signature of the `pub_decode` functional parameter of the 316 EVP_PKEY_asn1_set_public() function has changed so its `pub` argument is 317 now `const X509_PUBKEY *` instead of `X509_PUBKEY *`. 318 319 *David von Oheimb* 320 321 * The error return values from some control calls (ctrl) have changed. 322 323 *Paul Dale* 324 325 * A public key check is now performed during EVP_PKEY_derive_set_peer(). 326 327 *Shane Lontis* 328 329 * Many functions in the EVP_ namespace that are getters of values from 330 implementations or contexts were renamed to include get or get0 in their 331 names. Old names are provided as macro aliases for compatibility and 332 are not deprecated. 333 334 *Tomáš Mráz* 335 336 * The EVP_PKEY_CTRL_PKCS7_ENCRYPT, EVP_PKEY_CTRL_PKCS7_DECRYPT, 337 EVP_PKEY_CTRL_PKCS7_SIGN, EVP_PKEY_CTRL_CMS_ENCRYPT, 338 EVP_PKEY_CTRL_CMS_DECRYPT, and EVP_PKEY_CTRL_CMS_SIGN control operations 339 are deprecated. 340 341 *Tomáš Mráz* 342 343 * The EVP_PKEY_public_check() and EVP_PKEY_param_check() functions now work for 344 more key types. 345 346 * The output from the command line applications may have minor 347 changes. 348 349 *Paul Dale* 350 351 * The output from numerous "printing" may have minor changes. 352 353 *David von Oheimb* 354 355 * Windows thread synchronization uses read/write primitives (SRWLock) when 356 supported by the OS, otherwise CriticalSection continues to be used. 357 358 *Vincent Drake* 359 360 * Add filter BIO BIO_f_readbuffer() that allows BIO_tell() and BIO_seek() to 361 work on read only BIO source/sinks that do not support these functions. 362 This allows piping or redirection of a file BIO using stdin to be buffered 363 into memory. This is used internally in OSSL_DECODER_from_bio(). 364 365 *Shane Lontis* 366 367 * OSSL_STORE_INFO_get_type() may now return an additional value. In 1.1.1 368 this function would return one of the values OSSL_STORE_INFO_NAME, 369 OSSL_STORE_INFO_PKEY, OSSL_STORE_INFO_PARAMS, OSSL_STORE_INFO_CERT or 370 OSSL_STORE_INFO_CRL. Decoded public keys would previously have been reported 371 as type OSSL_STORE_INFO_PKEY in 1.1.1. In 3.0 decoded public keys are now 372 reported as having the new type OSSL_STORE_INFO_PUBKEY. Applications 373 using this function should be amended to handle the changed return value. 374 375 *Richard Levitte* 376 377 * Improved adherence to Enhanced Security Services (ESS, RFC 2634 and RFC 5035) 378 for the TSP and CMS Advanced Electronic Signatures (CAdES) implementations. 379 As required by RFC 5035 check both ESSCertID and ESSCertIDv2 if both present. 380 Correct the semantics of checking the validation chain in case ESSCertID{,v2} 381 contains more than one certificate identifier: This means that all 382 certificates referenced there MUST be part of the validation chain. 383 384 *David von Oheimb* 385 386 * The implementation of older EVP ciphers related to CAST, IDEA, SEED, RC2, RC4, 387 RC5, DESX and DES have been moved to the legacy provider. 388 389 *Matt Caswell* 390 391 * The implementation of the EVP digests MD2, MD4, MDC2, WHIRLPOOL and 392 RIPEMD-160 have been moved to the legacy provider. 393 394 *Matt Caswell* 395 396 * The deprecated function EVP_PKEY_get0() now returns NULL being called for a 397 provided key. 398 399 *Dmitry Belyavskiy* 400 401 * The deprecated functions EVP_PKEY_get0_RSA(), 402 EVP_PKEY_get0_DSA(), EVP_PKEY_get0_EC_KEY(), EVP_PKEY_get0_DH(), 403 EVP_PKEY_get0_hmac(), EVP_PKEY_get0_poly1305() and EVP_PKEY_get0_siphash() as 404 well as the similarly named "get1" functions behave differently in 405 OpenSSL 3.0. 406 407 *Matt Caswell* 408 409 * A number of functions handling low-level keys or engines were deprecated 410 including EVP_PKEY_set1_engine(), EVP_PKEY_get0_engine(), EVP_PKEY_assign(), 411 EVP_PKEY_get0(), EVP_PKEY_get0_hmac(), EVP_PKEY_get0_poly1305() and 412 EVP_PKEY_get0_siphash(). 413 414 *Matt Caswell* 415 416 * PKCS#5 PBKDF1 key derivation has been moved from PKCS5_PBE_keyivgen() into 417 the legacy crypto provider as an EVP_KDF. Applications requiring this KDF 418 will need to load the legacy crypto provider. This includes these PBE 419 algorithms which use this KDF: 420 - NID_pbeWithMD2AndDES_CBC 421 - NID_pbeWithMD5AndDES_CBC 422 - NID_pbeWithSHA1AndRC2_CBC 423 - NID_pbeWithMD2AndRC2_CBC 424 - NID_pbeWithMD5AndRC2_CBC 425 - NID_pbeWithSHA1AndDES_CBC 426 427 *Jon Spillett* 428 429 * Deprecated obsolete BIO_set_callback(), BIO_get_callback(), and 430 BIO_debug_callback() functions. 431 432 *Tomáš Mráz* 433 434 * Deprecated obsolete EVP_PKEY_CTX_get0_dh_kdf_ukm() and 435 EVP_PKEY_CTX_get0_ecdh_kdf_ukm() functions. 436 437 *Tomáš Mráz* 438 439 * The RAND_METHOD APIs have been deprecated. 440 441 *Paul Dale* 442 443 * The SRP APIs have been deprecated. 444 445 *Matt Caswell* 446 447 * Add a compile time option to prevent the caching of provider fetched 448 algorithms. This is enabled by including the no-cached-fetch option 449 at configuration time. 450 451 *Paul Dale* 452 453 * pkcs12 now uses defaults of PBKDF2, AES and SHA-256, with a MAC iteration 454 count of PKCS12_DEFAULT_ITER. 455 456 *Tomáš Mráz and Sahana Prasad* 457 458 * The openssl speed command does not use low-level API calls anymore. 459 460 *Tomáš Mráz* 461 462 * Parallel dual-prime 1024-bit modular exponentiation for AVX512_IFMA 463 capable processors. 464 465 *Ilya Albrekht, Sergey Kirillov, Andrey Matyukov (Intel Corp)* 466 467 * Combining the Configure options no-ec and no-dh no longer disables TLSv1.3. 468 469 *Matt Caswell* 470 471 * Implemented support for fully "pluggable" TLSv1.3 groups. This means that 472 providers may supply their own group implementations (using either the "key 473 exchange" or the "key encapsulation" methods) which will automatically be 474 detected and used by libssl. 475 476 *Matt Caswell, Nicola Tuveri* 477 478 * The undocumented function X509_certificate_type() has been deprecated; 479 480 *Rich Salz* 481 482 * Deprecated the obsolete BN_pseudo_rand() and BN_pseudo_rand_range(). 483 484 *Tomáš Mráz* 485 486 * Removed RSA padding mode for SSLv23 (which was only used for 487 SSLv2). This includes the functions RSA_padding_check_SSLv23() and 488 RSA_padding_add_SSLv23() and the `-ssl` option in the deprecated 489 `rsautl` command. 490 491 *Rich Salz* 492 493 * Deprecated the obsolete X9.31 RSA key generation related functions. 494 495 * While a callback function set via `SSL_CTX_set_cert_verify_callback()` 496 is not allowed to return a value > 1, this is no more taken as failure. 497 498 *Viktor Dukhovni and David von Oheimb* 499 500 * Deprecated the obsolete X9.31 RSA key generation related functions 501 BN_X931_generate_Xpq(), BN_X931_derive_prime_ex(), and 502 BN_X931_generate_prime_ex(). 503 504 *Tomáš Mráz* 505 506 * The default key generation method for the regular 2-prime RSA keys was 507 changed to the FIPS 186-4 B.3.6 method. 508 509 *Shane Lontis* 510 511 * Deprecated the BN_is_prime_ex() and BN_is_prime_fasttest_ex() functions. 512 513 *Kurt Roeckx* 514 515 * Deprecated EVP_MD_CTX_set_update_fn() and EVP_MD_CTX_update_fn(). 516 517 *Rich Salz* 518 519 * Deprecated the type OCSP_REQ_CTX and the functions OCSP_REQ_CTX_*() and 520 replaced with OSSL_HTTP_REQ_CTX and the functions OSSL_HTTP_REQ_CTX_*(). 521 522 *Rich Salz, Richard Levitte, and David von Oheimb* 523 524 * Deprecated `X509_http_nbio()` and `X509_CRL_http_nbio()`. 525 526 *David von Oheimb* 527 528 * Deprecated `OCSP_parse_url()`. 529 530 *David von Oheimb* 531 532 * Validation of SM2 keys has been separated from the validation of regular EC 533 keys. 534 535 *Nicola Tuveri* 536 537 * Behavior of the `pkey` app is changed, when using the `-check` or `-pubcheck` 538 switches: a validation failure triggers an early exit, returning a failure 539 exit status to the parent process. 540 541 *Nicola Tuveri* 542 543 * Changed behavior of SSL_CTX_set_ciphersuites() and SSL_set_ciphersuites() 544 to ignore unknown ciphers. 545 546 *Otto Hollmann* 547 548 * The `-cipher-commands` and `-digest-commands` options 549 of the command line utility `list` have been deprecated. 550 Instead use the `-cipher-algorithms` and `-digest-algorithms` options. 551 552 *Dmitry Belyavskiy* 553 554 * Added convenience functions for generating asymmetric key pairs: 555 The 'quick' one-shot (yet somewhat limited) function L<EVP_PKEY_Q_keygen(3)> 556 and macros for the most common cases: <EVP_RSA_gen(3)> and L<EVP_EC_gen(3)>. 557 558 *David von Oheimb* 559 560 * All of the low level EC_KEY functions have been deprecated. 561 562 *Shane Lontis, Paul Dale, Richard Levitte, and Tomáš Mráz* 563 564 * Deprecated all the libcrypto and libssl error string loading 565 functions. 566 567 *Richard Levitte* 568 569 * The functions SSL_CTX_set_tmp_dh_callback and SSL_set_tmp_dh_callback, as 570 well as the macros SSL_CTX_set_tmp_dh() and SSL_set_tmp_dh() have been 571 deprecated. 572 573 *Matt Caswell* 574 575 * The `-crypt` option to the `passwd` command line tool has been removed. 576 577 *Paul Dale* 578 579 * The -C option to the `x509`, `dhparam`, `dsaparam`, and `ecparam` commands 580 were removed. 581 582 *Rich Salz* 583 584 * Add support for AES Key Wrap inverse ciphers to the EVP layer. 585 586 *Shane Lontis* 587 588 * Deprecated EVP_PKEY_set1_tls_encodedpoint() and 589 EVP_PKEY_get1_tls_encodedpoint(). 590 591 *Matt Caswell* 592 593 * The security callback, which can be customised by application code, supports 594 the security operation SSL_SECOP_TMP_DH. One location of the "other" parameter 595 was incorrectly passing a DH object. It now passed an EVP_PKEY in all cases. 596 597 *Matt Caswell* 598 599 * Add PKCS7_get_octet_string() and PKCS7_type_is_other() to the public 600 interface. Their functionality remains unchanged. 601 602 *Jordan Montgomery* 603 604 * Added new option for 'openssl list', '-providers', which will display the 605 list of loaded providers, their names, version and status. It optionally 606 displays their gettable parameters. 607 608 *Paul Dale* 609 610 * Removed EVP_PKEY_set_alias_type(). 611 612 *Richard Levitte* 613 614 * Deprecated `EVP_PKEY_CTX_set_rsa_keygen_pubexp()` and introduced 615 `EVP_PKEY_CTX_set1_rsa_keygen_pubexp()`, which is now preferred. 616 617 *Jeremy Walch* 618 619 * Changed all "STACK" functions to be macros instead of inline functions. Macro 620 parameters are still checked for type safety at compile time via helper 621 inline functions. 622 623 *Matt Caswell* 624 625 * Remove the RAND_DRBG API 626 627 *Paul Dale and Matthias St. Pierre* 628 629 * Allow `SSL_set1_host()` and `SSL_add1_host()` to take IP literal addresses 630 as well as actual hostnames. 631 632 *David Woodhouse* 633 634 * The 'MinProtocol' and 'MaxProtocol' configuration commands now silently 635 ignore TLS protocol version bounds when configuring DTLS-based contexts, and 636 conversely, silently ignore DTLS protocol version bounds when configuring 637 TLS-based contexts. The commands can be repeated to set bounds of both 638 types. The same applies with the corresponding "min_protocol" and 639 "max_protocol" command-line switches, in case some application uses both TLS 640 and DTLS. 641 642 SSL_CTX instances that are created for a fixed protocol version (e.g. 643 `TLSv1_server_method()`) also silently ignore version bounds. Previously 644 attempts to apply bounds to these protocol versions would result in an 645 error. Now only the "version-flexible" SSL_CTX instances are subject to 646 limits in configuration files in command-line options. 647 648 *Viktor Dukhovni* 649 650 * Deprecated the `ENGINE` API. Engines should be replaced with providers 651 going forward. 652 653 *Paul Dale* 654 655 * Reworked the recorded ERR codes to make better space for system errors. 656 To distinguish them, the macro `ERR_SYSTEM_ERROR()` indicates if the 657 given code is a system error (true) or an OpenSSL error (false). 658 659 *Richard Levitte* 660 661 * Reworked the test perl framework to better allow parallel testing. 662 663 *Nicola Tuveri and David von Oheimb* 664 665 * Added ciphertext stealing algorithms AES-128-CBC-CTS, AES-192-CBC-CTS and 666 AES-256-CBC-CTS to the providers. CS1, CS2 and CS3 variants are supported. 667 668 *Shane Lontis* 669 670 * 'Configure' has been changed to figure out the configuration target if 671 none is given on the command line. Consequently, the 'config' script is 672 now only a mere wrapper. All documentation is changed to only mention 673 'Configure'. 674 675 *Rich Salz and Richard Levitte* 676 677 * Added a library context `OSSL_LIB_CTX` that applications as well as 678 other libraries can use to form a separate context within which 679 libcrypto operations are performed. 680 681 *Richard Levitte* 682 683 * Added various `_ex` functions to the OpenSSL API that support using 684 a non-default `OSSL_LIB_CTX`. 685 686 *OpenSSL team* 687 688 * Handshake now fails if Extended Master Secret extension is dropped 689 on renegotiation. 690 691 *Tomáš Mráz* 692 693 * Dropped interactive mode from the `openssl` program. 694 695 *Richard Levitte* 696 697 * Deprecated `EVP_PKEY_cmp()` and `EVP_PKEY_cmp_parameters()`. 698 699 *David von Oheimb and Shane Lontis* 700 701 * Deprecated `EC_METHOD_get_field_type()`. 702 703 *Billy Bob Brumley* 704 705 * Deprecated EC_GFp_simple_method(), EC_GFp_mont_method(), 706 EC_GF2m_simple_method(), EC_GFp_nist_method(), EC_GFp_nistp224_method() 707 EC_GFp_nistp256_method(), and EC_GFp_nistp521_method(). 708 709 *Billy Bob Brumley* 710 711 * Deprecated EC_GROUP_new(), EC_GROUP_method_of(), and EC_POINT_method_of(). 712 713 *Billy Bob Brumley* 714 715 * Add CAdES-BES signature verification support, mostly derived 716 from ESSCertIDv2 TS (RFC 5816) contribution by Marek Klein. 717 718 *Filipe Raimundo da Silva* 719 720 * Add CAdES-BES signature scheme and attributes support (RFC 5126) to CMS API. 721 722 *Antonio Iacono* 723 724 * Added the AuthEnvelopedData content type structure (RFC 5083) with AES-GCM 725 parameter (RFC 5084) for the Cryptographic Message Syntax (CMS). 726 727 *Jakub Zelenka* 728 729 * Deprecated EC_POINT_make_affine() and EC_POINTs_make_affine(). 730 731 *Billy Bob Brumley* 732 733 * Deprecated EC_GROUP_precompute_mult(), EC_GROUP_have_precompute_mult(), and 734 EC_KEY_precompute_mult(). 735 736 *Billy Bob Brumley* 737 738 * Deprecated EC_POINTs_mul(). 739 740 *Billy Bob Brumley* 741 742 * Removed FIPS_mode() and FIPS_mode_set(). 743 744 *Shane Lontis* 745 746 * The SSL option SSL_OP_IGNORE_UNEXPECTED_EOF is introduced. 747 748 *Dmitry Belyavskiy* 749 750 * Deprecated EC_POINT_set_Jprojective_coordinates_GFp() and 751 EC_POINT_get_Jprojective_coordinates_GFp(). 752 753 *Billy Bob Brumley* 754 755 * Added OSSL_PARAM_BLD to the public interface. This allows OSSL_PARAM 756 arrays to be more easily constructed via a series of utility functions. 757 Create a parameter builder using OSSL_PARAM_BLD_new(), add parameters using 758 the various push functions and finally convert to a passable OSSL_PARAM 759 array using OSSL_PARAM_BLD_to_param(). 760 761 *Paul Dale* 762 763 * The security strength of SHA1 and MD5 based signatures in TLS has been 764 reduced. 765 766 *Kurt Roeckx* 767 768 * Added EVP_PKEY_set_type_by_keymgmt(), to initialise an EVP_PKEY to 769 contain a provider side internal key. 770 771 *Richard Levitte* 772 773 * ASN1_verify(), ASN1_digest() and ASN1_sign() have been deprecated. 774 775 *Richard Levitte* 776 777 * Project text documents not yet having a proper file name extension 778 (`HACKING`, `LICENSE`, `NOTES*`, `README*`, `VERSION`) have been renamed to 779 `*.md` as far as reasonable, else `*.txt`, for better use with file managers. 780 781 *David von Oheimb* 782 783 * The main project documents (README, NEWS, CHANGES, INSTALL, SUPPORT) 784 have been converted to Markdown with the goal to produce documents 785 which not only look pretty when viewed online in the browser, but 786 remain well readable inside a plain text editor. 787 788 To achieve this goal, a 'minimalistic' Markdown style has been applied 789 which avoids formatting elements that interfere too much with the 790 reading flow in the text file. For example, it 791 792 * avoids [ATX headings][] and uses [setext headings][] instead 793 (which works for `<h1>` and `<h2>` headings only). 794 * avoids [inline links][] and uses [reference links][] instead. 795 * avoids [fenced code blocks][] and uses [indented code blocks][] instead. 796 797 [ATX headings]: https://github.github.com/gfm/#atx-headings 798 [setext headings]: https://github.github.com/gfm/#setext-headings 799 [inline links]: https://github.github.com/gfm/#inline-link 800 [reference links]: https://github.github.com/gfm/#reference-link 801 [fenced code blocks]: https://github.github.com/gfm/#fenced-code-blocks 802 [indented code blocks]: https://github.github.com/gfm/#indented-code-blocks 803 804 *Matthias St. Pierre* 805 806 * The test suite is changed to preserve results of each test recipe. 807 A new directory test-runs/ with subdirectories named like the 808 test recipes are created in the build tree for this purpose. 809 810 *Richard Levitte* 811 812 * Added an implementation of CMP and CRMF (RFC 4210, RFC 4211 RFC 6712). 813 This adds `crypto/cmp/`, `crpyto/crmf/`, `apps/cmp.c`, and `test/cmp_*`. 814 See L<openssl-cmp(1)> and L<OSSL_CMP_exec_IR_ses(3)> as starting points. 815 816 *David von Oheimb, Martin Peylo* 817 818 * Generalized the HTTP client code from `crypto/ocsp/` into `crpyto/http/`. 819 It supports arbitrary request and response content types, GET redirection, 820 TLS, connections via HTTP(S) proxies, connections and exchange via 821 user-defined BIOs (allowing implicit connections), persistent connections, 822 and timeout checks. See L<OSSL_HTTP_transfer(3)> etc. for details. 823 The legacy OCSP-focused (and only partly documented) API 824 is retained for backward compatibility, while most of it is deprecated. 825 826 *David von Oheimb* 827 828 * Added `util/check-format.pl`, a tool for checking adherence to the 829 OpenSSL coding style <https://www.openssl.org/policies/codingstyle.html>. 830 The checks performed are incomplete and yield some false positives. 831 Still the tool should be useful for detecting most typical glitches. 832 833 *David von Oheimb* 834 835 * `BIO_do_connect()` and `BIO_do_handshake()` have been extended: 836 If domain name resolution yields multiple IP addresses all of them are tried 837 after `connect()` failures. 838 839 *David von Oheimb* 840 841 * All of the low level RSA functions have been deprecated. 842 843 *Paul Dale* 844 845 * X509 certificates signed using SHA1 are no longer allowed at security 846 level 1 and above. 847 848 *Kurt Roeckx* 849 850 * The command line utilities dhparam, dsa, gendsa and dsaparam have been 851 modified to use PKEY APIs. These commands are now in maintenance mode 852 and no new features will be added to them. 853 854 *Paul Dale* 855 856 * The command line utility rsautl has been deprecated. 857 858 *Paul Dale* 859 860 * The command line utilities genrsa and rsa have been modified to use PKEY 861 APIs. They now write PKCS#8 keys by default. These commands are now in 862 maintenance mode and no new features will be added to them. 863 864 *Paul Dale* 865 866 * All of the low level DH functions have been deprecated. 867 868 *Paul Dale and Matt Caswell* 869 870 * All of the low level DSA functions have been deprecated. 871 872 *Paul Dale* 873 874 * Reworked the treatment of EC EVP_PKEYs with the SM2 curve to 875 automatically become EVP_PKEY_SM2 rather than EVP_PKEY_EC. 876 877 *Richard Levitte* 878 879 * Deprecated low level ECDH and ECDSA functions. 880 881 *Paul Dale* 882 883 * Deprecated EVP_PKEY_decrypt_old() and EVP_PKEY_encrypt_old(). 884 885 *Richard Levitte* 886 887 * Enhanced the documentation of EVP_PKEY_get_size(), EVP_PKEY_get_bits() 888 and EVP_PKEY_get_security_bits(). Especially EVP_PKEY_get_size() needed 889 a new formulation to include all the things it can be used for, 890 as well as words of caution. 891 892 *Richard Levitte* 893 894 * The SSL_CTX_set_tlsext_ticket_key_cb(3) function has been deprecated. 895 896 *Paul Dale* 897 898 * All of the low level HMAC functions have been deprecated. 899 900 *Paul Dale and David von Oheimb* 901 902 * Over two thousand fixes were made to the documentation, including: 903 - Common options (such as -rand/-writerand, TLS version control, etc) 904 were refactored and point to newly-enhanced descriptions in openssl.pod. 905 - Added style conformance for all options (with help from Richard Levitte), 906 documented all reported missing options, added a CI build to check 907 that all options are documented and that no unimplemented options 908 are documented. 909 - Documented some internals, such as all use of environment variables. 910 - Addressed all internal broken L<> references. 911 912 *Rich Salz* 913 914 * All of the low level CMAC functions have been deprecated. 915 916 *Paul Dale* 917 918 * The low-level MD2, MD4, MD5, MDC2, RIPEMD160 and Whirlpool digest 919 functions have been deprecated. 920 921 *Paul Dale and David von Oheimb* 922 923 * Corrected the documentation of the return values from the `EVP_DigestSign*` 924 set of functions. The documentation mentioned negative values for some 925 errors, but this was never the case, so the mention of negative values 926 was removed. 927 928 Code that followed the documentation and thereby check with something 929 like `EVP_DigestSignInit(...) <= 0` will continue to work undisturbed. 930 931 *Richard Levitte* 932 933 * All of the low level cipher functions have been deprecated. 934 935 *Matt Caswell and Paul Dale* 936 937 * Removed include/openssl/opensslconf.h.in and replaced it with 938 include/openssl/configuration.h.in, which differs in not including 939 <openssl/macros.h>. A short header include/openssl/opensslconf.h 940 was added to include both. 941 942 This allows internal hacks where one might need to modify the set 943 of configured macros, for example this if deprecated symbols are 944 still supposed to be available internally: 945 946 #include <openssl/configuration.h> 947 948 #undef OPENSSL_NO_DEPRECATED 949 #define OPENSSL_SUPPRESS_DEPRECATED 950 951 #include <openssl/macros.h> 952 953 This should not be used by applications that use the exported 954 symbols, as that will lead to linking errors. 955 956 *Richard Levitte* 957 958 * Fixed an overflow bug in the x64_64 Montgomery squaring procedure 959 used in exponentiation with 512-bit moduli. No EC algorithms are 960 affected. Analysis suggests that attacks against 2-prime RSA1024, 961 3-prime RSA1536, and DSA1024 as a result of this defect would be very 962 difficult to perform and are not believed likely. Attacks against DH512 963 are considered just feasible. However, for an attack the target would 964 have to re-use the DH512 private key, which is not recommended anyway. 965 Also applications directly using the low-level API BN_mod_exp may be 966 affected if they use BN_FLG_CONSTTIME. 967 ([CVE-2019-1551]) 968 969 *Andy Polyakov* 970 971 * Most memory-debug features have been deprecated, and the functionality 972 replaced with no-ops. 973 974 *Rich Salz* 975 976 * Added documentation for the STACK API. 977 978 *Rich Salz* 979 980 * Introduced a new method type and API, OSSL_ENCODER, to represent 981 generic encoders. These do the same sort of job that PEM writers 982 and d2i functions do, but with support for methods supplied by 983 providers, and the possibility for providers to support other 984 formats as well. 985 986 *Richard Levitte* 987 988 * Introduced a new method type and API, OSSL_DECODER, to represent 989 generic decoders. These do the same sort of job that PEM readers 990 and i2d functions do, but with support for methods supplied by 991 providers, and the possibility for providers to support other 992 formats as well. 993 994 *Richard Levitte* 995 996 * Added a .pragma directive to the syntax of configuration files, to 997 allow varying behavior in a supported and predictable manner. 998 Currently added pragma: 999 1000 .pragma dollarid:on 1001 1002 This allows dollar signs to be a keyword character unless it's 1003 followed by a opening brace or parenthesis. This is useful for 1004 platforms where dollar signs are commonly used in names, such as 1005 volume names and system directory names on VMS. 1006 1007 *Richard Levitte* 1008 1009 * Added functionality to create an EVP_PKEY from user data. 1010 1011 *Richard Levitte* 1012 1013 * Change the interpretation of the '--api' configuration option to 1014 mean that this is a desired API compatibility level with no 1015 further meaning. The previous interpretation, that this would 1016 also mean to remove all deprecated symbols up to and including 1017 the given version, no requires that 'no-deprecated' is also used 1018 in the configuration. 1019 1020 When building applications, the desired API compatibility level 1021 can be set with the OPENSSL_API_COMPAT macro like before. For 1022 API compatibility version below 3.0, the old style numerical 1023 value is valid as before, such as -DOPENSSL_API_COMPAT=0x10100000L. 1024 For version 3.0 and on, the value is expected to be the decimal 1025 value calculated from the major and minor version like this: 1026 1027 MAJOR * 10000 + MINOR * 100 1028 1029 Examples: 1030 1031 -DOPENSSL_API_COMPAT=30000 For 3.0 1032 -DOPENSSL_API_COMPAT=30200 For 3.2 1033 1034 To hide declarations that are deprecated up to and including the 1035 given API compatibility level, -DOPENSSL_NO_DEPRECATED must be 1036 given when building the application as well. 1037 1038 *Richard Levitte* 1039 1040 * Added the X509_LOOKUP_METHOD called X509_LOOKUP_store, to allow 1041 access to certificate and CRL stores via URIs and OSSL_STORE 1042 loaders. 1043 1044 This adds the following functions: 1045 1046 - X509_LOOKUP_store() 1047 - X509_STORE_load_file() 1048 - X509_STORE_load_path() 1049 - X509_STORE_load_store() 1050 - SSL_add_store_cert_subjects_to_stack() 1051 - SSL_CTX_set_default_verify_store() 1052 - SSL_CTX_load_verify_file() 1053 - SSL_CTX_load_verify_dir() 1054 - SSL_CTX_load_verify_store() 1055 1056 *Richard Levitte* 1057 1058 * Added a new method to gather entropy on VMS, based on SYS$GET_ENTROPY. 1059 The presence of this system service is determined at run-time. 1060 1061 *Richard Levitte* 1062 1063 * Added functionality to create an EVP_PKEY context based on data 1064 for methods from providers. This takes an algorithm name and a 1065 property query string and simply stores them, with the intent 1066 that any operation that uses this context will use those strings 1067 to fetch the needed methods implicitly, thereby making the port 1068 of application written for pre-3.0 OpenSSL easier. 1069 1070 *Richard Levitte* 1071 1072 * The undocumented function NCONF_WIN32() has been deprecated; for 1073 conversion details see the HISTORY section of doc/man5/config.pod 1074 1075 *Rich Salz* 1076 1077 * Introduced the new functions EVP_DigestSignInit_ex() and 1078 EVP_DigestVerifyInit_ex(). The macros EVP_DigestSignUpdate() and 1079 EVP_DigestVerifyUpdate() have been converted to functions. See the man 1080 pages for further details. 1081 1082 *Matt Caswell* 1083 1084 * Over two thousand fixes were made to the documentation, including: 1085 adding missing command flags, better style conformance, documentation 1086 of internals, etc. 1087 1088 *Rich Salz, Richard Levitte* 1089 1090 * s390x assembly pack: add hardware-support for P-256, P-384, P-521, 1091 X25519, X448, Ed25519 and Ed448. 1092 1093 *Patrick Steuer* 1094 1095 * Print all values for a PKCS#12 attribute with 'openssl pkcs12', not just 1096 the first value. 1097 1098 *Jon Spillett* 1099 1100 * Deprecated the public definition of `ERR_STATE` as well as the function 1101 `ERR_get_state()`. This is done in preparation of making `ERR_STATE` an 1102 opaque type. 1103 1104 *Richard Levitte* 1105 1106 * Added ERR functionality to give callers access to the stored function 1107 names that have replaced the older function code based functions. 1108 1109 New functions are ERR_peek_error_func(), ERR_peek_last_error_func(), 1110 ERR_peek_error_data(), ERR_peek_last_error_data(), ERR_get_error_all(), 1111 ERR_peek_error_all() and ERR_peek_last_error_all(). 1112 1113 Deprecate ERR functions ERR_get_error_line(), ERR_get_error_line_data(), 1114 ERR_peek_error_line_data(), ERR_peek_last_error_line_data() and 1115 ERR_func_error_string(). 1116 1117 *Richard Levitte* 1118 1119 * Extended testing to be verbose for failing tests only. The make variables 1120 VERBOSE_FAILURE or VF can be used to enable this: 1121 1122 $ make VF=1 test # Unix 1123 $ mms /macro=(VF=1) test ! OpenVMS 1124 $ nmake VF=1 test # Windows 1125 1126 *Richard Levitte* 1127 1128 * Added the `-copy_extensions` option to the `x509` command for use with 1129 `-req` and `-x509toreq`. When given with the `copy` or `copyall` argument, 1130 all extensions in the request are copied to the certificate or vice versa. 1131 1132 *David von Oheimb*, *Kirill Stefanenkov <kirill_stefanenkov@rambler.ru>* 1133 1134 * Added the `-copy_extensions` option to the `req` command for use with 1135 `-x509`. When given with the `copy` or `copyall` argument, 1136 all extensions in the certification request are copied to the certificate. 1137 1138 *David von Oheimb* 1139 1140 * The `x509`, `req`, and `ca` commands now make sure that X.509v3 certificates 1141 they generate are by default RFC 5280 compliant in the following sense: 1142 There is a subjectKeyIdentifier extension with a hash value of the public key 1143 and for not self-signed certs there is an authorityKeyIdentifier extension 1144 with a keyIdentifier field or issuer information identifying the signing key. 1145 This is done unless some configuration overrides the new default behavior, 1146 such as `subjectKeyIdentifier = none` and `authorityKeyIdentifier = none`. 1147 1148 *David von Oheimb* 1149 1150 * Added several checks to `X509_verify_cert()` according to requirements in 1151 RFC 5280 in case `X509_V_FLAG_X509_STRICT` is set 1152 (which may be done by using the CLI option `-x509_strict`): 1153 * The basicConstraints of CA certificates must be marked critical. 1154 * CA certificates must explicitly include the keyUsage extension. 1155 * If a pathlenConstraint is given the key usage keyCertSign must be allowed. 1156 * The issuer name of any certificate must not be empty. 1157 * The subject name of CA certs, certs with keyUsage crlSign, 1158 and certs without subjectAlternativeName must not be empty. 1159 * If a subjectAlternativeName extension is given it must not be empty. 1160 * The signatureAlgorithm field and the cert signature must be consistent. 1161 * Any given authorityKeyIdentifier and any given subjectKeyIdentifier 1162 must not be marked critical. 1163 * The authorityKeyIdentifier must be given for X.509v3 certs 1164 unless they are self-signed. 1165 * The subjectKeyIdentifier must be given for all X.509v3 CA certs. 1166 1167 *David von Oheimb* 1168 1169 * Certificate verification using `X509_verify_cert()` meanwhile rejects EC keys 1170 with explicit curve parameters (specifiedCurve) as required by RFC 5480. 1171 1172 *Tomáš Mráz* 1173 1174 * For built-in EC curves, ensure an EC_GROUP built from the curve name is 1175 used even when parsing explicit parameters, when loading a encoded key 1176 or calling `EC_GROUP_new_from_ecpkparameters()`/ 1177 `EC_GROUP_new_from_ecparameters()`. 1178 This prevents bypass of security hardening and performance gains, 1179 especially for curves with specialized EC_METHODs. 1180 By default, if a key encoded with explicit parameters is loaded and later 1181 encoded, the output is still encoded with explicit parameters, even if 1182 internally a "named" EC_GROUP is used for computation. 1183 1184 *Nicola Tuveri* 1185 1186 * Compute ECC cofactors if not provided during EC_GROUP construction. Before 1187 this change, EC_GROUP_set_generator would accept order and/or cofactor as 1188 NULL. After this change, only the cofactor parameter can be NULL. It also 1189 does some minimal sanity checks on the passed order. 1190 ([CVE-2019-1547]) 1191 1192 *Billy Bob Brumley* 1193 1194 * Fixed a padding oracle in PKCS7_dataDecode and CMS_decrypt_set1_pkey. 1195 An attack is simple, if the first CMS_recipientInfo is valid but the 1196 second CMS_recipientInfo is chosen ciphertext. If the second 1197 recipientInfo decodes to PKCS #1 v1.5 form plaintext, the correct 1198 encryption key will be replaced by garbage, and the message cannot be 1199 decoded, but if the RSA decryption fails, the correct encryption key is 1200 used and the recipient will not notice the attack. 1201 As a work around for this potential attack the length of the decrypted 1202 key must be equal to the cipher default key length, in case the 1203 certifiate is not given and all recipientInfo are tried out. 1204 The old behaviour can be re-enabled in the CMS code by setting the 1205 CMS_DEBUG_DECRYPT flag. 1206 1207 *Bernd Edlinger* 1208 1209 * Early start up entropy quality from the DEVRANDOM seed source has been 1210 improved for older Linux systems. The RAND subsystem will wait for 1211 /dev/random to be producing output before seeding from /dev/urandom. 1212 The seeded state is stored for future library initialisations using 1213 a system global shared memory segment. The shared memory identifier 1214 can be configured by defining OPENSSL_RAND_SEED_DEVRANDOM_SHM_ID to 1215 the desired value. The default identifier is 114. 1216 1217 *Paul Dale* 1218 1219 * Revised BN_generate_prime_ex to not avoid factors 2..17863 in p-1 1220 when primes for RSA keys are computed. 1221 Since we previously always generated primes == 2 (mod 3) for RSA keys, 1222 the 2-prime and 3-prime RSA modules were easy to distinguish, since 1223 `N = p*q = 1 (mod 3)`, but `N = p*q*r = 2 (mod 3)`. Therefore fingerprinting 1224 2-prime vs. 3-prime RSA keys was possible by computing N mod 3. 1225 This avoids possible fingerprinting of newly generated RSA modules. 1226 1227 *Bernd Edlinger* 1228 1229 * Correct the extended master secret constant on EBCDIC systems. Without this 1230 fix TLS connections between an EBCDIC system and a non-EBCDIC system that 1231 negotiate EMS will fail. Unfortunately this also means that TLS connections 1232 between EBCDIC systems with this fix, and EBCDIC systems without this 1233 fix will fail if they negotiate EMS. 1234 1235 *Matt Caswell* 1236 1237 * Changed the library initialisation so that the config file is now loaded 1238 by default. This was already the case for libssl. It now occurs for both 1239 libcrypto and libssl. Use the OPENSSL_INIT_NO_LOAD_CONFIG option to 1240 `OPENSSL_init_crypto()` to suppress automatic loading of a config file. 1241 1242 *Matt Caswell* 1243 1244 * Introduced new error raising macros, `ERR_raise()` and `ERR_raise_data()`, 1245 where the former acts as a replacement for `ERR_put_error()`, and the 1246 latter replaces the combination `ERR_put_error()` + `ERR_add_error_data()`. 1247 `ERR_raise_data()` adds more flexibility by taking a format string and 1248 an arbitrary number of arguments following it, to be processed with 1249 `BIO_snprintf()`. 1250 1251 *Richard Levitte* 1252 1253 * Introduced a new function, `OSSL_PROVIDER_available()`, which can be used 1254 to check if a named provider is loaded and available. When called, it 1255 will also activate all fallback providers if such are still present. 1256 1257 *Richard Levitte* 1258 1259 * Enforce a minimum DH modulus size of 512 bits. 1260 1261 *Bernd Edlinger* 1262 1263 * Changed DH parameters to generate the order q subgroup instead of 2q. 1264 Previously generated DH parameters are still accepted by DH_check 1265 but DH_generate_key works around that by clearing bit 0 of the 1266 private key for those. This avoids leaking bit 0 of the private key. 1267 1268 *Bernd Edlinger* 1269 1270 * Significantly reduce secure memory usage by the randomness pools. 1271 1272 *Paul Dale* 1273 1274 * `{CRYPTO,OPENSSL}_mem_debug_{push,pop}` are now no-ops and have been 1275 deprecated. 1276 1277 *Rich Salz* 1278 1279 * A new type, EVP_KEYEXCH, has been introduced to represent key exchange 1280 algorithms. An implementation of a key exchange algorithm can be obtained 1281 by using the function EVP_KEYEXCH_fetch(). An EVP_KEYEXCH algorithm can be 1282 used in a call to EVP_PKEY_derive_init_ex() which works in a similar way to 1283 the older EVP_PKEY_derive_init() function. See the man pages for the new 1284 functions for further details. 1285 1286 *Matt Caswell* 1287 1288 * The EVP_PKEY_CTX_set_dh_pad() macro has now been converted to a function. 1289 1290 *Matt Caswell* 1291 1292 * Removed the function names from error messages and deprecated the 1293 xxx_F_xxx define's. 1294 1295 *Richard Levitte* 1296 1297 * Removed NextStep support and the macro OPENSSL_UNISTD 1298 1299 *Rich Salz* 1300 1301 * Removed DES_check_key. Also removed OPENSSL_IMPLEMENT_GLOBAL, 1302 OPENSSL_GLOBAL_REF, OPENSSL_DECLARE_GLOBAL. 1303 Also removed "export var as function" capability; we do not export 1304 variables, only functions. 1305 1306 *Rich Salz* 1307 1308 * RC5_32_set_key has been changed to return an int type, with 0 indicating 1309 an error and 1 indicating success. In previous versions of OpenSSL this 1310 was a void type. If a key was set longer than the maximum possible this 1311 would crash. 1312 1313 *Matt Caswell* 1314 1315 * Support SM2 signing and verification schemes with X509 certificate. 1316 1317 *Paul Yang* 1318 1319 * Use SHA256 as the default digest for TS query in the `ts` app. 1320 1321 *Tomáš Mráz* 1322 1323 * Change PBKDF2 to conform to SP800-132 instead of the older PKCS5 RFC2898. 1324 1325 *Shane Lontis* 1326 1327 * Default cipher lists/suites are now available via a function, the 1328 #defines are deprecated. 1329 1330 *Todd Short* 1331 1332 * Add target VC-WIN32-UWP, VC-WIN64A-UWP, VC-WIN32-ARM-UWP and 1333 VC-WIN64-ARM-UWP in Windows OneCore target for making building libraries 1334 for Windows Store apps easier. Also, the "no-uplink" option has been added. 1335 1336 *Kenji Mouri* 1337 1338 * Join the directories crypto/x509 and crypto/x509v3 1339 1340 *Richard Levitte* 1341 1342 * Added command 'openssl kdf' that uses the EVP_KDF API. 1343 1344 *Shane Lontis* 1345 1346 * Added command 'openssl mac' that uses the EVP_MAC API. 1347 1348 *Shane Lontis* 1349 1350 * Added OPENSSL_info() to get diverse built-in OpenSSL data, such 1351 as default directories. Also added the command 'openssl info' 1352 for scripting purposes. 1353 1354 *Richard Levitte* 1355 1356 * The functions AES_ige_encrypt() and AES_bi_ige_encrypt() have been 1357 deprecated. 1358 1359 *Matt Caswell* 1360 1361 * Add prediction resistance to the DRBG reseeding process. 1362 1363 *Paul Dale* 1364 1365 * Limit the number of blocks in a data unit for AES-XTS to 2^20 as 1366 mandated by IEEE Std 1619-2018. 1367 1368 *Paul Dale* 1369 1370 * Added newline escaping functionality to a filename when using openssl dgst. 1371 This output format is to replicate the output format found in the `*sum` 1372 checksum programs. This aims to preserve backward compatibility. 1373 1374 *Matt Eaton, Richard Levitte, and Paul Dale* 1375 1376 * Removed the heartbeat message in DTLS feature, as it has very 1377 little usage and doesn't seem to fulfill a valuable purpose. 1378 The configuration option is now deprecated. 1379 1380 *Richard Levitte* 1381 1382 * Changed the output of 'openssl {digestname} < file' to display the 1383 digest name in its output. 1384 1385 *Richard Levitte* 1386 1387 * Added a new generic trace API which provides support for enabling 1388 instrumentation through trace output. 1389 1390 *Richard Levitte & Matthias St. Pierre* 1391 1392 * Added build tests for C++. These are generated files that only do one 1393 thing, to include one public OpenSSL head file each. This tests that 1394 the public header files can be usefully included in a C++ application. 1395 1396 This test isn't enabled by default. It can be enabled with the option 1397 'enable-buildtest-c++'. 1398 1399 *Richard Levitte* 1400 1401 * Added KB KDF (EVP_KDF_KB) to EVP_KDF. 1402 1403 *Robbie Harwood* 1404 1405 * Added SSH KDF (EVP_KDF_SSHKDF) and KRB5 KDF (EVP_KDF_KRB5KDF) to EVP_KDF. 1406 1407 *Simo Sorce* 1408 1409 * Added Single Step KDF (EVP_KDF_SS), X963 KDF, and X942 KDF to EVP_KDF. 1410 1411 *Shane Lontis* 1412 1413 * Added KMAC to EVP_MAC. 1414 1415 *Shane Lontis* 1416 1417 * Added property based algorithm implementation selection framework to 1418 the core. 1419 1420 *Paul Dale* 1421 1422 * Added SCA hardening for modular field inversion in EC_GROUP through 1423 a new dedicated field_inv() pointer in EC_METHOD. 1424 This also addresses a leakage affecting conversions from projective 1425 to affine coordinates. 1426 1427 *Billy Bob Brumley, Nicola Tuveri* 1428 1429 * Added EVP_KDF, an EVP layer KDF API, to simplify adding KDF and PRF 1430 implementations. This includes an EVP_PKEY to EVP_KDF bridge for 1431 those algorithms that were already supported through the EVP_PKEY API 1432 (scrypt, TLS1 PRF and HKDF). The low-level KDF functions for PBKDF2 1433 and scrypt are now wrappers that call EVP_KDF. 1434 1435 *David Makepeace* 1436 1437 * Build devcrypto engine as a dynamic engine. 1438 1439 *Eneas U de Queiroz* 1440 1441 * Add keyed BLAKE2 to EVP_MAC. 1442 1443 *Antoine Salon* 1444 1445 * Fix a bug in the computation of the endpoint-pair shared secret used 1446 by DTLS over SCTP. This breaks interoperability with older versions 1447 of OpenSSL like OpenSSL 1.1.0 and OpenSSL 1.0.2. There is a runtime 1448 switch SSL_MODE_DTLS_SCTP_LABEL_LENGTH_BUG (off by default) enabling 1449 interoperability with such broken implementations. However, enabling 1450 this switch breaks interoperability with correct implementations. 1451 1452 * Fix a use after free bug in d2i_X509_PUBKEY when overwriting a 1453 re-used X509_PUBKEY object if the second PUBKEY is malformed. 1454 1455 *Bernd Edlinger* 1456 1457 * Move strictness check from EVP_PKEY_asn1_new() to EVP_PKEY_asn1_add0(). 1458 1459 *Richard Levitte* 1460 1461 * Changed the license to the Apache License v2.0. 1462 1463 *Richard Levitte* 1464 1465 * Switch to a new version scheme using three numbers MAJOR.MINOR.PATCH. 1466 1467 - Major releases (indicated by incrementing the MAJOR release number) 1468 may introduce incompatible API/ABI changes. 1469 - Minor releases (indicated by incrementing the MINOR release number) 1470 may introduce new features but retain API/ABI compatibility. 1471 - Patch releases (indicated by incrementing the PATCH number) 1472 are intended for bug fixes and other improvements of existing 1473 features only (like improving performance or adding documentation) 1474 and retain API/ABI compatibility. 1475 1476 *Richard Levitte* 1477 1478 * Add support for RFC5297 SIV mode (siv128), including AES-SIV. 1479 1480 *Todd Short* 1481 1482 * Remove the 'dist' target and add a tarball building script. The 1483 'dist' target has fallen out of use, and it shouldn't be 1484 necessary to configure just to create a source distribution. 1485 1486 *Richard Levitte* 1487 1488 * Recreate the OS390-Unix config target. It no longer relies on a 1489 special script like it did for OpenSSL pre-1.1.0. 1490 1491 *Richard Levitte* 1492 1493 * Instead of having the source directories listed in Configure, add 1494 a 'build.info' keyword SUBDIRS to indicate what sub-directories to 1495 look into. 1496 1497 *Richard Levitte* 1498 1499 * Add GMAC to EVP_MAC. 1500 1501 *Paul Dale* 1502 1503 * Ported the HMAC, CMAC and SipHash EVP_PKEY_METHODs to EVP_MAC. 1504 1505 *Richard Levitte* 1506 1507 * Added EVP_MAC, an EVP layer MAC API, to simplify adding MAC 1508 implementations. This includes a generic EVP_PKEY to EVP_MAC bridge, 1509 to facilitate the continued use of MACs through raw private keys in 1510 functionality such as `EVP_DigestSign*` and `EVP_DigestVerify*`. 1511 1512 *Richard Levitte* 1513 1514 * Deprecate ECDH_KDF_X9_62(). 1515 1516 *Antoine Salon* 1517 1518 * Added EVP_PKEY_ECDH_KDF_X9_63 and ecdh_KDF_X9_63() as replacements for 1519 the EVP_PKEY_ECDH_KDF_X9_62 KDF type and ECDH_KDF_X9_62(). The old names 1520 are retained for backwards compatibility. 1521 1522 *Antoine Salon* 1523 1524 * AES-XTS mode now enforces that its two keys are different to mitigate 1525 the attacked described in "Efficient Instantiations of Tweakable 1526 Blockciphers and Refinements to Modes OCB and PMAC" by Phillip Rogaway. 1527 Details of this attack can be obtained from: 1528 <http://web.cs.ucdavis.edu/%7Erogaway/papers/offsets.pdf> 1529 1530 *Paul Dale* 1531 1532 * Rename the object files, i.e. give them other names than in previous 1533 versions. Their names now include the name of the final product, as 1534 well as its type mnemonic (bin, lib, shlib). 1535 1536 *Richard Levitte* 1537 1538 * Added new option for 'openssl list', '-objects', which will display the 1539 list of built in objects, i.e. OIDs with names. 1540 1541 *Richard Levitte* 1542 1543 * Added the options `-crl_lastupdate` and `-crl_nextupdate` to `openssl ca`, 1544 allowing the `lastUpdate` and `nextUpdate` fields in the generated CRL to 1545 be set explicitly. 1546 1547 *Chris Novakovic* 1548 1549 * Added support for Linux Kernel TLS data-path. The Linux Kernel data-path 1550 improves application performance by removing data copies and providing 1551 applications with zero-copy system calls such as sendfile and splice. 1552 1553 *Boris Pismenny* 1554 1555 * The SSL option SSL_OP_CLEANSE_PLAINTEXT is introduced. 1556 1557 *Martin Elshuber* 1558 1559 * `PKCS12_parse` now maintains the order of the parsed certificates 1560 when outputting them via `*ca` (rather than reversing it). 1561 1562 *David von Oheimb* 1563 1564 * Deprecated pthread fork support methods. 1565 1566 *Randall S. Becker* 1567 1568 * Added support for FFDHE key exchange in TLS 1.3. 1569 1570 *Raja Ashok* 1571 1572 * Added a new concept for OpenSSL plugability: providers. This 1573 functionality is designed to replace the ENGINE API and ENGINE 1574 implementations, and to be much more dynamic, allowing provider 1575 authors to introduce new algorithms among other things, as long as 1576 there's an API that supports the algorithm type. 1577 1578 With this concept comes a new core API for interaction between 1579 libcrypto and provider implementations. Public libcrypto functions 1580 that want to use providers do so through this core API. 1581 1582 The main documentation for this core API is found in 1583 doc/man7/provider.pod, doc/man7/provider-base.pod, and they in turn 1584 refer to other manuals describing the API specific for supported 1585 algorithm types (also called operations). 1586 1587 *The OpenSSL team* 1588 1589OpenSSL 1.1.1 1590------------- 1591 1592### Changes between 1.1.1l and 1.1.1m [xx XXX xxxx] 1593 1594 * Avoid loading of a dynamic engine twice. 1595 1596 *Bernd Edlinger* 1597 1598 * Prioritise DANE TLSA issuer certs over peer certs 1599 1600 *Viktor Dukhovni* 1601 1602 * Fixed random API for MacOS prior to 10.12 1603 1604 These MacOS versions don't support the CommonCrypto APIs 1605 1606 *Lenny Primak* 1607 1608### Changes between 1.1.1k and 1.1.1l [24 Aug 2021] 1609 1610 * Fixed an SM2 Decryption Buffer Overflow. 1611 1612 In order to decrypt SM2 encrypted data an application is expected to 1613 call the API function EVP_PKEY_decrypt(). Typically an application will 1614 call this function twice. The first time, on entry, the "out" parameter 1615 can be NULL and, on exit, the "outlen" parameter is populated with the 1616 buffer size required to hold the decrypted plaintext. The application 1617 can then allocate a sufficiently sized buffer and call EVP_PKEY_decrypt() 1618 again, but this time passing a non-NULL value for the "out" parameter. 1619 1620 A bug in the implementation of the SM2 decryption code means that the 1621 calculation of the buffer size required to hold the plaintext returned 1622 by the first call to EVP_PKEY_decrypt() can be smaller than the actual 1623 size required by the second call. This can lead to a buffer overflow 1624 when EVP_PKEY_decrypt() is called by the application a second time with 1625 a buffer that is too small. 1626 1627 A malicious attacker who is able present SM2 content for decryption to 1628 an application could cause attacker chosen data to overflow the buffer 1629 by up to a maximum of 62 bytes altering the contents of other data held 1630 after the buffer, possibly changing application behaviour or causing 1631 the application to crash. The location of the buffer is application 1632 dependent but is typically heap allocated. 1633 ([CVE-2021-3711]) 1634 1635 *Matt Caswell* 1636 1637 * Fixed various read buffer overruns processing ASN.1 strings 1638 1639 ASN.1 strings are represented internally within OpenSSL as an ASN1_STRING 1640 structure which contains a buffer holding the string data and a field 1641 holding the buffer length. This contrasts with normal C strings which 1642 are repesented as a buffer for the string data which is terminated 1643 with a NUL (0) byte. 1644 1645 Although not a strict requirement, ASN.1 strings that are parsed using 1646 OpenSSL's own "d2i" functions (and other similar parsing functions) as 1647 well as any string whose value has been set with the ASN1_STRING_set() 1648 function will additionally NUL terminate the byte array in the 1649 ASN1_STRING structure. 1650 1651 However, it is possible for applications to directly construct valid 1652 ASN1_STRING structures which do not NUL terminate the byte array by 1653 directly setting the "data" and "length" fields in the ASN1_STRING 1654 array. This can also happen by using the ASN1_STRING_set0() function. 1655 1656 Numerous OpenSSL functions that print ASN.1 data have been found to 1657 assume that the ASN1_STRING byte array will be NUL terminated, even 1658 though this is not guaranteed for strings that have been directly 1659 constructed. Where an application requests an ASN.1 structure to be 1660 printed, and where that ASN.1 structure contains ASN1_STRINGs that have 1661 been directly constructed by the application without NUL terminating 1662 the "data" field, then a read buffer overrun can occur. 1663 1664 The same thing can also occur during name constraints processing 1665 of certificates (for example if a certificate has been directly 1666 constructed by the application instead of loading it via the OpenSSL 1667 parsing functions, and the certificate contains non NUL terminated 1668 ASN1_STRING structures). It can also occur in the X509_get1_email(), 1669 X509_REQ_get1_email() and X509_get1_ocsp() functions. 1670 1671 If a malicious actor can cause an application to directly construct an 1672 ASN1_STRING and then process it through one of the affected OpenSSL 1673 functions then this issue could be hit. This might result in a crash 1674 (causing a Denial of Service attack). It could also result in the 1675 disclosure of private memory contents (such as private keys, or 1676 sensitive plaintext). 1677 ([CVE-2021-3712]) 1678 1679 *Matt Caswell* 1680 1681### Changes between 1.1.1j and 1.1.1k [25 Mar 2021] 1682 1683 * Fixed a problem with verifying a certificate chain when using the 1684 X509_V_FLAG_X509_STRICT flag. This flag enables additional security checks of 1685 the certificates present in a certificate chain. It is not set by default. 1686 1687 Starting from OpenSSL version 1.1.1h a check to disallow certificates in 1688 the chain that have explicitly encoded elliptic curve parameters was added 1689 as an additional strict check. 1690 1691 An error in the implementation of this check meant that the result of a 1692 previous check to confirm that certificates in the chain are valid CA 1693 certificates was overwritten. This effectively bypasses the check 1694 that non-CA certificates must not be able to issue other certificates. 1695 1696 If a "purpose" has been configured then there is a subsequent opportunity 1697 for checks that the certificate is a valid CA. All of the named "purpose" 1698 values implemented in libcrypto perform this check. Therefore, where 1699 a purpose is set the certificate chain will still be rejected even when the 1700 strict flag has been used. A purpose is set by default in libssl client and 1701 server certificate verification routines, but it can be overridden or 1702 removed by an application. 1703 1704 In order to be affected, an application must explicitly set the 1705 X509_V_FLAG_X509_STRICT verification flag and either not set a purpose 1706 for the certificate verification or, in the case of TLS client or server 1707 applications, override the default purpose. 1708 ([CVE-2021-3450]) 1709 1710 *Tomáš Mráz* 1711 1712 * Fixed an issue where an OpenSSL TLS server may crash if sent a maliciously 1713 crafted renegotiation ClientHello message from a client. If a TLSv1.2 1714 renegotiation ClientHello omits the signature_algorithms extension (where it 1715 was present in the initial ClientHello), but includes a 1716 signature_algorithms_cert extension then a NULL pointer dereference will 1717 result, leading to a crash and a denial of service attack. 1718 1719 A server is only vulnerable if it has TLSv1.2 and renegotiation enabled 1720 (which is the default configuration). OpenSSL TLS clients are not impacted by 1721 this issue. 1722 ([CVE-2021-3449]) 1723 1724 *Peter Kästle and Samuel Sapalski* 1725 1726### Changes between 1.1.1i and 1.1.1j [16 Feb 2021] 1727 1728 * Fixed the X509_issuer_and_serial_hash() function. It attempts to 1729 create a unique hash value based on the issuer and serial number data 1730 contained within an X509 certificate. However it was failing to correctly 1731 handle any errors that may occur while parsing the issuer field (which might 1732 occur if the issuer field is maliciously constructed). This may subsequently 1733 result in a NULL pointer deref and a crash leading to a potential denial of 1734 service attack. 1735 ([CVE-2021-23841]) 1736 1737 *Matt Caswell* 1738 1739 * Fixed the RSA_padding_check_SSLv23() function and the RSA_SSLV23_PADDING 1740 padding mode to correctly check for rollback attacks. This is considered a 1741 bug in OpenSSL 1.1.1 because it does not support SSLv2. In 1.0.2 this is 1742 CVE-2021-23839. 1743 1744 *Matt Caswell* 1745 1746 Fixed the EVP_CipherUpdate, EVP_EncryptUpdate and EVP_DecryptUpdate 1747 functions. Previously they could overflow the output length argument in some 1748 cases where the input length is close to the maximum permissable length for 1749 an integer on the platform. In such cases the return value from the function 1750 call would be 1 (indicating success), but the output length value would be 1751 negative. This could cause applications to behave incorrectly or crash. 1752 ([CVE-2021-23840]) 1753 1754 *Matt Caswell* 1755 1756 * Fixed SRP_Calc_client_key so that it runs in constant time. The previous 1757 implementation called BN_mod_exp without setting BN_FLG_CONSTTIME. This 1758 could be exploited in a side channel attack to recover the password. Since 1759 the attack is local host only this is outside of the current OpenSSL 1760 threat model and therefore no CVE is assigned. 1761 1762 Thanks to Mohammed Sabt and Daniel De Almeida Braga for reporting this 1763 issue. 1764 1765 *Matt Caswell* 1766 1767### Changes between 1.1.1h and 1.1.1i [8 Dec 2020] 1768 1769 * Fixed NULL pointer deref in the GENERAL_NAME_cmp function 1770 This function could crash if both GENERAL_NAMEs contain an EDIPARTYNAME. 1771 If an attacker can control both items being compared then this could lead 1772 to a possible denial of service attack. OpenSSL itself uses the 1773 GENERAL_NAME_cmp function for two purposes: 1774 1) Comparing CRL distribution point names between an available CRL and a 1775 CRL distribution point embedded in an X509 certificate 1776 2) When verifying that a timestamp response token signer matches the 1777 timestamp authority name (exposed via the API functions 1778 TS_RESP_verify_response and TS_RESP_verify_token) 1779 ([CVE-2020-1971]) 1780 1781 *Matt Caswell* 1782 1783### Changes between 1.1.1g and 1.1.1h [22 Sep 2020] 1784 1785 * Certificates with explicit curve parameters are now disallowed in 1786 verification chains if the X509_V_FLAG_X509_STRICT flag is used. 1787 1788 *Tomáš Mráz* 1789 1790 * The 'MinProtocol' and 'MaxProtocol' configuration commands now silently 1791 ignore TLS protocol version bounds when configuring DTLS-based contexts, and 1792 conversely, silently ignore DTLS protocol version bounds when configuring 1793 TLS-based contexts. The commands can be repeated to set bounds of both 1794 types. The same applies with the corresponding "min_protocol" and 1795 "max_protocol" command-line switches, in case some application uses both TLS 1796 and DTLS. 1797 1798 SSL_CTX instances that are created for a fixed protocol version (e.g. 1799 TLSv1_server_method()) also silently ignore version bounds. Previously 1800 attempts to apply bounds to these protocol versions would result in an 1801 error. Now only the "version-flexible" SSL_CTX instances are subject to 1802 limits in configuration files in command-line options. 1803 1804 *Viktor Dukhovni* 1805 1806 * Handshake now fails if Extended Master Secret extension is dropped 1807 on renegotiation. 1808 1809 *Tomáš Mráz* 1810 1811 * The Oracle Developer Studio compiler will start reporting deprecated APIs 1812 1813### Changes between 1.1.1f and 1.1.1g [21 Apr 2020] 1814 1815 * Fixed segmentation fault in SSL_check_chain() 1816 Server or client applications that call the SSL_check_chain() function 1817 during or after a TLS 1.3 handshake may crash due to a NULL pointer 1818 dereference as a result of incorrect handling of the 1819 "signature_algorithms_cert" TLS extension. The crash occurs if an invalid 1820 or unrecognised signature algorithm is received from the peer. This could 1821 be exploited by a malicious peer in a Denial of Service attack. 1822 ([CVE-2020-1967]) 1823 1824 *Benjamin Kaduk* 1825 1826 * Added AES consttime code for no-asm configurations 1827 an optional constant time support for AES was added 1828 when building openssl for no-asm. 1829 Enable with: ./config no-asm -DOPENSSL_AES_CONST_TIME 1830 Disable with: ./config no-asm -DOPENSSL_NO_AES_CONST_TIME 1831 At this time this feature is by default disabled. 1832 It will be enabled by default in 3.0. 1833 1834 *Bernd Edlinger* 1835 1836### Changes between 1.1.1e and 1.1.1f [31 Mar 2020] 1837 1838 * Revert the change of EOF detection while reading in libssl to avoid 1839 regressions in applications depending on the current way of reporting 1840 the EOF. As the existing method is not fully accurate the change to 1841 reporting the EOF via SSL_ERROR_SSL is kept on the current development 1842 branch and will be present in the 3.0 release. 1843 1844 *Tomáš Mráz* 1845 1846 * Revised BN_generate_prime_ex to not avoid factors 3..17863 in p-1 1847 when primes for RSA keys are computed. 1848 Since we previously always generated primes == 2 (mod 3) for RSA keys, 1849 the 2-prime and 3-prime RSA modules were easy to distinguish, since 1850 N = p*q = 1 (mod 3), but N = p*q*r = 2 (mod 3). Therefore fingerprinting 1851 2-prime vs. 3-prime RSA keys was possible by computing N mod 3. 1852 This avoids possible fingerprinting of newly generated RSA modules. 1853 1854 *Bernd Edlinger* 1855 1856### Changes between 1.1.1d and 1.1.1e [17 Mar 2020] 1857 1858 * Properly detect EOF while reading in libssl. Previously if we hit an EOF 1859 while reading in libssl then we would report an error back to the 1860 application (SSL_ERROR_SYSCALL) but errno would be 0. We now add 1861 an error to the stack (which means we instead return SSL_ERROR_SSL) and 1862 therefore give a hint as to what went wrong. 1863 1864 *Matt Caswell* 1865 1866 * Check that ed25519 and ed448 are allowed by the security level. Previously 1867 signature algorithms not using an MD were not being checked that they were 1868 allowed by the security level. 1869 1870 *Kurt Roeckx* 1871 1872 * Fixed SSL_get_servername() behaviour. The behaviour of SSL_get_servername() 1873 was not quite right. The behaviour was not consistent between resumption 1874 and normal handshakes, and also not quite consistent with historical 1875 behaviour. The behaviour in various scenarios has been clarified and 1876 it has been updated to make it match historical behaviour as closely as 1877 possible. 1878 1879 *Matt Caswell* 1880 1881 * *[VMS only]* The header files that the VMS compilers include automatically, 1882 `__DECC_INCLUDE_PROLOGUE.H` and `__DECC_INCLUDE_EPILOGUE.H`, use pragmas 1883 that the C++ compiler doesn't understand. This is a shortcoming in the 1884 compiler, but can be worked around with `__cplusplus` guards. 1885 1886 C++ applications that use OpenSSL libraries must be compiled using the 1887 qualifier `/NAMES=(AS_IS,SHORTENED)` to be able to use all the OpenSSL 1888 functions. Otherwise, only functions with symbols of less than 31 1889 characters can be used, as the linker will not be able to successfully 1890 resolve symbols with longer names. 1891 1892 *Richard Levitte* 1893 1894 * Added a new method to gather entropy on VMS, based on SYS$GET_ENTROPY. 1895 The presence of this system service is determined at run-time. 1896 1897 *Richard Levitte* 1898 1899 * Print all values for a PKCS#12 attribute with 'openssl pkcs12', not just 1900 the first value. 1901 1902 *Jon Spillett* 1903 1904### Changes between 1.1.1c and 1.1.1d [10 Sep 2019] 1905 1906 * Fixed a fork protection issue. OpenSSL 1.1.1 introduced a rewritten random 1907 number generator (RNG). This was intended to include protection in the 1908 event of a fork() system call in order to ensure that the parent and child 1909 processes did not share the same RNG state. However this protection was not 1910 being used in the default case. 1911 1912 A partial mitigation for this issue is that the output from a high 1913 precision timer is mixed into the RNG state so the likelihood of a parent 1914 and child process sharing state is significantly reduced. 1915 1916 If an application already calls OPENSSL_init_crypto() explicitly using 1917 OPENSSL_INIT_ATFORK then this problem does not occur at all. 1918 ([CVE-2019-1549]) 1919 1920 *Matthias St. Pierre* 1921 1922 * For built-in EC curves, ensure an EC_GROUP built from the curve name is 1923 used even when parsing explicit parameters, when loading a encoded key 1924 or calling `EC_GROUP_new_from_ecpkparameters()`/ 1925 `EC_GROUP_new_from_ecparameters()`. 1926 This prevents bypass of security hardening and performance gains, 1927 especially for curves with specialized EC_METHODs. 1928 By default, if a key encoded with explicit parameters is loaded and later 1929 encoded, the output is still encoded with explicit parameters, even if 1930 internally a "named" EC_GROUP is used for computation. 1931 1932 *Nicola Tuveri* 1933 1934 * Compute ECC cofactors if not provided during EC_GROUP construction. Before 1935 this change, EC_GROUP_set_generator would accept order and/or cofactor as 1936 NULL. After this change, only the cofactor parameter can be NULL. It also 1937 does some minimal sanity checks on the passed order. 1938 ([CVE-2019-1547]) 1939 1940 *Billy Bob Brumley* 1941 1942 * Fixed a padding oracle in PKCS7_dataDecode and CMS_decrypt_set1_pkey. 1943 An attack is simple, if the first CMS_recipientInfo is valid but the 1944 second CMS_recipientInfo is chosen ciphertext. If the second 1945 recipientInfo decodes to PKCS #1 v1.5 form plaintext, the correct 1946 encryption key will be replaced by garbage, and the message cannot be 1947 decoded, but if the RSA decryption fails, the correct encryption key is 1948 used and the recipient will not notice the attack. 1949 As a work around for this potential attack the length of the decrypted 1950 key must be equal to the cipher default key length, in case the 1951 certifiate is not given and all recipientInfo are tried out. 1952 The old behaviour can be re-enabled in the CMS code by setting the 1953 CMS_DEBUG_DECRYPT flag. 1954 ([CVE-2019-1563]) 1955 1956 *Bernd Edlinger* 1957 1958 * Early start up entropy quality from the DEVRANDOM seed source has been 1959 improved for older Linux systems. The RAND subsystem will wait for 1960 /dev/random to be producing output before seeding from /dev/urandom. 1961 The seeded state is stored for future library initialisations using 1962 a system global shared memory segment. The shared memory identifier 1963 can be configured by defining OPENSSL_RAND_SEED_DEVRANDOM_SHM_ID to 1964 the desired value. The default identifier is 114. 1965 1966 *Paul Dale* 1967 1968 * Correct the extended master secret constant on EBCDIC systems. Without this 1969 fix TLS connections between an EBCDIC system and a non-EBCDIC system that 1970 negotiate EMS will fail. Unfortunately this also means that TLS connections 1971 between EBCDIC systems with this fix, and EBCDIC systems without this 1972 fix will fail if they negotiate EMS. 1973 1974 *Matt Caswell* 1975 1976 * Use Windows installation paths in the mingw builds 1977 1978 Mingw isn't a POSIX environment per se, which means that Windows 1979 paths should be used for installation. 1980 ([CVE-2019-1552]) 1981 1982 *Richard Levitte* 1983 1984 * Changed DH_check to accept parameters with order q and 2q subgroups. 1985 With order 2q subgroups the bit 0 of the private key is not secret 1986 but DH_generate_key works around that by clearing bit 0 of the 1987 private key for those. This avoids leaking bit 0 of the private key. 1988 1989 *Bernd Edlinger* 1990 1991 * Significantly reduce secure memory usage by the randomness pools. 1992 1993 *Paul Dale* 1994 1995 * Revert the DEVRANDOM_WAIT feature for Linux systems 1996 1997 The DEVRANDOM_WAIT feature added a select() call to wait for the 1998 /dev/random device to become readable before reading from the 1999 /dev/urandom device. 2000 2001 It turned out that this change had negative side effects on 2002 performance which were not acceptable. After some discussion it 2003 was decided to revert this feature and leave it up to the OS 2004 resp. the platform maintainer to ensure a proper initialization 2005 during early boot time. 2006 2007 *Matthias St. Pierre* 2008 2009### Changes between 1.1.1b and 1.1.1c [28 May 2019] 2010 2011 * Add build tests for C++. These are generated files that only do one 2012 thing, to include one public OpenSSL head file each. This tests that 2013 the public header files can be usefully included in a C++ application. 2014 2015 This test isn't enabled by default. It can be enabled with the option 2016 'enable-buildtest-c++'. 2017 2018 *Richard Levitte* 2019 2020 * Enable SHA3 pre-hashing for ECDSA and DSA. 2021 2022 *Patrick Steuer* 2023 2024 * Change the default RSA, DSA and DH size to 2048 bit instead of 1024. 2025 This changes the size when using the `genpkey` command when no size is given. 2026 It fixes an omission in earlier changes that changed all RSA, DSA and DH 2027 generation commands to use 2048 bits by default. 2028 2029 *Kurt Roeckx* 2030 2031 * Reorganize the manual pages to consistently have RETURN VALUES, 2032 EXAMPLES, SEE ALSO and HISTORY come in that order, and adjust 2033 util/fix-doc-nits accordingly. 2034 2035 *Paul Yang, Joshua Lock* 2036 2037 * Add the missing accessor EVP_PKEY_get0_engine() 2038 2039 *Matt Caswell* 2040 2041 * Have commands like `s_client` and `s_server` output the signature scheme 2042 along with other cipher suite parameters when debugging. 2043 2044 *Lorinczy Zsigmond* 2045 2046 * Make OPENSSL_config() error agnostic again. 2047 2048 *Richard Levitte* 2049 2050 * Do the error handling in RSA decryption constant time. 2051 2052 *Bernd Edlinger* 2053 2054 * Prevent over long nonces in ChaCha20-Poly1305. 2055 2056 ChaCha20-Poly1305 is an AEAD cipher, and requires a unique nonce input 2057 for every encryption operation. RFC 7539 specifies that the nonce value 2058 (IV) should be 96 bits (12 bytes). OpenSSL allows a variable nonce length 2059 and front pads the nonce with 0 bytes if it is less than 12 2060 bytes. However it also incorrectly allows a nonce to be set of up to 16 2061 bytes. In this case only the last 12 bytes are significant and any 2062 additional leading bytes are ignored. 2063 2064 It is a requirement of using this cipher that nonce values are 2065 unique. Messages encrypted using a reused nonce value are susceptible to 2066 serious confidentiality and integrity attacks. If an application changes 2067 the default nonce length to be longer than 12 bytes and then makes a 2068 change to the leading bytes of the nonce expecting the new value to be a 2069 new unique nonce then such an application could inadvertently encrypt 2070 messages with a reused nonce. 2071 2072 Additionally the ignored bytes in a long nonce are not covered by the 2073 integrity guarantee of this cipher. Any application that relies on the 2074 integrity of these ignored leading bytes of a long nonce may be further 2075 affected. Any OpenSSL internal use of this cipher, including in SSL/TLS, 2076 is safe because no such use sets such a long nonce value. However user 2077 applications that use this cipher directly and set a non-default nonce 2078 length to be longer than 12 bytes may be vulnerable. 2079 2080 This issue was reported to OpenSSL on 16th of March 2019 by Joran Dirk 2081 Greef of Ronomon. 2082 ([CVE-2019-1543]) 2083 2084 *Matt Caswell* 2085 2086 * Add DEVRANDOM_WAIT feature for Linux systems 2087 2088 On older Linux systems where the getrandom() system call is not available, 2089 OpenSSL normally uses the /dev/urandom device for seeding its CSPRNG. 2090 Contrary to getrandom(), the /dev/urandom device will not block during 2091 early boot when the kernel CSPRNG has not been seeded yet. 2092 2093 To mitigate this known weakness, use select() to wait for /dev/random to 2094 become readable before reading from /dev/urandom. 2095 2096 * Ensure that SM2 only uses SM3 as digest algorithm 2097 2098 *Paul Yang* 2099 2100### Changes between 1.1.1a and 1.1.1b [26 Feb 2019] 2101 2102 * Change the info callback signals for the start and end of a post-handshake 2103 message exchange in TLSv1.3. In 1.1.1/1.1.1a we used SSL_CB_HANDSHAKE_START 2104 and SSL_CB_HANDSHAKE_DONE. Experience has shown that many applications get 2105 confused by this and assume that a TLSv1.2 renegotiation has started. This 2106 can break KeyUpdate handling. Instead we no longer signal the start and end 2107 of a post handshake message exchange (although the messages themselves are 2108 still signalled). This could break some applications that were expecting 2109 the old signals. However without this KeyUpdate is not usable for many 2110 applications. 2111 2112 *Matt Caswell* 2113 2114### Changes between 1.1.1 and 1.1.1a [20 Nov 2018] 2115 2116 * Timing vulnerability in DSA signature generation 2117 2118 The OpenSSL DSA signature algorithm has been shown to be vulnerable to a 2119 timing side channel attack. An attacker could use variations in the signing 2120 algorithm to recover the private key. 2121 2122 This issue was reported to OpenSSL on 16th October 2018 by Samuel Weiser. 2123 ([CVE-2018-0734]) 2124 2125 *Paul Dale* 2126 2127 * Timing vulnerability in ECDSA signature generation 2128 2129 The OpenSSL ECDSA signature algorithm has been shown to be vulnerable to a 2130 timing side channel attack. An attacker could use variations in the signing 2131 algorithm to recover the private key. 2132 2133 This issue was reported to OpenSSL on 25th October 2018 by Samuel Weiser. 2134 ([CVE-2018-0735]) 2135 2136 *Paul Dale* 2137 2138 * Fixed the issue that RAND_add()/RAND_seed() silently discards random input 2139 if its length exceeds 4096 bytes. The limit has been raised to a buffer size 2140 of two gigabytes and the error handling improved. 2141 2142 This issue was reported to OpenSSL by Dr. Falko Strenzke. It has been 2143 categorized as a normal bug, not a security issue, because the DRBG reseeds 2144 automatically and is fully functional even without additional randomness 2145 provided by the application. 2146 2147### Changes between 1.1.0i and 1.1.1 [11 Sep 2018] 2148 2149 * Add a new ClientHello callback. Provides a callback interface that gives 2150 the application the ability to adjust the nascent SSL object at the 2151 earliest stage of ClientHello processing, immediately after extensions have 2152 been collected but before they have been processed. In particular, this 2153 callback can adjust the supported TLS versions in response to the contents 2154 of the ClientHello 2155 2156 *Benjamin Kaduk* 2157 2158 * Add SM2 base algorithm support. 2159 2160 *Jack Lloyd* 2161 2162 * s390x assembly pack: add (improved) hardware-support for the following 2163 cryptographic primitives: sha3, shake, aes-gcm, aes-ccm, aes-ctr, aes-ofb, 2164 aes-cfb/cfb8, aes-ecb. 2165 2166 *Patrick Steuer* 2167 2168 * Make EVP_PKEY_asn1_new() a bit stricter about its input. A NULL pem_str 2169 parameter is no longer accepted, as it leads to a corrupt table. NULL 2170 pem_str is reserved for alias entries only. 2171 2172 *Richard Levitte* 2173 2174 * Use the new ec_scalar_mul_ladder scaffold to implement a specialized ladder 2175 step for prime curves. The new implementation is based on formulae from 2176 differential addition-and-doubling in homogeneous projective coordinates 2177 from Izu-Takagi "A fast parallel elliptic curve multiplication resistant 2178 against side channel attacks" and Brier-Joye "Weierstrass Elliptic Curves 2179 and Side-Channel Attacks" Eq. (8) for y-coordinate recovery, modified 2180 to work in projective coordinates. 2181 2182 *Billy Bob Brumley, Nicola Tuveri* 2183 2184 * Change generating and checking of primes so that the error rate of not 2185 being prime depends on the intended use based on the size of the input. 2186 For larger primes this will result in more rounds of Miller-Rabin. 2187 The maximal error rate for primes with more than 1080 bits is lowered 2188 to 2^-128. 2189 2190 *Kurt Roeckx, Annie Yousar* 2191 2192 * Increase the number of Miller-Rabin rounds for DSA key generating to 64. 2193 2194 *Kurt Roeckx* 2195 2196 * The 'tsget' script is renamed to 'tsget.pl', to avoid confusion when 2197 moving between systems, and to avoid confusion when a Windows build is 2198 done with mingw vs with MSVC. For POSIX installs, there's still a 2199 symlink or copy named 'tsget' to avoid that confusion as well. 2200 2201 *Richard Levitte* 2202 2203 * Revert blinding in ECDSA sign and instead make problematic addition 2204 length-invariant. Switch even to fixed-length Montgomery multiplication. 2205 2206 *Andy Polyakov* 2207 2208 * Use the new ec_scalar_mul_ladder scaffold to implement a specialized ladder 2209 step for binary curves. The new implementation is based on formulae from 2210 differential addition-and-doubling in mixed Lopez-Dahab projective 2211 coordinates, modified to independently blind the operands. 2212 2213 *Billy Bob Brumley, Sohaib ul Hassan, Nicola Tuveri* 2214 2215 * Add a scaffold to optionally enhance the Montgomery ladder implementation 2216 for `ec_scalar_mul_ladder` (formerly `ec_mul_consttime`) allowing 2217 EC_METHODs to implement their own specialized "ladder step", to take 2218 advantage of more favorable coordinate systems or more efficient 2219 differential addition-and-doubling algorithms. 2220 2221 *Billy Bob Brumley, Sohaib ul Hassan, Nicola Tuveri* 2222 2223 * Modified the random device based seed sources to keep the relevant 2224 file descriptors open rather than reopening them on each access. 2225 This allows such sources to operate in a chroot() jail without 2226 the associated device nodes being available. This behaviour can be 2227 controlled using RAND_keep_random_devices_open(). 2228 2229 *Paul Dale* 2230 2231 * Numerous side-channel attack mitigations have been applied. This may have 2232 performance impacts for some algorithms for the benefit of improved 2233 security. Specific changes are noted in this change log by their respective 2234 authors. 2235 2236 *Matt Caswell* 2237 2238 * AIX shared library support overhaul. Switch to AIX "natural" way of 2239 handling shared libraries, which means collecting shared objects of 2240 different versions and bitnesses in one common archive. This allows to 2241 mitigate conflict between 1.0 and 1.1 side-by-side installations. It 2242 doesn't affect the way 3rd party applications are linked, only how 2243 multi-version installation is managed. 2244 2245 *Andy Polyakov* 2246 2247 * Make ec_group_do_inverse_ord() more robust and available to other 2248 EC cryptosystems, so that irrespective of BN_FLG_CONSTTIME, SCA 2249 mitigations are applied to the fallback BN_mod_inverse(). 2250 When using this function rather than BN_mod_inverse() directly, new 2251 EC cryptosystem implementations are then safer-by-default. 2252 2253 *Billy Bob Brumley* 2254 2255 * Add coordinate blinding for EC_POINT and implement projective 2256 coordinate blinding for generic prime curves as a countermeasure to 2257 chosen point SCA attacks. 2258 2259 *Sohaib ul Hassan, Nicola Tuveri, Billy Bob Brumley* 2260 2261 * Add blinding to ECDSA and DSA signatures to protect against side channel 2262 attacks discovered by Keegan Ryan (NCC Group). 2263 2264 *Matt Caswell* 2265 2266 * Enforce checking in the `pkeyutl` command to ensure that the input 2267 length does not exceed the maximum supported digest length when performing 2268 a sign, verify or verifyrecover operation. 2269 2270 *Matt Caswell* 2271 2272 * SSL_MODE_AUTO_RETRY is enabled by default. Applications that use blocking 2273 I/O in combination with something like select() or poll() will hang. This 2274 can be turned off again using SSL_CTX_clear_mode(). 2275 Many applications do not properly handle non-application data records, and 2276 TLS 1.3 sends more of such records. Setting SSL_MODE_AUTO_RETRY works 2277 around the problems in those applications, but can also break some. 2278 It's recommended to read the manpages about SSL_read(), SSL_write(), 2279 SSL_get_error(), SSL_shutdown(), SSL_CTX_set_mode() and 2280 SSL_CTX_set_read_ahead() again. 2281 2282 *Kurt Roeckx* 2283 2284 * When unlocking a pass phrase protected PEM file or PKCS#8 container, we 2285 now allow empty (zero character) pass phrases. 2286 2287 *Richard Levitte* 2288 2289 * Apply blinding to binary field modular inversion and remove patent 2290 pending (OPENSSL_SUN_GF2M_DIV) BN_GF2m_mod_div implementation. 2291 2292 *Billy Bob Brumley* 2293 2294 * Deprecate ec2_mult.c and unify scalar multiplication code paths for 2295 binary and prime elliptic curves. 2296 2297 *Billy Bob Brumley* 2298 2299 * Remove ECDSA nonce padding: EC_POINT_mul is now responsible for 2300 constant time fixed point multiplication. 2301 2302 *Billy Bob Brumley* 2303 2304 * Revise elliptic curve scalar multiplication with timing attack 2305 defenses: ec_wNAF_mul redirects to a constant time implementation 2306 when computing fixed point and variable point multiplication (which 2307 in OpenSSL are mostly used with secret scalars in keygen, sign, 2308 ECDH derive operations). 2309 *Billy Bob Brumley, Nicola Tuveri, Cesar Pereida García, 2310 Sohaib ul Hassan* 2311 2312 * Updated CONTRIBUTING 2313 2314 *Rich Salz* 2315 2316 * Updated DRBG / RAND to request nonce and additional low entropy 2317 randomness from the system. 2318 2319 *Matthias St. Pierre* 2320 2321 * Updated 'openssl rehash' to use OpenSSL consistent default. 2322 2323 *Richard Levitte* 2324 2325 * Moved the load of the ssl_conf module to libcrypto, which helps 2326 loading engines that libssl uses before libssl is initialised. 2327 2328 *Matt Caswell* 2329 2330 * Added EVP_PKEY_sign() and EVP_PKEY_verify() for EdDSA 2331 2332 *Matt Caswell* 2333 2334 * Fixed X509_NAME_ENTRY_set to get multi-valued RDNs right in all cases. 2335 2336 *Ingo Schwarze, Rich Salz* 2337 2338 * Added output of accepting IP address and port for 'openssl s_server' 2339 2340 *Richard Levitte* 2341 2342 * Added a new API for TLSv1.3 ciphersuites: 2343 SSL_CTX_set_ciphersuites() 2344 SSL_set_ciphersuites() 2345 2346 *Matt Caswell* 2347 2348 * Memory allocation failures consistently add an error to the error 2349 stack. 2350 2351 *Rich Salz* 2352 2353 * Don't use OPENSSL_ENGINES and OPENSSL_CONF environment values 2354 in libcrypto when run as setuid/setgid. 2355 2356 *Bernd Edlinger* 2357 2358 * Load any config file by default when libssl is used. 2359 2360 *Matt Caswell* 2361 2362 * Added new public header file <openssl/rand_drbg.h> and documentation 2363 for the RAND_DRBG API. See manual page RAND_DRBG(7) for an overview. 2364 2365 *Matthias St. Pierre* 2366 2367 * QNX support removed (cannot find contributors to get their approval 2368 for the license change). 2369 2370 *Rich Salz* 2371 2372 * TLSv1.3 replay protection for early data has been implemented. See the 2373 SSL_read_early_data() man page for further details. 2374 2375 *Matt Caswell* 2376 2377 * Separated TLSv1.3 ciphersuite configuration out from TLSv1.2 ciphersuite 2378 configuration. TLSv1.3 ciphersuites are not compatible with TLSv1.2 and 2379 below. Similarly TLSv1.2 ciphersuites are not compatible with TLSv1.3. 2380 In order to avoid issues where legacy TLSv1.2 ciphersuite configuration 2381 would otherwise inadvertently disable all TLSv1.3 ciphersuites the 2382 configuration has been separated out. See the ciphers man page or the 2383 SSL_CTX_set_ciphersuites() man page for more information. 2384 2385 *Matt Caswell* 2386 2387 * On POSIX (BSD, Linux, ...) systems the ocsp(1) command running 2388 in responder mode now supports the new "-multi" option, which 2389 spawns the specified number of child processes to handle OCSP 2390 requests. The "-timeout" option now also limits the OCSP 2391 responder's patience to wait to receive the full client request 2392 on a newly accepted connection. Child processes are respawned 2393 as needed, and the CA index file is automatically reloaded 2394 when changed. This makes it possible to run the "ocsp" responder 2395 as a long-running service, making the OpenSSL CA somewhat more 2396 feature-complete. In this mode, most diagnostic messages logged 2397 after entering the event loop are logged via syslog(3) rather than 2398 written to stderr. 2399 2400 *Viktor Dukhovni* 2401 2402 * Added support for X448 and Ed448. Heavily based on original work by 2403 Mike Hamburg. 2404 2405 *Matt Caswell* 2406 2407 * Extend OSSL_STORE with capabilities to search and to narrow the set of 2408 objects loaded. This adds the functions OSSL_STORE_expect() and 2409 OSSL_STORE_find() as well as needed tools to construct searches and 2410 get the search data out of them. 2411 2412 *Richard Levitte* 2413 2414 * Support for TLSv1.3 added. Note that users upgrading from an earlier 2415 version of OpenSSL should review their configuration settings to ensure 2416 that they are still appropriate for TLSv1.3. For further information see: 2417 <https://wiki.openssl.org/index.php/TLS1.3> 2418 2419 *Matt Caswell* 2420 2421 * Grand redesign of the OpenSSL random generator 2422 2423 The default RAND method now utilizes an AES-CTR DRBG according to 2424 NIST standard SP 800-90Ar1. The new random generator is essentially 2425 a port of the default random generator from the OpenSSL FIPS 2.0 2426 object module. It is a hybrid deterministic random bit generator 2427 using an AES-CTR bit stream and which seeds and reseeds itself 2428 automatically using trusted system entropy sources. 2429 2430 Some of its new features are: 2431 - Support for multiple DRBG instances with seed chaining. 2432 - The default RAND method makes use of a DRBG. 2433 - There is a public and private DRBG instance. 2434 - The DRBG instances are fork-safe. 2435 - Keep all global DRBG instances on the secure heap if it is enabled. 2436 - The public and private DRBG instance are per thread for lock free 2437 operation 2438 2439 *Paul Dale, Benjamin Kaduk, Kurt Roeckx, Rich Salz, Matthias St. Pierre* 2440 2441 * Changed Configure so it only says what it does and doesn't dump 2442 so much data. Instead, ./configdata.pm should be used as a script 2443 to display all sorts of configuration data. 2444 2445 *Richard Levitte* 2446 2447 * Added processing of "make variables" to Configure. 2448 2449 *Richard Levitte* 2450 2451 * Added SHA512/224 and SHA512/256 algorithm support. 2452 2453 *Paul Dale* 2454 2455 * The last traces of Netware support, first removed in 1.1.0, have 2456 now been removed. 2457 2458 *Rich Salz* 2459 2460 * Get rid of Makefile.shared, and in the process, make the processing 2461 of certain files (rc.obj, or the .def/.map/.opt files produced from 2462 the ordinal files) more visible and hopefully easier to trace and 2463 debug (or make silent). 2464 2465 *Richard Levitte* 2466 2467 * Make it possible to have environment variable assignments as 2468 arguments to config / Configure. 2469 2470 *Richard Levitte* 2471 2472 * Add multi-prime RSA (RFC 8017) support. 2473 2474 *Paul Yang* 2475 2476 * Add SM3 implemented according to GB/T 32905-2016 2477 *Jack Lloyd <jack.lloyd@ribose.com>,* 2478 *Ronald Tse <ronald.tse@ribose.com>,* 2479 *Erick Borsboom <erick.borsboom@ribose.com>* 2480 2481 * Add 'Maximum Fragment Length' TLS extension negotiation and support 2482 as documented in RFC6066. 2483 Based on a patch from Tomasz Moń 2484 2485 *Filipe Raimundo da Silva* 2486 2487 * Add SM4 implemented according to GB/T 32907-2016. 2488 *Jack Lloyd <jack.lloyd@ribose.com>,* 2489 *Ronald Tse <ronald.tse@ribose.com>,* 2490 *Erick Borsboom <erick.borsboom@ribose.com>* 2491 2492 * Reimplement -newreq-nodes and ERR_error_string_n; the 2493 original author does not agree with the license change. 2494 2495 *Rich Salz* 2496 2497 * Add ARIA AEAD TLS support. 2498 2499 *Jon Spillett* 2500 2501 * Some macro definitions to support VS6 have been removed. Visual 2502 Studio 6 has not worked since 1.1.0 2503 2504 *Rich Salz* 2505 2506 * Add ERR_clear_last_mark(), to allow callers to clear the last mark 2507 without clearing the errors. 2508 2509 *Richard Levitte* 2510 2511 * Add "atfork" functions. If building on a system that without 2512 pthreads, see doc/man3/OPENSSL_fork_prepare.pod for application 2513 requirements. The RAND facility now uses/requires this. 2514 2515 *Rich Salz* 2516 2517 * Add SHA3. 2518 2519 *Andy Polyakov* 2520 2521 * The UI API becomes a permanent and integral part of libcrypto, i.e. 2522 not possible to disable entirely. However, it's still possible to 2523 disable the console reading UI method, UI_OpenSSL() (use UI_null() 2524 as a fallback). 2525 2526 To disable, configure with 'no-ui-console'. 'no-ui' is still 2527 possible to use as an alias. Check at compile time with the 2528 macro OPENSSL_NO_UI_CONSOLE. The macro OPENSSL_NO_UI is still 2529 possible to check and is an alias for OPENSSL_NO_UI_CONSOLE. 2530 2531 *Richard Levitte* 2532 2533 * Add a STORE module, which implements a uniform and URI based reader of 2534 stores that can contain keys, certificates, CRLs and numerous other 2535 objects. The main API is loosely based on a few stdio functions, 2536 and includes OSSL_STORE_open, OSSL_STORE_load, OSSL_STORE_eof, 2537 OSSL_STORE_error and OSSL_STORE_close. 2538 The implementation uses backends called "loaders" to implement arbitrary 2539 URI schemes. There is one built in "loader" for the 'file' scheme. 2540 2541 *Richard Levitte* 2542 2543 * Add devcrypto engine. This has been implemented against cryptodev-linux, 2544 then adjusted to work on FreeBSD 8.4 as well. 2545 Enable by configuring with 'enable-devcryptoeng'. This is done by default 2546 on BSD implementations, as cryptodev.h is assumed to exist on all of them. 2547 2548 *Richard Levitte* 2549 2550 * Module names can prefixed with OSSL_ or OPENSSL_. This affects 2551 util/mkerr.pl, which is adapted to allow those prefixes, leading to 2552 error code calls like this: 2553 2554 OSSL_FOOerr(OSSL_FOO_F_SOMETHING, OSSL_FOO_R_WHATEVER); 2555 2556 With this change, we claim the namespaces OSSL and OPENSSL in a manner 2557 that can be encoded in C. For the foreseeable future, this will only 2558 affect new modules. 2559 2560 *Richard Levitte and Tim Hudson* 2561 2562 * Removed BSD cryptodev engine. 2563 2564 *Rich Salz* 2565 2566 * Add a build target 'build_all_generated', to build all generated files 2567 and only that. This can be used to prepare everything that requires 2568 things like perl for a system that lacks perl and then move everything 2569 to that system and do the rest of the build there. 2570 2571 *Richard Levitte* 2572 2573 * In the UI interface, make it possible to duplicate the user data. This 2574 can be used by engines that need to retain the data for a longer time 2575 than just the call where this user data is passed. 2576 2577 *Richard Levitte* 2578 2579 * Ignore the '-named_curve auto' value for compatibility of applications 2580 with OpenSSL 1.0.2. 2581 2582 *Tomáš Mráz <tmraz@fedoraproject.org>* 2583 2584 * Fragmented SSL/TLS alerts are no longer accepted. An alert message is 2 2585 bytes long. In theory it is permissible in SSLv3 - TLSv1.2 to fragment such 2586 alerts across multiple records (some of which could be empty). In practice 2587 it make no sense to send an empty alert record, or to fragment one. TLSv1.3 2588 prohibits this altogether and other libraries (BoringSSL, NSS) do not 2589 support this at all. Supporting it adds significant complexity to the 2590 record layer, and its removal is unlikely to cause interoperability 2591 issues. 2592 2593 *Matt Caswell* 2594 2595 * Add the ASN.1 types INT32, UINT32, INT64, UINT64 and variants prefixed 2596 with Z. These are meant to replace LONG and ZLONG and to be size safe. 2597 The use of LONG and ZLONG is discouraged and scheduled for deprecation 2598 in OpenSSL 1.2.0. 2599 2600 *Richard Levitte* 2601 2602 * Add the 'z' and 'j' modifiers to BIO_printf() et al formatting string, 2603 'z' is to be used for [s]size_t, and 'j' - with [u]int64_t. 2604 2605 *Richard Levitte, Andy Polyakov* 2606 2607 * Add EC_KEY_get0_engine(), which does for EC_KEY what RSA_get0_engine() 2608 does for RSA, etc. 2609 2610 *Richard Levitte* 2611 2612 * Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target 2613 platform rather than 'mingw'. 2614 2615 *Richard Levitte* 2616 2617 * The functions X509_STORE_add_cert and X509_STORE_add_crl return 2618 success if they are asked to add an object which already exists 2619 in the store. This change cascades to other functions which load 2620 certificates and CRLs. 2621 2622 *Paul Dale* 2623 2624 * x86_64 assembly pack: annotate code with DWARF CFI directives to 2625 facilitate stack unwinding even from assembly subroutines. 2626 2627 *Andy Polyakov* 2628 2629 * Remove VAX C specific definitions of OPENSSL_EXPORT, OPENSSL_EXTERN. 2630 Also remove OPENSSL_GLOBAL entirely, as it became a no-op. 2631 2632 *Richard Levitte* 2633 2634 * Remove the VMS-specific reimplementation of gmtime from crypto/o_times.c. 2635 VMS C's RTL has a fully up to date gmtime() and gmtime_r() since V7.1, 2636 which is the minimum version we support. 2637 2638 *Richard Levitte* 2639 2640 * Certificate time validation (X509_cmp_time) enforces stricter 2641 compliance with RFC 5280. Fractional seconds and timezone offsets 2642 are no longer allowed. 2643 2644 *Emilia Käsper* 2645 2646 * Add support for ARIA 2647 2648 *Paul Dale* 2649 2650 * s_client will now send the Server Name Indication (SNI) extension by 2651 default unless the new "-noservername" option is used. The server name is 2652 based on the host provided to the "-connect" option unless overridden by 2653 using "-servername". 2654 2655 *Matt Caswell* 2656 2657 * Add support for SipHash 2658 2659 *Todd Short* 2660 2661 * OpenSSL now fails if it receives an unrecognised record type in TLS1.0 2662 or TLS1.1. Previously this only happened in SSLv3 and TLS1.2. This is to 2663 prevent issues where no progress is being made and the peer continually 2664 sends unrecognised record types, using up resources processing them. 2665 2666 *Matt Caswell* 2667 2668 * 'openssl passwd' can now produce SHA256 and SHA512 based output, 2669 using the algorithm defined in 2670 <https://www.akkadia.org/drepper/SHA-crypt.txt> 2671 2672 *Richard Levitte* 2673 2674 * Heartbeat support has been removed; the ABI is changed for now. 2675 2676 *Richard Levitte, Rich Salz* 2677 2678 * Support for SSL_OP_NO_ENCRYPT_THEN_MAC in SSL_CONF_cmd. 2679 2680 *Emilia Käsper* 2681 2682 * The RSA "null" method, which was partially supported to avoid patent 2683 issues, has been replaced to always returns NULL. 2684 2685 *Rich Salz* 2686 2687OpenSSL 1.1.0 2688------------- 2689 2690### Changes between 1.1.0k and 1.1.0l [10 Sep 2019] 2691 2692 * For built-in EC curves, ensure an EC_GROUP built from the curve name is 2693 used even when parsing explicit parameters, when loading a encoded key 2694 or calling `EC_GROUP_new_from_ecpkparameters()`/ 2695 `EC_GROUP_new_from_ecparameters()`. 2696 This prevents bypass of security hardening and performance gains, 2697 especially for curves with specialized EC_METHODs. 2698 By default, if a key encoded with explicit parameters is loaded and later 2699 encoded, the output is still encoded with explicit parameters, even if 2700 internally a "named" EC_GROUP is used for computation. 2701 2702 *Nicola Tuveri* 2703 2704 * Compute ECC cofactors if not provided during EC_GROUP construction. Before 2705 this change, EC_GROUP_set_generator would accept order and/or cofactor as 2706 NULL. After this change, only the cofactor parameter can be NULL. It also 2707 does some minimal sanity checks on the passed order. 2708 ([CVE-2019-1547]) 2709 2710 *Billy Bob Brumley* 2711 2712 * Fixed a padding oracle in PKCS7_dataDecode and CMS_decrypt_set1_pkey. 2713 An attack is simple, if the first CMS_recipientInfo is valid but the 2714 second CMS_recipientInfo is chosen ciphertext. If the second 2715 recipientInfo decodes to PKCS #1 v1.5 form plaintext, the correct 2716 encryption key will be replaced by garbage, and the message cannot be 2717 decoded, but if the RSA decryption fails, the correct encryption key is 2718 used and the recipient will not notice the attack. 2719 As a work around for this potential attack the length of the decrypted 2720 key must be equal to the cipher default key length, in case the 2721 certifiate is not given and all recipientInfo are tried out. 2722 The old behaviour can be re-enabled in the CMS code by setting the 2723 CMS_DEBUG_DECRYPT flag. 2724 ([CVE-2019-1563]) 2725 2726 *Bernd Edlinger* 2727 2728 * Use Windows installation paths in the mingw builds 2729 2730 Mingw isn't a POSIX environment per se, which means that Windows 2731 paths should be used for installation. 2732 ([CVE-2019-1552]) 2733 2734 *Richard Levitte* 2735 2736### Changes between 1.1.0j and 1.1.0k [28 May 2019] 2737 2738 * Change the default RSA, DSA and DH size to 2048 bit instead of 1024. 2739 This changes the size when using the `genpkey` command when no size is given. 2740 It fixes an omission in earlier changes that changed all RSA, DSA and DH 2741 generation commands to use 2048 bits by default. 2742 2743 *Kurt Roeckx* 2744 2745 * Prevent over long nonces in ChaCha20-Poly1305. 2746 2747 ChaCha20-Poly1305 is an AEAD cipher, and requires a unique nonce input 2748 for every encryption operation. RFC 7539 specifies that the nonce value 2749 (IV) should be 96 bits (12 bytes). OpenSSL allows a variable nonce length 2750 and front pads the nonce with 0 bytes if it is less than 12 2751 bytes. However it also incorrectly allows a nonce to be set of up to 16 2752 bytes. In this case only the last 12 bytes are significant and any 2753 additional leading bytes are ignored. 2754 2755 It is a requirement of using this cipher that nonce values are 2756 unique. Messages encrypted using a reused nonce value are susceptible to 2757 serious confidentiality and integrity attacks. If an application changes 2758 the default nonce length to be longer than 12 bytes and then makes a 2759 change to the leading bytes of the nonce expecting the new value to be a 2760 new unique nonce then such an application could inadvertently encrypt 2761 messages with a reused nonce. 2762 2763 Additionally the ignored bytes in a long nonce are not covered by the 2764 integrity guarantee of this cipher. Any application that relies on the 2765 integrity of these ignored leading bytes of a long nonce may be further 2766 affected. Any OpenSSL internal use of this cipher, including in SSL/TLS, 2767 is safe because no such use sets such a long nonce value. However user 2768 applications that use this cipher directly and set a non-default nonce 2769 length to be longer than 12 bytes may be vulnerable. 2770 2771 This issue was reported to OpenSSL on 16th of March 2019 by Joran Dirk 2772 Greef of Ronomon. 2773 ([CVE-2019-1543]) 2774 2775 *Matt Caswell* 2776 2777 * Added SCA hardening for modular field inversion in EC_GROUP through 2778 a new dedicated field_inv() pointer in EC_METHOD. 2779 This also addresses a leakage affecting conversions from projective 2780 to affine coordinates. 2781 2782 *Billy Bob Brumley, Nicola Tuveri* 2783 2784 * Fix a use after free bug in d2i_X509_PUBKEY when overwriting a 2785 re-used X509_PUBKEY object if the second PUBKEY is malformed. 2786 2787 *Bernd Edlinger* 2788 2789 * Move strictness check from EVP_PKEY_asn1_new() to EVP_PKEY_asn1_add0(). 2790 2791 *Richard Levitte* 2792 2793 * Remove the 'dist' target and add a tarball building script. The 2794 'dist' target has fallen out of use, and it shouldn't be 2795 necessary to configure just to create a source distribution. 2796 2797 *Richard Levitte* 2798 2799### Changes between 1.1.0i and 1.1.0j [20 Nov 2018] 2800 2801 * Timing vulnerability in DSA signature generation 2802 2803 The OpenSSL DSA signature algorithm has been shown to be vulnerable to a 2804 timing side channel attack. An attacker could use variations in the signing 2805 algorithm to recover the private key. 2806 2807 This issue was reported to OpenSSL on 16th October 2018 by Samuel Weiser. 2808 ([CVE-2018-0734]) 2809 2810 *Paul Dale* 2811 2812 * Timing vulnerability in ECDSA signature generation 2813 2814 The OpenSSL ECDSA signature algorithm has been shown to be vulnerable to a 2815 timing side channel attack. An attacker could use variations in the signing 2816 algorithm to recover the private key. 2817 2818 This issue was reported to OpenSSL on 25th October 2018 by Samuel Weiser. 2819 ([CVE-2018-0735]) 2820 2821 *Paul Dale* 2822 2823 * Add coordinate blinding for EC_POINT and implement projective 2824 coordinate blinding for generic prime curves as a countermeasure to 2825 chosen point SCA attacks. 2826 2827 *Sohaib ul Hassan, Nicola Tuveri, Billy Bob Brumley* 2828 2829### Changes between 1.1.0h and 1.1.0i [14 Aug 2018] 2830 2831 * Client DoS due to large DH parameter 2832 2833 During key agreement in a TLS handshake using a DH(E) based ciphersuite a 2834 malicious server can send a very large prime value to the client. This will 2835 cause the client to spend an unreasonably long period of time generating a 2836 key for this prime resulting in a hang until the client has finished. This 2837 could be exploited in a Denial Of Service attack. 2838 2839 This issue was reported to OpenSSL on 5th June 2018 by Guido Vranken 2840 ([CVE-2018-0732]) 2841 2842 *Guido Vranken* 2843 2844 * Cache timing vulnerability in RSA Key Generation 2845 2846 The OpenSSL RSA Key generation algorithm has been shown to be vulnerable to 2847 a cache timing side channel attack. An attacker with sufficient access to 2848 mount cache timing attacks during the RSA key generation process could 2849 recover the private key. 2850 2851 This issue was reported to OpenSSL on 4th April 2018 by Alejandro Cabrera 2852 Aldaya, Billy Brumley, Cesar Pereida Garcia and Luis Manuel Alvarez Tapia. 2853 ([CVE-2018-0737]) 2854 2855 *Billy Brumley* 2856 2857 * Make EVP_PKEY_asn1_new() a bit stricter about its input. A NULL pem_str 2858 parameter is no longer accepted, as it leads to a corrupt table. NULL 2859 pem_str is reserved for alias entries only. 2860 2861 *Richard Levitte* 2862 2863 * Revert blinding in ECDSA sign and instead make problematic addition 2864 length-invariant. Switch even to fixed-length Montgomery multiplication. 2865 2866 *Andy Polyakov* 2867 2868 * Change generating and checking of primes so that the error rate of not 2869 being prime depends on the intended use based on the size of the input. 2870 For larger primes this will result in more rounds of Miller-Rabin. 2871 The maximal error rate for primes with more than 1080 bits is lowered 2872 to 2^-128. 2873 2874 *Kurt Roeckx, Annie Yousar* 2875 2876 * Increase the number of Miller-Rabin rounds for DSA key generating to 64. 2877 2878 *Kurt Roeckx* 2879 2880 * Add blinding to ECDSA and DSA signatures to protect against side channel 2881 attacks discovered by Keegan Ryan (NCC Group). 2882 2883 *Matt Caswell* 2884 2885 * When unlocking a pass phrase protected PEM file or PKCS#8 container, we 2886 now allow empty (zero character) pass phrases. 2887 2888 *Richard Levitte* 2889 2890 * Certificate time validation (X509_cmp_time) enforces stricter 2891 compliance with RFC 5280. Fractional seconds and timezone offsets 2892 are no longer allowed. 2893 2894 *Emilia Käsper* 2895 2896 * Fixed a text canonicalisation bug in CMS 2897 2898 Where a CMS detached signature is used with text content the text goes 2899 through a canonicalisation process first prior to signing or verifying a 2900 signature. This process strips trailing space at the end of lines, converts 2901 line terminators to CRLF and removes additional trailing line terminators 2902 at the end of a file. A bug in the canonicalisation process meant that 2903 some characters, such as form-feed, were incorrectly treated as whitespace 2904 and removed. This is contrary to the specification (RFC5485). This fix 2905 could mean that detached text data signed with an earlier version of 2906 OpenSSL 1.1.0 may fail to verify using the fixed version, or text data 2907 signed with a fixed OpenSSL may fail to verify with an earlier version of 2908 OpenSSL 1.1.0. A workaround is to only verify the canonicalised text data 2909 and use the "-binary" flag (for the "cms" command line application) or set 2910 the SMIME_BINARY/PKCS7_BINARY/CMS_BINARY flags (if using CMS_verify()). 2911 2912 *Matt Caswell* 2913 2914### Changes between 1.1.0g and 1.1.0h [27 Mar 2018] 2915 2916 * Constructed ASN.1 types with a recursive definition could exceed the stack 2917 2918 Constructed ASN.1 types with a recursive definition (such as can be found 2919 in PKCS7) could eventually exceed the stack given malicious input with 2920 excessive recursion. This could result in a Denial Of Service attack. There 2921 are no such structures used within SSL/TLS that come from untrusted sources 2922 so this is considered safe. 2923 2924 This issue was reported to OpenSSL on 4th January 2018 by the OSS-fuzz 2925 project. 2926 ([CVE-2018-0739]) 2927 2928 *Matt Caswell* 2929 2930 * Incorrect CRYPTO_memcmp on HP-UX PA-RISC 2931 2932 Because of an implementation bug the PA-RISC CRYPTO_memcmp function is 2933 effectively reduced to only comparing the least significant bit of each 2934 byte. This allows an attacker to forge messages that would be considered as 2935 authenticated in an amount of tries lower than that guaranteed by the 2936 security claims of the scheme. The module can only be compiled by the 2937 HP-UX assembler, so that only HP-UX PA-RISC targets are affected. 2938 2939 This issue was reported to OpenSSL on 2nd March 2018 by Peter Waltenberg 2940 (IBM). 2941 ([CVE-2018-0733]) 2942 2943 *Andy Polyakov* 2944 2945 * Add a build target 'build_all_generated', to build all generated files 2946 and only that. This can be used to prepare everything that requires 2947 things like perl for a system that lacks perl and then move everything 2948 to that system and do the rest of the build there. 2949 2950 *Richard Levitte* 2951 2952 * Backport SSL_OP_NO_RENGOTIATION 2953 2954 OpenSSL 1.0.2 and below had the ability to disable renegotiation using the 2955 (undocumented) SSL3_FLAGS_NO_RENEGOTIATE_CIPHERS flag. Due to the opacity 2956 changes this is no longer possible in 1.1.0. Therefore the new 2957 SSL_OP_NO_RENEGOTIATION option from 1.1.1-dev has been backported to 2958 1.1.0 to provide equivalent functionality. 2959 2960 Note that if an application built against 1.1.0h headers (or above) is run 2961 using an older version of 1.1.0 (prior to 1.1.0h) then the option will be 2962 accepted but nothing will happen, i.e. renegotiation will not be prevented. 2963 2964 *Matt Caswell* 2965 2966 * Removed the OS390-Unix config target. It relied on a script that doesn't 2967 exist. 2968 2969 *Rich Salz* 2970 2971 * rsaz_1024_mul_avx2 overflow bug on x86_64 2972 2973 There is an overflow bug in the AVX2 Montgomery multiplication procedure 2974 used in exponentiation with 1024-bit moduli. No EC algorithms are affected. 2975 Analysis suggests that attacks against RSA and DSA as a result of this 2976 defect would be very difficult to perform and are not believed likely. 2977 Attacks against DH1024 are considered just feasible, because most of the 2978 work necessary to deduce information about a private key may be performed 2979 offline. The amount of resources required for such an attack would be 2980 significant. However, for an attack on TLS to be meaningful, the server 2981 would have to share the DH1024 private key among multiple clients, which is 2982 no longer an option since CVE-2016-0701. 2983 2984 This only affects processors that support the AVX2 but not ADX extensions 2985 like Intel Haswell (4th generation). 2986 2987 This issue was reported to OpenSSL by David Benjamin (Google). The issue 2988 was originally found via the OSS-Fuzz project. 2989 ([CVE-2017-3738]) 2990 2991 *Andy Polyakov* 2992 2993### Changes between 1.1.0f and 1.1.0g [2 Nov 2017] 2994 2995 * bn_sqrx8x_internal carry bug on x86_64 2996 2997 There is a carry propagating bug in the x86_64 Montgomery squaring 2998 procedure. No EC algorithms are affected. Analysis suggests that attacks 2999 against RSA and DSA as a result of this defect would be very difficult to 3000 perform and are not believed likely. Attacks against DH are considered just 3001 feasible (although very difficult) because most of the work necessary to 3002 deduce information about a private key may be performed offline. The amount 3003 of resources required for such an attack would be very significant and 3004 likely only accessible to a limited number of attackers. An attacker would 3005 additionally need online access to an unpatched system using the target 3006 private key in a scenario with persistent DH parameters and a private 3007 key that is shared between multiple clients. 3008 3009 This only affects processors that support the BMI1, BMI2 and ADX extensions 3010 like Intel Broadwell (5th generation) and later or AMD Ryzen. 3011 3012 This issue was reported to OpenSSL by the OSS-Fuzz project. 3013 ([CVE-2017-3736]) 3014 3015 *Andy Polyakov* 3016 3017 * Malformed X.509 IPAddressFamily could cause OOB read 3018 3019 If an X.509 certificate has a malformed IPAddressFamily extension, 3020 OpenSSL could do a one-byte buffer overread. The most likely result 3021 would be an erroneous display of the certificate in text format. 3022 3023 This issue was reported to OpenSSL by the OSS-Fuzz project. 3024 ([CVE-2017-3735]) 3025 3026 *Rich Salz* 3027 3028### Changes between 1.1.0e and 1.1.0f [25 May 2017] 3029 3030 * Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target 3031 platform rather than 'mingw'. 3032 3033 *Richard Levitte* 3034 3035 * Remove the VMS-specific reimplementation of gmtime from crypto/o_times.c. 3036 VMS C's RTL has a fully up to date gmtime() and gmtime_r() since V7.1, 3037 which is the minimum version we support. 3038 3039 *Richard Levitte* 3040 3041### Changes between 1.1.0d and 1.1.0e [16 Feb 2017] 3042 3043 * Encrypt-Then-Mac renegotiation crash 3044 3045 During a renegotiation handshake if the Encrypt-Then-Mac extension is 3046 negotiated where it was not in the original handshake (or vice-versa) then 3047 this can cause OpenSSL to crash (dependant on ciphersuite). Both clients 3048 and servers are affected. 3049 3050 This issue was reported to OpenSSL by Joe Orton (Red Hat). 3051 ([CVE-2017-3733]) 3052 3053 *Matt Caswell* 3054 3055### Changes between 1.1.0c and 1.1.0d [26 Jan 2017] 3056 3057 * Truncated packet could crash via OOB read 3058 3059 If one side of an SSL/TLS path is running on a 32-bit host and a specific 3060 cipher is being used, then a truncated packet can cause that host to 3061 perform an out-of-bounds read, usually resulting in a crash. 3062 3063 This issue was reported to OpenSSL by Robert Święcki of Google. 3064 ([CVE-2017-3731]) 3065 3066 *Andy Polyakov* 3067 3068 * Bad (EC)DHE parameters cause a client crash 3069 3070 If a malicious server supplies bad parameters for a DHE or ECDHE key 3071 exchange then this can result in the client attempting to dereference a 3072 NULL pointer leading to a client crash. This could be exploited in a Denial 3073 of Service attack. 3074 3075 This issue was reported to OpenSSL by Guido Vranken. 3076 ([CVE-2017-3730]) 3077 3078 *Matt Caswell* 3079 3080 * BN_mod_exp may produce incorrect results on x86_64 3081 3082 There is a carry propagating bug in the x86_64 Montgomery squaring 3083 procedure. No EC algorithms are affected. Analysis suggests that attacks 3084 against RSA and DSA as a result of this defect would be very difficult to 3085 perform and are not believed likely. Attacks against DH are considered just 3086 feasible (although very difficult) because most of the work necessary to 3087 deduce information about a private key may be performed offline. The amount 3088 of resources required for such an attack would be very significant and 3089 likely only accessible to a limited number of attackers. An attacker would 3090 additionally need online access to an unpatched system using the target 3091 private key in a scenario with persistent DH parameters and a private 3092 key that is shared between multiple clients. For example this can occur by 3093 default in OpenSSL DHE based SSL/TLS ciphersuites. Note: This issue is very 3094 similar to CVE-2015-3193 but must be treated as a separate problem. 3095 3096 This issue was reported to OpenSSL by the OSS-Fuzz project. 3097 ([CVE-2017-3732]) 3098 3099 *Andy Polyakov* 3100 3101### Changes between 1.1.0b and 1.1.0c [10 Nov 2016] 3102 3103 * ChaCha20/Poly1305 heap-buffer-overflow 3104 3105 TLS connections using `*-CHACHA20-POLY1305` ciphersuites are susceptible to 3106 a DoS attack by corrupting larger payloads. This can result in an OpenSSL 3107 crash. This issue is not considered to be exploitable beyond a DoS. 3108 3109 This issue was reported to OpenSSL by Robert Święcki (Google Security Team) 3110 ([CVE-2016-7054]) 3111 3112 *Richard Levitte* 3113 3114 * CMS Null dereference 3115 3116 Applications parsing invalid CMS structures can crash with a NULL pointer 3117 dereference. This is caused by a bug in the handling of the ASN.1 CHOICE 3118 type in OpenSSL 1.1.0 which can result in a NULL value being passed to the 3119 structure callback if an attempt is made to free certain invalid encodings. 3120 Only CHOICE structures using a callback which do not handle NULL value are 3121 affected. 3122 3123 This issue was reported to OpenSSL by Tyler Nighswander of ForAllSecure. 3124 ([CVE-2016-7053]) 3125 3126 *Stephen Henson* 3127 3128 * Montgomery multiplication may produce incorrect results 3129 3130 There is a carry propagating bug in the Broadwell-specific Montgomery 3131 multiplication procedure that handles input lengths divisible by, but 3132 longer than 256 bits. Analysis suggests that attacks against RSA, DSA 3133 and DH private keys are impossible. This is because the subroutine in 3134 question is not used in operations with the private key itself and an input 3135 of the attacker's direct choice. Otherwise the bug can manifest itself as 3136 transient authentication and key negotiation failures or reproducible 3137 erroneous outcome of public-key operations with specially crafted input. 3138 Among EC algorithms only Brainpool P-512 curves are affected and one 3139 presumably can attack ECDH key negotiation. Impact was not analyzed in 3140 detail, because pre-requisites for attack are considered unlikely. Namely 3141 multiple clients have to choose the curve in question and the server has to 3142 share the private key among them, neither of which is default behaviour. 3143 Even then only clients that chose the curve will be affected. 3144 3145 This issue was publicly reported as transient failures and was not 3146 initially recognized as a security issue. Thanks to Richard Morgan for 3147 providing reproducible case. 3148 ([CVE-2016-7055]) 3149 3150 *Andy Polyakov* 3151 3152 * Removed automatic addition of RPATH in shared libraries and executables, 3153 as this was a remainder from OpenSSL 1.0.x and isn't needed any more. 3154 3155 *Richard Levitte* 3156 3157### Changes between 1.1.0a and 1.1.0b [26 Sep 2016] 3158 3159 * Fix Use After Free for large message sizes 3160 3161 The patch applied to address CVE-2016-6307 resulted in an issue where if a 3162 message larger than approx 16k is received then the underlying buffer to 3163 store the incoming message is reallocated and moved. Unfortunately a 3164 dangling pointer to the old location is left which results in an attempt to 3165 write to the previously freed location. This is likely to result in a 3166 crash, however it could potentially lead to execution of arbitrary code. 3167 3168 This issue only affects OpenSSL 1.1.0a. 3169 3170 This issue was reported to OpenSSL by Robert Święcki. 3171 ([CVE-2016-6309]) 3172 3173 *Matt Caswell* 3174 3175### Changes between 1.1.0 and 1.1.0a [22 Sep 2016] 3176 3177 * OCSP Status Request extension unbounded memory growth 3178 3179 A malicious client can send an excessively large OCSP Status Request 3180 extension. If that client continually requests renegotiation, sending a 3181 large OCSP Status Request extension each time, then there will be unbounded 3182 memory growth on the server. This will eventually lead to a Denial Of 3183 Service attack through memory exhaustion. Servers with a default 3184 configuration are vulnerable even if they do not support OCSP. Builds using 3185 the "no-ocsp" build time option are not affected. 3186 3187 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 3188 ([CVE-2016-6304]) 3189 3190 *Matt Caswell* 3191 3192 * SSL_peek() hang on empty record 3193 3194 OpenSSL 1.1.0 SSL/TLS will hang during a call to SSL_peek() if the peer 3195 sends an empty record. This could be exploited by a malicious peer in a 3196 Denial Of Service attack. 3197 3198 This issue was reported to OpenSSL by Alex Gaynor. 3199 ([CVE-2016-6305]) 3200 3201 *Matt Caswell* 3202 3203 * Excessive allocation of memory in tls_get_message_header() and 3204 dtls1_preprocess_fragment() 3205 3206 A (D)TLS message includes 3 bytes for its length in the header for the 3207 message. This would allow for messages up to 16Mb in length. Messages of 3208 this length are excessive and OpenSSL includes a check to ensure that a 3209 peer is sending reasonably sized messages in order to avoid too much memory 3210 being consumed to service a connection. A flaw in the logic of version 3211 1.1.0 means that memory for the message is allocated too early, prior to 3212 the excessive message length check. Due to way memory is allocated in 3213 OpenSSL this could mean an attacker could force up to 21Mb to be allocated 3214 to service a connection. This could lead to a Denial of Service through 3215 memory exhaustion. However, the excessive message length check still takes 3216 place, and this would cause the connection to immediately fail. Assuming 3217 that the application calls SSL_free() on the failed connection in a timely 3218 manner then the 21Mb of allocated memory will then be immediately freed 3219 again. Therefore the excessive memory allocation will be transitory in 3220 nature. This then means that there is only a security impact if: 3221 3222 1) The application does not call SSL_free() in a timely manner in the event 3223 that the connection fails 3224 or 3225 2) The application is working in a constrained environment where there is 3226 very little free memory 3227 or 3228 3) The attacker initiates multiple connection attempts such that there are 3229 multiple connections in a state where memory has been allocated for the 3230 connection; SSL_free() has not yet been called; and there is insufficient 3231 memory to service the multiple requests. 3232 3233 Except in the instance of (1) above any Denial Of Service is likely to be 3234 transitory because as soon as the connection fails the memory is 3235 subsequently freed again in the SSL_free() call. However there is an 3236 increased risk during this period of application crashes due to the lack of 3237 memory - which would then mean a more serious Denial of Service. 3238 3239 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 3240 (CVE-2016-6307 and CVE-2016-6308) 3241 3242 *Matt Caswell* 3243 3244 * solaris-x86-cc, i.e. 32-bit configuration with vendor compiler, 3245 had to be removed. Primary reason is that vendor assembler can't 3246 assemble our modules with -KPIC flag. As result it, assembly 3247 support, was not even available as option. But its lack means 3248 lack of side-channel resistant code, which is incompatible with 3249 security by todays standards. Fortunately gcc is readily available 3250 prepackaged option, which we firmly point at... 3251 3252 *Andy Polyakov* 3253 3254### Changes between 1.0.2h and 1.1.0 [25 Aug 2016] 3255 3256 * Windows command-line tool supports UTF-8 opt-in option for arguments 3257 and console input. Setting OPENSSL_WIN32_UTF8 environment variable 3258 (to any value) allows Windows user to access PKCS#12 file generated 3259 with Windows CryptoAPI and protected with non-ASCII password, as well 3260 as files generated under UTF-8 locale on Linux also protected with 3261 non-ASCII password. 3262 3263 *Andy Polyakov* 3264 3265 * To mitigate the SWEET32 attack ([CVE-2016-2183]), 3DES cipher suites 3266 have been disabled by default and removed from DEFAULT, just like RC4. 3267 See the RC4 item below to re-enable both. 3268 3269 *Rich Salz* 3270 3271 * The method for finding the storage location for the Windows RAND seed file 3272 has changed. First we check %RANDFILE%. If that is not set then we check 3273 the directories %HOME%, %USERPROFILE% and %SYSTEMROOT% in that order. If 3274 all else fails we fall back to C:\. 3275 3276 *Matt Caswell* 3277 3278 * The EVP_EncryptUpdate() function has had its return type changed from void 3279 to int. A return of 0 indicates and error while a return of 1 indicates 3280 success. 3281 3282 *Matt Caswell* 3283 3284 * The flags RSA_FLAG_NO_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME and 3285 DH_FLAG_NO_EXP_CONSTTIME which previously provided the ability to switch 3286 off the constant time implementation for RSA, DSA and DH have been made 3287 no-ops and deprecated. 3288 3289 *Matt Caswell* 3290 3291 * Windows RAND implementation was simplified to only get entropy by 3292 calling CryptGenRandom(). Various other RAND-related tickets 3293 were also closed. 3294 3295 *Joseph Wylie Yandle, Rich Salz* 3296 3297 * The stack and lhash API's were renamed to start with `OPENSSL_SK_` 3298 and `OPENSSL_LH_`, respectively. The old names are available 3299 with API compatibility. They new names are now completely documented. 3300 3301 *Rich Salz* 3302 3303 * Unify TYPE_up_ref(obj) methods signature. 3304 SSL_CTX_up_ref(), SSL_up_ref(), X509_up_ref(), EVP_PKEY_up_ref(), 3305 X509_CRL_up_ref(), X509_OBJECT_up_ref_count() methods are now returning an 3306 int (instead of void) like all others TYPE_up_ref() methods. 3307 So now these methods also check the return value of CRYPTO_atomic_add(), 3308 and the validity of object reference counter. 3309 3310 *fdasilvayy@gmail.com* 3311 3312 * With Windows Visual Studio builds, the .pdb files are installed 3313 alongside the installed libraries and executables. For a static 3314 library installation, ossl_static.pdb is the associate compiler 3315 generated .pdb file to be used when linking programs. 3316 3317 *Richard Levitte* 3318 3319 * Remove openssl.spec. Packaging files belong with the packagers. 3320 3321 *Richard Levitte* 3322 3323 * Automatic Darwin/OSX configuration has had a refresh, it will now 3324 recognise x86_64 architectures automatically. You can still decide 3325 to build for a different bitness with the environment variable 3326 KERNEL_BITS (can be 32 or 64), for example: 3327 3328 KERNEL_BITS=32 ./config 3329 3330 *Richard Levitte* 3331 3332 * Change default algorithms in pkcs8 utility to use PKCS#5 v2.0, 3333 256 bit AES and HMAC with SHA256. 3334 3335 *Steve Henson* 3336 3337 * Remove support for MIPS o32 ABI on IRIX (and IRIX only). 3338 3339 *Andy Polyakov* 3340 3341 * Triple-DES ciphers have been moved from HIGH to MEDIUM. 3342 3343 *Rich Salz* 3344 3345 * To enable users to have their own config files and build file templates, 3346 Configure looks in the directory indicated by the environment variable 3347 OPENSSL_LOCAL_CONFIG_DIR as well as the in-source Configurations/ 3348 directory. On VMS, OPENSSL_LOCAL_CONFIG_DIR is expected to be a logical 3349 name and is used as is. 3350 3351 *Richard Levitte* 3352 3353 * The following datatypes were made opaque: X509_OBJECT, X509_STORE_CTX, 3354 X509_STORE, X509_LOOKUP, and X509_LOOKUP_METHOD. The unused type 3355 X509_CERT_FILE_CTX was removed. 3356 3357 *Rich Salz* 3358 3359 * "shared" builds are now the default. To create only static libraries use 3360 the "no-shared" Configure option. 3361 3362 *Matt Caswell* 3363 3364 * Remove the no-aes, no-hmac, no-rsa, no-sha and no-md5 Configure options. 3365 All of these option have not worked for some while and are fundamental 3366 algorithms. 3367 3368 *Matt Caswell* 3369 3370 * Make various cleanup routines no-ops and mark them as deprecated. Most 3371 global cleanup functions are no longer required because they are handled 3372 via auto-deinit (see OPENSSL_init_crypto and OPENSSL_init_ssl man pages). 3373 Explicitly de-initing can cause problems (e.g. where a library that uses 3374 OpenSSL de-inits, but an application is still using it). The affected 3375 functions are CONF_modules_free(), ENGINE_cleanup(), OBJ_cleanup(), 3376 EVP_cleanup(), BIO_sock_cleanup(), CRYPTO_cleanup_all_ex_data(), 3377 RAND_cleanup(), SSL_COMP_free_compression_methods(), ERR_free_strings() and 3378 COMP_zlib_cleanup(). 3379 3380 *Matt Caswell* 3381 3382 * --strict-warnings no longer enables runtime debugging options 3383 such as REF_DEBUG. Instead, debug options are automatically 3384 enabled with '--debug' builds. 3385 3386 *Andy Polyakov, Emilia Käsper* 3387 3388 * Made DH and DH_METHOD opaque. The structures for managing DH objects 3389 have been moved out of the public header files. New functions for managing 3390 these have been added. 3391 3392 *Matt Caswell* 3393 3394 * Made RSA and RSA_METHOD opaque. The structures for managing RSA 3395 objects have been moved out of the public header files. New 3396 functions for managing these have been added. 3397 3398 *Richard Levitte* 3399 3400 * Made DSA and DSA_METHOD opaque. The structures for managing DSA objects 3401 have been moved out of the public header files. New functions for managing 3402 these have been added. 3403 3404 *Matt Caswell* 3405 3406 * Made BIO and BIO_METHOD opaque. The structures for managing BIOs have been 3407 moved out of the public header files. New functions for managing these 3408 have been added. 3409 3410 *Matt Caswell* 3411 3412 * Removed no-rijndael as a config option. Rijndael is an old name for AES. 3413 3414 *Matt Caswell* 3415 3416 * Removed the mk1mf build scripts. 3417 3418 *Richard Levitte* 3419 3420 * Headers are now wrapped, if necessary, with OPENSSL_NO_xxx, so 3421 it is always safe to #include a header now. 3422 3423 *Rich Salz* 3424 3425 * Removed the aged BC-32 config and all its supporting scripts 3426 3427 *Richard Levitte* 3428 3429 * Removed support for Ultrix, Netware, and OS/2. 3430 3431 *Rich Salz* 3432 3433 * Add support for HKDF. 3434 3435 *Alessandro Ghedini* 3436 3437 * Add support for blake2b and blake2s 3438 3439 *Bill Cox* 3440 3441 * Added support for "pipelining". Ciphers that have the 3442 EVP_CIPH_FLAG_PIPELINE flag set have a capability to process multiple 3443 encryptions/decryptions simultaneously. There are currently no built-in 3444 ciphers with this property but the expectation is that engines will be able 3445 to offer it to significantly improve throughput. Support has been extended 3446 into libssl so that multiple records for a single connection can be 3447 processed in one go (for >=TLS 1.1). 3448 3449 *Matt Caswell* 3450 3451 * Added the AFALG engine. This is an async capable engine which is able to 3452 offload work to the Linux kernel. In this initial version it only supports 3453 AES128-CBC. The kernel must be version 4.1.0 or greater. 3454 3455 *Catriona Lucey* 3456 3457 * OpenSSL now uses a new threading API. It is no longer necessary to 3458 set locking callbacks to use OpenSSL in a multi-threaded environment. There 3459 are two supported threading models: pthreads and windows threads. It is 3460 also possible to configure OpenSSL at compile time for "no-threads". The 3461 old threading API should no longer be used. The functions have been 3462 replaced with "no-op" compatibility macros. 3463 3464 *Alessandro Ghedini, Matt Caswell* 3465 3466 * Modify behavior of ALPN to invoke callback after SNI/servername 3467 callback, such that updates to the SSL_CTX affect ALPN. 3468 3469 *Todd Short* 3470 3471 * Add SSL_CIPHER queries for authentication and key-exchange. 3472 3473 *Todd Short* 3474 3475 * Changes to the DEFAULT cipherlist: 3476 - Prefer (EC)DHE handshakes over plain RSA. 3477 - Prefer AEAD ciphers over legacy ciphers. 3478 - Prefer ECDSA over RSA when both certificates are available. 3479 - Prefer TLSv1.2 ciphers/PRF. 3480 - Remove DSS, SEED, IDEA, CAMELLIA, and AES-CCM from the 3481 default cipherlist. 3482 3483 *Emilia Käsper* 3484 3485 * Change the ECC default curve list to be this, in order: x25519, 3486 secp256r1, secp521r1, secp384r1. 3487 3488 *Rich Salz* 3489 3490 * RC4 based libssl ciphersuites are now classed as "weak" ciphers and are 3491 disabled by default. They can be re-enabled using the 3492 enable-weak-ssl-ciphers option to Configure. 3493 3494 *Matt Caswell* 3495 3496 * If the server has ALPN configured, but supports no protocols that the 3497 client advertises, send a fatal "no_application_protocol" alert. 3498 This behaviour is SHALL in RFC 7301, though it isn't universally 3499 implemented by other servers. 3500 3501 *Emilia Käsper* 3502 3503 * Add X25519 support. 3504 Add ASN.1 and EVP_PKEY methods for X25519. This includes support 3505 for public and private key encoding using the format documented in 3506 draft-ietf-curdle-pkix-02. The corresponding EVP_PKEY method supports 3507 key generation and key derivation. 3508 3509 TLS support complies with draft-ietf-tls-rfc4492bis-08 and uses 3510 X25519(29). 3511 3512 *Steve Henson* 3513 3514 * Deprecate SRP_VBASE_get_by_user. 3515 SRP_VBASE_get_by_user had inconsistent memory management behaviour. 3516 In order to fix an unavoidable memory leak ([CVE-2016-0798]), 3517 SRP_VBASE_get_by_user was changed to ignore the "fake user" SRP 3518 seed, even if the seed is configured. 3519 3520 Users should use SRP_VBASE_get1_by_user instead. Note that in 3521 SRP_VBASE_get1_by_user, caller must free the returned value. Note 3522 also that even though configuring the SRP seed attempts to hide 3523 invalid usernames by continuing the handshake with fake 3524 credentials, this behaviour is not constant time and no strong 3525 guarantees are made that the handshake is indistinguishable from 3526 that of a valid user. 3527 3528 *Emilia Käsper* 3529 3530 * Configuration change; it's now possible to build dynamic engines 3531 without having to build shared libraries and vice versa. This 3532 only applies to the engines in `engines/`, those in `crypto/engine/` 3533 will always be built into libcrypto (i.e. "static"). 3534 3535 Building dynamic engines is enabled by default; to disable, use 3536 the configuration option "disable-dynamic-engine". 3537 3538 The only requirements for building dynamic engines are the 3539 presence of the DSO module and building with position independent 3540 code, so they will also automatically be disabled if configuring 3541 with "disable-dso" or "disable-pic". 3542 3543 The macros OPENSSL_NO_STATIC_ENGINE and OPENSSL_NO_DYNAMIC_ENGINE 3544 are also taken away from openssl/opensslconf.h, as they are 3545 irrelevant. 3546 3547 *Richard Levitte* 3548 3549 * Configuration change; if there is a known flag to compile 3550 position independent code, it will always be applied on the 3551 libcrypto and libssl object files, and never on the application 3552 object files. This means other libraries that use routines from 3553 libcrypto / libssl can be made into shared libraries regardless 3554 of how OpenSSL was configured. 3555 3556 If this isn't desirable, the configuration options "disable-pic" 3557 or "no-pic" can be used to disable the use of PIC. This will 3558 also disable building shared libraries and dynamic engines. 3559 3560 *Richard Levitte* 3561 3562 * Removed JPAKE code. It was experimental and has no wide use. 3563 3564 *Rich Salz* 3565 3566 * The INSTALL_PREFIX Makefile variable has been renamed to 3567 DESTDIR. That makes for less confusion on what this variable 3568 is for. Also, the configuration option --install_prefix is 3569 removed. 3570 3571 *Richard Levitte* 3572 3573 * Heartbeat for TLS has been removed and is disabled by default 3574 for DTLS; configure with enable-heartbeats. Code that uses the 3575 old #define's might need to be updated. 3576 3577 *Emilia Käsper, Rich Salz* 3578 3579 * Rename REF_CHECK to REF_DEBUG. 3580 3581 *Rich Salz* 3582 3583 * New "unified" build system 3584 3585 The "unified" build system is aimed to be a common system for all 3586 platforms we support. With it comes new support for VMS. 3587 3588 This system builds supports building in a different directory tree 3589 than the source tree. It produces one Makefile (for unix family 3590 or lookalikes), or one descrip.mms (for VMS). 3591 3592 The source of information to make the Makefile / descrip.mms is 3593 small files called 'build.info', holding the necessary 3594 information for each directory with source to compile, and a 3595 template in Configurations, like unix-Makefile.tmpl or 3596 descrip.mms.tmpl. 3597 3598 With this change, the library names were also renamed on Windows 3599 and on VMS. They now have names that are closer to the standard 3600 on Unix, and include the major version number, and in certain 3601 cases, the architecture they are built for. See "Notes on shared 3602 libraries" in INSTALL. 3603 3604 We rely heavily on the perl module Text::Template. 3605 3606 *Richard Levitte* 3607 3608 * Added support for auto-initialisation and de-initialisation of the library. 3609 OpenSSL no longer requires explicit init or deinit routines to be called, 3610 except in certain circumstances. See the OPENSSL_init_crypto() and 3611 OPENSSL_init_ssl() man pages for further information. 3612 3613 *Matt Caswell* 3614 3615 * The arguments to the DTLSv1_listen function have changed. Specifically the 3616 "peer" argument is now expected to be a BIO_ADDR object. 3617 3618 * Rewrite of BIO networking library. The BIO library lacked consistent 3619 support of IPv6, and adding it required some more extensive 3620 modifications. This introduces the BIO_ADDR and BIO_ADDRINFO types, 3621 which hold all types of addresses and chains of address information. 3622 It also introduces a new API, with functions like BIO_socket, 3623 BIO_connect, BIO_listen, BIO_lookup and a rewrite of BIO_accept. 3624 The source/sink BIOs BIO_s_connect, BIO_s_accept and BIO_s_datagram 3625 have been adapted accordingly. 3626 3627 *Richard Levitte* 3628 3629 * RSA_padding_check_PKCS1_type_1 now accepts inputs with and without 3630 the leading 0-byte. 3631 3632 *Emilia Käsper* 3633 3634 * CRIME protection: disable compression by default, even if OpenSSL is 3635 compiled with zlib enabled. Applications can still enable compression 3636 by calling SSL_CTX_clear_options(ctx, SSL_OP_NO_COMPRESSION), or by 3637 using the SSL_CONF library to configure compression. 3638 3639 *Emilia Käsper* 3640 3641 * The signature of the session callback configured with 3642 SSL_CTX_sess_set_get_cb was changed. The read-only input buffer 3643 was explicitly marked as `const unsigned char*` instead of 3644 `unsigned char*`. 3645 3646 *Emilia Käsper* 3647 3648 * Always DPURIFY. Remove the use of uninitialized memory in the 3649 RNG, and other conditional uses of DPURIFY. This makes -DPURIFY a no-op. 3650 3651 *Emilia Käsper* 3652 3653 * Removed many obsolete configuration items, including 3654 DES_PTR, DES_RISC1, DES_RISC2, DES_INT 3655 MD2_CHAR, MD2_INT, MD2_LONG 3656 BF_PTR, BF_PTR2 3657 IDEA_SHORT, IDEA_LONG 3658 RC2_SHORT, RC2_LONG, RC4_LONG, RC4_CHUNK, RC4_INDEX 3659 3660 *Rich Salz, with advice from Andy Polyakov* 3661 3662 * Many BN internals have been moved to an internal header file. 3663 3664 *Rich Salz with help from Andy Polyakov* 3665 3666 * Configuration and writing out the results from it has changed. 3667 Files such as Makefile include/openssl/opensslconf.h and are now 3668 produced through general templates, such as Makefile.in and 3669 crypto/opensslconf.h.in and some help from the perl module 3670 Text::Template. 3671 3672 Also, the center of configuration information is no longer 3673 Makefile. Instead, Configure produces a perl module in 3674 configdata.pm which holds most of the config data (in the hash 3675 table %config), the target data that comes from the target 3676 configuration in one of the `Configurations/*.conf` files (in 3677 %target). 3678 3679 *Richard Levitte* 3680 3681 * To clarify their intended purposes, the Configure options 3682 --prefix and --openssldir change their semantics, and become more 3683 straightforward and less interdependent. 3684 3685 --prefix shall be used exclusively to give the location INSTALLTOP 3686 where programs, scripts, libraries, include files and manuals are 3687 going to be installed. The default is now /usr/local. 3688 3689 --openssldir shall be used exclusively to give the default 3690 location OPENSSLDIR where certificates, private keys, CRLs are 3691 managed. This is also where the default openssl.cnf gets 3692 installed. 3693 If the directory given with this option is a relative path, the 3694 values of both the --prefix value and the --openssldir value will 3695 be combined to become OPENSSLDIR. 3696 The default for --openssldir is INSTALLTOP/ssl. 3697 3698 Anyone who uses --openssldir to specify where OpenSSL is to be 3699 installed MUST change to use --prefix instead. 3700 3701 *Richard Levitte* 3702 3703 * The GOST engine was out of date and therefore it has been removed. An up 3704 to date GOST engine is now being maintained in an external repository. 3705 See: <https://wiki.openssl.org/index.php/Binaries>. Libssl still retains 3706 support for GOST ciphersuites (these are only activated if a GOST engine 3707 is present). 3708 3709 *Matt Caswell* 3710 3711 * EGD is no longer supported by default; use enable-egd when 3712 configuring. 3713 3714 *Ben Kaduk and Rich Salz* 3715 3716 * The distribution now has Makefile.in files, which are used to 3717 create Makefile's when Configure is run. *Configure must be run 3718 before trying to build now.* 3719 3720 *Rich Salz* 3721 3722 * The return value for SSL_CIPHER_description() for error conditions 3723 has changed. 3724 3725 *Rich Salz* 3726 3727 * Support for RFC6698/RFC7671 DANE TLSA peer authentication. 3728 3729 Obtaining and performing DNSSEC validation of TLSA records is 3730 the application's responsibility. The application provides 3731 the TLSA records of its choice to OpenSSL, and these are then 3732 used to authenticate the peer. 3733 3734 The TLSA records need not even come from DNS. They can, for 3735 example, be used to implement local end-entity certificate or 3736 trust-anchor "pinning", where the "pin" data takes the form 3737 of TLSA records, which can augment or replace verification 3738 based on the usual WebPKI public certification authorities. 3739 3740 *Viktor Dukhovni* 3741 3742 * Revert default OPENSSL_NO_DEPRECATED setting. Instead OpenSSL 3743 continues to support deprecated interfaces in default builds. 3744 However, applications are strongly advised to compile their 3745 source files with -DOPENSSL_API_COMPAT=0x10100000L, which hides 3746 the declarations of all interfaces deprecated in 0.9.8, 1.0.0 3747 or the 1.1.0 releases. 3748 3749 In environments in which all applications have been ported to 3750 not use any deprecated interfaces OpenSSL's Configure script 3751 should be used with the --api=1.1.0 option to entirely remove 3752 support for the deprecated features from the library and 3753 unconditionally disable them in the installed headers. 3754 Essentially the same effect can be achieved with the "no-deprecated" 3755 argument to Configure, except that this will always restrict 3756 the build to just the latest API, rather than a fixed API 3757 version. 3758 3759 As applications are ported to future revisions of the API, 3760 they should update their compile-time OPENSSL_API_COMPAT define 3761 accordingly, but in most cases should be able to continue to 3762 compile with later releases. 3763 3764 The OPENSSL_API_COMPAT versions for 1.0.0, and 0.9.8 are 3765 0x10000000L and 0x00908000L, respectively. However those 3766 versions did not support the OPENSSL_API_COMPAT feature, and 3767 so applications are not typically tested for explicit support 3768 of just the undeprecated features of either release. 3769 3770 *Viktor Dukhovni* 3771 3772 * Add support for setting the minimum and maximum supported protocol. 3773 It can bet set via the SSL_set_min_proto_version() and 3774 SSL_set_max_proto_version(), or via the SSL_CONF's MinProtocol and 3775 MaxProtocol. It's recommended to use the new APIs to disable 3776 protocols instead of disabling individual protocols using 3777 SSL_set_options() or SSL_CONF's Protocol. This change also 3778 removes support for disabling TLS 1.2 in the OpenSSL TLS 3779 client at compile time by defining OPENSSL_NO_TLS1_2_CLIENT. 3780 3781 *Kurt Roeckx* 3782 3783 * Support for ChaCha20 and Poly1305 added to libcrypto and libssl. 3784 3785 *Andy Polyakov* 3786 3787 * New EC_KEY_METHOD, this replaces the older ECDSA_METHOD and ECDH_METHOD 3788 and integrates ECDSA and ECDH functionality into EC. Implementations can 3789 now redirect key generation and no longer need to convert to or from 3790 ECDSA_SIG format. 3791 3792 Note: the ecdsa.h and ecdh.h headers are now no longer needed and just 3793 include the ec.h header file instead. 3794 3795 *Steve Henson* 3796 3797 * Remove support for all 40 and 56 bit ciphers. This includes all the export 3798 ciphers who are no longer supported and drops support the ephemeral RSA key 3799 exchange. The LOW ciphers currently doesn't have any ciphers in it. 3800 3801 *Kurt Roeckx* 3802 3803 * Made EVP_MD_CTX, EVP_MD, EVP_CIPHER_CTX, EVP_CIPHER and HMAC_CTX 3804 opaque. For HMAC_CTX, the following constructors and destructors 3805 were added: 3806 3807 HMAC_CTX *HMAC_CTX_new(void); 3808 void HMAC_CTX_free(HMAC_CTX *ctx); 3809 3810 For EVP_MD and EVP_CIPHER, complete APIs to create, fill and 3811 destroy such methods has been added. See EVP_MD_meth_new(3) and 3812 EVP_CIPHER_meth_new(3) for documentation. 3813 3814 Additional changes: 3815 1) `EVP_MD_CTX_cleanup()`, `EVP_CIPHER_CTX_cleanup()` and 3816 `HMAC_CTX_cleanup()` were removed. `HMAC_CTX_reset()` and 3817 `EVP_MD_CTX_reset()` should be called instead to reinitialise 3818 an already created structure. 3819 2) For consistency with the majority of our object creators and 3820 destructors, `EVP_MD_CTX_(create|destroy)` were renamed to 3821 `EVP_MD_CTX_(new|free)`. The old names are retained as macros 3822 for deprecated builds. 3823 3824 *Richard Levitte* 3825 3826 * Added ASYNC support. Libcrypto now includes the async sub-library to enable 3827 cryptographic operations to be performed asynchronously as long as an 3828 asynchronous capable engine is used. See the ASYNC_start_job() man page for 3829 further details. Libssl has also had this capability integrated with the 3830 introduction of the new mode SSL_MODE_ASYNC and associated error 3831 SSL_ERROR_WANT_ASYNC. See the SSL_CTX_set_mode() and SSL_get_error() man 3832 pages. This work was developed in partnership with Intel Corp. 3833 3834 *Matt Caswell* 3835 3836 * SSL_{CTX_}set_ecdh_auto() has been removed and ECDH is support is 3837 always enabled now. If you want to disable the support you should 3838 exclude it using the list of supported ciphers. This also means that the 3839 "-no_ecdhe" option has been removed from s_server. 3840 3841 *Kurt Roeckx* 3842 3843 * SSL_{CTX}_set_tmp_ecdh() which can set 1 EC curve now internally calls 3844 SSL_{CTX_}set1_curves() which can set a list. 3845 3846 *Kurt Roeckx* 3847 3848 * Remove support for SSL_{CTX_}set_tmp_ecdh_callback(). You should set the 3849 curve you want to support using SSL_{CTX_}set1_curves(). 3850 3851 *Kurt Roeckx* 3852 3853 * State machine rewrite. The state machine code has been significantly 3854 refactored in order to remove much duplication of code and solve issues 3855 with the old code (see [ssl/statem/README.md](ssl/statem/README.md) for 3856 further details). This change does have some associated API changes. 3857 Notably the SSL_state() function has been removed and replaced by 3858 SSL_get_state which now returns an "OSSL_HANDSHAKE_STATE" instead of an int. 3859 SSL_set_state() has been removed altogether. The previous handshake states 3860 defined in ssl.h and ssl3.h have also been removed. 3861 3862 *Matt Caswell* 3863 3864 * All instances of the string "ssleay" in the public API were replaced 3865 with OpenSSL (case-matching; e.g., OPENSSL_VERSION for #define's) 3866 Some error codes related to internal RSA_eay API's were renamed. 3867 3868 *Rich Salz* 3869 3870 * The demo files in crypto/threads were moved to demo/threads. 3871 3872 *Rich Salz* 3873 3874 * Removed obsolete engines: 4758cca, aep, atalla, cswift, nuron, gmp, 3875 sureware and ubsec. 3876 3877 *Matt Caswell, Rich Salz* 3878 3879 * New ASN.1 embed macro. 3880 3881 New ASN.1 macro ASN1_EMBED. This is the same as ASN1_SIMPLE except the 3882 structure is not allocated: it is part of the parent. That is instead of 3883 3884 FOO *x; 3885 3886 it must be: 3887 3888 FOO x; 3889 3890 This reduces memory fragmentation and make it impossible to accidentally 3891 set a mandatory field to NULL. 3892 3893 This currently only works for some fields specifically a SEQUENCE, CHOICE, 3894 or ASN1_STRING type which is part of a parent SEQUENCE. Since it is 3895 equivalent to ASN1_SIMPLE it cannot be tagged, OPTIONAL, SET OF or 3896 SEQUENCE OF. 3897 3898 *Steve Henson* 3899 3900 * Remove EVP_CHECK_DES_KEY, a compile-time option that never compiled. 3901 3902 *Emilia Käsper* 3903 3904 * Removed DES and RC4 ciphersuites from DEFAULT. Also removed RC2 although 3905 in 1.0.2 EXPORT was already removed and the only RC2 ciphersuite is also 3906 an EXPORT one. COMPLEMENTOFDEFAULT has been updated accordingly to add 3907 DES and RC4 ciphersuites. 3908 3909 *Matt Caswell* 3910 3911 * Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs. 3912 This changes the decoding behaviour for some invalid messages, 3913 though the change is mostly in the more lenient direction, and 3914 legacy behaviour is preserved as much as possible. 3915 3916 *Emilia Käsper* 3917 3918 * Fix no-stdio build. 3919 *David Woodhouse <David.Woodhouse@intel.com> and also* 3920 *Ivan Nestlerode <ivan.nestlerode@sonos.com>* 3921 3922 * New testing framework 3923 The testing framework has been largely rewritten and is now using 3924 perl and the perl modules Test::Harness and an extended variant of 3925 Test::More called OpenSSL::Test to do its work. All test scripts in 3926 test/ have been rewritten into test recipes, and all direct calls to 3927 executables in test/Makefile have become individual recipes using the 3928 simplified testing OpenSSL::Test::Simple. 3929 3930 For documentation on our testing modules, do: 3931 3932 perldoc test/testlib/OpenSSL/Test/Simple.pm 3933 perldoc test/testlib/OpenSSL/Test.pm 3934 3935 *Richard Levitte* 3936 3937 * Revamped memory debug; only -DCRYPTO_MDEBUG and -DCRYPTO_MDEBUG_ABORT 3938 are used; the latter aborts on memory leaks (usually checked on exit). 3939 Some undocumented "set malloc, etc., hooks" functions were removed 3940 and others were changed. All are now documented. 3941 3942 *Rich Salz* 3943 3944 * In DSA_generate_parameters_ex, if the provided seed is too short, 3945 return an error 3946 3947 *Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>* 3948 3949 * Rewrite PSK to support ECDHE_PSK, DHE_PSK and RSA_PSK. Add ciphersuites 3950 from RFC4279, RFC4785, RFC5487, RFC5489. 3951 3952 Thanks to Christian J. Dietrich and Giuseppe D'Angelo for the 3953 original RSA_PSK patch. 3954 3955 *Steve Henson* 3956 3957 * Dropped support for the SSL3_FLAGS_DELAY_CLIENT_FINISHED flag. This SSLeay 3958 era flag was never set throughout the codebase (only read). Also removed 3959 SSL3_FLAGS_POP_BUFFER which was only used if 3960 SSL3_FLAGS_DELAY_CLIENT_FINISHED was also set. 3961 3962 *Matt Caswell* 3963 3964 * Changed the default name options in the "ca", "crl", "req" and "x509" 3965 to be "oneline" instead of "compat". 3966 3967 *Richard Levitte* 3968 3969 * Remove SSL_OP_TLS_BLOCK_PADDING_BUG. This is SSLeay legacy, we're 3970 not aware of clients that still exhibit this bug, and the workaround 3971 hasn't been working properly for a while. 3972 3973 *Emilia Käsper* 3974 3975 * The return type of BIO_number_read() and BIO_number_written() as well as 3976 the corresponding num_read and num_write members in the BIO structure has 3977 changed from unsigned long to uint64_t. On platforms where an unsigned 3978 long is 32 bits (e.g. Windows) these counters could overflow if >4Gb is 3979 transferred. 3980 3981 *Matt Caswell* 3982 3983 * Given the pervasive nature of TLS extensions it is inadvisable to run 3984 OpenSSL without support for them. It also means that maintaining 3985 the OPENSSL_NO_TLSEXT option within the code is very invasive (and probably 3986 not well tested). Therefore the OPENSSL_NO_TLSEXT option has been removed. 3987 3988 *Matt Caswell* 3989 3990 * Removed support for the two export grade static DH ciphersuites 3991 EXP-DH-RSA-DES-CBC-SHA and EXP-DH-DSS-DES-CBC-SHA. These two ciphersuites 3992 were newly added (along with a number of other static DH ciphersuites) to 3993 1.0.2. However the two export ones have *never* worked since they were 3994 introduced. It seems strange in any case to be adding new export 3995 ciphersuites, and given "logjam" it also does not seem correct to fix them. 3996 3997 *Matt Caswell* 3998 3999 * Version negotiation has been rewritten. In particular SSLv23_method(), 4000 SSLv23_client_method() and SSLv23_server_method() have been deprecated, 4001 and turned into macros which simply call the new preferred function names 4002 TLS_method(), TLS_client_method() and TLS_server_method(). All new code 4003 should use the new names instead. Also as part of this change the ssl23.h 4004 header file has been removed. 4005 4006 *Matt Caswell* 4007 4008 * Support for Kerberos ciphersuites in TLS (RFC2712) has been removed. This 4009 code and the associated standard is no longer considered fit-for-purpose. 4010 4011 *Matt Caswell* 4012 4013 * RT2547 was closed. When generating a private key, try to make the 4014 output file readable only by the owner. This behavior change might 4015 be noticeable when interacting with other software. 4016 4017 * Documented all exdata functions. Added CRYPTO_free_ex_index. 4018 Added a test. 4019 4020 *Rich Salz* 4021 4022 * Added HTTP GET support to the ocsp command. 4023 4024 *Rich Salz* 4025 4026 * Changed default digest for the dgst and enc commands from MD5 to 4027 sha256 4028 4029 *Rich Salz* 4030 4031 * RAND_pseudo_bytes has been deprecated. Users should use RAND_bytes instead. 4032 4033 *Matt Caswell* 4034 4035 * Added support for TLS extended master secret from 4036 draft-ietf-tls-session-hash-03.txt. Thanks for Alfredo Pironti for an 4037 initial patch which was a great help during development. 4038 4039 *Steve Henson* 4040 4041 * All libssl internal structures have been removed from the public header 4042 files, and the OPENSSL_NO_SSL_INTERN option has been removed (since it is 4043 now redundant). Users should not attempt to access internal structures 4044 directly. Instead they should use the provided API functions. 4045 4046 *Matt Caswell* 4047 4048 * config has been changed so that by default OPENSSL_NO_DEPRECATED is used. 4049 Access to deprecated functions can be re-enabled by running config with 4050 "enable-deprecated". In addition applications wishing to use deprecated 4051 functions must define OPENSSL_USE_DEPRECATED. Note that this new behaviour 4052 will, by default, disable some transitive includes that previously existed 4053 in the header files (e.g. ec.h will no longer, by default, include bn.h) 4054 4055 *Matt Caswell* 4056 4057 * Added support for OCB mode. OpenSSL has been granted a patent license 4058 compatible with the OpenSSL license for use of OCB. Details are available 4059 at <https://www.openssl.org/source/OCB-patent-grant-OpenSSL.pdf>. Support 4060 for OCB can be removed by calling config with no-ocb. 4061 4062 *Matt Caswell* 4063 4064 * SSLv2 support has been removed. It still supports receiving a SSLv2 4065 compatible client hello. 4066 4067 *Kurt Roeckx* 4068 4069 * Increased the minimal RSA keysize from 256 to 512 bits [Rich Salz], 4070 done while fixing the error code for the key-too-small case. 4071 4072 *Annie Yousar <a.yousar@informatik.hu-berlin.de>* 4073 4074 * CA.sh has been removed; use CA.pl instead. 4075 4076 *Rich Salz* 4077 4078 * Removed old DES API. 4079 4080 *Rich Salz* 4081 4082 * Remove various unsupported platforms: 4083 Sony NEWS4 4084 BEOS and BEOS_R5 4085 NeXT 4086 SUNOS 4087 MPE/iX 4088 Sinix/ReliantUNIX RM400 4089 DGUX 4090 NCR 4091 Tandem 4092 Cray 4093 16-bit platforms such as WIN16 4094 4095 *Rich Salz* 4096 4097 * Clean up OPENSSL_NO_xxx #define's 4098 - Use setbuf() and remove OPENSSL_NO_SETVBUF_IONBF 4099 - Rename OPENSSL_SYSNAME_xxx to OPENSSL_SYS_xxx 4100 - OPENSSL_NO_EC{DH,DSA} merged into OPENSSL_NO_EC 4101 - OPENSSL_NO_RIPEMD160, OPENSSL_NO_RIPEMD merged into OPENSSL_NO_RMD160 4102 - OPENSSL_NO_FP_API merged into OPENSSL_NO_STDIO 4103 - Remove OPENSSL_NO_BIO OPENSSL_NO_BUFFER OPENSSL_NO_CHAIN_VERIFY 4104 OPENSSL_NO_EVP OPENSSL_NO_FIPS_ERR OPENSSL_NO_HASH_COMP 4105 OPENSSL_NO_LHASH OPENSSL_NO_OBJECT OPENSSL_NO_SPEED OPENSSL_NO_STACK 4106 OPENSSL_NO_X509 OPENSSL_NO_X509_VERIFY 4107 - Remove MS_STATIC; it's a relic from platforms <32 bits. 4108 4109 *Rich Salz* 4110 4111 * Cleaned up dead code 4112 Remove all but one '#ifdef undef' which is to be looked at. 4113 4114 *Rich Salz* 4115 4116 * Clean up calling of xxx_free routines. 4117 Just like free(), fix most of the xxx_free routines to accept 4118 NULL. Remove the non-null checks from callers. Save much code. 4119 4120 *Rich Salz* 4121 4122 * Add secure heap for storage of private keys (when possible). 4123 Add BIO_s_secmem(), CBIGNUM, etc. 4124 Contributed by Akamai Technologies under our Corporate CLA. 4125 4126 *Rich Salz* 4127 4128 * Experimental support for a new, fast, unbiased prime candidate generator, 4129 bn_probable_prime_dh_coprime(). Not currently used by any prime generator. 4130 4131 *Felix Laurie von Massenbach <felix@erbridge.co.uk>* 4132 4133 * New output format NSS in the sess_id command line tool. This allows 4134 exporting the session id and the master key in NSS keylog format. 4135 4136 *Martin Kaiser <martin@kaiser.cx>* 4137 4138 * Harmonize version and its documentation. -f flag is used to display 4139 compilation flags. 4140 4141 *mancha <mancha1@zoho.com>* 4142 4143 * Fix eckey_priv_encode so it immediately returns an error upon a failure 4144 in i2d_ECPrivateKey. Thanks to Ted Unangst for feedback on this issue. 4145 4146 *mancha <mancha1@zoho.com>* 4147 4148 * Fix some double frees. These are not thought to be exploitable. 4149 4150 *mancha <mancha1@zoho.com>* 4151 4152 * A missing bounds check in the handling of the TLS heartbeat extension 4153 can be used to reveal up to 64k of memory to a connected client or 4154 server. 4155 4156 Thanks for Neel Mehta of Google Security for discovering this bug and to 4157 Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for 4158 preparing the fix ([CVE-2014-0160]) 4159 4160 *Adam Langley, Bodo Moeller* 4161 4162 * Fix for the attack described in the paper "Recovering OpenSSL 4163 ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack" 4164 by Yuval Yarom and Naomi Benger. Details can be obtained from: 4165 <http://eprint.iacr.org/2014/140> 4166 4167 Thanks to Yuval Yarom and Naomi Benger for discovering this 4168 flaw and to Yuval Yarom for supplying a fix ([CVE-2014-0076]) 4169 4170 *Yuval Yarom and Naomi Benger* 4171 4172 * Use algorithm specific chains in SSL_CTX_use_certificate_chain_file(): 4173 this fixes a limitation in previous versions of OpenSSL. 4174 4175 *Steve Henson* 4176 4177 * Experimental encrypt-then-mac support. 4178 4179 Experimental support for encrypt then mac from 4180 draft-gutmann-tls-encrypt-then-mac-02.txt 4181 4182 To enable it set the appropriate extension number (0x42 for the test 4183 server) using e.g. -DTLSEXT_TYPE_encrypt_then_mac=0x42 4184 4185 For non-compliant peers (i.e. just about everything) this should have no 4186 effect. 4187 4188 WARNING: EXPERIMENTAL, SUBJECT TO CHANGE. 4189 4190 *Steve Henson* 4191 4192 * Add EVP support for key wrapping algorithms, to avoid problems with 4193 existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in 4194 the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap 4195 algorithms and include tests cases. 4196 4197 *Steve Henson* 4198 4199 * Extend CMS code to support RSA-PSS signatures and RSA-OAEP for 4200 enveloped data. 4201 4202 *Steve Henson* 4203 4204 * Extended RSA OAEP support via EVP_PKEY API. Options to specify digest, 4205 MGF1 digest and OAEP label. 4206 4207 *Steve Henson* 4208 4209 * Make openssl verify return errors. 4210 4211 *Chris Palmer <palmer@google.com> and Ben Laurie* 4212 4213 * New function ASN1_TIME_diff to calculate the difference between two 4214 ASN1_TIME structures or one structure and the current time. 4215 4216 *Steve Henson* 4217 4218 * Update fips_test_suite to support multiple command line options. New 4219 test to induce all self test errors in sequence and check expected 4220 failures. 4221 4222 *Steve Henson* 4223 4224 * Add FIPS_{rsa,dsa,ecdsa}_{sign,verify} functions which digest and 4225 sign or verify all in one operation. 4226 4227 *Steve Henson* 4228 4229 * Add fips_algvs: a multicall fips utility incorporating all the algorithm 4230 test programs and fips_test_suite. Includes functionality to parse 4231 the minimal script output of fipsalgest.pl directly. 4232 4233 *Steve Henson* 4234 4235 * Add authorisation parameter to FIPS_module_mode_set(). 4236 4237 *Steve Henson* 4238 4239 * Add FIPS selftest for ECDH algorithm using P-224 and B-233 curves. 4240 4241 *Steve Henson* 4242 4243 * Use separate DRBG fields for internal and external flags. New function 4244 FIPS_drbg_health_check() to perform on demand health checking. Add 4245 generation tests to fips_test_suite with reduced health check interval to 4246 demonstrate periodic health checking. Add "nodh" option to 4247 fips_test_suite to skip very slow DH test. 4248 4249 *Steve Henson* 4250 4251 * New function FIPS_get_cipherbynid() to lookup FIPS supported ciphers 4252 based on NID. 4253 4254 *Steve Henson* 4255 4256 * More extensive health check for DRBG checking many more failure modes. 4257 New function FIPS_selftest_drbg_all() to handle every possible DRBG 4258 combination: call this in fips_test_suite. 4259 4260 *Steve Henson* 4261 4262 * Add support for canonical generation of DSA parameter 'g'. See 4263 FIPS 186-3 A.2.3. 4264 4265 * Add support for HMAC DRBG from SP800-90. Update DRBG algorithm test and 4266 POST to handle HMAC cases. 4267 4268 *Steve Henson* 4269 4270 * Add functions FIPS_module_version() and FIPS_module_version_text() 4271 to return numerical and string versions of the FIPS module number. 4272 4273 *Steve Henson* 4274 4275 * Rename FIPS_mode_set and FIPS_mode to FIPS_module_mode_set and 4276 FIPS_module_mode. FIPS_mode and FIPS_mode_set will be implemented 4277 outside the validated module in the FIPS capable OpenSSL. 4278 4279 *Steve Henson* 4280 4281 * Minor change to DRBG entropy callback semantics. In some cases 4282 there is no multiple of the block length between min_len and 4283 max_len. Allow the callback to return more than max_len bytes 4284 of entropy but discard any extra: it is the callback's responsibility 4285 to ensure that the extra data discarded does not impact the 4286 requested amount of entropy. 4287 4288 *Steve Henson* 4289 4290 * Add PRNG security strength checks to RSA, DSA and ECDSA using 4291 information in FIPS186-3, SP800-57 and SP800-131A. 4292 4293 *Steve Henson* 4294 4295 * CCM support via EVP. Interface is very similar to GCM case except we 4296 must supply all data in one chunk (i.e. no update, final) and the 4297 message length must be supplied if AAD is used. Add algorithm test 4298 support. 4299 4300 *Steve Henson* 4301 4302 * Initial version of POST overhaul. Add POST callback to allow the status 4303 of POST to be monitored and/or failures induced. Modify fips_test_suite 4304 to use callback. Always run all selftests even if one fails. 4305 4306 *Steve Henson* 4307 4308 * XTS support including algorithm test driver in the fips_gcmtest program. 4309 Note: this does increase the maximum key length from 32 to 64 bytes but 4310 there should be no binary compatibility issues as existing applications 4311 will never use XTS mode. 4312 4313 *Steve Henson* 4314 4315 * Extensive reorganisation of FIPS PRNG behaviour. Remove all dependencies 4316 to OpenSSL RAND code and replace with a tiny FIPS RAND API which also 4317 performs algorithm blocking for unapproved PRNG types. Also do not 4318 set PRNG type in FIPS_mode_set(): leave this to the application. 4319 Add default OpenSSL DRBG handling: sets up FIPS PRNG and seeds with 4320 the standard OpenSSL PRNG: set additional data to a date time vector. 4321 4322 *Steve Henson* 4323 4324 * Rename old X9.31 PRNG functions of the form `FIPS_rand*` to `FIPS_x931*`. 4325 This shouldn't present any incompatibility problems because applications 4326 shouldn't be using these directly and any that are will need to rethink 4327 anyway as the X9.31 PRNG is now deprecated by FIPS 140-2 4328 4329 *Steve Henson* 4330 4331 * Extensive self tests and health checking required by SP800-90 DRBG. 4332 Remove strength parameter from FIPS_drbg_instantiate and always 4333 instantiate at maximum supported strength. 4334 4335 *Steve Henson* 4336 4337 * Add ECDH code to fips module and fips_ecdhvs for primitives only testing. 4338 4339 *Steve Henson* 4340 4341 * New algorithm test program fips_dhvs to handle DH primitives only testing. 4342 4343 *Steve Henson* 4344 4345 * New function DH_compute_key_padded() to compute a DH key and pad with 4346 leading zeroes if needed: this complies with SP800-56A et al. 4347 4348 *Steve Henson* 4349 4350 * Initial implementation of SP800-90 DRBGs for Hash and CTR. Not used by 4351 anything, incomplete, subject to change and largely untested at present. 4352 4353 *Steve Henson* 4354 4355 * Modify fipscanisteronly build option to only build the necessary object 4356 files by filtering FIPS_EX_OBJ through a perl script in crypto/Makefile. 4357 4358 *Steve Henson* 4359 4360 * Add experimental option FIPSSYMS to give all symbols in 4361 fipscanister.o and FIPS or fips prefix. This will avoid 4362 conflicts with future versions of OpenSSL. Add perl script 4363 util/fipsas.pl to preprocess assembly language source files 4364 and rename any affected symbols. 4365 4366 *Steve Henson* 4367 4368 * Add selftest checks and algorithm block of non-fips algorithms in 4369 FIPS mode. Remove DES2 from selftests. 4370 4371 *Steve Henson* 4372 4373 * Add ECDSA code to fips module. Add tiny fips_ecdsa_check to just 4374 return internal method without any ENGINE dependencies. Add new 4375 tiny fips sign and verify functions. 4376 4377 *Steve Henson* 4378 4379 * New build option no-ec2m to disable characteristic 2 code. 4380 4381 *Steve Henson* 4382 4383 * New build option "fipscanisteronly". This only builds fipscanister.o 4384 and (currently) associated fips utilities. Uses the file Makefile.fips 4385 instead of Makefile.org as the prototype. 4386 4387 *Steve Henson* 4388 4389 * Add some FIPS mode restrictions to GCM. Add internal IV generator. 4390 Update fips_gcmtest to use IV generator. 4391 4392 *Steve Henson* 4393 4394 * Initial, experimental EVP support for AES-GCM. AAD can be input by 4395 setting output buffer to NULL. The `*Final` function must be 4396 called although it will not retrieve any additional data. The tag 4397 can be set or retrieved with a ctrl. The IV length is by default 12 4398 bytes (96 bits) but can be set to an alternative value. If the IV 4399 length exceeds the maximum IV length (currently 16 bytes) it cannot be 4400 set before the key. 4401 4402 *Steve Henson* 4403 4404 * New flag in ciphers: EVP_CIPH_FLAG_CUSTOM_CIPHER. This means the 4405 underlying do_cipher function handles all cipher semantics itself 4406 including padding and finalisation. This is useful if (for example) 4407 an ENGINE cipher handles block padding itself. The behaviour of 4408 do_cipher is subtly changed if this flag is set: the return value 4409 is the number of characters written to the output buffer (zero is 4410 no longer an error code) or a negative error code. Also if the 4411 input buffer is NULL and length 0 finalisation should be performed. 4412 4413 *Steve Henson* 4414 4415 * If a candidate issuer certificate is already part of the constructed 4416 path ignore it: new debug notification X509_V_ERR_PATH_LOOP for this case. 4417 4418 *Steve Henson* 4419 4420 * Improve forward-security support: add functions 4421 4422 void SSL_CTX_set_not_resumable_session_callback( 4423 SSL_CTX *ctx, int (*cb)(SSL *ssl, int is_forward_secure)) 4424 void SSL_set_not_resumable_session_callback( 4425 SSL *ssl, int (*cb)(SSL *ssl, int is_forward_secure)) 4426 4427 for use by SSL/TLS servers; the callback function will be called whenever a 4428 new session is created, and gets to decide whether the session may be 4429 cached to make it resumable (return 0) or not (return 1). (As by the 4430 SSL/TLS protocol specifications, the session_id sent by the server will be 4431 empty to indicate that the session is not resumable; also, the server will 4432 not generate RFC 4507 (RFC 5077) session tickets.) 4433 4434 A simple reasonable callback implementation is to return is_forward_secure. 4435 This parameter will be set to 1 or 0 depending on the ciphersuite selected 4436 by the SSL/TLS server library, indicating whether it can provide forward 4437 security. 4438 4439 *Emilia Käsper <emilia.kasper@esat.kuleuven.be> (Google)* 4440 4441 * New -verify_name option in command line utilities to set verification 4442 parameters by name. 4443 4444 *Steve Henson* 4445 4446 * Initial CMAC implementation. WARNING: EXPERIMENTAL, API MAY CHANGE. 4447 Add CMAC pkey methods. 4448 4449 *Steve Henson* 4450 4451 * Experimental renegotiation in s_server -www mode. If the client 4452 browses /reneg connection is renegotiated. If /renegcert it is 4453 renegotiated requesting a certificate. 4454 4455 *Steve Henson* 4456 4457 * Add an "external" session cache for debugging purposes to s_server. This 4458 should help trace issues which normally are only apparent in deployed 4459 multi-process servers. 4460 4461 *Steve Henson* 4462 4463 * Extensive audit of libcrypto with DEBUG_UNUSED. Fix many cases where 4464 return value is ignored. NB. The functions RAND_add(), RAND_seed(), 4465 BIO_set_cipher() and some obscure PEM functions were changed so they 4466 can now return an error. The RAND changes required a change to the 4467 RAND_METHOD structure. 4468 4469 *Steve Henson* 4470 4471 * New macro `__owur` for "OpenSSL Warn Unused Result". This makes use of 4472 a gcc attribute to warn if the result of a function is ignored. This 4473 is enable if DEBUG_UNUSED is set. Add to several functions in evp.h 4474 whose return value is often ignored. 4475 4476 *Steve Henson* 4477 4478 * New -noct, -requestct, -requirect and -ctlogfile options for s_client. 4479 These allow SCTs (signed certificate timestamps) to be requested and 4480 validated when establishing a connection. 4481 4482 *Rob Percival <robpercival@google.com>* 4483 4484OpenSSL 1.0.2 4485------------- 4486 4487### Changes between 1.0.2s and 1.0.2t [10 Sep 2019] 4488 4489 * For built-in EC curves, ensure an EC_GROUP built from the curve name is 4490 used even when parsing explicit parameters, when loading a encoded key 4491 or calling `EC_GROUP_new_from_ecpkparameters()`/ 4492 `EC_GROUP_new_from_ecparameters()`. 4493 This prevents bypass of security hardening and performance gains, 4494 especially for curves with specialized EC_METHODs. 4495 By default, if a key encoded with explicit parameters is loaded and later 4496 encoded, the output is still encoded with explicit parameters, even if 4497 internally a "named" EC_GROUP is used for computation. 4498 4499 *Nicola Tuveri* 4500 4501 * Compute ECC cofactors if not provided during EC_GROUP construction. Before 4502 this change, EC_GROUP_set_generator would accept order and/or cofactor as 4503 NULL. After this change, only the cofactor parameter can be NULL. It also 4504 does some minimal sanity checks on the passed order. 4505 ([CVE-2019-1547]) 4506 4507 *Billy Bob Brumley* 4508 4509 * Fixed a padding oracle in PKCS7_dataDecode and CMS_decrypt_set1_pkey. 4510 An attack is simple, if the first CMS_recipientInfo is valid but the 4511 second CMS_recipientInfo is chosen ciphertext. If the second 4512 recipientInfo decodes to PKCS #1 v1.5 form plaintext, the correct 4513 encryption key will be replaced by garbage, and the message cannot be 4514 decoded, but if the RSA decryption fails, the correct encryption key is 4515 used and the recipient will not notice the attack. 4516 As a work around for this potential attack the length of the decrypted 4517 key must be equal to the cipher default key length, in case the 4518 certifiate is not given and all recipientInfo are tried out. 4519 The old behaviour can be re-enabled in the CMS code by setting the 4520 CMS_DEBUG_DECRYPT flag. 4521 ([CVE-2019-1563]) 4522 4523 *Bernd Edlinger* 4524 4525 * Document issue with installation paths in diverse Windows builds 4526 4527 '/usr/local/ssl' is an unsafe prefix for location to install OpenSSL 4528 binaries and run-time config file. 4529 ([CVE-2019-1552]) 4530 4531 *Richard Levitte* 4532 4533### Changes between 1.0.2r and 1.0.2s [28 May 2019] 4534 4535 * Change the default RSA, DSA and DH size to 2048 bit instead of 1024. 4536 This changes the size when using the `genpkey` command when no size is given. 4537 It fixes an omission in earlier changes that changed all RSA, DSA and DH 4538 generation commands to use 2048 bits by default. 4539 4540 *Kurt Roeckx* 4541 4542 * Add FIPS support for Android Arm 64-bit 4543 4544 Support for Android Arm 64-bit was added to the OpenSSL FIPS Object 4545 Module in Version 2.0.10. For some reason, the corresponding target 4546 'android64-aarch64' was missing OpenSSL 1.0.2, whence it could not be 4547 built with FIPS support on Android Arm 64-bit. This omission has been 4548 fixed. 4549 4550 *Matthias St. Pierre* 4551 4552### Changes between 1.0.2q and 1.0.2r [26 Feb 2019] 4553 4554 * 0-byte record padding oracle 4555 4556 If an application encounters a fatal protocol error and then calls 4557 SSL_shutdown() twice (once to send a close_notify, and once to receive one) 4558 then OpenSSL can respond differently to the calling application if a 0 byte 4559 record is received with invalid padding compared to if a 0 byte record is 4560 received with an invalid MAC. If the application then behaves differently 4561 based on that in a way that is detectable to the remote peer, then this 4562 amounts to a padding oracle that could be used to decrypt data. 4563 4564 In order for this to be exploitable "non-stitched" ciphersuites must be in 4565 use. Stitched ciphersuites are optimised implementations of certain 4566 commonly used ciphersuites. Also the application must call SSL_shutdown() 4567 twice even if a protocol error has occurred (applications should not do 4568 this but some do anyway). 4569 4570 This issue was discovered by Juraj Somorovsky, Robert Merget and Nimrod 4571 Aviram, with additional investigation by Steven Collison and Andrew 4572 Hourselt. It was reported to OpenSSL on 10th December 2018. 4573 ([CVE-2019-1559]) 4574 4575 *Matt Caswell* 4576 4577 * Move strictness check from EVP_PKEY_asn1_new() to EVP_PKEY_asn1_add0(). 4578 4579 *Richard Levitte* 4580 4581### Changes between 1.0.2p and 1.0.2q [20 Nov 2018] 4582 4583 * Microarchitecture timing vulnerability in ECC scalar multiplication 4584 4585 OpenSSL ECC scalar multiplication, used in e.g. ECDSA and ECDH, has been 4586 shown to be vulnerable to a microarchitecture timing side channel attack. 4587 An attacker with sufficient access to mount local timing attacks during 4588 ECDSA signature generation could recover the private key. 4589 4590 This issue was reported to OpenSSL on 26th October 2018 by Alejandro 4591 Cabrera Aldaya, Billy Brumley, Sohaib ul Hassan, Cesar Pereida Garcia and 4592 Nicola Tuveri. 4593 ([CVE-2018-5407]) 4594 4595 *Billy Brumley* 4596 4597 * Timing vulnerability in DSA signature generation 4598 4599 The OpenSSL DSA signature algorithm has been shown to be vulnerable to a 4600 timing side channel attack. An attacker could use variations in the signing 4601 algorithm to recover the private key. 4602 4603 This issue was reported to OpenSSL on 16th October 2018 by Samuel Weiser. 4604 ([CVE-2018-0734]) 4605 4606 *Paul Dale* 4607 4608 * Resolve a compatibility issue in EC_GROUP handling with the FIPS Object 4609 Module, accidentally introduced while backporting security fixes from the 4610 development branch and hindering the use of ECC in FIPS mode. 4611 4612 *Nicola Tuveri* 4613 4614### Changes between 1.0.2o and 1.0.2p [14 Aug 2018] 4615 4616 * Client DoS due to large DH parameter 4617 4618 During key agreement in a TLS handshake using a DH(E) based ciphersuite a 4619 malicious server can send a very large prime value to the client. This will 4620 cause the client to spend an unreasonably long period of time generating a 4621 key for this prime resulting in a hang until the client has finished. This 4622 could be exploited in a Denial Of Service attack. 4623 4624 This issue was reported to OpenSSL on 5th June 2018 by Guido Vranken 4625 ([CVE-2018-0732]) 4626 4627 *Guido Vranken* 4628 4629 * Cache timing vulnerability in RSA Key Generation 4630 4631 The OpenSSL RSA Key generation algorithm has been shown to be vulnerable to 4632 a cache timing side channel attack. An attacker with sufficient access to 4633 mount cache timing attacks during the RSA key generation process could 4634 recover the private key. 4635 4636 This issue was reported to OpenSSL on 4th April 2018 by Alejandro Cabrera 4637 Aldaya, Billy Brumley, Cesar Pereida Garcia and Luis Manuel Alvarez Tapia. 4638 ([CVE-2018-0737]) 4639 4640 *Billy Brumley* 4641 4642 * Make EVP_PKEY_asn1_new() a bit stricter about its input. A NULL pem_str 4643 parameter is no longer accepted, as it leads to a corrupt table. NULL 4644 pem_str is reserved for alias entries only. 4645 4646 *Richard Levitte* 4647 4648 * Revert blinding in ECDSA sign and instead make problematic addition 4649 length-invariant. Switch even to fixed-length Montgomery multiplication. 4650 4651 *Andy Polyakov* 4652 4653 * Change generating and checking of primes so that the error rate of not 4654 being prime depends on the intended use based on the size of the input. 4655 For larger primes this will result in more rounds of Miller-Rabin. 4656 The maximal error rate for primes with more than 1080 bits is lowered 4657 to 2^-128. 4658 4659 *Kurt Roeckx, Annie Yousar* 4660 4661 * Increase the number of Miller-Rabin rounds for DSA key generating to 64. 4662 4663 *Kurt Roeckx* 4664 4665 * Add blinding to ECDSA and DSA signatures to protect against side channel 4666 attacks discovered by Keegan Ryan (NCC Group). 4667 4668 *Matt Caswell* 4669 4670 * When unlocking a pass phrase protected PEM file or PKCS#8 container, we 4671 now allow empty (zero character) pass phrases. 4672 4673 *Richard Levitte* 4674 4675 * Certificate time validation (X509_cmp_time) enforces stricter 4676 compliance with RFC 5280. Fractional seconds and timezone offsets 4677 are no longer allowed. 4678 4679 *Emilia Käsper* 4680 4681### Changes between 1.0.2n and 1.0.2o [27 Mar 2018] 4682 4683 * Constructed ASN.1 types with a recursive definition could exceed the stack 4684 4685 Constructed ASN.1 types with a recursive definition (such as can be found 4686 in PKCS7) could eventually exceed the stack given malicious input with 4687 excessive recursion. This could result in a Denial Of Service attack. There 4688 are no such structures used within SSL/TLS that come from untrusted sources 4689 so this is considered safe. 4690 4691 This issue was reported to OpenSSL on 4th January 2018 by the OSS-fuzz 4692 project. 4693 ([CVE-2018-0739]) 4694 4695 *Matt Caswell* 4696 4697### Changes between 1.0.2m and 1.0.2n [7 Dec 2017] 4698 4699 * Read/write after SSL object in error state 4700 4701 OpenSSL 1.0.2 (starting from version 1.0.2b) introduced an "error state" 4702 mechanism. The intent was that if a fatal error occurred during a handshake 4703 then OpenSSL would move into the error state and would immediately fail if 4704 you attempted to continue the handshake. This works as designed for the 4705 explicit handshake functions (SSL_do_handshake(), SSL_accept() and 4706 SSL_connect()), however due to a bug it does not work correctly if 4707 SSL_read() or SSL_write() is called directly. In that scenario, if the 4708 handshake fails then a fatal error will be returned in the initial function 4709 call. If SSL_read()/SSL_write() is subsequently called by the application 4710 for the same SSL object then it will succeed and the data is passed without 4711 being decrypted/encrypted directly from the SSL/TLS record layer. 4712 4713 In order to exploit this issue an application bug would have to be present 4714 that resulted in a call to SSL_read()/SSL_write() being issued after having 4715 already received a fatal error. 4716 4717 This issue was reported to OpenSSL by David Benjamin (Google). 4718 ([CVE-2017-3737]) 4719 4720 *Matt Caswell* 4721 4722 * rsaz_1024_mul_avx2 overflow bug on x86_64 4723 4724 There is an overflow bug in the AVX2 Montgomery multiplication procedure 4725 used in exponentiation with 1024-bit moduli. No EC algorithms are affected. 4726 Analysis suggests that attacks against RSA and DSA as a result of this 4727 defect would be very difficult to perform and are not believed likely. 4728 Attacks against DH1024 are considered just feasible, because most of the 4729 work necessary to deduce information about a private key may be performed 4730 offline. The amount of resources required for such an attack would be 4731 significant. However, for an attack on TLS to be meaningful, the server 4732 would have to share the DH1024 private key among multiple clients, which is 4733 no longer an option since CVE-2016-0701. 4734 4735 This only affects processors that support the AVX2 but not ADX extensions 4736 like Intel Haswell (4th generation). 4737 4738 This issue was reported to OpenSSL by David Benjamin (Google). The issue 4739 was originally found via the OSS-Fuzz project. 4740 ([CVE-2017-3738]) 4741 4742 *Andy Polyakov* 4743 4744### Changes between 1.0.2l and 1.0.2m [2 Nov 2017] 4745 4746 * bn_sqrx8x_internal carry bug on x86_64 4747 4748 There is a carry propagating bug in the x86_64 Montgomery squaring 4749 procedure. No EC algorithms are affected. Analysis suggests that attacks 4750 against RSA and DSA as a result of this defect would be very difficult to 4751 perform and are not believed likely. Attacks against DH are considered just 4752 feasible (although very difficult) because most of the work necessary to 4753 deduce information about a private key may be performed offline. The amount 4754 of resources required for such an attack would be very significant and 4755 likely only accessible to a limited number of attackers. An attacker would 4756 additionally need online access to an unpatched system using the target 4757 private key in a scenario with persistent DH parameters and a private 4758 key that is shared between multiple clients. 4759 4760 This only affects processors that support the BMI1, BMI2 and ADX extensions 4761 like Intel Broadwell (5th generation) and later or AMD Ryzen. 4762 4763 This issue was reported to OpenSSL by the OSS-Fuzz project. 4764 ([CVE-2017-3736]) 4765 4766 *Andy Polyakov* 4767 4768 * Malformed X.509 IPAddressFamily could cause OOB read 4769 4770 If an X.509 certificate has a malformed IPAddressFamily extension, 4771 OpenSSL could do a one-byte buffer overread. The most likely result 4772 would be an erroneous display of the certificate in text format. 4773 4774 This issue was reported to OpenSSL by the OSS-Fuzz project. 4775 4776 *Rich Salz* 4777 4778### Changes between 1.0.2k and 1.0.2l [25 May 2017] 4779 4780 * Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target 4781 platform rather than 'mingw'. 4782 4783 *Richard Levitte* 4784 4785### Changes between 1.0.2j and 1.0.2k [26 Jan 2017] 4786 4787 * Truncated packet could crash via OOB read 4788 4789 If one side of an SSL/TLS path is running on a 32-bit host and a specific 4790 cipher is being used, then a truncated packet can cause that host to 4791 perform an out-of-bounds read, usually resulting in a crash. 4792 4793 This issue was reported to OpenSSL by Robert Święcki of Google. 4794 ([CVE-2017-3731]) 4795 4796 *Andy Polyakov* 4797 4798 * BN_mod_exp may produce incorrect results on x86_64 4799 4800 There is a carry propagating bug in the x86_64 Montgomery squaring 4801 procedure. No EC algorithms are affected. Analysis suggests that attacks 4802 against RSA and DSA as a result of this defect would be very difficult to 4803 perform and are not believed likely. Attacks against DH are considered just 4804 feasible (although very difficult) because most of the work necessary to 4805 deduce information about a private key may be performed offline. The amount 4806 of resources required for such an attack would be very significant and 4807 likely only accessible to a limited number of attackers. An attacker would 4808 additionally need online access to an unpatched system using the target 4809 private key in a scenario with persistent DH parameters and a private 4810 key that is shared between multiple clients. For example this can occur by 4811 default in OpenSSL DHE based SSL/TLS ciphersuites. Note: This issue is very 4812 similar to CVE-2015-3193 but must be treated as a separate problem. 4813 4814 This issue was reported to OpenSSL by the OSS-Fuzz project. 4815 ([CVE-2017-3732]) 4816 4817 *Andy Polyakov* 4818 4819 * Montgomery multiplication may produce incorrect results 4820 4821 There is a carry propagating bug in the Broadwell-specific Montgomery 4822 multiplication procedure that handles input lengths divisible by, but 4823 longer than 256 bits. Analysis suggests that attacks against RSA, DSA 4824 and DH private keys are impossible. This is because the subroutine in 4825 question is not used in operations with the private key itself and an input 4826 of the attacker's direct choice. Otherwise the bug can manifest itself as 4827 transient authentication and key negotiation failures or reproducible 4828 erroneous outcome of public-key operations with specially crafted input. 4829 Among EC algorithms only Brainpool P-512 curves are affected and one 4830 presumably can attack ECDH key negotiation. Impact was not analyzed in 4831 detail, because pre-requisites for attack are considered unlikely. Namely 4832 multiple clients have to choose the curve in question and the server has to 4833 share the private key among them, neither of which is default behaviour. 4834 Even then only clients that chose the curve will be affected. 4835 4836 This issue was publicly reported as transient failures and was not 4837 initially recognized as a security issue. Thanks to Richard Morgan for 4838 providing reproducible case. 4839 ([CVE-2016-7055]) 4840 4841 *Andy Polyakov* 4842 4843 * OpenSSL now fails if it receives an unrecognised record type in TLS1.0 4844 or TLS1.1. Previously this only happened in SSLv3 and TLS1.2. This is to 4845 prevent issues where no progress is being made and the peer continually 4846 sends unrecognised record types, using up resources processing them. 4847 4848 *Matt Caswell* 4849 4850### Changes between 1.0.2i and 1.0.2j [26 Sep 2016] 4851 4852 * Missing CRL sanity check 4853 4854 A bug fix which included a CRL sanity check was added to OpenSSL 1.1.0 4855 but was omitted from OpenSSL 1.0.2i. As a result any attempt to use 4856 CRLs in OpenSSL 1.0.2i will crash with a null pointer exception. 4857 4858 This issue only affects the OpenSSL 1.0.2i 4859 ([CVE-2016-7052]) 4860 4861 *Matt Caswell* 4862 4863### Changes between 1.0.2h and 1.0.2i [22 Sep 2016] 4864 4865 * OCSP Status Request extension unbounded memory growth 4866 4867 A malicious client can send an excessively large OCSP Status Request 4868 extension. If that client continually requests renegotiation, sending a 4869 large OCSP Status Request extension each time, then there will be unbounded 4870 memory growth on the server. This will eventually lead to a Denial Of 4871 Service attack through memory exhaustion. Servers with a default 4872 configuration are vulnerable even if they do not support OCSP. Builds using 4873 the "no-ocsp" build time option are not affected. 4874 4875 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 4876 ([CVE-2016-6304]) 4877 4878 *Matt Caswell* 4879 4880 * In order to mitigate the SWEET32 attack, the DES ciphers were moved from 4881 HIGH to MEDIUM. 4882 4883 This issue was reported to OpenSSL Karthikeyan Bhargavan and Gaetan 4884 Leurent (INRIA) 4885 ([CVE-2016-2183]) 4886 4887 *Rich Salz* 4888 4889 * OOB write in MDC2_Update() 4890 4891 An overflow can occur in MDC2_Update() either if called directly or 4892 through the EVP_DigestUpdate() function using MDC2. If an attacker 4893 is able to supply very large amounts of input data after a previous 4894 call to EVP_EncryptUpdate() with a partial block then a length check 4895 can overflow resulting in a heap corruption. 4896 4897 The amount of data needed is comparable to SIZE_MAX which is impractical 4898 on most platforms. 4899 4900 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 4901 ([CVE-2016-6303]) 4902 4903 *Stephen Henson* 4904 4905 * Malformed SHA512 ticket DoS 4906 4907 If a server uses SHA512 for TLS session ticket HMAC it is vulnerable to a 4908 DoS attack where a malformed ticket will result in an OOB read which will 4909 ultimately crash. 4910 4911 The use of SHA512 in TLS session tickets is comparatively rare as it requires 4912 a custom server callback and ticket lookup mechanism. 4913 4914 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 4915 ([CVE-2016-6302]) 4916 4917 *Stephen Henson* 4918 4919 * OOB write in BN_bn2dec() 4920 4921 The function BN_bn2dec() does not check the return value of BN_div_word(). 4922 This can cause an OOB write if an application uses this function with an 4923 overly large BIGNUM. This could be a problem if an overly large certificate 4924 or CRL is printed out from an untrusted source. TLS is not affected because 4925 record limits will reject an oversized certificate before it is parsed. 4926 4927 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 4928 ([CVE-2016-2182]) 4929 4930 *Stephen Henson* 4931 4932 * OOB read in TS_OBJ_print_bio() 4933 4934 The function TS_OBJ_print_bio() misuses OBJ_obj2txt(): the return value is 4935 the total length the OID text representation would use and not the amount 4936 of data written. This will result in OOB reads when large OIDs are 4937 presented. 4938 4939 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 4940 ([CVE-2016-2180]) 4941 4942 *Stephen Henson* 4943 4944 * Pointer arithmetic undefined behaviour 4945 4946 Avoid some undefined pointer arithmetic 4947 4948 A common idiom in the codebase is to check limits in the following manner: 4949 "p + len > limit" 4950 4951 Where "p" points to some malloc'd data of SIZE bytes and 4952 limit == p + SIZE 4953 4954 "len" here could be from some externally supplied data (e.g. from a TLS 4955 message). 4956 4957 The rules of C pointer arithmetic are such that "p + len" is only well 4958 defined where len <= SIZE. Therefore the above idiom is actually 4959 undefined behaviour. 4960 4961 For example this could cause problems if some malloc implementation 4962 provides an address for "p" such that "p + len" actually overflows for 4963 values of len that are too big and therefore p + len < limit. 4964 4965 This issue was reported to OpenSSL by Guido Vranken 4966 ([CVE-2016-2177]) 4967 4968 *Matt Caswell* 4969 4970 * Constant time flag not preserved in DSA signing 4971 4972 Operations in the DSA signing algorithm should run in constant time in 4973 order to avoid side channel attacks. A flaw in the OpenSSL DSA 4974 implementation means that a non-constant time codepath is followed for 4975 certain operations. This has been demonstrated through a cache-timing 4976 attack to be sufficient for an attacker to recover the private DSA key. 4977 4978 This issue was reported by César Pereida (Aalto University), Billy Brumley 4979 (Tampere University of Technology), and Yuval Yarom (The University of 4980 Adelaide and NICTA). 4981 ([CVE-2016-2178]) 4982 4983 *César Pereida* 4984 4985 * DTLS buffered message DoS 4986 4987 In a DTLS connection where handshake messages are delivered out-of-order 4988 those messages that OpenSSL is not yet ready to process will be buffered 4989 for later use. Under certain circumstances, a flaw in the logic means that 4990 those messages do not get removed from the buffer even though the handshake 4991 has been completed. An attacker could force up to approx. 15 messages to 4992 remain in the buffer when they are no longer required. These messages will 4993 be cleared when the DTLS connection is closed. The default maximum size for 4994 a message is 100k. Therefore the attacker could force an additional 1500k 4995 to be consumed per connection. By opening many simulataneous connections an 4996 attacker could cause a DoS attack through memory exhaustion. 4997 4998 This issue was reported to OpenSSL by Quan Luo. 4999 ([CVE-2016-2179]) 5000 5001 *Matt Caswell* 5002 5003 * DTLS replay protection DoS 5004 5005 A flaw in the DTLS replay attack protection mechanism means that records 5006 that arrive for future epochs update the replay protection "window" before 5007 the MAC for the record has been validated. This could be exploited by an 5008 attacker by sending a record for the next epoch (which does not have to 5009 decrypt or have a valid MAC), with a very large sequence number. This means 5010 that all subsequent legitimate packets are dropped causing a denial of 5011 service for a specific DTLS connection. 5012 5013 This issue was reported to OpenSSL by the OCAP audit team. 5014 ([CVE-2016-2181]) 5015 5016 *Matt Caswell* 5017 5018 * Certificate message OOB reads 5019 5020 In OpenSSL 1.0.2 and earlier some missing message length checks can result 5021 in OOB reads of up to 2 bytes beyond an allocated buffer. There is a 5022 theoretical DoS risk but this has not been observed in practice on common 5023 platforms. 5024 5025 The messages affected are client certificate, client certificate request 5026 and server certificate. As a result the attack can only be performed 5027 against a client or a server which enables client authentication. 5028 5029 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 5030 ([CVE-2016-6306]) 5031 5032 *Stephen Henson* 5033 5034### Changes between 1.0.2g and 1.0.2h [3 May 2016] 5035 5036 * Prevent padding oracle in AES-NI CBC MAC check 5037 5038 A MITM attacker can use a padding oracle attack to decrypt traffic 5039 when the connection uses an AES CBC cipher and the server support 5040 AES-NI. 5041 5042 This issue was introduced as part of the fix for Lucky 13 padding 5043 attack ([CVE-2013-0169]). The padding check was rewritten to be in 5044 constant time by making sure that always the same bytes are read and 5045 compared against either the MAC or padding bytes. But it no longer 5046 checked that there was enough data to have both the MAC and padding 5047 bytes. 5048 5049 This issue was reported by Juraj Somorovsky using TLS-Attacker. 5050 5051 *Kurt Roeckx* 5052 5053 * Fix EVP_EncodeUpdate overflow 5054 5055 An overflow can occur in the EVP_EncodeUpdate() function which is used for 5056 Base64 encoding of binary data. If an attacker is able to supply very large 5057 amounts of input data then a length check can overflow resulting in a heap 5058 corruption. 5059 5060 Internally to OpenSSL the EVP_EncodeUpdate() function is primarily used by 5061 the `PEM_write_bio*` family of functions. These are mainly used within the 5062 OpenSSL command line applications, so any application which processes data 5063 from an untrusted source and outputs it as a PEM file should be considered 5064 vulnerable to this issue. User applications that call these APIs directly 5065 with large amounts of untrusted data may also be vulnerable. 5066 5067 This issue was reported by Guido Vranken. 5068 ([CVE-2016-2105]) 5069 5070 *Matt Caswell* 5071 5072 * Fix EVP_EncryptUpdate overflow 5073 5074 An overflow can occur in the EVP_EncryptUpdate() function. If an attacker 5075 is able to supply very large amounts of input data after a previous call to 5076 EVP_EncryptUpdate() with a partial block then a length check can overflow 5077 resulting in a heap corruption. Following an analysis of all OpenSSL 5078 internal usage of the EVP_EncryptUpdate() function all usage is one of two 5079 forms. The first form is where the EVP_EncryptUpdate() call is known to be 5080 the first called function after an EVP_EncryptInit(), and therefore that 5081 specific call must be safe. The second form is where the length passed to 5082 EVP_EncryptUpdate() can be seen from the code to be some small value and 5083 therefore there is no possibility of an overflow. Since all instances are 5084 one of these two forms, it is believed that there can be no overflows in 5085 internal code due to this problem. It should be noted that 5086 EVP_DecryptUpdate() can call EVP_EncryptUpdate() in certain code paths. 5087 Also EVP_CipherUpdate() is a synonym for EVP_EncryptUpdate(). All instances 5088 of these calls have also been analysed too and it is believed there are no 5089 instances in internal usage where an overflow could occur. 5090 5091 This issue was reported by Guido Vranken. 5092 ([CVE-2016-2106]) 5093 5094 *Matt Caswell* 5095 5096 * Prevent ASN.1 BIO excessive memory allocation 5097 5098 When ASN.1 data is read from a BIO using functions such as d2i_CMS_bio() 5099 a short invalid encoding can cause allocation of large amounts of memory 5100 potentially consuming excessive resources or exhausting memory. 5101 5102 Any application parsing untrusted data through d2i BIO functions is 5103 affected. The memory based functions such as d2i_X509() are *not* affected. 5104 Since the memory based functions are used by the TLS library, TLS 5105 applications are not affected. 5106 5107 This issue was reported by Brian Carpenter. 5108 ([CVE-2016-2109]) 5109 5110 *Stephen Henson* 5111 5112 * EBCDIC overread 5113 5114 ASN1 Strings that are over 1024 bytes can cause an overread in applications 5115 using the X509_NAME_oneline() function on EBCDIC systems. This could result 5116 in arbitrary stack data being returned in the buffer. 5117 5118 This issue was reported by Guido Vranken. 5119 ([CVE-2016-2176]) 5120 5121 *Matt Caswell* 5122 5123 * Modify behavior of ALPN to invoke callback after SNI/servername 5124 callback, such that updates to the SSL_CTX affect ALPN. 5125 5126 *Todd Short* 5127 5128 * Remove LOW from the DEFAULT cipher list. This removes singles DES from the 5129 default. 5130 5131 *Kurt Roeckx* 5132 5133 * Only remove the SSLv2 methods with the no-ssl2-method option. When the 5134 methods are enabled and ssl2 is disabled the methods return NULL. 5135 5136 *Kurt Roeckx* 5137 5138### Changes between 1.0.2f and 1.0.2g [1 Mar 2016] 5139 5140* Disable weak ciphers in SSLv3 and up in default builds of OpenSSL. 5141 Builds that are not configured with "enable-weak-ssl-ciphers" will not 5142 provide any "EXPORT" or "LOW" strength ciphers. 5143 5144 *Viktor Dukhovni* 5145 5146* Disable SSLv2 default build, default negotiation and weak ciphers. SSLv2 5147 is by default disabled at build-time. Builds that are not configured with 5148 "enable-ssl2" will not support SSLv2. Even if "enable-ssl2" is used, 5149 users who want to negotiate SSLv2 via the version-flexible SSLv23_method() 5150 will need to explicitly call either of: 5151 5152 SSL_CTX_clear_options(ctx, SSL_OP_NO_SSLv2); 5153 or 5154 SSL_clear_options(ssl, SSL_OP_NO_SSLv2); 5155 5156 as appropriate. Even if either of those is used, or the application 5157 explicitly uses the version-specific SSLv2_method() or its client and 5158 server variants, SSLv2 ciphers vulnerable to exhaustive search key 5159 recovery have been removed. Specifically, the SSLv2 40-bit EXPORT 5160 ciphers, and SSLv2 56-bit DES are no longer available. 5161 ([CVE-2016-0800]) 5162 5163 *Viktor Dukhovni* 5164 5165 * Fix a double-free in DSA code 5166 5167 A double free bug was discovered when OpenSSL parses malformed DSA private 5168 keys and could lead to a DoS attack or memory corruption for applications 5169 that receive DSA private keys from untrusted sources. This scenario is 5170 considered rare. 5171 5172 This issue was reported to OpenSSL by Adam Langley(Google/BoringSSL) using 5173 libFuzzer. 5174 ([CVE-2016-0705]) 5175 5176 *Stephen Henson* 5177 5178 * Disable SRP fake user seed to address a server memory leak. 5179 5180 Add a new method SRP_VBASE_get1_by_user that handles the seed properly. 5181 5182 SRP_VBASE_get_by_user had inconsistent memory management behaviour. 5183 In order to fix an unavoidable memory leak, SRP_VBASE_get_by_user 5184 was changed to ignore the "fake user" SRP seed, even if the seed 5185 is configured. 5186 5187 Users should use SRP_VBASE_get1_by_user instead. Note that in 5188 SRP_VBASE_get1_by_user, caller must free the returned value. Note 5189 also that even though configuring the SRP seed attempts to hide 5190 invalid usernames by continuing the handshake with fake 5191 credentials, this behaviour is not constant time and no strong 5192 guarantees are made that the handshake is indistinguishable from 5193 that of a valid user. 5194 ([CVE-2016-0798]) 5195 5196 *Emilia Käsper* 5197 5198 * Fix BN_hex2bn/BN_dec2bn NULL pointer deref/heap corruption 5199 5200 In the BN_hex2bn function the number of hex digits is calculated using an 5201 int value `i`. Later `bn_expand` is called with a value of `i * 4`. For 5202 large values of `i` this can result in `bn_expand` not allocating any 5203 memory because `i * 4` is negative. This can leave the internal BIGNUM data 5204 field as NULL leading to a subsequent NULL ptr deref. For very large values 5205 of `i`, the calculation `i * 4` could be a positive value smaller than `i`. 5206 In this case memory is allocated to the internal BIGNUM data field, but it 5207 is insufficiently sized leading to heap corruption. A similar issue exists 5208 in BN_dec2bn. This could have security consequences if BN_hex2bn/BN_dec2bn 5209 is ever called by user applications with very large untrusted hex/dec data. 5210 This is anticipated to be a rare occurrence. 5211 5212 All OpenSSL internal usage of these functions use data that is not expected 5213 to be untrusted, e.g. config file data or application command line 5214 arguments. If user developed applications generate config file data based 5215 on untrusted data then it is possible that this could also lead to security 5216 consequences. This is also anticipated to be rare. 5217 5218 This issue was reported to OpenSSL by Guido Vranken. 5219 ([CVE-2016-0797]) 5220 5221 *Matt Caswell* 5222 5223 * Fix memory issues in `BIO_*printf` functions 5224 5225 The internal `fmtstr` function used in processing a "%s" format string in 5226 the `BIO_*printf` functions could overflow while calculating the length of a 5227 string and cause an OOB read when printing very long strings. 5228 5229 Additionally the internal `doapr_outch` function can attempt to write to an 5230 OOB memory location (at an offset from the NULL pointer) in the event of a 5231 memory allocation failure. In 1.0.2 and below this could be caused where 5232 the size of a buffer to be allocated is greater than INT_MAX. E.g. this 5233 could be in processing a very long "%s" format string. Memory leaks can 5234 also occur. 5235 5236 The first issue may mask the second issue dependent on compiler behaviour. 5237 These problems could enable attacks where large amounts of untrusted data 5238 is passed to the `BIO_*printf` functions. If applications use these functions 5239 in this way then they could be vulnerable. OpenSSL itself uses these 5240 functions when printing out human-readable dumps of ASN.1 data. Therefore 5241 applications that print this data could be vulnerable if the data is from 5242 untrusted sources. OpenSSL command line applications could also be 5243 vulnerable where they print out ASN.1 data, or if untrusted data is passed 5244 as command line arguments. 5245 5246 Libssl is not considered directly vulnerable. Additionally certificates etc 5247 received via remote connections via libssl are also unlikely to be able to 5248 trigger these issues because of message size limits enforced within libssl. 5249 5250 This issue was reported to OpenSSL Guido Vranken. 5251 ([CVE-2016-0799]) 5252 5253 *Matt Caswell* 5254 5255 * Side channel attack on modular exponentiation 5256 5257 A side-channel attack was found which makes use of cache-bank conflicts on 5258 the Intel Sandy-Bridge microarchitecture which could lead to the recovery 5259 of RSA keys. The ability to exploit this issue is limited as it relies on 5260 an attacker who has control of code in a thread running on the same 5261 hyper-threaded core as the victim thread which is performing decryptions. 5262 5263 This issue was reported to OpenSSL by Yuval Yarom, The University of 5264 Adelaide and NICTA, Daniel Genkin, Technion and Tel Aviv University, and 5265 Nadia Heninger, University of Pennsylvania with more information at 5266 <http://cachebleed.info>. 5267 ([CVE-2016-0702]) 5268 5269 *Andy Polyakov* 5270 5271 * Change the `req` command to generate a 2048-bit RSA/DSA key by default, 5272 if no keysize is specified with default_bits. This fixes an 5273 omission in an earlier change that changed all RSA/DSA key generation 5274 commands to use 2048 bits by default. 5275 5276 *Emilia Käsper* 5277 5278### Changes between 1.0.2e and 1.0.2f [28 Jan 2016] 5279 5280 * DH small subgroups 5281 5282 Historically OpenSSL only ever generated DH parameters based on "safe" 5283 primes. More recently (in version 1.0.2) support was provided for 5284 generating X9.42 style parameter files such as those required for RFC 5114 5285 support. The primes used in such files may not be "safe". Where an 5286 application is using DH configured with parameters based on primes that are 5287 not "safe" then an attacker could use this fact to find a peer's private 5288 DH exponent. This attack requires that the attacker complete multiple 5289 handshakes in which the peer uses the same private DH exponent. For example 5290 this could be used to discover a TLS server's private DH exponent if it's 5291 reusing the private DH exponent or it's using a static DH ciphersuite. 5292 5293 OpenSSL provides the option SSL_OP_SINGLE_DH_USE for ephemeral DH (DHE) in 5294 TLS. It is not on by default. If the option is not set then the server 5295 reuses the same private DH exponent for the life of the server process and 5296 would be vulnerable to this attack. It is believed that many popular 5297 applications do set this option and would therefore not be at risk. 5298 5299 The fix for this issue adds an additional check where a "q" parameter is 5300 available (as is the case in X9.42 based parameters). This detects the 5301 only known attack, and is the only possible defense for static DH 5302 ciphersuites. This could have some performance impact. 5303 5304 Additionally the SSL_OP_SINGLE_DH_USE option has been switched on by 5305 default and cannot be disabled. This could have some performance impact. 5306 5307 This issue was reported to OpenSSL by Antonio Sanso (Adobe). 5308 ([CVE-2016-0701]) 5309 5310 *Matt Caswell* 5311 5312 * SSLv2 doesn't block disabled ciphers 5313 5314 A malicious client can negotiate SSLv2 ciphers that have been disabled on 5315 the server and complete SSLv2 handshakes even if all SSLv2 ciphers have 5316 been disabled, provided that the SSLv2 protocol was not also disabled via 5317 SSL_OP_NO_SSLv2. 5318 5319 This issue was reported to OpenSSL on 26th December 2015 by Nimrod Aviram 5320 and Sebastian Schinzel. 5321 ([CVE-2015-3197]) 5322 5323 *Viktor Dukhovni* 5324 5325### Changes between 1.0.2d and 1.0.2e [3 Dec 2015] 5326 5327 * BN_mod_exp may produce incorrect results on x86_64 5328 5329 There is a carry propagating bug in the x86_64 Montgomery squaring 5330 procedure. No EC algorithms are affected. Analysis suggests that attacks 5331 against RSA and DSA as a result of this defect would be very difficult to 5332 perform and are not believed likely. Attacks against DH are considered just 5333 feasible (although very difficult) because most of the work necessary to 5334 deduce information about a private key may be performed offline. The amount 5335 of resources required for such an attack would be very significant and 5336 likely only accessible to a limited number of attackers. An attacker would 5337 additionally need online access to an unpatched system using the target 5338 private key in a scenario with persistent DH parameters and a private 5339 key that is shared between multiple clients. For example this can occur by 5340 default in OpenSSL DHE based SSL/TLS ciphersuites. 5341 5342 This issue was reported to OpenSSL by Hanno Böck. 5343 ([CVE-2015-3193]) 5344 5345 *Andy Polyakov* 5346 5347 * Certificate verify crash with missing PSS parameter 5348 5349 The signature verification routines will crash with a NULL pointer 5350 dereference if presented with an ASN.1 signature using the RSA PSS 5351 algorithm and absent mask generation function parameter. Since these 5352 routines are used to verify certificate signature algorithms this can be 5353 used to crash any certificate verification operation and exploited in a 5354 DoS attack. Any application which performs certificate verification is 5355 vulnerable including OpenSSL clients and servers which enable client 5356 authentication. 5357 5358 This issue was reported to OpenSSL by Loïc Jonas Etienne (Qnective AG). 5359 ([CVE-2015-3194]) 5360 5361 *Stephen Henson* 5362 5363 * X509_ATTRIBUTE memory leak 5364 5365 When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak 5366 memory. This structure is used by the PKCS#7 and CMS routines so any 5367 application which reads PKCS#7 or CMS data from untrusted sources is 5368 affected. SSL/TLS is not affected. 5369 5370 This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using 5371 libFuzzer. 5372 ([CVE-2015-3195]) 5373 5374 *Stephen Henson* 5375 5376 * Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs. 5377 This changes the decoding behaviour for some invalid messages, 5378 though the change is mostly in the more lenient direction, and 5379 legacy behaviour is preserved as much as possible. 5380 5381 *Emilia Käsper* 5382 5383 * In DSA_generate_parameters_ex, if the provided seed is too short, 5384 return an error 5385 5386 *Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>* 5387 5388### Changes between 1.0.2c and 1.0.2d [9 Jul 2015] 5389 5390 * Alternate chains certificate forgery 5391 5392 During certificate verification, OpenSSL will attempt to find an 5393 alternative certificate chain if the first attempt to build such a chain 5394 fails. An error in the implementation of this logic can mean that an 5395 attacker could cause certain checks on untrusted certificates to be 5396 bypassed, such as the CA flag, enabling them to use a valid leaf 5397 certificate to act as a CA and "issue" an invalid certificate. 5398 5399 This issue was reported to OpenSSL by Adam Langley/David Benjamin 5400 (Google/BoringSSL). 5401 5402 *Matt Caswell* 5403 5404### Changes between 1.0.2b and 1.0.2c [12 Jun 2015] 5405 5406 * Fix HMAC ABI incompatibility. The previous version introduced an ABI 5407 incompatibility in the handling of HMAC. The previous ABI has now been 5408 restored. 5409 5410 *Matt Caswell* 5411 5412### Changes between 1.0.2a and 1.0.2b [11 Jun 2015] 5413 5414 * Malformed ECParameters causes infinite loop 5415 5416 When processing an ECParameters structure OpenSSL enters an infinite loop 5417 if the curve specified is over a specially malformed binary polynomial 5418 field. 5419 5420 This can be used to perform denial of service against any 5421 system which processes public keys, certificate requests or 5422 certificates. This includes TLS clients and TLS servers with 5423 client authentication enabled. 5424 5425 This issue was reported to OpenSSL by Joseph Barr-Pixton. 5426 ([CVE-2015-1788]) 5427 5428 *Andy Polyakov* 5429 5430 * Exploitable out-of-bounds read in X509_cmp_time 5431 5432 X509_cmp_time does not properly check the length of the ASN1_TIME 5433 string and can read a few bytes out of bounds. In addition, 5434 X509_cmp_time accepts an arbitrary number of fractional seconds in the 5435 time string. 5436 5437 An attacker can use this to craft malformed certificates and CRLs of 5438 various sizes and potentially cause a segmentation fault, resulting in 5439 a DoS on applications that verify certificates or CRLs. TLS clients 5440 that verify CRLs are affected. TLS clients and servers with client 5441 authentication enabled may be affected if they use custom verification 5442 callbacks. 5443 5444 This issue was reported to OpenSSL by Robert Swiecki (Google), and 5445 independently by Hanno Böck. 5446 ([CVE-2015-1789]) 5447 5448 *Emilia Käsper* 5449 5450 * PKCS7 crash with missing EnvelopedContent 5451 5452 The PKCS#7 parsing code does not handle missing inner EncryptedContent 5453 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs 5454 with missing content and trigger a NULL pointer dereference on parsing. 5455 5456 Applications that decrypt PKCS#7 data or otherwise parse PKCS#7 5457 structures from untrusted sources are affected. OpenSSL clients and 5458 servers are not affected. 5459 5460 This issue was reported to OpenSSL by Michal Zalewski (Google). 5461 ([CVE-2015-1790]) 5462 5463 *Emilia Käsper* 5464 5465 * CMS verify infinite loop with unknown hash function 5466 5467 When verifying a signedData message the CMS code can enter an infinite loop 5468 if presented with an unknown hash function OID. This can be used to perform 5469 denial of service against any system which verifies signedData messages using 5470 the CMS code. 5471 This issue was reported to OpenSSL by Johannes Bauer. 5472 ([CVE-2015-1792]) 5473 5474 *Stephen Henson* 5475 5476 * Race condition handling NewSessionTicket 5477 5478 If a NewSessionTicket is received by a multi-threaded client when attempting to 5479 reuse a previous ticket then a race condition can occur potentially leading to 5480 a double free of the ticket data. 5481 ([CVE-2015-1791]) 5482 5483 *Matt Caswell* 5484 5485 * Only support 256-bit or stronger elliptic curves with the 5486 'ecdh_auto' setting (server) or by default (client). Of supported 5487 curves, prefer P-256 (both). 5488 5489 *Emilia Kasper* 5490 5491### Changes between 1.0.2 and 1.0.2a [19 Mar 2015] 5492 5493 * ClientHello sigalgs DoS fix 5494 5495 If a client connects to an OpenSSL 1.0.2 server and renegotiates with an 5496 invalid signature algorithms extension a NULL pointer dereference will 5497 occur. This can be exploited in a DoS attack against the server. 5498 5499 This issue was was reported to OpenSSL by David Ramos of Stanford 5500 University. 5501 ([CVE-2015-0291]) 5502 5503 *Stephen Henson and Matt Caswell* 5504 5505 * Multiblock corrupted pointer fix 5506 5507 OpenSSL 1.0.2 introduced the "multiblock" performance improvement. This 5508 feature only applies on 64 bit x86 architecture platforms that support AES 5509 NI instructions. A defect in the implementation of "multiblock" can cause 5510 OpenSSL's internal write buffer to become incorrectly set to NULL when 5511 using non-blocking IO. Typically, when the user application is using a 5512 socket BIO for writing, this will only result in a failed connection. 5513 However if some other BIO is used then it is likely that a segmentation 5514 fault will be triggered, thus enabling a potential DoS attack. 5515 5516 This issue was reported to OpenSSL by Daniel Danner and Rainer Mueller. 5517 ([CVE-2015-0290]) 5518 5519 *Matt Caswell* 5520 5521 * Segmentation fault in DTLSv1_listen fix 5522 5523 The DTLSv1_listen function is intended to be stateless and processes the 5524 initial ClientHello from many peers. It is common for user code to loop 5525 over the call to DTLSv1_listen until a valid ClientHello is received with 5526 an associated cookie. A defect in the implementation of DTLSv1_listen means 5527 that state is preserved in the SSL object from one invocation to the next 5528 that can lead to a segmentation fault. Errors processing the initial 5529 ClientHello can trigger this scenario. An example of such an error could be 5530 that a DTLS1.0 only client is attempting to connect to a DTLS1.2 only 5531 server. 5532 5533 This issue was reported to OpenSSL by Per Allansson. 5534 ([CVE-2015-0207]) 5535 5536 *Matt Caswell* 5537 5538 * Segmentation fault in ASN1_TYPE_cmp fix 5539 5540 The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is 5541 made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check 5542 certificate signature algorithm consistency this can be used to crash any 5543 certificate verification operation and exploited in a DoS attack. Any 5544 application which performs certificate verification is vulnerable including 5545 OpenSSL clients and servers which enable client authentication. 5546 ([CVE-2015-0286]) 5547 5548 *Stephen Henson* 5549 5550 * Segmentation fault for invalid PSS parameters fix 5551 5552 The signature verification routines will crash with a NULL pointer 5553 dereference if presented with an ASN.1 signature using the RSA PSS 5554 algorithm and invalid parameters. Since these routines are used to verify 5555 certificate signature algorithms this can be used to crash any 5556 certificate verification operation and exploited in a DoS attack. Any 5557 application which performs certificate verification is vulnerable including 5558 OpenSSL clients and servers which enable client authentication. 5559 5560 This issue was was reported to OpenSSL by Brian Carpenter. 5561 ([CVE-2015-0208]) 5562 5563 *Stephen Henson* 5564 5565 * ASN.1 structure reuse memory corruption fix 5566 5567 Reusing a structure in ASN.1 parsing may allow an attacker to cause 5568 memory corruption via an invalid write. Such reuse is and has been 5569 strongly discouraged and is believed to be rare. 5570 5571 Applications that parse structures containing CHOICE or ANY DEFINED BY 5572 components may be affected. Certificate parsing (d2i_X509 and related 5573 functions) are however not affected. OpenSSL clients and servers are 5574 not affected. 5575 ([CVE-2015-0287]) 5576 5577 *Stephen Henson* 5578 5579 * PKCS7 NULL pointer dereferences fix 5580 5581 The PKCS#7 parsing code does not handle missing outer ContentInfo 5582 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with 5583 missing content and trigger a NULL pointer dereference on parsing. 5584 5585 Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or 5586 otherwise parse PKCS#7 structures from untrusted sources are 5587 affected. OpenSSL clients and servers are not affected. 5588 5589 This issue was reported to OpenSSL by Michal Zalewski (Google). 5590 ([CVE-2015-0289]) 5591 5592 *Emilia Käsper* 5593 5594 * DoS via reachable assert in SSLv2 servers fix 5595 5596 A malicious client can trigger an OPENSSL_assert (i.e., an abort) in 5597 servers that both support SSLv2 and enable export cipher suites by sending 5598 a specially crafted SSLv2 CLIENT-MASTER-KEY message. 5599 5600 This issue was discovered by Sean Burford (Google) and Emilia Käsper 5601 (OpenSSL development team). 5602 ([CVE-2015-0293]) 5603 5604 *Emilia Käsper* 5605 5606 * Empty CKE with client auth and DHE fix 5607 5608 If client auth is used then a server can seg fault in the event of a DHE 5609 ciphersuite being selected and a zero length ClientKeyExchange message 5610 being sent by the client. This could be exploited in a DoS attack. 5611 ([CVE-2015-1787]) 5612 5613 *Matt Caswell* 5614 5615 * Handshake with unseeded PRNG fix 5616 5617 Under certain conditions an OpenSSL 1.0.2 client can complete a handshake 5618 with an unseeded PRNG. The conditions are: 5619 - The client is on a platform where the PRNG has not been seeded 5620 automatically, and the user has not seeded manually 5621 - A protocol specific client method version has been used (i.e. not 5622 SSL_client_methodv23) 5623 - A ciphersuite is used that does not require additional random data from 5624 the PRNG beyond the initial ClientHello client random (e.g. PSK-RC4-SHA). 5625 5626 If the handshake succeeds then the client random that has been used will 5627 have been generated from a PRNG with insufficient entropy and therefore the 5628 output may be predictable. 5629 5630 For example using the following command with an unseeded openssl will 5631 succeed on an unpatched platform: 5632 5633 openssl s_client -psk 1a2b3c4d -tls1_2 -cipher PSK-RC4-SHA 5634 ([CVE-2015-0285]) 5635 5636 *Matt Caswell* 5637 5638 * Use After Free following d2i_ECPrivatekey error fix 5639 5640 A malformed EC private key file consumed via the d2i_ECPrivateKey function 5641 could cause a use after free condition. This, in turn, could cause a double 5642 free in several private key parsing functions (such as d2i_PrivateKey 5643 or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption 5644 for applications that receive EC private keys from untrusted 5645 sources. This scenario is considered rare. 5646 5647 This issue was discovered by the BoringSSL project and fixed in their 5648 commit 517073cd4b. 5649 ([CVE-2015-0209]) 5650 5651 *Matt Caswell* 5652 5653 * X509_to_X509_REQ NULL pointer deref fix 5654 5655 The function X509_to_X509_REQ will crash with a NULL pointer dereference if 5656 the certificate key is invalid. This function is rarely used in practice. 5657 5658 This issue was discovered by Brian Carpenter. 5659 ([CVE-2015-0288]) 5660 5661 *Stephen Henson* 5662 5663 * Removed the export ciphers from the DEFAULT ciphers 5664 5665 *Kurt Roeckx* 5666 5667### Changes between 1.0.1l and 1.0.2 [22 Jan 2015] 5668 5669 * Facilitate "universal" ARM builds targeting range of ARM ISAs, e.g. 5670 ARMv5 through ARMv8, as opposite to "locking" it to single one. 5671 So far those who have to target multiple platforms would compromise 5672 and argue that binary targeting say ARMv5 would still execute on 5673 ARMv8. "Universal" build resolves this compromise by providing 5674 near-optimal performance even on newer platforms. 5675 5676 *Andy Polyakov* 5677 5678 * Accelerated NIST P-256 elliptic curve implementation for x86_64 5679 (other platforms pending). 5680 5681 *Shay Gueron & Vlad Krasnov (Intel Corp), Andy Polyakov* 5682 5683 * Add support for the SignedCertificateTimestampList certificate and 5684 OCSP response extensions from RFC6962. 5685 5686 *Rob Stradling* 5687 5688 * Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.) 5689 for corner cases. (Certain input points at infinity could lead to 5690 bogus results, with non-infinity inputs mapped to infinity too.) 5691 5692 *Bodo Moeller* 5693 5694 * Initial support for PowerISA 2.0.7, first implemented in POWER8. 5695 This covers AES, SHA256/512 and GHASH. "Initial" means that most 5696 common cases are optimized and there still is room for further 5697 improvements. Vector Permutation AES for Altivec is also added. 5698 5699 *Andy Polyakov* 5700 5701 * Add support for little-endian ppc64 Linux target. 5702 5703 *Marcelo Cerri (IBM)* 5704 5705 * Initial support for AMRv8 ISA crypto extensions. This covers AES, 5706 SHA1, SHA256 and GHASH. "Initial" means that most common cases 5707 are optimized and there still is room for further improvements. 5708 Both 32- and 64-bit modes are supported. 5709 5710 *Andy Polyakov, Ard Biesheuvel (Linaro)* 5711 5712 * Improved ARMv7 NEON support. 5713 5714 *Andy Polyakov* 5715 5716 * Support for SPARC Architecture 2011 crypto extensions, first 5717 implemented in SPARC T4. This covers AES, DES, Camellia, SHA1, 5718 SHA256/512, MD5, GHASH and modular exponentiation. 5719 5720 *Andy Polyakov, David Miller* 5721 5722 * Accelerated modular exponentiation for Intel processors, a.k.a. 5723 RSAZ. 5724 5725 *Shay Gueron & Vlad Krasnov (Intel Corp)* 5726 5727 * Support for new and upcoming Intel processors, including AVX2, 5728 BMI and SHA ISA extensions. This includes additional "stitched" 5729 implementations, AESNI-SHA256 and GCM, and multi-buffer support 5730 for TLS encrypt. 5731 5732 This work was sponsored by Intel Corp. 5733 5734 *Andy Polyakov* 5735 5736 * Support for DTLS 1.2. This adds two sets of DTLS methods: DTLS_*_method() 5737 supports both DTLS 1.2 and 1.0 and should use whatever version the peer 5738 supports and DTLSv1_2_*_method() which supports DTLS 1.2 only. 5739 5740 *Steve Henson* 5741 5742 * Use algorithm specific chains in SSL_CTX_use_certificate_chain_file(): 5743 this fixes a limitation in previous versions of OpenSSL. 5744 5745 *Steve Henson* 5746 5747 * Extended RSA OAEP support via EVP_PKEY API. Options to specify digest, 5748 MGF1 digest and OAEP label. 5749 5750 *Steve Henson* 5751 5752 * Add EVP support for key wrapping algorithms, to avoid problems with 5753 existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in 5754 the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap 5755 algorithms and include tests cases. 5756 5757 *Steve Henson* 5758 5759 * Add functions to allocate and set the fields of an ECDSA_METHOD 5760 structure. 5761 5762 *Douglas E. Engert, Steve Henson* 5763 5764 * New functions OPENSSL_gmtime_diff and ASN1_TIME_diff to find the 5765 difference in days and seconds between two tm or ASN1_TIME structures. 5766 5767 *Steve Henson* 5768 5769 * Add -rev test option to s_server to just reverse order of characters 5770 received by client and send back to server. Also prints an abbreviated 5771 summary of the connection parameters. 5772 5773 *Steve Henson* 5774 5775 * New option -brief for s_client and s_server to print out a brief summary 5776 of connection parameters. 5777 5778 *Steve Henson* 5779 5780 * Add callbacks for arbitrary TLS extensions. 5781 5782 *Trevor Perrin <trevp@trevp.net> and Ben Laurie* 5783 5784 * New option -crl_download in several openssl utilities to download CRLs 5785 from CRLDP extension in certificates. 5786 5787 *Steve Henson* 5788 5789 * New options -CRL and -CRLform for s_client and s_server for CRLs. 5790 5791 *Steve Henson* 5792 5793 * New function X509_CRL_diff to generate a delta CRL from the difference 5794 of two full CRLs. Add support to "crl" utility. 5795 5796 *Steve Henson* 5797 5798 * New functions to set lookup_crls function and to retrieve 5799 X509_STORE from X509_STORE_CTX. 5800 5801 *Steve Henson* 5802 5803 * Print out deprecated issuer and subject unique ID fields in 5804 certificates. 5805 5806 *Steve Henson* 5807 5808 * Extend OCSP I/O functions so they can be used for simple general purpose 5809 HTTP as well as OCSP. New wrapper function which can be used to download 5810 CRLs using the OCSP API. 5811 5812 *Steve Henson* 5813 5814 * Delegate command line handling in s_client/s_server to SSL_CONF APIs. 5815 5816 *Steve Henson* 5817 5818 * `SSL_CONF*` functions. These provide a common framework for application 5819 configuration using configuration files or command lines. 5820 5821 *Steve Henson* 5822 5823 * SSL/TLS tracing code. This parses out SSL/TLS records using the 5824 message callback and prints the results. Needs compile time option 5825 "enable-ssl-trace". New options to s_client and s_server to enable 5826 tracing. 5827 5828 *Steve Henson* 5829 5830 * New ctrl and macro to retrieve supported points extensions. 5831 Print out extension in s_server and s_client. 5832 5833 *Steve Henson* 5834 5835 * New functions to retrieve certificate signature and signature 5836 OID NID. 5837 5838 *Steve Henson* 5839 5840 * Add functions to retrieve and manipulate the raw cipherlist sent by a 5841 client to OpenSSL. 5842 5843 *Steve Henson* 5844 5845 * New Suite B modes for TLS code. These use and enforce the requirements 5846 of RFC6460: restrict ciphersuites, only permit Suite B algorithms and 5847 only use Suite B curves. The Suite B modes can be set by using the 5848 strings "SUITEB128", "SUITEB192" or "SUITEB128ONLY" for the cipherstring. 5849 5850 *Steve Henson* 5851 5852 * New chain verification flags for Suite B levels of security. Check 5853 algorithms are acceptable when flags are set in X509_verify_cert. 5854 5855 *Steve Henson* 5856 5857 * Make tls1_check_chain return a set of flags indicating checks passed 5858 by a certificate chain. Add additional tests to handle client 5859 certificates: checks for matching certificate type and issuer name 5860 comparison. 5861 5862 *Steve Henson* 5863 5864 * If an attempt is made to use a signature algorithm not in the peer 5865 preference list abort the handshake. If client has no suitable 5866 signature algorithms in response to a certificate request do not 5867 use the certificate. 5868 5869 *Steve Henson* 5870 5871 * If server EC tmp key is not in client preference list abort handshake. 5872 5873 *Steve Henson* 5874 5875 * Add support for certificate stores in CERT structure. This makes it 5876 possible to have different stores per SSL structure or one store in 5877 the parent SSL_CTX. Include distinct stores for certificate chain 5878 verification and chain building. New ctrl SSL_CTRL_BUILD_CERT_CHAIN 5879 to build and store a certificate chain in CERT structure: returning 5880 an error if the chain cannot be built: this will allow applications 5881 to test if a chain is correctly configured. 5882 5883 Note: if the CERT based stores are not set then the parent SSL_CTX 5884 store is used to retain compatibility with existing behaviour. 5885 5886 *Steve Henson* 5887 5888 * New function ssl_set_client_disabled to set a ciphersuite disabled 5889 mask based on the current session, check mask when sending client 5890 hello and checking the requested ciphersuite. 5891 5892 *Steve Henson* 5893 5894 * New ctrls to retrieve and set certificate types in a certificate 5895 request message. Print out received values in s_client. If certificate 5896 types is not set with custom values set sensible values based on 5897 supported signature algorithms. 5898 5899 *Steve Henson* 5900 5901 * Support for distinct client and server supported signature algorithms. 5902 5903 *Steve Henson* 5904 5905 * Add certificate callback. If set this is called whenever a certificate 5906 is required by client or server. An application can decide which 5907 certificate chain to present based on arbitrary criteria: for example 5908 supported signature algorithms. Add very simple example to s_server. 5909 This fixes many of the problems and restrictions of the existing client 5910 certificate callback: for example you can now clear an existing 5911 certificate and specify the whole chain. 5912 5913 *Steve Henson* 5914 5915 * Add new "valid_flags" field to CERT_PKEY structure which determines what 5916 the certificate can be used for (if anything). Set valid_flags field 5917 in new tls1_check_chain function. Simplify ssl_set_cert_masks which used 5918 to have similar checks in it. 5919 5920 Add new "cert_flags" field to CERT structure and include a "strict mode". 5921 This enforces some TLS certificate requirements (such as only permitting 5922 certificate signature algorithms contained in the supported algorithms 5923 extension) which some implementations ignore: this option should be used 5924 with caution as it could cause interoperability issues. 5925 5926 *Steve Henson* 5927 5928 * Update and tidy signature algorithm extension processing. Work out 5929 shared signature algorithms based on preferences and peer algorithms 5930 and print them out in s_client and s_server. Abort handshake if no 5931 shared signature algorithms. 5932 5933 *Steve Henson* 5934 5935 * Add new functions to allow customised supported signature algorithms 5936 for SSL and SSL_CTX structures. Add options to s_client and s_server 5937 to support them. 5938 5939 *Steve Henson* 5940 5941 * New function SSL_certs_clear() to delete all references to certificates 5942 from an SSL structure. Before this once a certificate had been added 5943 it couldn't be removed. 5944 5945 *Steve Henson* 5946 5947 * Integrate hostname, email address and IP address checking with certificate 5948 verification. New verify options supporting checking in openssl utility. 5949 5950 *Steve Henson* 5951 5952 * Fixes and wildcard matching support to hostname and email checking 5953 functions. Add manual page. 5954 5955 *Florian Weimer (Red Hat Product Security Team)* 5956 5957 * New functions to check a hostname email or IP address against a 5958 certificate. Add options x509 utility to print results of checks against 5959 a certificate. 5960 5961 *Steve Henson* 5962 5963 * Fix OCSP checking. 5964 5965 *Rob Stradling <rob.stradling@comodo.com> and Ben Laurie* 5966 5967 * Initial experimental support for explicitly trusted non-root CAs. 5968 OpenSSL still tries to build a complete chain to a root but if an 5969 intermediate CA has a trust setting included that is used. The first 5970 setting is used: whether to trust (e.g., -addtrust option to the x509 5971 utility) or reject. 5972 5973 *Steve Henson* 5974 5975 * Add -trusted_first option which attempts to find certificates in the 5976 trusted store even if an untrusted chain is also supplied. 5977 5978 *Steve Henson* 5979 5980 * MIPS assembly pack updates: support for MIPS32r2 and SmartMIPS ASE, 5981 platform support for Linux and Android. 5982 5983 *Andy Polyakov* 5984 5985 * Support for linux-x32, ILP32 environment in x86_64 framework. 5986 5987 *Andy Polyakov* 5988 5989 * Experimental multi-implementation support for FIPS capable OpenSSL. 5990 When in FIPS mode the approved implementations are used as normal, 5991 when not in FIPS mode the internal unapproved versions are used instead. 5992 This means that the FIPS capable OpenSSL isn't forced to use the 5993 (often lower performance) FIPS implementations outside FIPS mode. 5994 5995 *Steve Henson* 5996 5997 * Transparently support X9.42 DH parameters when calling 5998 PEM_read_bio_DHparameters. This means existing applications can handle 5999 the new parameter format automatically. 6000 6001 *Steve Henson* 6002 6003 * Initial experimental support for X9.42 DH parameter format: mainly 6004 to support use of 'q' parameter for RFC5114 parameters. 6005 6006 *Steve Henson* 6007 6008 * Add DH parameters from RFC5114 including test data to dhtest. 6009 6010 *Steve Henson* 6011 6012 * Support for automatic EC temporary key parameter selection. If enabled 6013 the most preferred EC parameters are automatically used instead of 6014 hardcoded fixed parameters. Now a server just has to call: 6015 SSL_CTX_set_ecdh_auto(ctx, 1) and the server will automatically 6016 support ECDH and use the most appropriate parameters. 6017 6018 *Steve Henson* 6019 6020 * Enhance and tidy EC curve and point format TLS extension code. Use 6021 static structures instead of allocation if default values are used. 6022 New ctrls to set curves we wish to support and to retrieve shared curves. 6023 Print out shared curves in s_server. New options to s_server and s_client 6024 to set list of supported curves. 6025 6026 *Steve Henson* 6027 6028 * New ctrls to retrieve supported signature algorithms and 6029 supported curve values as an array of NIDs. Extend openssl utility 6030 to print out received values. 6031 6032 *Steve Henson* 6033 6034 * Add new APIs EC_curve_nist2nid and EC_curve_nid2nist which convert 6035 between NIDs and the more common NIST names such as "P-256". Enhance 6036 ecparam utility and ECC method to recognise the NIST names for curves. 6037 6038 *Steve Henson* 6039 6040 * Enhance SSL/TLS certificate chain handling to support different 6041 chains for each certificate instead of one chain in the parent SSL_CTX. 6042 6043 *Steve Henson* 6044 6045 * Support for fixed DH ciphersuite client authentication: where both 6046 server and client use DH certificates with common parameters. 6047 6048 *Steve Henson* 6049 6050 * Support for fixed DH ciphersuites: those requiring DH server 6051 certificates. 6052 6053 *Steve Henson* 6054 6055 * New function i2d_re_X509_tbs for re-encoding the TBS portion of 6056 the certificate. 6057 Note: Related 1.0.2-beta specific macros X509_get_cert_info, 6058 X509_CINF_set_modified, X509_CINF_get_issuer, X509_CINF_get_extensions and 6059 X509_CINF_get_signature were reverted post internal team review. 6060 6061OpenSSL 1.0.1 6062------------- 6063 6064### Changes between 1.0.1t and 1.0.1u [22 Sep 2016] 6065 6066 * OCSP Status Request extension unbounded memory growth 6067 6068 A malicious client can send an excessively large OCSP Status Request 6069 extension. If that client continually requests renegotiation, sending a 6070 large OCSP Status Request extension each time, then there will be unbounded 6071 memory growth on the server. This will eventually lead to a Denial Of 6072 Service attack through memory exhaustion. Servers with a default 6073 configuration are vulnerable even if they do not support OCSP. Builds using 6074 the "no-ocsp" build time option are not affected. 6075 6076 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 6077 ([CVE-2016-6304]) 6078 6079 *Matt Caswell* 6080 6081 * In order to mitigate the SWEET32 attack, the DES ciphers were moved from 6082 HIGH to MEDIUM. 6083 6084 This issue was reported to OpenSSL Karthikeyan Bhargavan and Gaetan 6085 Leurent (INRIA) 6086 ([CVE-2016-2183]) 6087 6088 *Rich Salz* 6089 6090 * OOB write in MDC2_Update() 6091 6092 An overflow can occur in MDC2_Update() either if called directly or 6093 through the EVP_DigestUpdate() function using MDC2. If an attacker 6094 is able to supply very large amounts of input data after a previous 6095 call to EVP_EncryptUpdate() with a partial block then a length check 6096 can overflow resulting in a heap corruption. 6097 6098 The amount of data needed is comparable to SIZE_MAX which is impractical 6099 on most platforms. 6100 6101 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 6102 ([CVE-2016-6303]) 6103 6104 *Stephen Henson* 6105 6106 * Malformed SHA512 ticket DoS 6107 6108 If a server uses SHA512 for TLS session ticket HMAC it is vulnerable to a 6109 DoS attack where a malformed ticket will result in an OOB read which will 6110 ultimately crash. 6111 6112 The use of SHA512 in TLS session tickets is comparatively rare as it requires 6113 a custom server callback and ticket lookup mechanism. 6114 6115 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 6116 ([CVE-2016-6302]) 6117 6118 *Stephen Henson* 6119 6120 * OOB write in BN_bn2dec() 6121 6122 The function BN_bn2dec() does not check the return value of BN_div_word(). 6123 This can cause an OOB write if an application uses this function with an 6124 overly large BIGNUM. This could be a problem if an overly large certificate 6125 or CRL is printed out from an untrusted source. TLS is not affected because 6126 record limits will reject an oversized certificate before it is parsed. 6127 6128 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 6129 ([CVE-2016-2182]) 6130 6131 *Stephen Henson* 6132 6133 * OOB read in TS_OBJ_print_bio() 6134 6135 The function TS_OBJ_print_bio() misuses OBJ_obj2txt(): the return value is 6136 the total length the OID text representation would use and not the amount 6137 of data written. This will result in OOB reads when large OIDs are 6138 presented. 6139 6140 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 6141 ([CVE-2016-2180]) 6142 6143 *Stephen Henson* 6144 6145 * Pointer arithmetic undefined behaviour 6146 6147 Avoid some undefined pointer arithmetic 6148 6149 A common idiom in the codebase is to check limits in the following manner: 6150 "p + len > limit" 6151 6152 Where "p" points to some malloc'd data of SIZE bytes and 6153 limit == p + SIZE 6154 6155 "len" here could be from some externally supplied data (e.g. from a TLS 6156 message). 6157 6158 The rules of C pointer arithmetic are such that "p + len" is only well 6159 defined where len <= SIZE. Therefore the above idiom is actually 6160 undefined behaviour. 6161 6162 For example this could cause problems if some malloc implementation 6163 provides an address for "p" such that "p + len" actually overflows for 6164 values of len that are too big and therefore p + len < limit. 6165 6166 This issue was reported to OpenSSL by Guido Vranken 6167 ([CVE-2016-2177]) 6168 6169 *Matt Caswell* 6170 6171 * Constant time flag not preserved in DSA signing 6172 6173 Operations in the DSA signing algorithm should run in constant time in 6174 order to avoid side channel attacks. A flaw in the OpenSSL DSA 6175 implementation means that a non-constant time codepath is followed for 6176 certain operations. This has been demonstrated through a cache-timing 6177 attack to be sufficient for an attacker to recover the private DSA key. 6178 6179 This issue was reported by César Pereida (Aalto University), Billy Brumley 6180 (Tampere University of Technology), and Yuval Yarom (The University of 6181 Adelaide and NICTA). 6182 ([CVE-2016-2178]) 6183 6184 *César Pereida* 6185 6186 * DTLS buffered message DoS 6187 6188 In a DTLS connection where handshake messages are delivered out-of-order 6189 those messages that OpenSSL is not yet ready to process will be buffered 6190 for later use. Under certain circumstances, a flaw in the logic means that 6191 those messages do not get removed from the buffer even though the handshake 6192 has been completed. An attacker could force up to approx. 15 messages to 6193 remain in the buffer when they are no longer required. These messages will 6194 be cleared when the DTLS connection is closed. The default maximum size for 6195 a message is 100k. Therefore the attacker could force an additional 1500k 6196 to be consumed per connection. By opening many simulataneous connections an 6197 attacker could cause a DoS attack through memory exhaustion. 6198 6199 This issue was reported to OpenSSL by Quan Luo. 6200 ([CVE-2016-2179]) 6201 6202 *Matt Caswell* 6203 6204 * DTLS replay protection DoS 6205 6206 A flaw in the DTLS replay attack protection mechanism means that records 6207 that arrive for future epochs update the replay protection "window" before 6208 the MAC for the record has been validated. This could be exploited by an 6209 attacker by sending a record for the next epoch (which does not have to 6210 decrypt or have a valid MAC), with a very large sequence number. This means 6211 that all subsequent legitimate packets are dropped causing a denial of 6212 service for a specific DTLS connection. 6213 6214 This issue was reported to OpenSSL by the OCAP audit team. 6215 ([CVE-2016-2181]) 6216 6217 *Matt Caswell* 6218 6219 * Certificate message OOB reads 6220 6221 In OpenSSL 1.0.2 and earlier some missing message length checks can result 6222 in OOB reads of up to 2 bytes beyond an allocated buffer. There is a 6223 theoretical DoS risk but this has not been observed in practice on common 6224 platforms. 6225 6226 The messages affected are client certificate, client certificate request 6227 and server certificate. As a result the attack can only be performed 6228 against a client or a server which enables client authentication. 6229 6230 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 6231 ([CVE-2016-6306]) 6232 6233 *Stephen Henson* 6234 6235### Changes between 1.0.1s and 1.0.1t [3 May 2016] 6236 6237 * Prevent padding oracle in AES-NI CBC MAC check 6238 6239 A MITM attacker can use a padding oracle attack to decrypt traffic 6240 when the connection uses an AES CBC cipher and the server support 6241 AES-NI. 6242 6243 This issue was introduced as part of the fix for Lucky 13 padding 6244 attack ([CVE-2013-0169]). The padding check was rewritten to be in 6245 constant time by making sure that always the same bytes are read and 6246 compared against either the MAC or padding bytes. But it no longer 6247 checked that there was enough data to have both the MAC and padding 6248 bytes. 6249 6250 This issue was reported by Juraj Somorovsky using TLS-Attacker. 6251 ([CVE-2016-2107]) 6252 6253 *Kurt Roeckx* 6254 6255 * Fix EVP_EncodeUpdate overflow 6256 6257 An overflow can occur in the EVP_EncodeUpdate() function which is used for 6258 Base64 encoding of binary data. If an attacker is able to supply very large 6259 amounts of input data then a length check can overflow resulting in a heap 6260 corruption. 6261 6262 Internally to OpenSSL the EVP_EncodeUpdate() function is primarly used by 6263 the `PEM_write_bio*` family of functions. These are mainly used within the 6264 OpenSSL command line applications, so any application which processes data 6265 from an untrusted source and outputs it as a PEM file should be considered 6266 vulnerable to this issue. User applications that call these APIs directly 6267 with large amounts of untrusted data may also be vulnerable. 6268 6269 This issue was reported by Guido Vranken. 6270 ([CVE-2016-2105]) 6271 6272 *Matt Caswell* 6273 6274 * Fix EVP_EncryptUpdate overflow 6275 6276 An overflow can occur in the EVP_EncryptUpdate() function. If an attacker 6277 is able to supply very large amounts of input data after a previous call to 6278 EVP_EncryptUpdate() with a partial block then a length check can overflow 6279 resulting in a heap corruption. Following an analysis of all OpenSSL 6280 internal usage of the EVP_EncryptUpdate() function all usage is one of two 6281 forms. The first form is where the EVP_EncryptUpdate() call is known to be 6282 the first called function after an EVP_EncryptInit(), and therefore that 6283 specific call must be safe. The second form is where the length passed to 6284 EVP_EncryptUpdate() can be seen from the code to be some small value and 6285 therefore there is no possibility of an overflow. Since all instances are 6286 one of these two forms, it is believed that there can be no overflows in 6287 internal code due to this problem. It should be noted that 6288 EVP_DecryptUpdate() can call EVP_EncryptUpdate() in certain code paths. 6289 Also EVP_CipherUpdate() is a synonym for EVP_EncryptUpdate(). All instances 6290 of these calls have also been analysed too and it is believed there are no 6291 instances in internal usage where an overflow could occur. 6292 6293 This issue was reported by Guido Vranken. 6294 ([CVE-2016-2106]) 6295 6296 *Matt Caswell* 6297 6298 * Prevent ASN.1 BIO excessive memory allocation 6299 6300 When ASN.1 data is read from a BIO using functions such as d2i_CMS_bio() 6301 a short invalid encoding can casuse allocation of large amounts of memory 6302 potentially consuming excessive resources or exhausting memory. 6303 6304 Any application parsing untrusted data through d2i BIO functions is 6305 affected. The memory based functions such as d2i_X509() are *not* affected. 6306 Since the memory based functions are used by the TLS library, TLS 6307 applications are not affected. 6308 6309 This issue was reported by Brian Carpenter. 6310 ([CVE-2016-2109]) 6311 6312 *Stephen Henson* 6313 6314 * EBCDIC overread 6315 6316 ASN1 Strings that are over 1024 bytes can cause an overread in applications 6317 using the X509_NAME_oneline() function on EBCDIC systems. This could result 6318 in arbitrary stack data being returned in the buffer. 6319 6320 This issue was reported by Guido Vranken. 6321 ([CVE-2016-2176]) 6322 6323 *Matt Caswell* 6324 6325 * Modify behavior of ALPN to invoke callback after SNI/servername 6326 callback, such that updates to the SSL_CTX affect ALPN. 6327 6328 *Todd Short* 6329 6330 * Remove LOW from the DEFAULT cipher list. This removes singles DES from the 6331 default. 6332 6333 *Kurt Roeckx* 6334 6335 * Only remove the SSLv2 methods with the no-ssl2-method option. When the 6336 methods are enabled and ssl2 is disabled the methods return NULL. 6337 6338 *Kurt Roeckx* 6339 6340### Changes between 1.0.1r and 1.0.1s [1 Mar 2016] 6341 6342* Disable weak ciphers in SSLv3 and up in default builds of OpenSSL. 6343 Builds that are not configured with "enable-weak-ssl-ciphers" will not 6344 provide any "EXPORT" or "LOW" strength ciphers. 6345 6346 *Viktor Dukhovni* 6347 6348* Disable SSLv2 default build, default negotiation and weak ciphers. SSLv2 6349 is by default disabled at build-time. Builds that are not configured with 6350 "enable-ssl2" will not support SSLv2. Even if "enable-ssl2" is used, 6351 users who want to negotiate SSLv2 via the version-flexible SSLv23_method() 6352 will need to explicitly call either of: 6353 6354 SSL_CTX_clear_options(ctx, SSL_OP_NO_SSLv2); 6355 or 6356 SSL_clear_options(ssl, SSL_OP_NO_SSLv2); 6357 6358 as appropriate. Even if either of those is used, or the application 6359 explicitly uses the version-specific SSLv2_method() or its client and 6360 server variants, SSLv2 ciphers vulnerable to exhaustive search key 6361 recovery have been removed. Specifically, the SSLv2 40-bit EXPORT 6362 ciphers, and SSLv2 56-bit DES are no longer available. 6363 ([CVE-2016-0800]) 6364 6365 *Viktor Dukhovni* 6366 6367 * Fix a double-free in DSA code 6368 6369 A double free bug was discovered when OpenSSL parses malformed DSA private 6370 keys and could lead to a DoS attack or memory corruption for applications 6371 that receive DSA private keys from untrusted sources. This scenario is 6372 considered rare. 6373 6374 This issue was reported to OpenSSL by Adam Langley(Google/BoringSSL) using 6375 libFuzzer. 6376 ([CVE-2016-0705]) 6377 6378 *Stephen Henson* 6379 6380 * Disable SRP fake user seed to address a server memory leak. 6381 6382 Add a new method SRP_VBASE_get1_by_user that handles the seed properly. 6383 6384 SRP_VBASE_get_by_user had inconsistent memory management behaviour. 6385 In order to fix an unavoidable memory leak, SRP_VBASE_get_by_user 6386 was changed to ignore the "fake user" SRP seed, even if the seed 6387 is configured. 6388 6389 Users should use SRP_VBASE_get1_by_user instead. Note that in 6390 SRP_VBASE_get1_by_user, caller must free the returned value. Note 6391 also that even though configuring the SRP seed attempts to hide 6392 invalid usernames by continuing the handshake with fake 6393 credentials, this behaviour is not constant time and no strong 6394 guarantees are made that the handshake is indistinguishable from 6395 that of a valid user. 6396 ([CVE-2016-0798]) 6397 6398 *Emilia Käsper* 6399 6400 * Fix BN_hex2bn/BN_dec2bn NULL pointer deref/heap corruption 6401 6402 In the BN_hex2bn function the number of hex digits is calculated using an 6403 int value `i`. Later `bn_expand` is called with a value of `i * 4`. For 6404 large values of `i` this can result in `bn_expand` not allocating any 6405 memory because `i * 4` is negative. This can leave the internal BIGNUM data 6406 field as NULL leading to a subsequent NULL ptr deref. For very large values 6407 of `i`, the calculation `i * 4` could be a positive value smaller than `i`. 6408 In this case memory is allocated to the internal BIGNUM data field, but it 6409 is insufficiently sized leading to heap corruption. A similar issue exists 6410 in BN_dec2bn. This could have security consequences if BN_hex2bn/BN_dec2bn 6411 is ever called by user applications with very large untrusted hex/dec data. 6412 This is anticipated to be a rare occurrence. 6413 6414 All OpenSSL internal usage of these functions use data that is not expected 6415 to be untrusted, e.g. config file data or application command line 6416 arguments. If user developed applications generate config file data based 6417 on untrusted data then it is possible that this could also lead to security 6418 consequences. This is also anticipated to be rare. 6419 6420 This issue was reported to OpenSSL by Guido Vranken. 6421 ([CVE-2016-0797]) 6422 6423 *Matt Caswell* 6424 6425 * Fix memory issues in `BIO_*printf` functions 6426 6427 The internal `fmtstr` function used in processing a "%s" format string in 6428 the `BIO_*printf` functions could overflow while calculating the length of a 6429 string and cause an OOB read when printing very long strings. 6430 6431 Additionally the internal `doapr_outch` function can attempt to write to an 6432 OOB memory location (at an offset from the NULL pointer) in the event of a 6433 memory allocation failure. In 1.0.2 and below this could be caused where 6434 the size of a buffer to be allocated is greater than INT_MAX. E.g. this 6435 could be in processing a very long "%s" format string. Memory leaks can 6436 also occur. 6437 6438 The first issue may mask the second issue dependent on compiler behaviour. 6439 These problems could enable attacks where large amounts of untrusted data 6440 is passed to the `BIO_*printf` functions. If applications use these functions 6441 in this way then they could be vulnerable. OpenSSL itself uses these 6442 functions when printing out human-readable dumps of ASN.1 data. Therefore 6443 applications that print this data could be vulnerable if the data is from 6444 untrusted sources. OpenSSL command line applications could also be 6445 vulnerable where they print out ASN.1 data, or if untrusted data is passed 6446 as command line arguments. 6447 6448 Libssl is not considered directly vulnerable. Additionally certificates etc 6449 received via remote connections via libssl are also unlikely to be able to 6450 trigger these issues because of message size limits enforced within libssl. 6451 6452 This issue was reported to OpenSSL Guido Vranken. 6453 ([CVE-2016-0799]) 6454 6455 *Matt Caswell* 6456 6457 * Side channel attack on modular exponentiation 6458 6459 A side-channel attack was found which makes use of cache-bank conflicts on 6460 the Intel Sandy-Bridge microarchitecture which could lead to the recovery 6461 of RSA keys. The ability to exploit this issue is limited as it relies on 6462 an attacker who has control of code in a thread running on the same 6463 hyper-threaded core as the victim thread which is performing decryptions. 6464 6465 This issue was reported to OpenSSL by Yuval Yarom, The University of 6466 Adelaide and NICTA, Daniel Genkin, Technion and Tel Aviv University, and 6467 Nadia Heninger, University of Pennsylvania with more information at 6468 <http://cachebleed.info>. 6469 ([CVE-2016-0702]) 6470 6471 *Andy Polyakov* 6472 6473 * Change the req command to generate a 2048-bit RSA/DSA key by default, 6474 if no keysize is specified with default_bits. This fixes an 6475 omission in an earlier change that changed all RSA/DSA key generation 6476 commands to use 2048 bits by default. 6477 6478 *Emilia Käsper* 6479 6480### Changes between 1.0.1q and 1.0.1r [28 Jan 2016] 6481 6482 * Protection for DH small subgroup attacks 6483 6484 As a precautionary measure the SSL_OP_SINGLE_DH_USE option has been 6485 switched on by default and cannot be disabled. This could have some 6486 performance impact. 6487 6488 *Matt Caswell* 6489 6490 * SSLv2 doesn't block disabled ciphers 6491 6492 A malicious client can negotiate SSLv2 ciphers that have been disabled on 6493 the server and complete SSLv2 handshakes even if all SSLv2 ciphers have 6494 been disabled, provided that the SSLv2 protocol was not also disabled via 6495 SSL_OP_NO_SSLv2. 6496 6497 This issue was reported to OpenSSL on 26th December 2015 by Nimrod Aviram 6498 and Sebastian Schinzel. 6499 ([CVE-2015-3197]) 6500 6501 *Viktor Dukhovni* 6502 6503 * Reject DH handshakes with parameters shorter than 1024 bits. 6504 6505 *Kurt Roeckx* 6506 6507### Changes between 1.0.1p and 1.0.1q [3 Dec 2015] 6508 6509 * Certificate verify crash with missing PSS parameter 6510 6511 The signature verification routines will crash with a NULL pointer 6512 dereference if presented with an ASN.1 signature using the RSA PSS 6513 algorithm and absent mask generation function parameter. Since these 6514 routines are used to verify certificate signature algorithms this can be 6515 used to crash any certificate verification operation and exploited in a 6516 DoS attack. Any application which performs certificate verification is 6517 vulnerable including OpenSSL clients and servers which enable client 6518 authentication. 6519 6520 This issue was reported to OpenSSL by Loïc Jonas Etienne (Qnective AG). 6521 ([CVE-2015-3194]) 6522 6523 *Stephen Henson* 6524 6525 * X509_ATTRIBUTE memory leak 6526 6527 When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak 6528 memory. This structure is used by the PKCS#7 and CMS routines so any 6529 application which reads PKCS#7 or CMS data from untrusted sources is 6530 affected. SSL/TLS is not affected. 6531 6532 This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using 6533 libFuzzer. 6534 ([CVE-2015-3195]) 6535 6536 *Stephen Henson* 6537 6538 * Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs. 6539 This changes the decoding behaviour for some invalid messages, 6540 though the change is mostly in the more lenient direction, and 6541 legacy behaviour is preserved as much as possible. 6542 6543 *Emilia Käsper* 6544 6545 * In DSA_generate_parameters_ex, if the provided seed is too short, 6546 use a random seed, as already documented. 6547 6548 *Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>* 6549 6550### Changes between 1.0.1o and 1.0.1p [9 Jul 2015] 6551 6552 * Alternate chains certificate forgery 6553 6554 During certificate verfification, OpenSSL will attempt to find an 6555 alternative certificate chain if the first attempt to build such a chain 6556 fails. An error in the implementation of this logic can mean that an 6557 attacker could cause certain checks on untrusted certificates to be 6558 bypassed, such as the CA flag, enabling them to use a valid leaf 6559 certificate to act as a CA and "issue" an invalid certificate. 6560 6561 This issue was reported to OpenSSL by Adam Langley/David Benjamin 6562 (Google/BoringSSL). 6563 ([CVE-2015-1793]) 6564 6565 *Matt Caswell* 6566 6567 * Race condition handling PSK identify hint 6568 6569 If PSK identity hints are received by a multi-threaded client then 6570 the values are wrongly updated in the parent SSL_CTX structure. This can 6571 result in a race condition potentially leading to a double free of the 6572 identify hint data. 6573 ([CVE-2015-3196]) 6574 6575 *Stephen Henson* 6576 6577### Changes between 1.0.1n and 1.0.1o [12 Jun 2015] 6578 6579 * Fix HMAC ABI incompatibility. The previous version introduced an ABI 6580 incompatibility in the handling of HMAC. The previous ABI has now been 6581 restored. 6582 6583### Changes between 1.0.1m and 1.0.1n [11 Jun 2015] 6584 6585 * Malformed ECParameters causes infinite loop 6586 6587 When processing an ECParameters structure OpenSSL enters an infinite loop 6588 if the curve specified is over a specially malformed binary polynomial 6589 field. 6590 6591 This can be used to perform denial of service against any 6592 system which processes public keys, certificate requests or 6593 certificates. This includes TLS clients and TLS servers with 6594 client authentication enabled. 6595 6596 This issue was reported to OpenSSL by Joseph Barr-Pixton. 6597 ([CVE-2015-1788]) 6598 6599 *Andy Polyakov* 6600 6601 * Exploitable out-of-bounds read in X509_cmp_time 6602 6603 X509_cmp_time does not properly check the length of the ASN1_TIME 6604 string and can read a few bytes out of bounds. In addition, 6605 X509_cmp_time accepts an arbitrary number of fractional seconds in the 6606 time string. 6607 6608 An attacker can use this to craft malformed certificates and CRLs of 6609 various sizes and potentially cause a segmentation fault, resulting in 6610 a DoS on applications that verify certificates or CRLs. TLS clients 6611 that verify CRLs are affected. TLS clients and servers with client 6612 authentication enabled may be affected if they use custom verification 6613 callbacks. 6614 6615 This issue was reported to OpenSSL by Robert Swiecki (Google), and 6616 independently by Hanno Böck. 6617 ([CVE-2015-1789]) 6618 6619 *Emilia Käsper* 6620 6621 * PKCS7 crash with missing EnvelopedContent 6622 6623 The PKCS#7 parsing code does not handle missing inner EncryptedContent 6624 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs 6625 with missing content and trigger a NULL pointer dereference on parsing. 6626 6627 Applications that decrypt PKCS#7 data or otherwise parse PKCS#7 6628 structures from untrusted sources are affected. OpenSSL clients and 6629 servers are not affected. 6630 6631 This issue was reported to OpenSSL by Michal Zalewski (Google). 6632 ([CVE-2015-1790]) 6633 6634 *Emilia Käsper* 6635 6636 * CMS verify infinite loop with unknown hash function 6637 6638 When verifying a signedData message the CMS code can enter an infinite loop 6639 if presented with an unknown hash function OID. This can be used to perform 6640 denial of service against any system which verifies signedData messages using 6641 the CMS code. 6642 This issue was reported to OpenSSL by Johannes Bauer. 6643 ([CVE-2015-1792]) 6644 6645 *Stephen Henson* 6646 6647 * Race condition handling NewSessionTicket 6648 6649 If a NewSessionTicket is received by a multi-threaded client when attempting to 6650 reuse a previous ticket then a race condition can occur potentially leading to 6651 a double free of the ticket data. 6652 ([CVE-2015-1791]) 6653 6654 *Matt Caswell* 6655 6656 * Reject DH handshakes with parameters shorter than 768 bits. 6657 6658 *Kurt Roeckx and Emilia Kasper* 6659 6660 * dhparam: generate 2048-bit parameters by default. 6661 6662 *Kurt Roeckx and Emilia Kasper* 6663 6664### Changes between 1.0.1l and 1.0.1m [19 Mar 2015] 6665 6666 * Segmentation fault in ASN1_TYPE_cmp fix 6667 6668 The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is 6669 made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check 6670 certificate signature algorithm consistency this can be used to crash any 6671 certificate verification operation and exploited in a DoS attack. Any 6672 application which performs certificate verification is vulnerable including 6673 OpenSSL clients and servers which enable client authentication. 6674 ([CVE-2015-0286]) 6675 6676 *Stephen Henson* 6677 6678 * ASN.1 structure reuse memory corruption fix 6679 6680 Reusing a structure in ASN.1 parsing may allow an attacker to cause 6681 memory corruption via an invalid write. Such reuse is and has been 6682 strongly discouraged and is believed to be rare. 6683 6684 Applications that parse structures containing CHOICE or ANY DEFINED BY 6685 components may be affected. Certificate parsing (d2i_X509 and related 6686 functions) are however not affected. OpenSSL clients and servers are 6687 not affected. 6688 ([CVE-2015-0287]) 6689 6690 *Stephen Henson* 6691 6692 * PKCS7 NULL pointer dereferences fix 6693 6694 The PKCS#7 parsing code does not handle missing outer ContentInfo 6695 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with 6696 missing content and trigger a NULL pointer dereference on parsing. 6697 6698 Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or 6699 otherwise parse PKCS#7 structures from untrusted sources are 6700 affected. OpenSSL clients and servers are not affected. 6701 6702 This issue was reported to OpenSSL by Michal Zalewski (Google). 6703 ([CVE-2015-0289]) 6704 6705 *Emilia Käsper* 6706 6707 * DoS via reachable assert in SSLv2 servers fix 6708 6709 A malicious client can trigger an OPENSSL_assert (i.e., an abort) in 6710 servers that both support SSLv2 and enable export cipher suites by sending 6711 a specially crafted SSLv2 CLIENT-MASTER-KEY message. 6712 6713 This issue was discovered by Sean Burford (Google) and Emilia Käsper 6714 (OpenSSL development team). 6715 ([CVE-2015-0293]) 6716 6717 *Emilia Käsper* 6718 6719 * Use After Free following d2i_ECPrivatekey error fix 6720 6721 A malformed EC private key file consumed via the d2i_ECPrivateKey function 6722 could cause a use after free condition. This, in turn, could cause a double 6723 free in several private key parsing functions (such as d2i_PrivateKey 6724 or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption 6725 for applications that receive EC private keys from untrusted 6726 sources. This scenario is considered rare. 6727 6728 This issue was discovered by the BoringSSL project and fixed in their 6729 commit 517073cd4b. 6730 ([CVE-2015-0209]) 6731 6732 *Matt Caswell* 6733 6734 * X509_to_X509_REQ NULL pointer deref fix 6735 6736 The function X509_to_X509_REQ will crash with a NULL pointer dereference if 6737 the certificate key is invalid. This function is rarely used in practice. 6738 6739 This issue was discovered by Brian Carpenter. 6740 ([CVE-2015-0288]) 6741 6742 *Stephen Henson* 6743 6744 * Removed the export ciphers from the DEFAULT ciphers 6745 6746 *Kurt Roeckx* 6747 6748### Changes between 1.0.1k and 1.0.1l [15 Jan 2015] 6749 6750 * Build fixes for the Windows and OpenVMS platforms 6751 6752 *Matt Caswell and Richard Levitte* 6753 6754### Changes between 1.0.1j and 1.0.1k [8 Jan 2015] 6755 6756 * Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS 6757 message can cause a segmentation fault in OpenSSL due to a NULL pointer 6758 dereference. This could lead to a Denial Of Service attack. Thanks to 6759 Markus Stenberg of Cisco Systems, Inc. for reporting this issue. 6760 ([CVE-2014-3571]) 6761 6762 *Steve Henson* 6763 6764 * Fix DTLS memory leak in dtls1_buffer_record. A memory leak can occur in the 6765 dtls1_buffer_record function under certain conditions. In particular this 6766 could occur if an attacker sent repeated DTLS records with the same 6767 sequence number but for the next epoch. The memory leak could be exploited 6768 by an attacker in a Denial of Service attack through memory exhaustion. 6769 Thanks to Chris Mueller for reporting this issue. 6770 ([CVE-2015-0206]) 6771 6772 *Matt Caswell* 6773 6774 * Fix issue where no-ssl3 configuration sets method to NULL. When openssl is 6775 built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl 6776 method would be set to NULL which could later result in a NULL pointer 6777 dereference. Thanks to Frank Schmirler for reporting this issue. 6778 ([CVE-2014-3569]) 6779 6780 *Kurt Roeckx* 6781 6782 * Abort handshake if server key exchange message is omitted for ephemeral 6783 ECDH ciphersuites. 6784 6785 Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for 6786 reporting this issue. 6787 ([CVE-2014-3572]) 6788 6789 *Steve Henson* 6790 6791 * Remove non-export ephemeral RSA code on client and server. This code 6792 violated the TLS standard by allowing the use of temporary RSA keys in 6793 non-export ciphersuites and could be used by a server to effectively 6794 downgrade the RSA key length used to a value smaller than the server 6795 certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at 6796 INRIA or reporting this issue. 6797 ([CVE-2015-0204]) 6798 6799 *Steve Henson* 6800 6801 * Fixed issue where DH client certificates are accepted without verification. 6802 An OpenSSL server will accept a DH certificate for client authentication 6803 without the certificate verify message. This effectively allows a client to 6804 authenticate without the use of a private key. This only affects servers 6805 which trust a client certificate authority which issues certificates 6806 containing DH keys: these are extremely rare and hardly ever encountered. 6807 Thanks for Karthikeyan Bhargavan of the PROSECCO team at INRIA or reporting 6808 this issue. 6809 ([CVE-2015-0205]) 6810 6811 *Steve Henson* 6812 6813 * Ensure that the session ID context of an SSL is updated when its 6814 SSL_CTX is updated via SSL_set_SSL_CTX. 6815 6816 The session ID context is typically set from the parent SSL_CTX, 6817 and can vary with the CTX. 6818 6819 *Adam Langley* 6820 6821 * Fix various certificate fingerprint issues. 6822 6823 By using non-DER or invalid encodings outside the signed portion of a 6824 certificate the fingerprint can be changed without breaking the signature. 6825 Although no details of the signed portion of the certificate can be changed 6826 this can cause problems with some applications: e.g. those using the 6827 certificate fingerprint for blacklists. 6828 6829 1. Reject signatures with non zero unused bits. 6830 6831 If the BIT STRING containing the signature has non zero unused bits reject 6832 the signature. All current signature algorithms require zero unused bits. 6833 6834 2. Check certificate algorithm consistency. 6835 6836 Check the AlgorithmIdentifier inside TBS matches the one in the 6837 certificate signature. NB: this will result in signature failure 6838 errors for some broken certificates. 6839 6840 Thanks to Konrad Kraszewski from Google for reporting this issue. 6841 6842 3. Check DSA/ECDSA signatures use DER. 6843 6844 Re-encode DSA/ECDSA signatures and compare with the original received 6845 signature. Return an error if there is a mismatch. 6846 6847 This will reject various cases including garbage after signature 6848 (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS 6849 program for discovering this case) and use of BER or invalid ASN.1 INTEGERs 6850 (negative or with leading zeroes). 6851 6852 Further analysis was conducted and fixes were developed by Stephen Henson 6853 of the OpenSSL core team. 6854 6855 ([CVE-2014-8275]) 6856 6857 *Steve Henson* 6858 6859 * Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect 6860 results on some platforms, including x86_64. This bug occurs at random 6861 with a very low probability, and is not known to be exploitable in any 6862 way, though its exact impact is difficult to determine. Thanks to Pieter 6863 Wuille (Blockstream) who reported this issue and also suggested an initial 6864 fix. Further analysis was conducted by the OpenSSL development team and 6865 Adam Langley of Google. The final fix was developed by Andy Polyakov of 6866 the OpenSSL core team. 6867 ([CVE-2014-3570]) 6868 6869 *Andy Polyakov* 6870 6871 * Do not resume sessions on the server if the negotiated protocol 6872 version does not match the session's version. Resuming with a different 6873 version, while not strictly forbidden by the RFC, is of questionable 6874 sanity and breaks all known clients. 6875 6876 *David Benjamin, Emilia Käsper* 6877 6878 * Tighten handling of the ChangeCipherSpec (CCS) message: reject 6879 early CCS messages during renegotiation. (Note that because 6880 renegotiation is encrypted, this early CCS was not exploitable.) 6881 6882 *Emilia Käsper* 6883 6884 * Tighten client-side session ticket handling during renegotiation: 6885 ensure that the client only accepts a session ticket if the server sends 6886 the extension anew in the ServerHello. Previously, a TLS client would 6887 reuse the old extension state and thus accept a session ticket if one was 6888 announced in the initial ServerHello. 6889 6890 Similarly, ensure that the client requires a session ticket if one 6891 was advertised in the ServerHello. Previously, a TLS client would 6892 ignore a missing NewSessionTicket message. 6893 6894 *Emilia Käsper* 6895 6896### Changes between 1.0.1i and 1.0.1j [15 Oct 2014] 6897 6898 * SRTP Memory Leak. 6899 6900 A flaw in the DTLS SRTP extension parsing code allows an attacker, who 6901 sends a carefully crafted handshake message, to cause OpenSSL to fail 6902 to free up to 64k of memory causing a memory leak. This could be 6903 exploited in a Denial Of Service attack. This issue affects OpenSSL 6904 1.0.1 server implementations for both SSL/TLS and DTLS regardless of 6905 whether SRTP is used or configured. Implementations of OpenSSL that 6906 have been compiled with OPENSSL_NO_SRTP defined are not affected. 6907 6908 The fix was developed by the OpenSSL team. 6909 ([CVE-2014-3513]) 6910 6911 *OpenSSL team* 6912 6913 * Session Ticket Memory Leak. 6914 6915 When an OpenSSL SSL/TLS/DTLS server receives a session ticket the 6916 integrity of that ticket is first verified. In the event of a session 6917 ticket integrity check failing, OpenSSL will fail to free memory 6918 causing a memory leak. By sending a large number of invalid session 6919 tickets an attacker could exploit this issue in a Denial Of Service 6920 attack. 6921 ([CVE-2014-3567]) 6922 6923 *Steve Henson* 6924 6925 * Build option no-ssl3 is incomplete. 6926 6927 When OpenSSL is configured with "no-ssl3" as a build option, servers 6928 could accept and complete a SSL 3.0 handshake, and clients could be 6929 configured to send them. 6930 ([CVE-2014-3568]) 6931 6932 *Akamai and the OpenSSL team* 6933 6934 * Add support for TLS_FALLBACK_SCSV. 6935 Client applications doing fallback retries should call 6936 SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV). 6937 ([CVE-2014-3566]) 6938 6939 *Adam Langley, Bodo Moeller* 6940 6941 * Add additional DigestInfo checks. 6942 6943 Re-encode DigestInto in DER and check against the original when 6944 verifying RSA signature: this will reject any improperly encoded 6945 DigestInfo structures. 6946 6947 Note: this is a precautionary measure and no attacks are currently known. 6948 6949 *Steve Henson* 6950 6951### Changes between 1.0.1h and 1.0.1i [6 Aug 2014] 6952 6953 * Fix SRP buffer overrun vulnerability. Invalid parameters passed to the 6954 SRP code can be overrun an internal buffer. Add sanity check that 6955 g, A, B < N to SRP code. 6956 6957 Thanks to Sean Devlin and Watson Ladd of Cryptography Services, NCC 6958 Group for discovering this issue. 6959 ([CVE-2014-3512]) 6960 6961 *Steve Henson* 6962 6963 * A flaw in the OpenSSL SSL/TLS server code causes the server to negotiate 6964 TLS 1.0 instead of higher protocol versions when the ClientHello message 6965 is badly fragmented. This allows a man-in-the-middle attacker to force a 6966 downgrade to TLS 1.0 even if both the server and the client support a 6967 higher protocol version, by modifying the client's TLS records. 6968 6969 Thanks to David Benjamin and Adam Langley (Google) for discovering and 6970 researching this issue. 6971 ([CVE-2014-3511]) 6972 6973 *David Benjamin* 6974 6975 * OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject 6976 to a denial of service attack. A malicious server can crash the client 6977 with a null pointer dereference (read) by specifying an anonymous (EC)DH 6978 ciphersuite and sending carefully crafted handshake messages. 6979 6980 Thanks to Felix Gröbert (Google) for discovering and researching this 6981 issue. 6982 ([CVE-2014-3510]) 6983 6984 *Emilia Käsper* 6985 6986 * By sending carefully crafted DTLS packets an attacker could cause openssl 6987 to leak memory. This can be exploited through a Denial of Service attack. 6988 Thanks to Adam Langley for discovering and researching this issue. 6989 ([CVE-2014-3507]) 6990 6991 *Adam Langley* 6992 6993 * An attacker can force openssl to consume large amounts of memory whilst 6994 processing DTLS handshake messages. This can be exploited through a 6995 Denial of Service attack. 6996 Thanks to Adam Langley for discovering and researching this issue. 6997 ([CVE-2014-3506]) 6998 6999 *Adam Langley* 7000 7001 * An attacker can force an error condition which causes openssl to crash 7002 whilst processing DTLS packets due to memory being freed twice. This 7003 can be exploited through a Denial of Service attack. 7004 Thanks to Adam Langley and Wan-Teh Chang for discovering and researching 7005 this issue. 7006 ([CVE-2014-3505]) 7007 7008 *Adam Langley* 7009 7010 * If a multithreaded client connects to a malicious server using a resumed 7011 session and the server sends an ec point format extension it could write 7012 up to 255 bytes to freed memory. 7013 7014 Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this 7015 issue. 7016 ([CVE-2014-3509]) 7017 7018 *Gabor Tyukasz* 7019 7020 * A malicious server can crash an OpenSSL client with a null pointer 7021 dereference (read) by specifying an SRP ciphersuite even though it was not 7022 properly negotiated with the client. This can be exploited through a 7023 Denial of Service attack. 7024 7025 Thanks to Joonas Kuorilehto and Riku Hietamäki (Codenomicon) for 7026 discovering and researching this issue. 7027 ([CVE-2014-5139]) 7028 7029 *Steve Henson* 7030 7031 * A flaw in OBJ_obj2txt may cause pretty printing functions such as 7032 X509_name_oneline, X509_name_print_ex et al. to leak some information 7033 from the stack. Applications may be affected if they echo pretty printing 7034 output to the attacker. 7035 7036 Thanks to Ivan Fratric (Google) for discovering this issue. 7037 ([CVE-2014-3508]) 7038 7039 *Emilia Käsper, and Steve Henson* 7040 7041 * Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.) 7042 for corner cases. (Certain input points at infinity could lead to 7043 bogus results, with non-infinity inputs mapped to infinity too.) 7044 7045 *Bodo Moeller* 7046 7047### Changes between 1.0.1g and 1.0.1h [5 Jun 2014] 7048 7049 * Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted 7050 handshake can force the use of weak keying material in OpenSSL 7051 SSL/TLS clients and servers. 7052 7053 Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and 7054 researching this issue. ([CVE-2014-0224]) 7055 7056 *KIKUCHI Masashi, Steve Henson* 7057 7058 * Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an 7059 OpenSSL DTLS client the code can be made to recurse eventually crashing 7060 in a DoS attack. 7061 7062 Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue. 7063 ([CVE-2014-0221]) 7064 7065 *Imre Rad, Steve Henson* 7066 7067 * Fix DTLS invalid fragment vulnerability. A buffer overrun attack can 7068 be triggered by sending invalid DTLS fragments to an OpenSSL DTLS 7069 client or server. This is potentially exploitable to run arbitrary 7070 code on a vulnerable client or server. 7071 7072 Thanks to Jüri Aedla for reporting this issue. ([CVE-2014-0195]) 7073 7074 *Jüri Aedla, Steve Henson* 7075 7076 * Fix bug in TLS code where clients enable anonymous ECDH ciphersuites 7077 are subject to a denial of service attack. 7078 7079 Thanks to Felix Gröbert and Ivan Fratric at Google for discovering 7080 this issue. ([CVE-2014-3470]) 7081 7082 *Felix Gröbert, Ivan Fratric, Steve Henson* 7083 7084 * Harmonize version and its documentation. -f flag is used to display 7085 compilation flags. 7086 7087 *mancha <mancha1@zoho.com>* 7088 7089 * Fix eckey_priv_encode so it immediately returns an error upon a failure 7090 in i2d_ECPrivateKey. 7091 7092 *mancha <mancha1@zoho.com>* 7093 7094 * Fix some double frees. These are not thought to be exploitable. 7095 7096 *mancha <mancha1@zoho.com>* 7097 7098### Changes between 1.0.1f and 1.0.1g [7 Apr 2014] 7099 7100 * A missing bounds check in the handling of the TLS heartbeat extension 7101 can be used to reveal up to 64k of memory to a connected client or 7102 server. 7103 7104 Thanks for Neel Mehta of Google Security for discovering this bug and to 7105 Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for 7106 preparing the fix ([CVE-2014-0160]) 7107 7108 *Adam Langley, Bodo Moeller* 7109 7110 * Fix for the attack described in the paper "Recovering OpenSSL 7111 ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack" 7112 by Yuval Yarom and Naomi Benger. Details can be obtained from: 7113 <http://eprint.iacr.org/2014/140> 7114 7115 Thanks to Yuval Yarom and Naomi Benger for discovering this 7116 flaw and to Yuval Yarom for supplying a fix ([CVE-2014-0076]) 7117 7118 *Yuval Yarom and Naomi Benger* 7119 7120 * TLS pad extension: draft-agl-tls-padding-03 7121 7122 Workaround for the "TLS hang bug" (see FAQ and PR#2771): if the 7123 TLS client Hello record length value would otherwise be > 255 and 7124 less that 512 pad with a dummy extension containing zeroes so it 7125 is at least 512 bytes long. 7126 7127 *Adam Langley, Steve Henson* 7128 7129### Changes between 1.0.1e and 1.0.1f [6 Jan 2014] 7130 7131 * Fix for TLS record tampering bug. A carefully crafted invalid 7132 handshake could crash OpenSSL with a NULL pointer exception. 7133 Thanks to Anton Johansson for reporting this issues. 7134 ([CVE-2013-4353]) 7135 7136 * Keep original DTLS digest and encryption contexts in retransmission 7137 structures so we can use the previous session parameters if they need 7138 to be resent. ([CVE-2013-6450]) 7139 7140 *Steve Henson* 7141 7142 * Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which 7143 avoids preferring ECDHE-ECDSA ciphers when the client appears to be 7144 Safari on OS X. Safari on OS X 10.8..10.8.3 advertises support for 7145 several ECDHE-ECDSA ciphers, but fails to negotiate them. The bug 7146 is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing 7147 10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer. 7148 7149 *Rob Stradling, Adam Langley* 7150 7151### Changes between 1.0.1d and 1.0.1e [11 Feb 2013] 7152 7153 * Correct fix for CVE-2013-0169. The original didn't work on AES-NI 7154 supporting platforms or when small records were transferred. 7155 7156 *Andy Polyakov, Steve Henson* 7157 7158### Changes between 1.0.1c and 1.0.1d [5 Feb 2013] 7159 7160 * Make the decoding of SSLv3, TLS and DTLS CBC records constant time. 7161 7162 This addresses the flaw in CBC record processing discovered by 7163 Nadhem Alfardan and Kenny Paterson. Details of this attack can be found 7164 at: <http://www.isg.rhul.ac.uk/tls/> 7165 7166 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 7167 Security Group at Royal Holloway, University of London 7168 (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and 7169 Emilia Käsper for the initial patch. 7170 ([CVE-2013-0169]) 7171 7172 *Emilia Käsper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson* 7173 7174 * Fix flaw in AESNI handling of TLS 1.2 and 1.1 records for CBC mode 7175 ciphersuites which can be exploited in a denial of service attack. 7176 Thanks go to and to Adam Langley <agl@chromium.org> for discovering 7177 and detecting this bug and to Wolfgang Ettlinger 7178 <wolfgang.ettlinger@gmail.com> for independently discovering this issue. 7179 ([CVE-2012-2686]) 7180 7181 *Adam Langley* 7182 7183 * Return an error when checking OCSP signatures when key is NULL. 7184 This fixes a DoS attack. ([CVE-2013-0166]) 7185 7186 *Steve Henson* 7187 7188 * Make openssl verify return errors. 7189 7190 *Chris Palmer <palmer@google.com> and Ben Laurie* 7191 7192 * Call OCSP Stapling callback after ciphersuite has been chosen, so 7193 the right response is stapled. Also change SSL_get_certificate() 7194 so it returns the certificate actually sent. 7195 See <http://rt.openssl.org/Ticket/Display.html?id=2836>. 7196 7197 *Rob Stradling <rob.stradling@comodo.com>* 7198 7199 * Fix possible deadlock when decoding public keys. 7200 7201 *Steve Henson* 7202 7203 * Don't use TLS 1.0 record version number in initial client hello 7204 if renegotiating. 7205 7206 *Steve Henson* 7207 7208### Changes between 1.0.1b and 1.0.1c [10 May 2012] 7209 7210 * Sanity check record length before skipping explicit IV in TLS 7211 1.2, 1.1 and DTLS to fix DoS attack. 7212 7213 Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic 7214 fuzzing as a service testing platform. 7215 ([CVE-2012-2333]) 7216 7217 *Steve Henson* 7218 7219 * Initialise tkeylen properly when encrypting CMS messages. 7220 Thanks to Solar Designer of Openwall for reporting this issue. 7221 7222 *Steve Henson* 7223 7224 * In FIPS mode don't try to use composite ciphers as they are not 7225 approved. 7226 7227 *Steve Henson* 7228 7229### Changes between 1.0.1a and 1.0.1b [26 Apr 2012] 7230 7231 * OpenSSL 1.0.0 sets SSL_OP_ALL to 0x80000FFFL and OpenSSL 1.0.1 and 7232 1.0.1a set SSL_OP_NO_TLSv1_1 to 0x00000400L which would unfortunately 7233 mean any application compiled against OpenSSL 1.0.0 headers setting 7234 SSL_OP_ALL would also set SSL_OP_NO_TLSv1_1, unintentionally disabling 7235 TLS 1.1 also. Fix this by changing the value of SSL_OP_NO_TLSv1_1 to 7236 0x10000000L Any application which was previously compiled against 7237 OpenSSL 1.0.1 or 1.0.1a headers and which cares about SSL_OP_NO_TLSv1_1 7238 will need to be recompiled as a result. Letting be results in 7239 inability to disable specifically TLS 1.1 and in client context, 7240 in unlike event, limit maximum offered version to TLS 1.0 [see below]. 7241 7242 *Steve Henson* 7243 7244 * In order to ensure interoperability SSL_OP_NO_protocolX does not 7245 disable just protocol X, but all protocols above X *if* there are 7246 protocols *below* X still enabled. In more practical terms it means 7247 that if application wants to disable TLS1.0 in favor of TLS1.1 and 7248 above, it's not sufficient to pass `SSL_OP_NO_TLSv1`, one has to pass 7249 `SSL_OP_NO_TLSv1|SSL_OP_NO_SSLv3|SSL_OP_NO_SSLv2`. This applies to 7250 client side. 7251 7252 *Andy Polyakov* 7253 7254### Changes between 1.0.1 and 1.0.1a [19 Apr 2012] 7255 7256 * Check for potentially exploitable overflows in asn1_d2i_read_bio 7257 BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer 7258 in CRYPTO_realloc_clean. 7259 7260 Thanks to Tavis Ormandy, Google Security Team, for discovering this 7261 issue and to Adam Langley <agl@chromium.org> for fixing it. 7262 ([CVE-2012-2110]) 7263 7264 *Adam Langley (Google), Tavis Ormandy, Google Security Team* 7265 7266 * Don't allow TLS 1.2 SHA-256 ciphersuites in TLS 1.0, 1.1 connections. 7267 7268 *Adam Langley* 7269 7270 * Workarounds for some broken servers that "hang" if a client hello 7271 record length exceeds 255 bytes. 7272 7273 1. Do not use record version number > TLS 1.0 in initial client 7274 hello: some (but not all) hanging servers will now work. 7275 2. If we set OPENSSL_MAX_TLS1_2_CIPHER_LENGTH this will truncate 7276 the number of ciphers sent in the client hello. This should be 7277 set to an even number, such as 50, for example by passing: 7278 -DOPENSSL_MAX_TLS1_2_CIPHER_LENGTH=50 to config or Configure. 7279 Most broken servers should now work. 7280 3. If all else fails setting OPENSSL_NO_TLS1_2_CLIENT will disable 7281 TLS 1.2 client support entirely. 7282 7283 *Steve Henson* 7284 7285 * Fix SEGV in Vector Permutation AES module observed in OpenSSH. 7286 7287 *Andy Polyakov* 7288 7289### Changes between 1.0.0h and 1.0.1 [14 Mar 2012] 7290 7291 * Add compatibility with old MDC2 signatures which use an ASN1 OCTET 7292 STRING form instead of a DigestInfo. 7293 7294 *Steve Henson* 7295 7296 * The format used for MDC2 RSA signatures is inconsistent between EVP 7297 and the RSA_sign/RSA_verify functions. This was made more apparent when 7298 OpenSSL used RSA_sign/RSA_verify for some RSA signatures in particular 7299 those which went through EVP_PKEY_METHOD in 1.0.0 and later. Detect 7300 the correct format in RSA_verify so both forms transparently work. 7301 7302 *Steve Henson* 7303 7304 * Some servers which support TLS 1.0 can choke if we initially indicate 7305 support for TLS 1.2 and later renegotiate using TLS 1.0 in the RSA 7306 encrypted premaster secret. As a workaround use the maximum permitted 7307 client version in client hello, this should keep such servers happy 7308 and still work with previous versions of OpenSSL. 7309 7310 *Steve Henson* 7311 7312 * Add support for TLS/DTLS heartbeats. 7313 7314 *Robin Seggelmann <seggelmann@fh-muenster.de>* 7315 7316 * Add support for SCTP. 7317 7318 *Robin Seggelmann <seggelmann@fh-muenster.de>* 7319 7320 * Improved PRNG seeding for VOS. 7321 7322 *Paul Green <Paul.Green@stratus.com>* 7323 7324 * Extensive assembler packs updates, most notably: 7325 7326 - x86[_64]: AES-NI, PCLMULQDQ, RDRAND support; 7327 - x86[_64]: SSSE3 support (SHA1, vector-permutation AES); 7328 - x86_64: bit-sliced AES implementation; 7329 - ARM: NEON support, contemporary platforms optimizations; 7330 - s390x: z196 support; 7331 - `*`: GHASH and GF(2^m) multiplication implementations; 7332 7333 *Andy Polyakov* 7334 7335 * Make TLS-SRP code conformant with RFC 5054 API cleanup 7336 (removal of unnecessary code) 7337 7338 *Peter Sylvester <peter.sylvester@edelweb.fr>* 7339 7340 * Add TLS key material exporter from RFC 5705. 7341 7342 *Eric Rescorla* 7343 7344 * Add DTLS-SRTP negotiation from RFC 5764. 7345 7346 *Eric Rescorla* 7347 7348 * Add Next Protocol Negotiation, 7349 <http://tools.ietf.org/html/draft-agl-tls-nextprotoneg-00>. Can be 7350 disabled with a no-npn flag to config or Configure. Code donated 7351 by Google. 7352 7353 *Adam Langley <agl@google.com> and Ben Laurie* 7354 7355 * Add optional 64-bit optimized implementations of elliptic curves NIST-P224, 7356 NIST-P256, NIST-P521, with constant-time single point multiplication on 7357 typical inputs. Compiler support for the nonstandard type `__uint128_t` is 7358 required to use this (present in gcc 4.4 and later, for 64-bit builds). 7359 Code made available under Apache License version 2.0. 7360 7361 Specify "enable-ec_nistp_64_gcc_128" on the Configure (or config) command 7362 line to include this in your build of OpenSSL, and run "make depend" (or 7363 "make update"). This enables the following EC_METHODs: 7364 7365 EC_GFp_nistp224_method() 7366 EC_GFp_nistp256_method() 7367 EC_GFp_nistp521_method() 7368 7369 EC_GROUP_new_by_curve_name() will automatically use these (while 7370 EC_GROUP_new_curve_GFp() currently prefers the more flexible 7371 implementations). 7372 7373 *Emilia Käsper, Adam Langley, Bodo Moeller (Google)* 7374 7375 * Use type ossl_ssize_t instead of ssize_t which isn't available on 7376 all platforms. Move ssize_t definition from e_os.h to the public 7377 header file e_os2.h as it now appears in public header file cms.h 7378 7379 *Steve Henson* 7380 7381 * New -sigopt option to the ca, req and x509 utilities. Additional 7382 signature parameters can be passed using this option and in 7383 particular PSS. 7384 7385 *Steve Henson* 7386 7387 * Add RSA PSS signing function. This will generate and set the 7388 appropriate AlgorithmIdentifiers for PSS based on those in the 7389 corresponding EVP_MD_CTX structure. No application support yet. 7390 7391 *Steve Henson* 7392 7393 * Support for companion algorithm specific ASN1 signing routines. 7394 New function ASN1_item_sign_ctx() signs a pre-initialised 7395 EVP_MD_CTX structure and sets AlgorithmIdentifiers based on 7396 the appropriate parameters. 7397 7398 *Steve Henson* 7399 7400 * Add new algorithm specific ASN1 verification initialisation function 7401 to EVP_PKEY_ASN1_METHOD: this is not in EVP_PKEY_METHOD since the ASN1 7402 handling will be the same no matter what EVP_PKEY_METHOD is used. 7403 Add a PSS handler to support verification of PSS signatures: checked 7404 against a number of sample certificates. 7405 7406 *Steve Henson* 7407 7408 * Add signature printing for PSS. Add PSS OIDs. 7409 7410 *Steve Henson, Martin Kaiser <lists@kaiser.cx>* 7411 7412 * Add algorithm specific signature printing. An individual ASN1 method 7413 can now print out signatures instead of the standard hex dump. 7414 7415 More complex signatures (e.g. PSS) can print out more meaningful 7416 information. Include DSA version that prints out the signature 7417 parameters r, s. 7418 7419 *Steve Henson* 7420 7421 * Password based recipient info support for CMS library: implementing 7422 RFC3211. 7423 7424 *Steve Henson* 7425 7426 * Split password based encryption into PBES2 and PBKDF2 functions. This 7427 neatly separates the code into cipher and PBE sections and is required 7428 for some algorithms that split PBES2 into separate pieces (such as 7429 password based CMS). 7430 7431 *Steve Henson* 7432 7433 * Session-handling fixes: 7434 - Fix handling of connections that are resuming with a session ID, 7435 but also support Session Tickets. 7436 - Fix a bug that suppressed issuing of a new ticket if the client 7437 presented a ticket with an expired session. 7438 - Try to set the ticket lifetime hint to something reasonable. 7439 - Make tickets shorter by excluding irrelevant information. 7440 - On the client side, don't ignore renewed tickets. 7441 7442 *Adam Langley, Bodo Moeller (Google)* 7443 7444 * Fix PSK session representation. 7445 7446 *Bodo Moeller* 7447 7448 * Add RC4-MD5 and AESNI-SHA1 "stitched" implementations. 7449 7450 This work was sponsored by Intel. 7451 7452 *Andy Polyakov* 7453 7454 * Add GCM support to TLS library. Some custom code is needed to split 7455 the IV between the fixed (from PRF) and explicit (from TLS record) 7456 portions. This adds all GCM ciphersuites supported by RFC5288 and 7457 RFC5289. Generalise some `AES*` cipherstrings to include GCM and 7458 add a special AESGCM string for GCM only. 7459 7460 *Steve Henson* 7461 7462 * Expand range of ctrls for AES GCM. Permit setting invocation 7463 field on decrypt and retrieval of invocation field only on encrypt. 7464 7465 *Steve Henson* 7466 7467 * Add HMAC ECC ciphersuites from RFC5289. Include SHA384 PRF support. 7468 As required by RFC5289 these ciphersuites cannot be used if for 7469 versions of TLS earlier than 1.2. 7470 7471 *Steve Henson* 7472 7473 * For FIPS capable OpenSSL interpret a NULL default public key method 7474 as unset and return the appropriate default but do *not* set the default. 7475 This means we can return the appropriate method in applications that 7476 switch between FIPS and non-FIPS modes. 7477 7478 *Steve Henson* 7479 7480 * Redirect HMAC and CMAC operations to FIPS module in FIPS mode. If an 7481 ENGINE is used then we cannot handle that in the FIPS module so we 7482 keep original code iff non-FIPS operations are allowed. 7483 7484 *Steve Henson* 7485 7486 * Add -attime option to openssl utilities. 7487 7488 *Peter Eckersley <pde@eff.org>, Ben Laurie and Steve Henson* 7489 7490 * Redirect DSA and DH operations to FIPS module in FIPS mode. 7491 7492 *Steve Henson* 7493 7494 * Redirect ECDSA and ECDH operations to FIPS module in FIPS mode. Also use 7495 FIPS EC methods unconditionally for now. 7496 7497 *Steve Henson* 7498 7499 * New build option no-ec2m to disable characteristic 2 code. 7500 7501 *Steve Henson* 7502 7503 * Backport libcrypto audit of return value checking from 1.1.0-dev; not 7504 all cases can be covered as some introduce binary incompatibilities. 7505 7506 *Steve Henson* 7507 7508 * Redirect RSA operations to FIPS module including keygen, 7509 encrypt, decrypt, sign and verify. Block use of non FIPS RSA methods. 7510 7511 *Steve Henson* 7512 7513 * Add similar low-level API blocking to ciphers. 7514 7515 *Steve Henson* 7516 7517 * low-level digest APIs are not approved in FIPS mode: any attempt 7518 to use these will cause a fatal error. Applications that *really* want 7519 to use them can use the `private_*` version instead. 7520 7521 *Steve Henson* 7522 7523 * Redirect cipher operations to FIPS module for FIPS builds. 7524 7525 *Steve Henson* 7526 7527 * Redirect digest operations to FIPS module for FIPS builds. 7528 7529 *Steve Henson* 7530 7531 * Update build system to add "fips" flag which will link in fipscanister.o 7532 for static and shared library builds embedding a signature if needed. 7533 7534 *Steve Henson* 7535 7536 * Output TLS supported curves in preference order instead of numerical 7537 order. This is currently hardcoded for the highest order curves first. 7538 This should be configurable so applications can judge speed vs strength. 7539 7540 *Steve Henson* 7541 7542 * Add TLS v1.2 server support for client authentication. 7543 7544 *Steve Henson* 7545 7546 * Add support for FIPS mode in ssl library: disable SSLv3, non-FIPS ciphers 7547 and enable MD5. 7548 7549 *Steve Henson* 7550 7551 * Functions FIPS_mode_set() and FIPS_mode() which call the underlying 7552 FIPS modules versions. 7553 7554 *Steve Henson* 7555 7556 * Add TLS v1.2 client side support for client authentication. Keep cache 7557 of handshake records longer as we don't know the hash algorithm to use 7558 until after the certificate request message is received. 7559 7560 *Steve Henson* 7561 7562 * Initial TLS v1.2 client support. Add a default signature algorithms 7563 extension including all the algorithms we support. Parse new signature 7564 format in client key exchange. Relax some ECC signing restrictions for 7565 TLS v1.2 as indicated in RFC5246. 7566 7567 *Steve Henson* 7568 7569 * Add server support for TLS v1.2 signature algorithms extension. Switch 7570 to new signature format when needed using client digest preference. 7571 All server ciphersuites should now work correctly in TLS v1.2. No client 7572 support yet and no support for client certificates. 7573 7574 *Steve Henson* 7575 7576 * Initial TLS v1.2 support. Add new SHA256 digest to ssl code, switch 7577 to SHA256 for PRF when using TLS v1.2 and later. Add new SHA256 based 7578 ciphersuites. At present only RSA key exchange ciphersuites work with 7579 TLS v1.2. Add new option for TLS v1.2 replacing the old and obsolete 7580 SSL_OP_PKCS1_CHECK flags with SSL_OP_NO_TLSv1_2. New TLSv1.2 methods 7581 and version checking. 7582 7583 *Steve Henson* 7584 7585 * New option OPENSSL_NO_SSL_INTERN. If an application can be compiled 7586 with this defined it will not be affected by any changes to ssl internal 7587 structures. Add several utility functions to allow openssl application 7588 to work with OPENSSL_NO_SSL_INTERN defined. 7589 7590 *Steve Henson* 7591 7592 * A long standing patch to add support for SRP from EdelWeb (Peter 7593 Sylvester and Christophe Renou) was integrated. 7594 *Christophe Renou <christophe.renou@edelweb.fr>, Peter Sylvester 7595 <peter.sylvester@edelweb.fr>, Tom Wu <tjw@cs.stanford.edu>, and 7596 Ben Laurie* 7597 7598 * Add functions to copy EVP_PKEY_METHOD and retrieve flags and id. 7599 7600 *Steve Henson* 7601 7602 * Permit abbreviated handshakes when renegotiating using the function 7603 SSL_renegotiate_abbreviated(). 7604 7605 *Robin Seggelmann <seggelmann@fh-muenster.de>* 7606 7607 * Add call to ENGINE_register_all_complete() to 7608 ENGINE_load_builtin_engines(), so some implementations get used 7609 automatically instead of needing explicit application support. 7610 7611 *Steve Henson* 7612 7613 * Add support for TLS key exporter as described in RFC5705. 7614 7615 *Robin Seggelmann <seggelmann@fh-muenster.de>, Steve Henson* 7616 7617 * Initial TLSv1.1 support. Since TLSv1.1 is very similar to TLS v1.0 only 7618 a few changes are required: 7619 7620 Add SSL_OP_NO_TLSv1_1 flag. 7621 Add TLSv1_1 methods. 7622 Update version checking logic to handle version 1.1. 7623 Add explicit IV handling (ported from DTLS code). 7624 Add command line options to s_client/s_server. 7625 7626 *Steve Henson* 7627 7628OpenSSL 1.0.0 7629------------- 7630 7631### Changes between 1.0.0s and 1.0.0t [3 Dec 2015] 7632 7633 * X509_ATTRIBUTE memory leak 7634 7635 When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak 7636 memory. This structure is used by the PKCS#7 and CMS routines so any 7637 application which reads PKCS#7 or CMS data from untrusted sources is 7638 affected. SSL/TLS is not affected. 7639 7640 This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using 7641 libFuzzer. 7642 ([CVE-2015-3195]) 7643 7644 *Stephen Henson* 7645 7646 * Race condition handling PSK identify hint 7647 7648 If PSK identity hints are received by a multi-threaded client then 7649 the values are wrongly updated in the parent SSL_CTX structure. This can 7650 result in a race condition potentially leading to a double free of the 7651 identify hint data. 7652 ([CVE-2015-3196]) 7653 7654 *Stephen Henson* 7655 7656### Changes between 1.0.0r and 1.0.0s [11 Jun 2015] 7657 7658 * Malformed ECParameters causes infinite loop 7659 7660 When processing an ECParameters structure OpenSSL enters an infinite loop 7661 if the curve specified is over a specially malformed binary polynomial 7662 field. 7663 7664 This can be used to perform denial of service against any 7665 system which processes public keys, certificate requests or 7666 certificates. This includes TLS clients and TLS servers with 7667 client authentication enabled. 7668 7669 This issue was reported to OpenSSL by Joseph Barr-Pixton. 7670 ([CVE-2015-1788]) 7671 7672 *Andy Polyakov* 7673 7674 * Exploitable out-of-bounds read in X509_cmp_time 7675 7676 X509_cmp_time does not properly check the length of the ASN1_TIME 7677 string and can read a few bytes out of bounds. In addition, 7678 X509_cmp_time accepts an arbitrary number of fractional seconds in the 7679 time string. 7680 7681 An attacker can use this to craft malformed certificates and CRLs of 7682 various sizes and potentially cause a segmentation fault, resulting in 7683 a DoS on applications that verify certificates or CRLs. TLS clients 7684 that verify CRLs are affected. TLS clients and servers with client 7685 authentication enabled may be affected if they use custom verification 7686 callbacks. 7687 7688 This issue was reported to OpenSSL by Robert Swiecki (Google), and 7689 independently by Hanno Böck. 7690 ([CVE-2015-1789]) 7691 7692 *Emilia Käsper* 7693 7694 * PKCS7 crash with missing EnvelopedContent 7695 7696 The PKCS#7 parsing code does not handle missing inner EncryptedContent 7697 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs 7698 with missing content and trigger a NULL pointer dereference on parsing. 7699 7700 Applications that decrypt PKCS#7 data or otherwise parse PKCS#7 7701 structures from untrusted sources are affected. OpenSSL clients and 7702 servers are not affected. 7703 7704 This issue was reported to OpenSSL by Michal Zalewski (Google). 7705 ([CVE-2015-1790]) 7706 7707 *Emilia Käsper* 7708 7709 * CMS verify infinite loop with unknown hash function 7710 7711 When verifying a signedData message the CMS code can enter an infinite loop 7712 if presented with an unknown hash function OID. This can be used to perform 7713 denial of service against any system which verifies signedData messages using 7714 the CMS code. 7715 This issue was reported to OpenSSL by Johannes Bauer. 7716 ([CVE-2015-1792]) 7717 7718 *Stephen Henson* 7719 7720 * Race condition handling NewSessionTicket 7721 7722 If a NewSessionTicket is received by a multi-threaded client when attempting to 7723 reuse a previous ticket then a race condition can occur potentially leading to 7724 a double free of the ticket data. 7725 ([CVE-2015-1791]) 7726 7727 *Matt Caswell* 7728 7729### Changes between 1.0.0q and 1.0.0r [19 Mar 2015] 7730 7731 * Segmentation fault in ASN1_TYPE_cmp fix 7732 7733 The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is 7734 made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check 7735 certificate signature algorithm consistency this can be used to crash any 7736 certificate verification operation and exploited in a DoS attack. Any 7737 application which performs certificate verification is vulnerable including 7738 OpenSSL clients and servers which enable client authentication. 7739 ([CVE-2015-0286]) 7740 7741 *Stephen Henson* 7742 7743 * ASN.1 structure reuse memory corruption fix 7744 7745 Reusing a structure in ASN.1 parsing may allow an attacker to cause 7746 memory corruption via an invalid write. Such reuse is and has been 7747 strongly discouraged and is believed to be rare. 7748 7749 Applications that parse structures containing CHOICE or ANY DEFINED BY 7750 components may be affected. Certificate parsing (d2i_X509 and related 7751 functions) are however not affected. OpenSSL clients and servers are 7752 not affected. 7753 ([CVE-2015-0287]) 7754 7755 *Stephen Henson* 7756 7757 * PKCS7 NULL pointer dereferences fix 7758 7759 The PKCS#7 parsing code does not handle missing outer ContentInfo 7760 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with 7761 missing content and trigger a NULL pointer dereference on parsing. 7762 7763 Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or 7764 otherwise parse PKCS#7 structures from untrusted sources are 7765 affected. OpenSSL clients and servers are not affected. 7766 7767 This issue was reported to OpenSSL by Michal Zalewski (Google). 7768 ([CVE-2015-0289]) 7769 7770 *Emilia Käsper* 7771 7772 * DoS via reachable assert in SSLv2 servers fix 7773 7774 A malicious client can trigger an OPENSSL_assert (i.e., an abort) in 7775 servers that both support SSLv2 and enable export cipher suites by sending 7776 a specially crafted SSLv2 CLIENT-MASTER-KEY message. 7777 7778 This issue was discovered by Sean Burford (Google) and Emilia Käsper 7779 (OpenSSL development team). 7780 ([CVE-2015-0293]) 7781 7782 *Emilia Käsper* 7783 7784 * Use After Free following d2i_ECPrivatekey error fix 7785 7786 A malformed EC private key file consumed via the d2i_ECPrivateKey function 7787 could cause a use after free condition. This, in turn, could cause a double 7788 free in several private key parsing functions (such as d2i_PrivateKey 7789 or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption 7790 for applications that receive EC private keys from untrusted 7791 sources. This scenario is considered rare. 7792 7793 This issue was discovered by the BoringSSL project and fixed in their 7794 commit 517073cd4b. 7795 ([CVE-2015-0209]) 7796 7797 *Matt Caswell* 7798 7799 * X509_to_X509_REQ NULL pointer deref fix 7800 7801 The function X509_to_X509_REQ will crash with a NULL pointer dereference if 7802 the certificate key is invalid. This function is rarely used in practice. 7803 7804 This issue was discovered by Brian Carpenter. 7805 ([CVE-2015-0288]) 7806 7807 *Stephen Henson* 7808 7809 * Removed the export ciphers from the DEFAULT ciphers 7810 7811 *Kurt Roeckx* 7812 7813### Changes between 1.0.0p and 1.0.0q [15 Jan 2015] 7814 7815 * Build fixes for the Windows and OpenVMS platforms 7816 7817 *Matt Caswell and Richard Levitte* 7818 7819### Changes between 1.0.0o and 1.0.0p [8 Jan 2015] 7820 7821 * Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS 7822 message can cause a segmentation fault in OpenSSL due to a NULL pointer 7823 dereference. This could lead to a Denial Of Service attack. Thanks to 7824 Markus Stenberg of Cisco Systems, Inc. for reporting this issue. 7825 ([CVE-2014-3571]) 7826 7827 *Steve Henson* 7828 7829 * Fix DTLS memory leak in dtls1_buffer_record. A memory leak can occur in the 7830 dtls1_buffer_record function under certain conditions. In particular this 7831 could occur if an attacker sent repeated DTLS records with the same 7832 sequence number but for the next epoch. The memory leak could be exploited 7833 by an attacker in a Denial of Service attack through memory exhaustion. 7834 Thanks to Chris Mueller for reporting this issue. 7835 ([CVE-2015-0206]) 7836 7837 *Matt Caswell* 7838 7839 * Fix issue where no-ssl3 configuration sets method to NULL. When openssl is 7840 built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl 7841 method would be set to NULL which could later result in a NULL pointer 7842 dereference. Thanks to Frank Schmirler for reporting this issue. 7843 ([CVE-2014-3569]) 7844 7845 *Kurt Roeckx* 7846 7847 * Abort handshake if server key exchange message is omitted for ephemeral 7848 ECDH ciphersuites. 7849 7850 Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for 7851 reporting this issue. 7852 ([CVE-2014-3572]) 7853 7854 *Steve Henson* 7855 7856 * Remove non-export ephemeral RSA code on client and server. This code 7857 violated the TLS standard by allowing the use of temporary RSA keys in 7858 non-export ciphersuites and could be used by a server to effectively 7859 downgrade the RSA key length used to a value smaller than the server 7860 certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at 7861 INRIA or reporting this issue. 7862 ([CVE-2015-0204]) 7863 7864 *Steve Henson* 7865 7866 * Fixed issue where DH client certificates are accepted without verification. 7867 An OpenSSL server will accept a DH certificate for client authentication 7868 without the certificate verify message. This effectively allows a client to 7869 authenticate without the use of a private key. This only affects servers 7870 which trust a client certificate authority which issues certificates 7871 containing DH keys: these are extremely rare and hardly ever encountered. 7872 Thanks for Karthikeyan Bhargavan of the PROSECCO team at INRIA or reporting 7873 this issue. 7874 ([CVE-2015-0205]) 7875 7876 *Steve Henson* 7877 7878 * Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect 7879 results on some platforms, including x86_64. This bug occurs at random 7880 with a very low probability, and is not known to be exploitable in any 7881 way, though its exact impact is difficult to determine. Thanks to Pieter 7882 Wuille (Blockstream) who reported this issue and also suggested an initial 7883 fix. Further analysis was conducted by the OpenSSL development team and 7884 Adam Langley of Google. The final fix was developed by Andy Polyakov of 7885 the OpenSSL core team. 7886 ([CVE-2014-3570]) 7887 7888 *Andy Polyakov* 7889 7890 * Fix various certificate fingerprint issues. 7891 7892 By using non-DER or invalid encodings outside the signed portion of a 7893 certificate the fingerprint can be changed without breaking the signature. 7894 Although no details of the signed portion of the certificate can be changed 7895 this can cause problems with some applications: e.g. those using the 7896 certificate fingerprint for blacklists. 7897 7898 1. Reject signatures with non zero unused bits. 7899 7900 If the BIT STRING containing the signature has non zero unused bits reject 7901 the signature. All current signature algorithms require zero unused bits. 7902 7903 2. Check certificate algorithm consistency. 7904 7905 Check the AlgorithmIdentifier inside TBS matches the one in the 7906 certificate signature. NB: this will result in signature failure 7907 errors for some broken certificates. 7908 7909 Thanks to Konrad Kraszewski from Google for reporting this issue. 7910 7911 3. Check DSA/ECDSA signatures use DER. 7912 7913 Reencode DSA/ECDSA signatures and compare with the original received 7914 signature. Return an error if there is a mismatch. 7915 7916 This will reject various cases including garbage after signature 7917 (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS 7918 program for discovering this case) and use of BER or invalid ASN.1 INTEGERs 7919 (negative or with leading zeroes). 7920 7921 Further analysis was conducted and fixes were developed by Stephen Henson 7922 of the OpenSSL core team. 7923 7924 ([CVE-2014-8275]) 7925 7926 *Steve Henson* 7927 7928### Changes between 1.0.0n and 1.0.0o [15 Oct 2014] 7929 7930 * Session Ticket Memory Leak. 7931 7932 When an OpenSSL SSL/TLS/DTLS server receives a session ticket the 7933 integrity of that ticket is first verified. In the event of a session 7934 ticket integrity check failing, OpenSSL will fail to free memory 7935 causing a memory leak. By sending a large number of invalid session 7936 tickets an attacker could exploit this issue in a Denial Of Service 7937 attack. 7938 ([CVE-2014-3567]) 7939 7940 *Steve Henson* 7941 7942 * Build option no-ssl3 is incomplete. 7943 7944 When OpenSSL is configured with "no-ssl3" as a build option, servers 7945 could accept and complete a SSL 3.0 handshake, and clients could be 7946 configured to send them. 7947 ([CVE-2014-3568]) 7948 7949 *Akamai and the OpenSSL team* 7950 7951 * Add support for TLS_FALLBACK_SCSV. 7952 Client applications doing fallback retries should call 7953 SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV). 7954 ([CVE-2014-3566]) 7955 7956 *Adam Langley, Bodo Moeller* 7957 7958 * Add additional DigestInfo checks. 7959 7960 Reencode DigestInto in DER and check against the original when 7961 verifying RSA signature: this will reject any improperly encoded 7962 DigestInfo structures. 7963 7964 Note: this is a precautionary measure and no attacks are currently known. 7965 7966 *Steve Henson* 7967 7968### Changes between 1.0.0m and 1.0.0n [6 Aug 2014] 7969 7970 * OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject 7971 to a denial of service attack. A malicious server can crash the client 7972 with a null pointer dereference (read) by specifying an anonymous (EC)DH 7973 ciphersuite and sending carefully crafted handshake messages. 7974 7975 Thanks to Felix Gröbert (Google) for discovering and researching this 7976 issue. 7977 ([CVE-2014-3510]) 7978 7979 *Emilia Käsper* 7980 7981 * By sending carefully crafted DTLS packets an attacker could cause openssl 7982 to leak memory. This can be exploited through a Denial of Service attack. 7983 Thanks to Adam Langley for discovering and researching this issue. 7984 ([CVE-2014-3507]) 7985 7986 *Adam Langley* 7987 7988 * An attacker can force openssl to consume large amounts of memory whilst 7989 processing DTLS handshake messages. This can be exploited through a 7990 Denial of Service attack. 7991 Thanks to Adam Langley for discovering and researching this issue. 7992 ([CVE-2014-3506]) 7993 7994 *Adam Langley* 7995 7996 * An attacker can force an error condition which causes openssl to crash 7997 whilst processing DTLS packets due to memory being freed twice. This 7998 can be exploited through a Denial of Service attack. 7999 Thanks to Adam Langley and Wan-Teh Chang for discovering and researching 8000 this issue. 8001 ([CVE-2014-3505]) 8002 8003 *Adam Langley* 8004 8005 * If a multithreaded client connects to a malicious server using a resumed 8006 session and the server sends an ec point format extension it could write 8007 up to 255 bytes to freed memory. 8008 8009 Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this 8010 issue. 8011 ([CVE-2014-3509]) 8012 8013 *Gabor Tyukasz* 8014 8015 * A flaw in OBJ_obj2txt may cause pretty printing functions such as 8016 X509_name_oneline, X509_name_print_ex et al. to leak some information 8017 from the stack. Applications may be affected if they echo pretty printing 8018 output to the attacker. 8019 8020 Thanks to Ivan Fratric (Google) for discovering this issue. 8021 ([CVE-2014-3508]) 8022 8023 *Emilia Käsper, and Steve Henson* 8024 8025 * Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.) 8026 for corner cases. (Certain input points at infinity could lead to 8027 bogus results, with non-infinity inputs mapped to infinity too.) 8028 8029 *Bodo Moeller* 8030 8031### Changes between 1.0.0l and 1.0.0m [5 Jun 2014] 8032 8033 * Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted 8034 handshake can force the use of weak keying material in OpenSSL 8035 SSL/TLS clients and servers. 8036 8037 Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and 8038 researching this issue. ([CVE-2014-0224]) 8039 8040 *KIKUCHI Masashi, Steve Henson* 8041 8042 * Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an 8043 OpenSSL DTLS client the code can be made to recurse eventually crashing 8044 in a DoS attack. 8045 8046 Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue. 8047 ([CVE-2014-0221]) 8048 8049 *Imre Rad, Steve Henson* 8050 8051 * Fix DTLS invalid fragment vulnerability. A buffer overrun attack can 8052 be triggered by sending invalid DTLS fragments to an OpenSSL DTLS 8053 client or server. This is potentially exploitable to run arbitrary 8054 code on a vulnerable client or server. 8055 8056 Thanks to Jüri Aedla for reporting this issue. ([CVE-2014-0195]) 8057 8058 *Jüri Aedla, Steve Henson* 8059 8060 * Fix bug in TLS code where clients enable anonymous ECDH ciphersuites 8061 are subject to a denial of service attack. 8062 8063 Thanks to Felix Gröbert and Ivan Fratric at Google for discovering 8064 this issue. ([CVE-2014-3470]) 8065 8066 *Felix Gröbert, Ivan Fratric, Steve Henson* 8067 8068 * Harmonize version and its documentation. -f flag is used to display 8069 compilation flags. 8070 8071 *mancha <mancha1@zoho.com>* 8072 8073 * Fix eckey_priv_encode so it immediately returns an error upon a failure 8074 in i2d_ECPrivateKey. 8075 8076 *mancha <mancha1@zoho.com>* 8077 8078 * Fix some double frees. These are not thought to be exploitable. 8079 8080 *mancha <mancha1@zoho.com>* 8081 8082 * Fix for the attack described in the paper "Recovering OpenSSL 8083 ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack" 8084 by Yuval Yarom and Naomi Benger. Details can be obtained from: 8085 <http://eprint.iacr.org/2014/140> 8086 8087 Thanks to Yuval Yarom and Naomi Benger for discovering this 8088 flaw and to Yuval Yarom for supplying a fix ([CVE-2014-0076]) 8089 8090 *Yuval Yarom and Naomi Benger* 8091 8092### Changes between 1.0.0k and 1.0.0l [6 Jan 2014] 8093 8094 * Keep original DTLS digest and encryption contexts in retransmission 8095 structures so we can use the previous session parameters if they need 8096 to be resent. ([CVE-2013-6450]) 8097 8098 *Steve Henson* 8099 8100 * Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which 8101 avoids preferring ECDHE-ECDSA ciphers when the client appears to be 8102 Safari on OS X. Safari on OS X 10.8..10.8.3 advertises support for 8103 several ECDHE-ECDSA ciphers, but fails to negotiate them. The bug 8104 is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing 8105 10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer. 8106 8107 *Rob Stradling, Adam Langley* 8108 8109### Changes between 1.0.0j and 1.0.0k [5 Feb 2013] 8110 8111 * Make the decoding of SSLv3, TLS and DTLS CBC records constant time. 8112 8113 This addresses the flaw in CBC record processing discovered by 8114 Nadhem Alfardan and Kenny Paterson. Details of this attack can be found 8115 at: <http://www.isg.rhul.ac.uk/tls/> 8116 8117 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 8118 Security Group at Royal Holloway, University of London 8119 (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and 8120 Emilia Käsper for the initial patch. 8121 ([CVE-2013-0169]) 8122 8123 *Emilia Käsper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson* 8124 8125 * Return an error when checking OCSP signatures when key is NULL. 8126 This fixes a DoS attack. ([CVE-2013-0166]) 8127 8128 *Steve Henson* 8129 8130 * Call OCSP Stapling callback after ciphersuite has been chosen, so 8131 the right response is stapled. Also change SSL_get_certificate() 8132 so it returns the certificate actually sent. 8133 See <http://rt.openssl.org/Ticket/Display.html?id=2836>. 8134 (This is a backport) 8135 8136 *Rob Stradling <rob.stradling@comodo.com>* 8137 8138 * Fix possible deadlock when decoding public keys. 8139 8140 *Steve Henson* 8141 8142### Changes between 1.0.0i and 1.0.0j [10 May 2012] 8143 8144[NB: OpenSSL 1.0.0i and later 1.0.0 patch levels were released after 8145OpenSSL 1.0.1.] 8146 8147 * Sanity check record length before skipping explicit IV in DTLS 8148 to fix DoS attack. 8149 8150 Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic 8151 fuzzing as a service testing platform. 8152 ([CVE-2012-2333]) 8153 8154 *Steve Henson* 8155 8156 * Initialise tkeylen properly when encrypting CMS messages. 8157 Thanks to Solar Designer of Openwall for reporting this issue. 8158 8159 *Steve Henson* 8160 8161### Changes between 1.0.0h and 1.0.0i [19 Apr 2012] 8162 8163 * Check for potentially exploitable overflows in asn1_d2i_read_bio 8164 BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer 8165 in CRYPTO_realloc_clean. 8166 8167 Thanks to Tavis Ormandy, Google Security Team, for discovering this 8168 issue and to Adam Langley <agl@chromium.org> for fixing it. 8169 ([CVE-2012-2110]) 8170 8171 *Adam Langley (Google), Tavis Ormandy, Google Security Team* 8172 8173### Changes between 1.0.0g and 1.0.0h [12 Mar 2012] 8174 8175 * Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness 8176 in CMS and PKCS7 code. When RSA decryption fails use a random key for 8177 content decryption and always return the same error. Note: this attack 8178 needs on average 2^20 messages so it only affects automated senders. The 8179 old behaviour can be re-enabled in the CMS code by setting the 8180 CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where 8181 an MMA defence is not necessary. 8182 Thanks to Ivan Nestlerode <inestlerode@us.ibm.com> for discovering 8183 this issue. ([CVE-2012-0884]) 8184 8185 *Steve Henson* 8186 8187 * Fix CVE-2011-4619: make sure we really are receiving a 8188 client hello before rejecting multiple SGC restarts. Thanks to 8189 Ivan Nestlerode <inestlerode@us.ibm.com> for discovering this bug. 8190 8191 *Steve Henson* 8192 8193### Changes between 1.0.0f and 1.0.0g [18 Jan 2012] 8194 8195 * Fix for DTLS DoS issue introduced by fix for CVE-2011-4109. 8196 Thanks to Antonio Martin, Enterprise Secure Access Research and 8197 Development, Cisco Systems, Inc. for discovering this bug and 8198 preparing a fix. ([CVE-2012-0050]) 8199 8200 *Antonio Martin* 8201 8202### Changes between 1.0.0e and 1.0.0f [4 Jan 2012] 8203 8204 * Nadhem Alfardan and Kenny Paterson have discovered an extension 8205 of the Vaudenay padding oracle attack on CBC mode encryption 8206 which enables an efficient plaintext recovery attack against 8207 the OpenSSL implementation of DTLS. Their attack exploits timing 8208 differences arising during decryption processing. A research 8209 paper describing this attack can be found at: 8210 <http://www.isg.rhul.ac.uk/~kp/dtls.pdf> 8211 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 8212 Security Group at Royal Holloway, University of London 8213 (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann 8214 <seggelmann@fh-muenster.de> and Michael Tuexen <tuexen@fh-muenster.de> 8215 for preparing the fix. ([CVE-2011-4108]) 8216 8217 *Robin Seggelmann, Michael Tuexen* 8218 8219 * Clear bytes used for block padding of SSL 3.0 records. 8220 ([CVE-2011-4576]) 8221 8222 *Adam Langley (Google)* 8223 8224 * Only allow one SGC handshake restart for SSL/TLS. Thanks to George 8225 Kadianakis <desnacked@gmail.com> for discovering this issue and 8226 Adam Langley for preparing the fix. ([CVE-2011-4619]) 8227 8228 *Adam Langley (Google)* 8229 8230 * Check parameters are not NULL in GOST ENGINE. ([CVE-2012-0027]) 8231 8232 *Andrey Kulikov <amdeich@gmail.com>* 8233 8234 * Prevent malformed RFC3779 data triggering an assertion failure. 8235 Thanks to Andrew Chi, BBN Technologies, for discovering the flaw 8236 and Rob Austein <sra@hactrn.net> for fixing it. ([CVE-2011-4577]) 8237 8238 *Rob Austein <sra@hactrn.net>* 8239 8240 * Improved PRNG seeding for VOS. 8241 8242 *Paul Green <Paul.Green@stratus.com>* 8243 8244 * Fix ssl_ciph.c set-up race. 8245 8246 *Adam Langley (Google)* 8247 8248 * Fix spurious failures in ecdsatest.c. 8249 8250 *Emilia Käsper (Google)* 8251 8252 * Fix the BIO_f_buffer() implementation (which was mixing different 8253 interpretations of the `..._len` fields). 8254 8255 *Adam Langley (Google)* 8256 8257 * Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than 8258 BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent 8259 threads won't reuse the same blinding coefficients. 8260 8261 This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING 8262 lock to call BN_BLINDING_invert_ex, and avoids one use of 8263 BN_BLINDING_update for each BN_BLINDING structure (previously, 8264 the last update always remained unused). 8265 8266 *Emilia Käsper (Google)* 8267 8268 * In ssl3_clear, preserve s3->init_extra along with s3->rbuf. 8269 8270 *Bob Buckholz (Google)* 8271 8272### Changes between 1.0.0d and 1.0.0e [6 Sep 2011] 8273 8274 * Fix bug where CRLs with nextUpdate in the past are sometimes accepted 8275 by initialising X509_STORE_CTX properly. ([CVE-2011-3207]) 8276 8277 *Kaspar Brand <ossl@velox.ch>* 8278 8279 * Fix SSL memory handling for (EC)DH ciphersuites, in particular 8280 for multi-threaded use of ECDH. ([CVE-2011-3210]) 8281 8282 *Adam Langley (Google)* 8283 8284 * Fix x509_name_ex_d2i memory leak on bad inputs. 8285 8286 *Bodo Moeller* 8287 8288 * Remove hard coded ecdsaWithSHA1 signature tests in ssl code and check 8289 signature public key algorithm by using OID xref utilities instead. 8290 Before this you could only use some ECC ciphersuites with SHA1 only. 8291 8292 *Steve Henson* 8293 8294 * Add protection against ECDSA timing attacks as mentioned in the paper 8295 by Billy Bob Brumley and Nicola Tuveri, see: 8296 <http://eprint.iacr.org/2011/232.pdf> 8297 8298 *Billy Bob Brumley and Nicola Tuveri* 8299 8300### Changes between 1.0.0c and 1.0.0d [8 Feb 2011] 8301 8302 * Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014 8303 8304 *Neel Mehta, Adam Langley, Bodo Moeller (Google)* 8305 8306 * Fix bug in string printing code: if *any* escaping is enabled we must 8307 escape the escape character (backslash) or the resulting string is 8308 ambiguous. 8309 8310 *Steve Henson* 8311 8312### Changes between 1.0.0b and 1.0.0c [2 Dec 2010] 8313 8314 * Disable code workaround for ancient and obsolete Netscape browsers 8315 and servers: an attacker can use it in a ciphersuite downgrade attack. 8316 Thanks to Martin Rex for discovering this bug. CVE-2010-4180 8317 8318 *Steve Henson* 8319 8320 * Fixed J-PAKE implementation error, originally discovered by 8321 Sebastien Martini, further info and confirmation from Stefan 8322 Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252 8323 8324 *Ben Laurie* 8325 8326### Changes between 1.0.0a and 1.0.0b [16 Nov 2010] 8327 8328 * Fix extension code to avoid race conditions which can result in a buffer 8329 overrun vulnerability: resumed sessions must not be modified as they can 8330 be shared by multiple threads. CVE-2010-3864 8331 8332 *Steve Henson* 8333 8334 * Fix WIN32 build system to correctly link an ENGINE directory into 8335 a DLL. 8336 8337 *Steve Henson* 8338 8339### Changes between 1.0.0 and 1.0.0a [01 Jun 2010] 8340 8341 * Check return value of int_rsa_verify in pkey_rsa_verifyrecover 8342 ([CVE-2010-1633]) 8343 8344 *Steve Henson, Peter-Michael Hager <hager@dortmund.net>* 8345 8346### Changes between 0.9.8n and 1.0.0 [29 Mar 2010] 8347 8348 * Add "missing" function EVP_CIPHER_CTX_copy(). This copies a cipher 8349 context. The operation can be customised via the ctrl mechanism in 8350 case ENGINEs want to include additional functionality. 8351 8352 *Steve Henson* 8353 8354 * Tolerate yet another broken PKCS#8 key format: private key value negative. 8355 8356 *Steve Henson* 8357 8358 * Add new -subject_hash_old and -issuer_hash_old options to x509 utility to 8359 output hashes compatible with older versions of OpenSSL. 8360 8361 *Willy Weisz <weisz@vcpc.univie.ac.at>* 8362 8363 * Fix compression algorithm handling: if resuming a session use the 8364 compression algorithm of the resumed session instead of determining 8365 it from client hello again. Don't allow server to change algorithm. 8366 8367 *Steve Henson* 8368 8369 * Add load_crls() function to commands tidying load_certs() too. Add option 8370 to verify utility to allow additional CRLs to be included. 8371 8372 *Steve Henson* 8373 8374 * Update OCSP request code to permit adding custom headers to the request: 8375 some responders need this. 8376 8377 *Steve Henson* 8378 8379 * The function EVP_PKEY_sign() returns <=0 on error: check return code 8380 correctly. 8381 8382 *Julia Lawall <julia@diku.dk>* 8383 8384 * Update verify callback code in `apps/s_cb.c` and `apps/verify.c`, it 8385 needlessly dereferenced structures, used obsolete functions and 8386 didn't handle all updated verify codes correctly. 8387 8388 *Steve Henson* 8389 8390 * Disable MD2 in the default configuration. 8391 8392 *Steve Henson* 8393 8394 * In BIO_pop() and BIO_push() use the ctrl argument (which was NULL) to 8395 indicate the initial BIO being pushed or popped. This makes it possible 8396 to determine whether the BIO is the one explicitly called or as a result 8397 of the ctrl being passed down the chain. Fix BIO_pop() and SSL BIOs so 8398 it handles reference counts correctly and doesn't zero out the I/O bio 8399 when it is not being explicitly popped. WARNING: applications which 8400 included workarounds for the old buggy behaviour will need to be modified 8401 or they could free up already freed BIOs. 8402 8403 *Steve Henson* 8404 8405 * Extend the uni2asc/asc2uni => OPENSSL_uni2asc/OPENSSL_asc2uni 8406 renaming to all platforms (within the 0.9.8 branch, this was 8407 done conditionally on Netware platforms to avoid a name clash). 8408 8409 *Guenter <lists@gknw.net>* 8410 8411 * Add ECDHE and PSK support to DTLS. 8412 8413 *Michael Tuexen <tuexen@fh-muenster.de>* 8414 8415 * Add CHECKED_STACK_OF macro to safestack.h, otherwise safestack can't 8416 be used on C++. 8417 8418 *Steve Henson* 8419 8420 * Add "missing" function EVP_MD_flags() (without this the only way to 8421 retrieve a digest flags is by accessing the structure directly. Update 8422 `EVP_MD_do_all*()` and `EVP_CIPHER_do_all*()` to include the name a digest 8423 or cipher is registered as in the "from" argument. Print out all 8424 registered digests in the dgst usage message instead of manually 8425 attempting to work them out. 8426 8427 *Steve Henson* 8428 8429 * If no SSLv2 ciphers are used don't use an SSLv2 compatible client hello: 8430 this allows the use of compression and extensions. Change default cipher 8431 string to remove SSLv2 ciphersuites. This effectively avoids ancient SSLv2 8432 by default unless an application cipher string requests it. 8433 8434 *Steve Henson* 8435 8436 * Alter match criteria in PKCS12_parse(). It used to try to use local 8437 key ids to find matching certificates and keys but some PKCS#12 files 8438 don't follow the (somewhat unwritten) rules and this strategy fails. 8439 Now just gather all certificates together and the first private key 8440 then look for the first certificate that matches the key. 8441 8442 *Steve Henson* 8443 8444 * Support use of registered digest and cipher names for dgst and cipher 8445 commands instead of having to add each one as a special case. So now 8446 you can do: 8447 8448 openssl sha256 foo 8449 8450 as well as: 8451 8452 openssl dgst -sha256 foo 8453 8454 and this works for ENGINE based algorithms too. 8455 8456 *Steve Henson* 8457 8458 * Update Gost ENGINE to support parameter files. 8459 8460 *Victor B. Wagner <vitus@cryptocom.ru>* 8461 8462 * Support GeneralizedTime in ca utility. 8463 8464 *Oliver Martin <oliver@volatilevoid.net>, Steve Henson* 8465 8466 * Enhance the hash format used for certificate directory links. The new 8467 form uses the canonical encoding (meaning equivalent names will work 8468 even if they aren't identical) and uses SHA1 instead of MD5. This form 8469 is incompatible with the older format and as a result c_rehash should 8470 be used to rebuild symbolic links. 8471 8472 *Steve Henson* 8473 8474 * Make PKCS#8 the default write format for private keys, replacing the 8475 traditional format. This form is standardised, more secure and doesn't 8476 include an implicit MD5 dependency. 8477 8478 *Steve Henson* 8479 8480 * Add a $gcc_devteam_warn option to Configure. The idea is that any code 8481 committed to OpenSSL should pass this lot as a minimum. 8482 8483 *Steve Henson* 8484 8485 * Add session ticket override functionality for use by EAP-FAST. 8486 8487 *Jouni Malinen <j@w1.fi>* 8488 8489 * Modify HMAC functions to return a value. Since these can be implemented 8490 in an ENGINE errors can occur. 8491 8492 *Steve Henson* 8493 8494 * Type-checked OBJ_bsearch_ex. 8495 8496 *Ben Laurie* 8497 8498 * Type-checked OBJ_bsearch. Also some constification necessitated 8499 by type-checking. Still to come: TXT_DB, bsearch(?), 8500 OBJ_bsearch_ex, qsort, CRYPTO_EX_DATA, ASN1_VALUE, ASN1_STRING, 8501 CONF_VALUE. 8502 8503 *Ben Laurie* 8504 8505 * New function OPENSSL_gmtime_adj() to add a specific number of days and 8506 seconds to a tm structure directly, instead of going through OS 8507 specific date routines. This avoids any issues with OS routines such 8508 as the year 2038 bug. New `*_adj()` functions for ASN1 time structures 8509 and X509_time_adj_ex() to cover the extended range. The existing 8510 X509_time_adj() is still usable and will no longer have any date issues. 8511 8512 *Steve Henson* 8513 8514 * Delta CRL support. New use deltas option which will attempt to locate 8515 and search any appropriate delta CRLs available. 8516 8517 This work was sponsored by Google. 8518 8519 *Steve Henson* 8520 8521 * Support for CRLs partitioned by reason code. Reorganise CRL processing 8522 code and add additional score elements. Validate alternate CRL paths 8523 as part of the CRL checking and indicate a new error "CRL path validation 8524 error" in this case. Applications wanting additional details can use 8525 the verify callback and check the new "parent" field. If this is not 8526 NULL CRL path validation is taking place. Existing applications won't 8527 see this because it requires extended CRL support which is off by 8528 default. 8529 8530 This work was sponsored by Google. 8531 8532 *Steve Henson* 8533 8534 * Support for freshest CRL extension. 8535 8536 This work was sponsored by Google. 8537 8538 *Steve Henson* 8539 8540 * Initial indirect CRL support. Currently only supported in the CRLs 8541 passed directly and not via lookup. Process certificate issuer 8542 CRL entry extension and lookup CRL entries by bother issuer name 8543 and serial number. Check and process CRL issuer entry in IDP extension. 8544 8545 This work was sponsored by Google. 8546 8547 *Steve Henson* 8548 8549 * Add support for distinct certificate and CRL paths. The CRL issuer 8550 certificate is validated separately in this case. Only enabled if 8551 an extended CRL support flag is set: this flag will enable additional 8552 CRL functionality in future. 8553 8554 This work was sponsored by Google. 8555 8556 *Steve Henson* 8557 8558 * Add support for policy mappings extension. 8559 8560 This work was sponsored by Google. 8561 8562 *Steve Henson* 8563 8564 * Fixes to pathlength constraint, self issued certificate handling, 8565 policy processing to align with RFC3280 and PKITS tests. 8566 8567 This work was sponsored by Google. 8568 8569 *Steve Henson* 8570 8571 * Support for name constraints certificate extension. DN, email, DNS 8572 and URI types are currently supported. 8573 8574 This work was sponsored by Google. 8575 8576 *Steve Henson* 8577 8578 * To cater for systems that provide a pointer-based thread ID rather 8579 than numeric, deprecate the current numeric thread ID mechanism and 8580 replace it with a structure and associated callback type. This 8581 mechanism allows a numeric "hash" to be extracted from a thread ID in 8582 either case, and on platforms where pointers are larger than 'long', 8583 mixing is done to help ensure the numeric 'hash' is usable even if it 8584 can't be guaranteed unique. The default mechanism is to use "&errno" 8585 as a pointer-based thread ID to distinguish between threads. 8586 8587 Applications that want to provide their own thread IDs should now use 8588 CRYPTO_THREADID_set_callback() to register a callback that will call 8589 either CRYPTO_THREADID_set_numeric() or CRYPTO_THREADID_set_pointer(). 8590 8591 Note that ERR_remove_state() is now deprecated, because it is tied 8592 to the assumption that thread IDs are numeric. ERR_remove_state(0) 8593 to free the current thread's error state should be replaced by 8594 ERR_remove_thread_state(NULL). 8595 8596 (This new approach replaces the functions CRYPTO_set_idptr_callback(), 8597 CRYPTO_get_idptr_callback(), and CRYPTO_thread_idptr() that existed in 8598 OpenSSL 0.9.9-dev between June 2006 and August 2008. Also, if an 8599 application was previously providing a numeric thread callback that 8600 was inappropriate for distinguishing threads, then uniqueness might 8601 have been obtained with &errno that happened immediately in the 8602 intermediate development versions of OpenSSL; this is no longer the 8603 case, the numeric thread callback will now override the automatic use 8604 of &errno.) 8605 8606 *Geoff Thorpe, with help from Bodo Moeller* 8607 8608 * Initial support for different CRL issuing certificates. This covers a 8609 simple case where the self issued certificates in the chain exist and 8610 the real CRL issuer is higher in the existing chain. 8611 8612 This work was sponsored by Google. 8613 8614 *Steve Henson* 8615 8616 * Removed effectively defunct crypto/store from the build. 8617 8618 *Ben Laurie* 8619 8620 * Revamp of STACK to provide stronger type-checking. Still to come: 8621 TXT_DB, bsearch(?), OBJ_bsearch, qsort, CRYPTO_EX_DATA, ASN1_VALUE, 8622 ASN1_STRING, CONF_VALUE. 8623 8624 *Ben Laurie* 8625 8626 * Add a new SSL_MODE_RELEASE_BUFFERS mode flag to release unused buffer 8627 RAM on SSL connections. This option can save about 34k per idle SSL. 8628 8629 *Nick Mathewson* 8630 8631 * Revamp of LHASH to provide stronger type-checking. Still to come: 8632 STACK, TXT_DB, bsearch, qsort. 8633 8634 *Ben Laurie* 8635 8636 * Initial support for Cryptographic Message Syntax (aka CMS) based 8637 on RFC3850, RFC3851 and RFC3852. New cms directory and cms utility, 8638 support for data, signedData, compressedData, digestedData and 8639 encryptedData, envelopedData types included. Scripts to check against 8640 RFC4134 examples draft and interop and consistency checks of many 8641 content types and variants. 8642 8643 *Steve Henson* 8644 8645 * Add options to enc utility to support use of zlib compression BIO. 8646 8647 *Steve Henson* 8648 8649 * Extend mk1mf to support importing of options and assembly language 8650 files from Configure script, currently only included in VC-WIN32. 8651 The assembly language rules can now optionally generate the source 8652 files from the associated perl scripts. 8653 8654 *Steve Henson* 8655 8656 * Implement remaining functionality needed to support GOST ciphersuites. 8657 Interop testing has been performed using CryptoPro implementations. 8658 8659 *Victor B. Wagner <vitus@cryptocom.ru>* 8660 8661 * s390x assembler pack. 8662 8663 *Andy Polyakov* 8664 8665 * ARMv4 assembler pack. ARMv4 refers to v4 and later ISA, not CPU 8666 "family." 8667 8668 *Andy Polyakov* 8669 8670 * Implement Opaque PRF Input TLS extension as specified in 8671 draft-rescorla-tls-opaque-prf-input-00.txt. Since this is not an 8672 official specification yet and no extension type assignment by 8673 IANA exists, this extension (for now) will have to be explicitly 8674 enabled when building OpenSSL by providing the extension number 8675 to use. For example, specify an option 8676 8677 -DTLSEXT_TYPE_opaque_prf_input=0x9527 8678 8679 to the "config" or "Configure" script to enable the extension, 8680 assuming extension number 0x9527 (which is a completely arbitrary 8681 and unofficial assignment based on the MD5 hash of the Internet 8682 Draft). Note that by doing so, you potentially lose 8683 interoperability with other TLS implementations since these might 8684 be using the same extension number for other purposes. 8685 8686 SSL_set_tlsext_opaque_prf_input(ssl, src, len) is used to set the 8687 opaque PRF input value to use in the handshake. This will create 8688 an internal copy of the length-'len' string at 'src', and will 8689 return non-zero for success. 8690 8691 To get more control and flexibility, provide a callback function 8692 by using 8693 8694 SSL_CTX_set_tlsext_opaque_prf_input_callback(ctx, cb) 8695 SSL_CTX_set_tlsext_opaque_prf_input_callback_arg(ctx, arg) 8696 8697 where 8698 8699 int (*cb)(SSL *, void *peerinput, size_t len, void *arg); 8700 void *arg; 8701 8702 Callback function 'cb' will be called in handshakes, and is 8703 expected to use SSL_set_tlsext_opaque_prf_input() as appropriate. 8704 Argument 'arg' is for application purposes (the value as given to 8705 SSL_CTX_set_tlsext_opaque_prf_input_callback_arg() will directly 8706 be provided to the callback function). The callback function 8707 has to return non-zero to report success: usually 1 to use opaque 8708 PRF input just if possible, or 2 to enforce use of the opaque PRF 8709 input. In the latter case, the library will abort the handshake 8710 if opaque PRF input is not successfully negotiated. 8711 8712 Arguments 'peerinput' and 'len' given to the callback function 8713 will always be NULL and 0 in the case of a client. A server will 8714 see the client's opaque PRF input through these variables if 8715 available (NULL and 0 otherwise). Note that if the server 8716 provides an opaque PRF input, the length must be the same as the 8717 length of the client's opaque PRF input. 8718 8719 Note that the callback function will only be called when creating 8720 a new session (session resumption can resume whatever was 8721 previously negotiated), and will not be called in SSL 2.0 8722 handshakes; thus, SSL_CTX_set_options(ctx, SSL_OP_NO_SSLv2) or 8723 SSL_set_options(ssl, SSL_OP_NO_SSLv2) is especially recommended 8724 for applications that need to enforce opaque PRF input. 8725 8726 *Bodo Moeller* 8727 8728 * Update ssl code to support digests other than SHA1+MD5 for handshake 8729 MAC. 8730 8731 *Victor B. Wagner <vitus@cryptocom.ru>* 8732 8733 * Add RFC4507 support to OpenSSL. This includes the corrections in 8734 RFC4507bis. The encrypted ticket format is an encrypted encoded 8735 SSL_SESSION structure, that way new session features are automatically 8736 supported. 8737 8738 If a client application caches session in an SSL_SESSION structure 8739 support is transparent because tickets are now stored in the encoded 8740 SSL_SESSION. 8741 8742 The SSL_CTX structure automatically generates keys for ticket 8743 protection in servers so again support should be possible 8744 with no application modification. 8745 8746 If a client or server wishes to disable RFC4507 support then the option 8747 SSL_OP_NO_TICKET can be set. 8748 8749 Add a TLS extension debugging callback to allow the contents of any client 8750 or server extensions to be examined. 8751 8752 This work was sponsored by Google. 8753 8754 *Steve Henson* 8755 8756 * Final changes to avoid use of pointer pointer casts in OpenSSL. 8757 OpenSSL should now compile cleanly on gcc 4.2 8758 8759 *Peter Hartley <pdh@utter.chaos.org.uk>, Steve Henson* 8760 8761 * Update SSL library to use new EVP_PKEY MAC API. Include generic MAC 8762 support including streaming MAC support: this is required for GOST 8763 ciphersuite support. 8764 8765 *Victor B. Wagner <vitus@cryptocom.ru>, Steve Henson* 8766 8767 * Add option -stream to use PKCS#7 streaming in smime utility. New 8768 function i2d_PKCS7_bio_stream() and PEM_write_PKCS7_bio_stream() 8769 to output in BER and PEM format. 8770 8771 *Steve Henson* 8772 8773 * Experimental support for use of HMAC via EVP_PKEY interface. This 8774 allows HMAC to be handled via the `EVP_DigestSign*()` interface. The 8775 EVP_PKEY "key" in this case is the HMAC key, potentially allowing 8776 ENGINE support for HMAC keys which are unextractable. New -mac and 8777 -macopt options to dgst utility. 8778 8779 *Steve Henson* 8780 8781 * New option -sigopt to dgst utility. Update dgst to use 8782 `EVP_Digest{Sign,Verify}*`. These two changes make it possible to use 8783 alternative signing parameters such as X9.31 or PSS in the dgst 8784 utility. 8785 8786 *Steve Henson* 8787 8788 * Change ssl_cipher_apply_rule(), the internal function that does 8789 the work each time a ciphersuite string requests enabling 8790 ("foo+bar"), moving ("+foo+bar"), disabling ("-foo+bar", or 8791 removing ("!foo+bar") a class of ciphersuites: Now it maintains 8792 the order of disabled ciphersuites such that those ciphersuites 8793 that most recently went from enabled to disabled not only stay 8794 in order with respect to each other, but also have higher priority 8795 than other disabled ciphersuites the next time ciphersuites are 8796 enabled again. 8797 8798 This means that you can now say, e.g., "PSK:-PSK:HIGH" to enable 8799 the same ciphersuites as with "HIGH" alone, but in a specific 8800 order where the PSK ciphersuites come first (since they are the 8801 most recently disabled ciphersuites when "HIGH" is parsed). 8802 8803 Also, change ssl_create_cipher_list() (using this new 8804 functionality) such that between otherwise identical 8805 ciphersuites, ephemeral ECDH is preferred over ephemeral DH in 8806 the default order. 8807 8808 *Bodo Moeller* 8809 8810 * Change ssl_create_cipher_list() so that it automatically 8811 arranges the ciphersuites in reasonable order before starting 8812 to process the rule string. Thus, the definition for "DEFAULT" 8813 (SSL_DEFAULT_CIPHER_LIST) now is just "ALL:!aNULL:!eNULL", but 8814 remains equivalent to `"AES:ALL:!aNULL:!eNULL:+aECDH:+kRSA:+RC4:@STRENGTH"`. 8815 This makes it much easier to arrive at a reasonable default order 8816 in applications for which anonymous ciphers are OK (meaning 8817 that you can't actually use DEFAULT). 8818 8819 *Bodo Moeller; suggested by Victor Duchovni* 8820 8821 * Split the SSL/TLS algorithm mask (as used for ciphersuite string 8822 processing) into multiple integers instead of setting 8823 "SSL_MKEY_MASK" bits, "SSL_AUTH_MASK" bits, "SSL_ENC_MASK", 8824 "SSL_MAC_MASK", and "SSL_SSL_MASK" bits all in a single integer. 8825 (These masks as well as the individual bit definitions are hidden 8826 away into the non-exported interface ssl/ssl_locl.h, so this 8827 change to the definition of the SSL_CIPHER structure shouldn't 8828 affect applications.) This give us more bits for each of these 8829 categories, so there is no longer a need to coagulate AES128 and 8830 AES256 into a single algorithm bit, and to coagulate Camellia128 8831 and Camellia256 into a single algorithm bit, which has led to all 8832 kinds of kludges. 8833 8834 Thus, among other things, the kludge introduced in 0.9.7m and 8835 0.9.8e for masking out AES256 independently of AES128 or masking 8836 out Camellia256 independently of AES256 is not needed here in 0.9.9. 8837 8838 With the change, we also introduce new ciphersuite aliases that 8839 so far were missing: "AES128", "AES256", "CAMELLIA128", and 8840 "CAMELLIA256". 8841 8842 *Bodo Moeller* 8843 8844 * Add support for dsa-with-SHA224 and dsa-with-SHA256. 8845 Use the leftmost N bytes of the signature input if the input is 8846 larger than the prime q (with N being the size in bytes of q). 8847 8848 *Nils Larsch* 8849 8850 * Very *very* experimental PKCS#7 streaming encoder support. Nothing uses 8851 it yet and it is largely untested. 8852 8853 *Steve Henson* 8854 8855 * Add support for the ecdsa-with-SHA224/256/384/512 signature types. 8856 8857 *Nils Larsch* 8858 8859 * Initial incomplete changes to avoid need for function casts in OpenSSL 8860 some compilers (gcc 4.2 and later) reject their use. Safestack is 8861 reimplemented. Update ASN1 to avoid use of legacy functions. 8862 8863 *Steve Henson* 8864 8865 * Win32/64 targets are linked with Winsock2. 8866 8867 *Andy Polyakov* 8868 8869 * Add an X509_CRL_METHOD structure to allow CRL processing to be redirected 8870 to external functions. This can be used to increase CRL handling 8871 efficiency especially when CRLs are very large by (for example) storing 8872 the CRL revoked certificates in a database. 8873 8874 *Steve Henson* 8875 8876 * Overhaul of by_dir code. Add support for dynamic loading of CRLs so 8877 new CRLs added to a directory can be used. New command line option 8878 -verify_return_error to s_client and s_server. This causes real errors 8879 to be returned by the verify callback instead of carrying on no matter 8880 what. This reflects the way a "real world" verify callback would behave. 8881 8882 *Steve Henson* 8883 8884 * GOST engine, supporting several GOST algorithms and public key formats. 8885 Kindly donated by Cryptocom. 8886 8887 *Cryptocom* 8888 8889 * Partial support for Issuing Distribution Point CRL extension. CRLs 8890 partitioned by DP are handled but no indirect CRL or reason partitioning 8891 (yet). Complete overhaul of CRL handling: now the most suitable CRL is 8892 selected via a scoring technique which handles IDP and AKID in CRLs. 8893 8894 *Steve Henson* 8895 8896 * New X509_STORE_CTX callbacks lookup_crls() and lookup_certs() which 8897 will ultimately be used for all verify operations: this will remove the 8898 X509_STORE dependency on certificate verification and allow alternative 8899 lookup methods. X509_STORE based implementations of these two callbacks. 8900 8901 *Steve Henson* 8902 8903 * Allow multiple CRLs to exist in an X509_STORE with matching issuer names. 8904 Modify get_crl() to find a valid (unexpired) CRL if possible. 8905 8906 *Steve Henson* 8907 8908 * New function X509_CRL_match() to check if two CRLs are identical. Normally 8909 this would be called X509_CRL_cmp() but that name is already used by 8910 a function that just compares CRL issuer names. Cache several CRL 8911 extensions in X509_CRL structure and cache CRLDP in X509. 8912 8913 *Steve Henson* 8914 8915 * Store a "canonical" representation of X509_NAME structure (ASN1 Name) 8916 this maps equivalent X509_NAME structures into a consistent structure. 8917 Name comparison can then be performed rapidly using memcmp(). 8918 8919 *Steve Henson* 8920 8921 * Non-blocking OCSP request processing. Add -timeout option to ocsp 8922 utility. 8923 8924 *Steve Henson* 8925 8926 * Allow digests to supply their own micalg string for S/MIME type using 8927 the ctrl EVP_MD_CTRL_MICALG. 8928 8929 *Steve Henson* 8930 8931 * During PKCS7 signing pass the PKCS7 SignerInfo structure to the 8932 EVP_PKEY_METHOD before and after signing via the EVP_PKEY_CTRL_PKCS7_SIGN 8933 ctrl. It can then customise the structure before and/or after signing 8934 if necessary. 8935 8936 *Steve Henson* 8937 8938 * New function OBJ_add_sigid() to allow application defined signature OIDs 8939 to be added to OpenSSLs internal tables. New function OBJ_sigid_free() 8940 to free up any added signature OIDs. 8941 8942 *Steve Henson* 8943 8944 * New functions EVP_CIPHER_do_all(), EVP_CIPHER_do_all_sorted(), 8945 EVP_MD_do_all() and EVP_MD_do_all_sorted() to enumerate internal 8946 digest and cipher tables. New options added to openssl utility: 8947 list-message-digest-algorithms and list-cipher-algorithms. 8948 8949 *Steve Henson* 8950 8951 * Change the array representation of binary polynomials: the list 8952 of degrees of non-zero coefficients is now terminated with -1. 8953 Previously it was terminated with 0, which was also part of the 8954 value; thus, the array representation was not applicable to 8955 polynomials where t^0 has coefficient zero. This change makes 8956 the array representation useful in a more general context. 8957 8958 *Douglas Stebila* 8959 8960 * Various modifications and fixes to SSL/TLS cipher string 8961 handling. For ECC, the code now distinguishes between fixed ECDH 8962 with RSA certificates on the one hand and with ECDSA certificates 8963 on the other hand, since these are separate ciphersuites. The 8964 unused code for Fortezza ciphersuites has been removed. 8965 8966 For consistency with EDH, ephemeral ECDH is now called "EECDH" 8967 (not "ECDHE"). For consistency with the code for DH 8968 certificates, use of ECDH certificates is now considered ECDH 8969 authentication, not RSA or ECDSA authentication (the latter is 8970 merely the CA's signing algorithm and not actively used in the 8971 protocol). 8972 8973 The temporary ciphersuite alias "ECCdraft" is no longer 8974 available, and ECC ciphersuites are no longer excluded from "ALL" 8975 and "DEFAULT". The following aliases now exist for RFC 4492 8976 ciphersuites, most of these by analogy with the DH case: 8977 8978 kECDHr - ECDH cert, signed with RSA 8979 kECDHe - ECDH cert, signed with ECDSA 8980 kECDH - ECDH cert (signed with either RSA or ECDSA) 8981 kEECDH - ephemeral ECDH 8982 ECDH - ECDH cert or ephemeral ECDH 8983 8984 aECDH - ECDH cert 8985 aECDSA - ECDSA cert 8986 ECDSA - ECDSA cert 8987 8988 AECDH - anonymous ECDH 8989 EECDH - non-anonymous ephemeral ECDH (equivalent to "kEECDH:-AECDH") 8990 8991 *Bodo Moeller* 8992 8993 * Add additional S/MIME capabilities for AES and GOST ciphers if supported. 8994 Use correct micalg parameters depending on digest(s) in signed message. 8995 8996 *Steve Henson* 8997 8998 * Add engine support for EVP_PKEY_ASN1_METHOD. Add functions to process 8999 an ENGINE asn1 method. Support ENGINE lookups in the ASN1 code. 9000 9001 *Steve Henson* 9002 9003 * Initial engine support for EVP_PKEY_METHOD. New functions to permit 9004 an engine to register a method. Add ENGINE lookups for methods and 9005 functional reference processing. 9006 9007 *Steve Henson* 9008 9009 * New functions `EVP_Digest{Sign,Verify)*`. These are enhanced versions of 9010 `EVP_{Sign,Verify}*` which allow an application to customise the signature 9011 process. 9012 9013 *Steve Henson* 9014 9015 * New -resign option to smime utility. This adds one or more signers 9016 to an existing PKCS#7 signedData structure. Also -md option to use an 9017 alternative message digest algorithm for signing. 9018 9019 *Steve Henson* 9020 9021 * Tidy up PKCS#7 routines and add new functions to make it easier to 9022 create PKCS7 structures containing multiple signers. Update smime 9023 application to support multiple signers. 9024 9025 *Steve Henson* 9026 9027 * New -macalg option to pkcs12 utility to allow setting of an alternative 9028 digest MAC. 9029 9030 *Steve Henson* 9031 9032 * Initial support for PKCS#5 v2.0 PRFs other than default SHA1 HMAC. 9033 Reorganize PBE internals to lookup from a static table using NIDs, 9034 add support for HMAC PBE OID translation. Add a EVP_CIPHER ctrl: 9035 EVP_CTRL_PBE_PRF_NID this allows a cipher to specify an alternative 9036 PRF which will be automatically used with PBES2. 9037 9038 *Steve Henson* 9039 9040 * Replace the algorithm specific calls to generate keys in "req" with the 9041 new API. 9042 9043 *Steve Henson* 9044 9045 * Update PKCS#7 enveloped data routines to use new API. This is now 9046 supported by any public key method supporting the encrypt operation. A 9047 ctrl is added to allow the public key algorithm to examine or modify 9048 the PKCS#7 RecipientInfo structure if it needs to: for RSA this is 9049 a no op. 9050 9051 *Steve Henson* 9052 9053 * Add a ctrl to asn1 method to allow a public key algorithm to express 9054 a default digest type to use. In most cases this will be SHA1 but some 9055 algorithms (such as GOST) need to specify an alternative digest. The 9056 return value indicates how strong the preference is 1 means optional and 9057 2 is mandatory (that is it is the only supported type). Modify 9058 ASN1_item_sign() to accept a NULL digest argument to indicate it should 9059 use the default md. Update openssl utilities to use the default digest 9060 type for signing if it is not explicitly indicated. 9061 9062 *Steve Henson* 9063 9064 * Use OID cross reference table in ASN1_sign() and ASN1_verify(). New 9065 EVP_MD flag EVP_MD_FLAG_PKEY_METHOD_SIGNATURE. This uses the relevant 9066 signing method from the key type. This effectively removes the link 9067 between digests and public key types. 9068 9069 *Steve Henson* 9070 9071 * Add an OID cross reference table and utility functions. Its purpose is to 9072 translate between signature OIDs such as SHA1WithrsaEncryption and SHA1, 9073 rsaEncryption. This will allow some of the algorithm specific hackery 9074 needed to use the correct OID to be removed. 9075 9076 *Steve Henson* 9077 9078 * Remove algorithm specific dependencies when setting PKCS7_SIGNER_INFO 9079 structures for PKCS7_sign(). They are now set up by the relevant public 9080 key ASN1 method. 9081 9082 *Steve Henson* 9083 9084 * Add provisional EC pkey method with support for ECDSA and ECDH. 9085 9086 *Steve Henson* 9087 9088 * Add support for key derivation (agreement) in the API, DH method and 9089 pkeyutl. 9090 9091 *Steve Henson* 9092 9093 * Add DSA pkey method and DH pkey methods, extend DH ASN1 method to support 9094 public and private key formats. As a side effect these add additional 9095 command line functionality not previously available: DSA signatures can be 9096 generated and verified using pkeyutl and DH key support and generation in 9097 pkey, genpkey. 9098 9099 *Steve Henson* 9100 9101 * BeOS support. 9102 9103 *Oliver Tappe <zooey@hirschkaefer.de>* 9104 9105 * New make target "install_html_docs" installs HTML renditions of the 9106 manual pages. 9107 9108 *Oliver Tappe <zooey@hirschkaefer.de>* 9109 9110 * New utility "genpkey" this is analogous to "genrsa" etc except it can 9111 generate keys for any algorithm. Extend and update EVP_PKEY_METHOD to 9112 support key and parameter generation and add initial key generation 9113 functionality for RSA. 9114 9115 *Steve Henson* 9116 9117 * Add functions for main EVP_PKEY_method operations. The undocumented 9118 functions `EVP_PKEY_{encrypt,decrypt}` have been renamed to 9119 `EVP_PKEY_{encrypt,decrypt}_old`. 9120 9121 *Steve Henson* 9122 9123 * Initial definitions for EVP_PKEY_METHOD. This will be a high level public 9124 key API, doesn't do much yet. 9125 9126 *Steve Henson* 9127 9128 * New function EVP_PKEY_asn1_get0_info() to retrieve information about 9129 public key algorithms. New option to openssl utility: 9130 "list-public-key-algorithms" to print out info. 9131 9132 *Steve Henson* 9133 9134 * Implement the Supported Elliptic Curves Extension for 9135 ECC ciphersuites from draft-ietf-tls-ecc-12.txt. 9136 9137 *Douglas Stebila* 9138 9139 * Don't free up OIDs in OBJ_cleanup() if they are in use by EVP_MD or 9140 EVP_CIPHER structures to avoid later problems in EVP_cleanup(). 9141 9142 *Steve Henson* 9143 9144 * New utilities pkey and pkeyparam. These are similar to algorithm specific 9145 utilities such as rsa, dsa, dsaparam etc except they process any key 9146 type. 9147 9148 *Steve Henson* 9149 9150 * Transfer public key printing routines to EVP_PKEY_ASN1_METHOD. New 9151 functions EVP_PKEY_print_public(), EVP_PKEY_print_private(), 9152 EVP_PKEY_print_param() to print public key data from an EVP_PKEY 9153 structure. 9154 9155 *Steve Henson* 9156 9157 * Initial support for pluggable public key ASN1. 9158 De-spaghettify the public key ASN1 handling. Move public and private 9159 key ASN1 handling to a new EVP_PKEY_ASN1_METHOD structure. Relocate 9160 algorithm specific handling to a single module within the relevant 9161 algorithm directory. Add functions to allow (near) opaque processing 9162 of public and private key structures. 9163 9164 *Steve Henson* 9165 9166 * Implement the Supported Point Formats Extension for 9167 ECC ciphersuites from draft-ietf-tls-ecc-12.txt. 9168 9169 *Douglas Stebila* 9170 9171 * Add initial support for RFC 4279 PSK TLS ciphersuites. Add members 9172 for the psk identity [hint] and the psk callback functions to the 9173 SSL_SESSION, SSL and SSL_CTX structure. 9174 9175 New ciphersuites: 9176 PSK-RC4-SHA, PSK-3DES-EDE-CBC-SHA, PSK-AES128-CBC-SHA, 9177 PSK-AES256-CBC-SHA 9178 9179 New functions: 9180 SSL_CTX_use_psk_identity_hint 9181 SSL_get_psk_identity_hint 9182 SSL_get_psk_identity 9183 SSL_use_psk_identity_hint 9184 9185 *Mika Kousa and Pasi Eronen of Nokia Corporation* 9186 9187 * Add RFC 3161 compliant time stamp request creation, response generation 9188 and response verification functionality. 9189 9190 *Zoltán Glózik <zglozik@opentsa.org>, The OpenTSA Project* 9191 9192 * Add initial support for TLS extensions, specifically for the server_name 9193 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now 9194 have new members for a host name. The SSL data structure has an 9195 additional member `SSL_CTX *initial_ctx` so that new sessions can be 9196 stored in that context to allow for session resumption, even after the 9197 SSL has been switched to a new SSL_CTX in reaction to a client's 9198 server_name extension. 9199 9200 New functions (subject to change): 9201 9202 SSL_get_servername() 9203 SSL_get_servername_type() 9204 SSL_set_SSL_CTX() 9205 9206 New CTRL codes and macros (subject to change): 9207 9208 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB 9209 - SSL_CTX_set_tlsext_servername_callback() 9210 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG 9211 - SSL_CTX_set_tlsext_servername_arg() 9212 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name() 9213 9214 openssl s_client has a new '-servername ...' option. 9215 9216 openssl s_server has new options '-servername_host ...', '-cert2 ...', 9217 '-key2 ...', '-servername_fatal' (subject to change). This allows 9218 testing the HostName extension for a specific single host name ('-cert' 9219 and '-key' remain fallbacks for handshakes without HostName 9220 negotiation). If the unrecognized_name alert has to be sent, this by 9221 default is a warning; it becomes fatal with the '-servername_fatal' 9222 option. 9223 9224 *Peter Sylvester, Remy Allais, Christophe Renou* 9225 9226 * Whirlpool hash implementation is added. 9227 9228 *Andy Polyakov* 9229 9230 * BIGNUM code on 64-bit SPARCv9 targets is switched from bn(64,64) to 9231 bn(64,32). Because of instruction set limitations it doesn't have 9232 any negative impact on performance. This was done mostly in order 9233 to make it possible to share assembler modules, such as bn_mul_mont 9234 implementations, between 32- and 64-bit builds without hassle. 9235 9236 *Andy Polyakov* 9237 9238 * Move code previously exiled into file crypto/ec/ec2_smpt.c 9239 to ec2_smpl.c, and no longer require the OPENSSL_EC_BIN_PT_COMP 9240 macro. 9241 9242 *Bodo Moeller* 9243 9244 * New candidate for BIGNUM assembler implementation, bn_mul_mont, 9245 dedicated Montgomery multiplication procedure, is introduced. 9246 BN_MONT_CTX is modified to allow bn_mul_mont to reach for higher 9247 "64-bit" performance on certain 32-bit targets. 9248 9249 *Andy Polyakov* 9250 9251 * New option SSL_OP_NO_COMP to disable use of compression selectively 9252 in SSL structures. New SSL ctrl to set maximum send fragment size. 9253 Save memory by setting the I/O buffer sizes dynamically instead of 9254 using the maximum available value. 9255 9256 *Steve Henson* 9257 9258 * New option -V for 'openssl ciphers'. This prints the ciphersuite code 9259 in addition to the text details. 9260 9261 *Bodo Moeller* 9262 9263 * Very, very preliminary EXPERIMENTAL support for printing of general 9264 ASN1 structures. This currently produces rather ugly output and doesn't 9265 handle several customised structures at all. 9266 9267 *Steve Henson* 9268 9269 * Integrated support for PVK file format and some related formats such 9270 as MS PUBLICKEYBLOB and PRIVATEKEYBLOB. Command line switches to support 9271 these in the 'rsa' and 'dsa' utilities. 9272 9273 *Steve Henson* 9274 9275 * Support for PKCS#1 RSAPublicKey format on rsa utility command line. 9276 9277 *Steve Henson* 9278 9279 * Remove the ancient ASN1_METHOD code. This was only ever used in one 9280 place for the (very old) "NETSCAPE" format certificates which are now 9281 handled using new ASN1 code equivalents. 9282 9283 *Steve Henson* 9284 9285 * Let the TLSv1_method() etc. functions return a 'const' SSL_METHOD 9286 pointer and make the SSL_METHOD parameter in SSL_CTX_new, 9287 SSL_CTX_set_ssl_version and SSL_set_ssl_method 'const'. 9288 9289 *Nils Larsch* 9290 9291 * Modify CRL distribution points extension code to print out previously 9292 unsupported fields. Enhance extension setting code to allow setting of 9293 all fields. 9294 9295 *Steve Henson* 9296 9297 * Add print and set support for Issuing Distribution Point CRL extension. 9298 9299 *Steve Henson* 9300 9301 * Change 'Configure' script to enable Camellia by default. 9302 9303 *NTT* 9304 9305OpenSSL 0.9.x 9306------------- 9307 9308### Changes between 0.9.8m and 0.9.8n [24 Mar 2010] 9309 9310 * When rejecting SSL/TLS records due to an incorrect version number, never 9311 update s->server with a new major version number. As of 9312 - OpenSSL 0.9.8m if 'short' is a 16-bit type, 9313 - OpenSSL 0.9.8f if 'short' is longer than 16 bits, 9314 the previous behavior could result in a read attempt at NULL when 9315 receiving specific incorrect SSL/TLS records once record payload 9316 protection is active. ([CVE-2010-0740]) 9317 9318 *Bodo Moeller, Adam Langley <agl@chromium.org>* 9319 9320 * Fix for CVE-2010-0433 where some kerberos enabled versions of OpenSSL 9321 could be crashed if the relevant tables were not present (e.g. chrooted). 9322 9323 *Tomas Hoger <thoger@redhat.com>* 9324 9325### Changes between 0.9.8l and 0.9.8m [25 Feb 2010] 9326 9327 * Always check bn_wexpand() return values for failure. ([CVE-2009-3245]) 9328 9329 *Martin Olsson, Neel Mehta* 9330 9331 * Fix X509_STORE locking: Every 'objs' access requires a lock (to 9332 accommodate for stack sorting, always a write lock!). 9333 9334 *Bodo Moeller* 9335 9336 * On some versions of WIN32 Heap32Next is very slow. This can cause 9337 excessive delays in the RAND_poll(): over a minute. As a workaround 9338 include a time check in the inner Heap32Next loop too. 9339 9340 *Steve Henson* 9341 9342 * The code that handled flushing of data in SSL/TLS originally used the 9343 BIO_CTRL_INFO ctrl to see if any data was pending first. This caused 9344 the problem outlined in PR#1949. The fix suggested there however can 9345 trigger problems with buggy BIO_CTRL_WPENDING (e.g. some versions 9346 of Apache). So instead simplify the code to flush unconditionally. 9347 This should be fine since flushing with no data to flush is a no op. 9348 9349 *Steve Henson* 9350 9351 * Handle TLS versions 2.0 and later properly and correctly use the 9352 highest version of TLS/SSL supported. Although TLS >= 2.0 is some way 9353 off ancient servers have a habit of sticking around for a while... 9354 9355 *Steve Henson* 9356 9357 * Modify compression code so it frees up structures without using the 9358 ex_data callbacks. This works around a problem where some applications 9359 call CRYPTO_cleanup_all_ex_data() before application exit (e.g. when 9360 restarting) then use compression (e.g. SSL with compression) later. 9361 This results in significant per-connection memory leaks and 9362 has caused some security issues including CVE-2008-1678 and 9363 CVE-2009-4355. 9364 9365 *Steve Henson* 9366 9367 * Constify crypto/cast (i.e., <openssl/cast.h>): a CAST_KEY doesn't 9368 change when encrypting or decrypting. 9369 9370 *Bodo Moeller* 9371 9372 * Add option SSL_OP_LEGACY_SERVER_CONNECT which will allow clients to 9373 connect and renegotiate with servers which do not support RI. 9374 Until RI is more widely deployed this option is enabled by default. 9375 9376 *Steve Henson* 9377 9378 * Add "missing" ssl ctrls to clear options and mode. 9379 9380 *Steve Henson* 9381 9382 * If client attempts to renegotiate and doesn't support RI respond with 9383 a no_renegotiation alert as required by RFC5746. Some renegotiating 9384 TLS clients will continue a connection gracefully when they receive 9385 the alert. Unfortunately OpenSSL mishandled this alert and would hang 9386 waiting for a server hello which it will never receive. Now we treat a 9387 received no_renegotiation alert as a fatal error. This is because 9388 applications requesting a renegotiation might well expect it to succeed 9389 and would have no code in place to handle the server denying it so the 9390 only safe thing to do is to terminate the connection. 9391 9392 *Steve Henson* 9393 9394 * Add ctrl macro SSL_get_secure_renegotiation_support() which returns 1 if 9395 peer supports secure renegotiation and 0 otherwise. Print out peer 9396 renegotiation support in s_client/s_server. 9397 9398 *Steve Henson* 9399 9400 * Replace the highly broken and deprecated SPKAC certification method with 9401 the updated NID creation version. This should correctly handle UTF8. 9402 9403 *Steve Henson* 9404 9405 * Implement RFC5746. Re-enable renegotiation but require the extension 9406 as needed. Unfortunately, SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION 9407 turns out to be a bad idea. It has been replaced by 9408 SSL_OP_ALLOW_UNSAFE_LEGACY_RENEGOTIATION which can be set with 9409 SSL_CTX_set_options(). This is really not recommended unless you 9410 know what you are doing. 9411 9412 *Eric Rescorla <ekr@networkresonance.com>, Ben Laurie, Steve Henson* 9413 9414 * Fixes to stateless session resumption handling. Use initial_ctx when 9415 issuing and attempting to decrypt tickets in case it has changed during 9416 servername handling. Use a non-zero length session ID when attempting 9417 stateless session resumption: this makes it possible to determine if 9418 a resumption has occurred immediately after receiving server hello 9419 (several places in OpenSSL subtly assume this) instead of later in 9420 the handshake. 9421 9422 *Steve Henson* 9423 9424 * The functions ENGINE_ctrl(), OPENSSL_isservice(), 9425 CMS_get1_RecipientRequest() and RAND_bytes() can return <=0 on error 9426 fixes for a few places where the return code is not checked 9427 correctly. 9428 9429 *Julia Lawall <julia@diku.dk>* 9430 9431 * Add --strict-warnings option to Configure script to include devteam 9432 warnings in other configurations. 9433 9434 *Steve Henson* 9435 9436 * Add support for --libdir option and LIBDIR variable in makefiles. This 9437 makes it possible to install openssl libraries in locations which 9438 have names other than "lib", for example "/usr/lib64" which some 9439 systems need. 9440 9441 *Steve Henson, based on patch from Jeremy Utley* 9442 9443 * Don't allow the use of leading 0x80 in OIDs. This is a violation of 9444 X690 8.9.12 and can produce some misleading textual output of OIDs. 9445 9446 *Steve Henson, reported by Dan Kaminsky* 9447 9448 * Delete MD2 from algorithm tables. This follows the recommendation in 9449 several standards that it is not used in new applications due to 9450 several cryptographic weaknesses. For binary compatibility reasons 9451 the MD2 API is still compiled in by default. 9452 9453 *Steve Henson* 9454 9455 * Add compression id to {d2i,i2d}_SSL_SESSION so it is correctly saved 9456 and restored. 9457 9458 *Steve Henson* 9459 9460 * Rename uni2asc and asc2uni functions to OPENSSL_uni2asc and 9461 OPENSSL_asc2uni conditionally on Netware platforms to avoid a name 9462 clash. 9463 9464 *Guenter <lists@gknw.net>* 9465 9466 * Fix the server certificate chain building code to use X509_verify_cert(), 9467 it used to have an ad-hoc builder which was unable to cope with anything 9468 other than a simple chain. 9469 9470 *David Woodhouse <dwmw2@infradead.org>, Steve Henson* 9471 9472 * Don't check self signed certificate signatures in X509_verify_cert() 9473 by default (a flag can override this): it just wastes time without 9474 adding any security. As a useful side effect self signed root CAs 9475 with non-FIPS digests are now usable in FIPS mode. 9476 9477 *Steve Henson* 9478 9479 * In dtls1_process_out_of_seq_message() the check if the current message 9480 is already buffered was missing. For every new message was memory 9481 allocated, allowing an attacker to perform an denial of service attack 9482 with sending out of seq handshake messages until there is no memory 9483 left. Additionally every future message was buffered, even if the 9484 sequence number made no sense and would be part of another handshake. 9485 So only messages with sequence numbers less than 10 in advance will be 9486 buffered. ([CVE-2009-1378]) 9487 9488 *Robin Seggelmann, discovered by Daniel Mentz* 9489 9490 * Records are buffered if they arrive with a future epoch to be 9491 processed after finishing the corresponding handshake. There is 9492 currently no limitation to this buffer allowing an attacker to perform 9493 a DOS attack with sending records with future epochs until there is no 9494 memory left. This patch adds the pqueue_size() function to determine 9495 the size of a buffer and limits the record buffer to 100 entries. 9496 ([CVE-2009-1377]) 9497 9498 *Robin Seggelmann, discovered by Daniel Mentz* 9499 9500 * Keep a copy of frag->msg_header.frag_len so it can be used after the 9501 parent structure is freed. ([CVE-2009-1379]) 9502 9503 *Daniel Mentz* 9504 9505 * Handle non-blocking I/O properly in SSL_shutdown() call. 9506 9507 *Darryl Miles <darryl-mailinglists@netbauds.net>* 9508 9509 * Add `2.5.4.*` OIDs 9510 9511 *Ilya O. <vrghost@gmail.com>* 9512 9513### Changes between 0.9.8k and 0.9.8l [5 Nov 2009] 9514 9515 * Disable renegotiation completely - this fixes a severe security 9516 problem ([CVE-2009-3555]) at the cost of breaking all 9517 renegotiation. Renegotiation can be re-enabled by setting 9518 SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION in s3->flags at 9519 run-time. This is really not recommended unless you know what 9520 you're doing. 9521 9522 *Ben Laurie* 9523 9524### Changes between 0.9.8j and 0.9.8k [25 Mar 2009] 9525 9526 * Don't set val to NULL when freeing up structures, it is freed up by 9527 underlying code. If `sizeof(void *) > sizeof(long)` this can result in 9528 zeroing past the valid field. ([CVE-2009-0789]) 9529 9530 *Paolo Ganci <Paolo.Ganci@AdNovum.CH>* 9531 9532 * Fix bug where return value of CMS_SignerInfo_verify_content() was not 9533 checked correctly. This would allow some invalid signed attributes to 9534 appear to verify correctly. ([CVE-2009-0591]) 9535 9536 *Ivan Nestlerode <inestlerode@us.ibm.com>* 9537 9538 * Reject UniversalString and BMPString types with invalid lengths. This 9539 prevents a crash in ASN1_STRING_print_ex() which assumes the strings have 9540 a legal length. ([CVE-2009-0590]) 9541 9542 *Steve Henson* 9543 9544 * Set S/MIME signing as the default purpose rather than setting it 9545 unconditionally. This allows applications to override it at the store 9546 level. 9547 9548 *Steve Henson* 9549 9550 * Permit restricted recursion of ASN1 strings. This is needed in practice 9551 to handle some structures. 9552 9553 *Steve Henson* 9554 9555 * Improve efficiency of mem_gets: don't search whole buffer each time 9556 for a '\n' 9557 9558 *Jeremy Shapiro <jnshapir@us.ibm.com>* 9559 9560 * New -hex option for openssl rand. 9561 9562 *Matthieu Herrb* 9563 9564 * Print out UTF8String and NumericString when parsing ASN1. 9565 9566 *Steve Henson* 9567 9568 * Support NumericString type for name components. 9569 9570 *Steve Henson* 9571 9572 * Allow CC in the environment to override the automatically chosen 9573 compiler. Note that nothing is done to ensure flags work with the 9574 chosen compiler. 9575 9576 *Ben Laurie* 9577 9578### Changes between 0.9.8i and 0.9.8j [07 Jan 2009] 9579 9580 * Properly check EVP_VerifyFinal() and similar return values 9581 ([CVE-2008-5077]). 9582 9583 *Ben Laurie, Bodo Moeller, Google Security Team* 9584 9585 * Enable TLS extensions by default. 9586 9587 *Ben Laurie* 9588 9589 * Allow the CHIL engine to be loaded, whether the application is 9590 multithreaded or not. (This does not release the developer from the 9591 obligation to set up the dynamic locking callbacks.) 9592 9593 *Sander Temme <sander@temme.net>* 9594 9595 * Use correct exit code if there is an error in dgst command. 9596 9597 *Steve Henson; problem pointed out by Roland Dirlewanger* 9598 9599 * Tweak Configure so that you need to say "experimental-jpake" to enable 9600 JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications. 9601 9602 *Bodo Moeller* 9603 9604 * Add experimental JPAKE support, including demo authentication in 9605 s_client and s_server. 9606 9607 *Ben Laurie* 9608 9609 * Set the comparison function in v3_addr_canonize(). 9610 9611 *Rob Austein <sra@hactrn.net>* 9612 9613 * Add support for XMPP STARTTLS in s_client. 9614 9615 *Philip Paeps <philip@freebsd.org>* 9616 9617 * Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior 9618 to ensure that even with this option, only ciphersuites in the 9619 server's preference list will be accepted. (Note that the option 9620 applies only when resuming a session, so the earlier behavior was 9621 just about the algorithm choice for symmetric cryptography.) 9622 9623 *Bodo Moeller* 9624 9625### Changes between 0.9.8h and 0.9.8i [15 Sep 2008] 9626 9627 * Fix NULL pointer dereference if a DTLS server received 9628 ChangeCipherSpec as first record ([CVE-2009-1386]). 9629 9630 *PR #1679* 9631 9632 * Fix a state transition in s3_srvr.c and d1_srvr.c 9633 (was using SSL3_ST_CW_CLNT_HELLO_B, should be `..._ST_SW_SRVR_...`). 9634 9635 *Nagendra Modadugu* 9636 9637 * The fix in 0.9.8c that supposedly got rid of unsafe 9638 double-checked locking was incomplete for RSA blinding, 9639 addressing just one layer of what turns out to have been 9640 doubly unsafe triple-checked locking. 9641 9642 So now fix this for real by retiring the MONT_HELPER macro 9643 in crypto/rsa/rsa_eay.c. 9644 9645 *Bodo Moeller; problem pointed out by Marius Schilder* 9646 9647 * Various precautionary measures: 9648 9649 - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h). 9650 9651 - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c). 9652 (NB: This would require knowledge of the secret session ticket key 9653 to exploit, in which case you'd be SOL either way.) 9654 9655 - Change bn_nist.c so that it will properly handle input BIGNUMs 9656 outside the expected range. 9657 9658 - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG 9659 builds. 9660 9661 *Neel Mehta, Bodo Moeller* 9662 9663 * Allow engines to be "soft loaded" - i.e. optionally don't die if 9664 the load fails. Useful for distros. 9665 9666 *Ben Laurie and the FreeBSD team* 9667 9668 * Add support for Local Machine Keyset attribute in PKCS#12 files. 9669 9670 *Steve Henson* 9671 9672 * Fix BN_GF2m_mod_arr() top-bit cleanup code. 9673 9674 *Huang Ying* 9675 9676 * Expand ENGINE to support engine supplied SSL client certificate functions. 9677 9678 This work was sponsored by Logica. 9679 9680 *Steve Henson* 9681 9682 * Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows 9683 keystores. Support for SSL/TLS client authentication too. 9684 Not compiled unless enable-capieng specified to Configure. 9685 9686 This work was sponsored by Logica. 9687 9688 *Steve Henson* 9689 9690 * Fix bug in X509_ATTRIBUTE creation: don't set attribute using 9691 ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain 9692 attribute creation routines such as certificate requests and PKCS#12 9693 files. 9694 9695 *Steve Henson* 9696 9697### Changes between 0.9.8g and 0.9.8h [28 May 2008] 9698 9699 * Fix flaw if 'Server Key exchange message' is omitted from a TLS 9700 handshake which could lead to a client crash as found using the 9701 Codenomicon TLS test suite ([CVE-2008-1672]) 9702 9703 *Steve Henson, Mark Cox* 9704 9705 * Fix double free in TLS server name extensions which could lead to 9706 a remote crash found by Codenomicon TLS test suite ([CVE-2008-0891]) 9707 9708 *Joe Orton* 9709 9710 * Clear error queue in SSL_CTX_use_certificate_chain_file() 9711 9712 Clear the error queue to ensure that error entries left from 9713 older function calls do not interfere with the correct operation. 9714 9715 *Lutz Jaenicke, Erik de Castro Lopo* 9716 9717 * Remove root CA certificates of commercial CAs: 9718 9719 The OpenSSL project does not recommend any specific CA and does not 9720 have any policy with respect to including or excluding any CA. 9721 Therefore it does not make any sense to ship an arbitrary selection 9722 of root CA certificates with the OpenSSL software. 9723 9724 *Lutz Jaenicke* 9725 9726 * RSA OAEP patches to fix two separate invalid memory reads. 9727 The first one involves inputs when 'lzero' is greater than 9728 'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes 9729 before the beginning of from). The second one involves inputs where 9730 the 'db' section contains nothing but zeroes (there is a one-byte 9731 invalid read after the end of 'db'). 9732 9733 *Ivan Nestlerode <inestlerode@us.ibm.com>* 9734 9735 * Partial backport from 0.9.9-dev: 9736 9737 Introduce bn_mul_mont (dedicated Montgomery multiplication 9738 procedure) as a candidate for BIGNUM assembler implementation. 9739 While 0.9.9-dev uses assembler for various architectures, only 9740 x86_64 is available by default here in the 0.9.8 branch, and 9741 32-bit x86 is available through a compile-time setting. 9742 9743 To try the 32-bit x86 assembler implementation, use Configure 9744 option "enable-montasm" (which exists only for this backport). 9745 9746 As "enable-montasm" for 32-bit x86 disclaims code stability 9747 anyway, in this constellation we activate additional code 9748 backported from 0.9.9-dev for further performance improvements, 9749 namely BN_from_montgomery_word. (To enable this otherwise, 9750 e.g. x86_64, try `-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD`.) 9751 9752 *Andy Polyakov (backport partially by Bodo Moeller)* 9753 9754 * Add TLS session ticket callback. This allows an application to set 9755 TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed 9756 values. This is useful for key rollover for example where several key 9757 sets may exist with different names. 9758 9759 *Steve Henson* 9760 9761 * Reverse ENGINE-internal logic for caching default ENGINE handles. 9762 This was broken until now in 0.9.8 releases, such that the only way 9763 a registered ENGINE could be used (assuming it initialises 9764 successfully on the host) was to explicitly set it as the default 9765 for the relevant algorithms. This is in contradiction with 0.9.7 9766 behaviour and the documentation. With this fix, when an ENGINE is 9767 registered into a given algorithm's table of implementations, the 9768 'uptodate' flag is reset so that auto-discovery will be used next 9769 time a new context for that algorithm attempts to select an 9770 implementation. 9771 9772 *Ian Lister (tweaked by Geoff Thorpe)* 9773 9774 * Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9 9775 implementation in the following ways: 9776 9777 Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be 9778 hard coded. 9779 9780 Lack of BER streaming support means one pass streaming processing is 9781 only supported if data is detached: setting the streaming flag is 9782 ignored for embedded content. 9783 9784 CMS support is disabled by default and must be explicitly enabled 9785 with the enable-cms configuration option. 9786 9787 *Steve Henson* 9788 9789 * Update the GMP engine glue to do direct copies between BIGNUM and 9790 mpz_t when openssl and GMP use the same limb size. Otherwise the 9791 existing "conversion via a text string export" trick is still used. 9792 9793 *Paul Sheer <paulsheer@gmail.com>* 9794 9795 * Zlib compression BIO. This is a filter BIO which compressed and 9796 uncompresses any data passed through it. 9797 9798 *Steve Henson* 9799 9800 * Add AES_wrap_key() and AES_unwrap_key() functions to implement 9801 RFC3394 compatible AES key wrapping. 9802 9803 *Steve Henson* 9804 9805 * Add utility functions to handle ASN1 structures. ASN1_STRING_set0(): 9806 sets string data without copying. X509_ALGOR_set0() and 9807 X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier) 9808 data. Attribute function X509at_get0_data_by_OBJ(): retrieves data 9809 from an X509_ATTRIBUTE structure optionally checking it occurs only 9810 once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied 9811 data. 9812 9813 *Steve Henson* 9814 9815 * Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set() 9816 to get the expected BN_FLG_CONSTTIME behavior. 9817 9818 *Bodo Moeller (Google)* 9819 9820 * Netware support: 9821 9822 - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets 9823 - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT) 9824 - added some more tests to do_tests.pl 9825 - fixed RunningProcess usage so that it works with newer LIBC NDKs too 9826 - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency 9827 - added new Configure targets netware-clib-bsdsock, netware-clib-gcc, 9828 netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc 9829 - various changes to netware.pl to enable gcc-cross builds on Win32 9830 platform 9831 - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD) 9832 - various changes to fix missing prototype warnings 9833 - fixed x86nasm.pl to create correct asm files for NASM COFF output 9834 - added AES, WHIRLPOOL and CPUID assembler code to build files 9835 - added missing AES assembler make rules to mk1mf.pl 9836 - fixed order of includes in `apps/ocsp.c` so that `e_os.h` settings apply 9837 9838 *Guenter Knauf <eflash@gmx.net>* 9839 9840 * Implement certificate status request TLS extension defined in RFC3546. 9841 A client can set the appropriate parameters and receive the encoded 9842 OCSP response via a callback. A server can query the supplied parameters 9843 and set the encoded OCSP response in the callback. Add simplified examples 9844 to s_client and s_server. 9845 9846 *Steve Henson* 9847 9848### Changes between 0.9.8f and 0.9.8g [19 Oct 2007] 9849 9850 * Fix various bugs: 9851 + Binary incompatibility of ssl_ctx_st structure 9852 + DTLS interoperation with non-compliant servers 9853 + Don't call get_session_cb() without proposed session 9854 + Fix ia64 assembler code 9855 9856 *Andy Polyakov, Steve Henson* 9857 9858### Changes between 0.9.8e and 0.9.8f [11 Oct 2007] 9859 9860 * DTLS Handshake overhaul. There were longstanding issues with 9861 OpenSSL DTLS implementation, which were making it impossible for 9862 RFC 4347 compliant client to communicate with OpenSSL server. 9863 Unfortunately just fixing these incompatibilities would "cut off" 9864 pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e 9865 server keeps tolerating non RFC compliant syntax. The opposite is 9866 not true, 0.9.8f client can not communicate with earlier server. 9867 This update even addresses CVE-2007-4995. 9868 9869 *Andy Polyakov* 9870 9871 * Changes to avoid need for function casts in OpenSSL: some compilers 9872 (gcc 4.2 and later) reject their use. 9873 *Kurt Roeckx <kurt@roeckx.be>, Peter Hartley <pdh@utter.chaos.org.uk>, 9874 Steve Henson* 9875 9876 * Add RFC4507 support to OpenSSL. This includes the corrections in 9877 RFC4507bis. The encrypted ticket format is an encrypted encoded 9878 SSL_SESSION structure, that way new session features are automatically 9879 supported. 9880 9881 If a client application caches session in an SSL_SESSION structure 9882 support is transparent because tickets are now stored in the encoded 9883 SSL_SESSION. 9884 9885 The SSL_CTX structure automatically generates keys for ticket 9886 protection in servers so again support should be possible 9887 with no application modification. 9888 9889 If a client or server wishes to disable RFC4507 support then the option 9890 SSL_OP_NO_TICKET can be set. 9891 9892 Add a TLS extension debugging callback to allow the contents of any client 9893 or server extensions to be examined. 9894 9895 This work was sponsored by Google. 9896 9897 *Steve Henson* 9898 9899 * Add initial support for TLS extensions, specifically for the server_name 9900 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now 9901 have new members for a host name. The SSL data structure has an 9902 additional member `SSL_CTX *initial_ctx` so that new sessions can be 9903 stored in that context to allow for session resumption, even after the 9904 SSL has been switched to a new SSL_CTX in reaction to a client's 9905 server_name extension. 9906 9907 New functions (subject to change): 9908 9909 SSL_get_servername() 9910 SSL_get_servername_type() 9911 SSL_set_SSL_CTX() 9912 9913 New CTRL codes and macros (subject to change): 9914 9915 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB 9916 - SSL_CTX_set_tlsext_servername_callback() 9917 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG 9918 - SSL_CTX_set_tlsext_servername_arg() 9919 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name() 9920 9921 openssl s_client has a new '-servername ...' option. 9922 9923 openssl s_server has new options '-servername_host ...', '-cert2 ...', 9924 '-key2 ...', '-servername_fatal' (subject to change). This allows 9925 testing the HostName extension for a specific single host name ('-cert' 9926 and '-key' remain fallbacks for handshakes without HostName 9927 negotiation). If the unrecognized_name alert has to be sent, this by 9928 default is a warning; it becomes fatal with the '-servername_fatal' 9929 option. 9930 9931 *Peter Sylvester, Remy Allais, Christophe Renou, Steve Henson* 9932 9933 * Add AES and SSE2 assembly language support to VC++ build. 9934 9935 *Steve Henson* 9936 9937 * Mitigate attack on final subtraction in Montgomery reduction. 9938 9939 *Andy Polyakov* 9940 9941 * Fix crypto/ec/ec_mult.c to work properly with scalars of value 0 9942 (which previously caused an internal error). 9943 9944 *Bodo Moeller* 9945 9946 * Squeeze another 10% out of IGE mode when in != out. 9947 9948 *Ben Laurie* 9949 9950 * AES IGE mode speedup. 9951 9952 *Dean Gaudet (Google)* 9953 9954 * Add the Korean symmetric 128-bit cipher SEED (see 9955 <http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp>) and 9956 add SEED ciphersuites from RFC 4162: 9957 9958 TLS_RSA_WITH_SEED_CBC_SHA = "SEED-SHA" 9959 TLS_DHE_DSS_WITH_SEED_CBC_SHA = "DHE-DSS-SEED-SHA" 9960 TLS_DHE_RSA_WITH_SEED_CBC_SHA = "DHE-RSA-SEED-SHA" 9961 TLS_DH_anon_WITH_SEED_CBC_SHA = "ADH-SEED-SHA" 9962 9963 To minimize changes between patchlevels in the OpenSSL 0.9.8 9964 series, SEED remains excluded from compilation unless OpenSSL 9965 is configured with 'enable-seed'. 9966 9967 *KISA, Bodo Moeller* 9968 9969 * Mitigate branch prediction attacks, which can be practical if a 9970 single processor is shared, allowing a spy process to extract 9971 information. For detailed background information, see 9972 <http://eprint.iacr.org/2007/039> (O. Aciicmez, S. Gueron, 9973 J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL 9974 and Necessary Software Countermeasures"). The core of the change 9975 are new versions BN_div_no_branch() and 9976 BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(), 9977 respectively, which are slower, but avoid the security-relevant 9978 conditional branches. These are automatically called by BN_div() 9979 and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one 9980 of the input BIGNUMs. Also, BN_is_bit_set() has been changed to 9981 remove a conditional branch. 9982 9983 BN_FLG_CONSTTIME is the new name for the previous 9984 BN_FLG_EXP_CONSTTIME flag, since it now affects more than just 9985 modular exponentiation. (Since OpenSSL 0.9.7h, setting this flag 9986 in the exponent causes BN_mod_exp_mont() to use the alternative 9987 implementation in BN_mod_exp_mont_consttime().) The old name 9988 remains as a deprecated alias. 9989 9990 Similarly, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general 9991 RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses 9992 constant-time implementations for more than just exponentiation. 9993 Here too the old name is kept as a deprecated alias. 9994 9995 BN_BLINDING_new() will now use BN_dup() for the modulus so that 9996 the BN_BLINDING structure gets an independent copy of the 9997 modulus. This means that the previous `BIGNUM *m` argument to 9998 BN_BLINDING_new() and to BN_BLINDING_create_param() now 9999 essentially becomes `const BIGNUM *m`, although we can't actually 10000 change this in the header file before 0.9.9. It allows 10001 RSA_setup_blinding() to use BN_with_flags() on the modulus to 10002 enable BN_FLG_CONSTTIME. 10003 10004 *Matthew D Wood (Intel Corp)* 10005 10006 * In the SSL/TLS server implementation, be strict about session ID 10007 context matching (which matters if an application uses a single 10008 external cache for different purposes). Previously, 10009 out-of-context reuse was forbidden only if SSL_VERIFY_PEER was 10010 set. This did ensure strict client verification, but meant that, 10011 with applications using a single external cache for quite 10012 different requirements, clients could circumvent ciphersuite 10013 restrictions for a given session ID context by starting a session 10014 in a different context. 10015 10016 *Bodo Moeller* 10017 10018 * Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 10019 a ciphersuite string such as "DEFAULT:RSA" cannot enable 10020 authentication-only ciphersuites. 10021 10022 *Bodo Moeller* 10023 10024 * Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was 10025 not complete and could lead to a possible single byte overflow 10026 ([CVE-2007-5135]) [Ben Laurie] 10027 10028### Changes between 0.9.8d and 0.9.8e [23 Feb 2007] 10029 10030 * Since AES128 and AES256 (and similarly Camellia128 and 10031 Camellia256) share a single mask bit in the logic of 10032 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 10033 kludge to work properly if AES128 is available and AES256 isn't 10034 (or if Camellia128 is available and Camellia256 isn't). 10035 10036 *Victor Duchovni* 10037 10038 * Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c 10039 (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters): 10040 When a point or a seed is encoded in a BIT STRING, we need to 10041 prevent the removal of trailing zero bits to get the proper DER 10042 encoding. (By default, crypto/asn1/a_bitstr.c assumes the case 10043 of a NamedBitList, for which trailing 0 bits need to be removed.) 10044 10045 *Bodo Moeller* 10046 10047 * Have SSL/TLS server implementation tolerate "mismatched" record 10048 protocol version while receiving ClientHello even if the 10049 ClientHello is fragmented. (The server can't insist on the 10050 particular protocol version it has chosen before the ServerHello 10051 message has informed the client about his choice.) 10052 10053 *Bodo Moeller* 10054 10055 * Add RFC 3779 support. 10056 10057 *Rob Austein for ARIN, Ben Laurie* 10058 10059 * Load error codes if they are not already present instead of using a 10060 static variable. This allows them to be cleanly unloaded and reloaded. 10061 Improve header file function name parsing. 10062 10063 *Steve Henson* 10064 10065 * extend SMTP and IMAP protocol emulation in s_client to use EHLO 10066 or CAPABILITY handshake as required by RFCs. 10067 10068 *Goetz Babin-Ebell* 10069 10070### Changes between 0.9.8c and 0.9.8d [28 Sep 2006] 10071 10072 * Introduce limits to prevent malicious keys being able to 10073 cause a denial of service. ([CVE-2006-2940]) 10074 10075 *Steve Henson, Bodo Moeller* 10076 10077 * Fix ASN.1 parsing of certain invalid structures that can result 10078 in a denial of service. ([CVE-2006-2937]) [Steve Henson] 10079 10080 * Fix buffer overflow in SSL_get_shared_ciphers() function. 10081 ([CVE-2006-3738]) [Tavis Ormandy and Will Drewry, Google Security Team] 10082 10083 * Fix SSL client code which could crash if connecting to a 10084 malicious SSLv2 server. ([CVE-2006-4343]) 10085 10086 *Tavis Ormandy and Will Drewry, Google Security Team* 10087 10088 * Since 0.9.8b, ciphersuite strings naming explicit ciphersuites 10089 match only those. Before that, "AES256-SHA" would be interpreted 10090 as a pattern and match "AES128-SHA" too (since AES128-SHA got 10091 the same strength classification in 0.9.7h) as we currently only 10092 have a single AES bit in the ciphersuite description bitmap. 10093 That change, however, also applied to ciphersuite strings such as 10094 "RC4-MD5" that intentionally matched multiple ciphersuites -- 10095 namely, SSL 2.0 ciphersuites in addition to the more common ones 10096 from SSL 3.0/TLS 1.0. 10097 10098 So we change the selection algorithm again: Naming an explicit 10099 ciphersuite selects this one ciphersuite, and any other similar 10100 ciphersuite (same bitmap) from *other* protocol versions. 10101 Thus, "RC4-MD5" again will properly select both the SSL 2.0 10102 ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite. 10103 10104 Since SSL 2.0 does not have any ciphersuites for which the 10105 128/256 bit distinction would be relevant, this works for now. 10106 The proper fix will be to use different bits for AES128 and 10107 AES256, which would have avoided the problems from the beginning; 10108 however, bits are scarce, so we can only do this in a new release 10109 (not just a patchlevel) when we can change the SSL_CIPHER 10110 definition to split the single 'unsigned long mask' bitmap into 10111 multiple values to extend the available space. 10112 10113 *Bodo Moeller* 10114 10115### Changes between 0.9.8b and 0.9.8c [05 Sep 2006] 10116 10117 * Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 10118 ([CVE-2006-4339]) [Ben Laurie and Google Security Team] 10119 10120 * Add AES IGE and biIGE modes. 10121 10122 *Ben Laurie* 10123 10124 * Change the Unix randomness entropy gathering to use poll() when 10125 possible instead of select(), since the latter has some 10126 undesirable limitations. 10127 10128 *Darryl Miles via Richard Levitte and Bodo Moeller* 10129 10130 * Disable "ECCdraft" ciphersuites more thoroughly. Now special 10131 treatment in ssl/ssl_ciph.s makes sure that these ciphersuites 10132 cannot be implicitly activated as part of, e.g., the "AES" alias. 10133 However, please upgrade to OpenSSL 0.9.9[-dev] for 10134 non-experimental use of the ECC ciphersuites to get TLS extension 10135 support, which is required for curve and point format negotiation 10136 to avoid potential handshake problems. 10137 10138 *Bodo Moeller* 10139 10140 * Disable rogue ciphersuites: 10141 10142 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 10143 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 10144 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 10145 10146 The latter two were purportedly from 10147 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 10148 appear there. 10149 10150 Also deactivate the remaining ciphersuites from 10151 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 10152 unofficial, and the ID has long expired. 10153 10154 *Bodo Moeller* 10155 10156 * Fix RSA blinding Heisenbug (problems sometimes occurred on 10157 dual-core machines) and other potential thread-safety issues. 10158 10159 *Bodo Moeller* 10160 10161 * Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key 10162 versions), which is now available for royalty-free use 10163 (see <http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html>). 10164 Also, add Camellia TLS ciphersuites from RFC 4132. 10165 10166 To minimize changes between patchlevels in the OpenSSL 0.9.8 10167 series, Camellia remains excluded from compilation unless OpenSSL 10168 is configured with 'enable-camellia'. 10169 10170 *NTT* 10171 10172 * Disable the padding bug check when compression is in use. The padding 10173 bug check assumes the first packet is of even length, this is not 10174 necessarily true if compression is enabled and can result in false 10175 positives causing handshake failure. The actual bug test is ancient 10176 code so it is hoped that implementations will either have fixed it by 10177 now or any which still have the bug do not support compression. 10178 10179 *Steve Henson* 10180 10181### Changes between 0.9.8a and 0.9.8b [04 May 2006] 10182 10183 * When applying a cipher rule check to see if string match is an explicit 10184 cipher suite and only match that one cipher suite if it is. 10185 10186 *Steve Henson* 10187 10188 * Link in manifests for VC++ if needed. 10189 10190 *Austin Ziegler <halostatue@gmail.com>* 10191 10192 * Update support for ECC-based TLS ciphersuites according to 10193 draft-ietf-tls-ecc-12.txt with proposed changes (but without 10194 TLS extensions, which are supported starting with the 0.9.9 10195 branch, not in the OpenSSL 0.9.8 branch). 10196 10197 *Douglas Stebila* 10198 10199 * New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support 10200 opaque EVP_CIPHER_CTX handling. 10201 10202 *Steve Henson* 10203 10204 * Fixes and enhancements to zlib compression code. We now only use 10205 "zlib1.dll" and use the default `__cdecl` calling convention on Win32 10206 to conform with the standards mentioned here: 10207 <http://www.zlib.net/DLL_FAQ.txt> 10208 Static zlib linking now works on Windows and the new --with-zlib-include 10209 --with-zlib-lib options to Configure can be used to supply the location 10210 of the headers and library. Gracefully handle case where zlib library 10211 can't be loaded. 10212 10213 *Steve Henson* 10214 10215 * Several fixes and enhancements to the OID generation code. The old code 10216 sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't 10217 handle numbers larger than ULONG_MAX, truncated printing and had a 10218 non standard OBJ_obj2txt() behaviour. 10219 10220 *Steve Henson* 10221 10222 * Add support for building of engines under engine/ as shared libraries 10223 under VC++ build system. 10224 10225 *Steve Henson* 10226 10227 * Corrected the numerous bugs in the Win32 path splitter in DSO. 10228 Hopefully, we will not see any false combination of paths any more. 10229 10230 *Richard Levitte* 10231 10232### Changes between 0.9.8 and 0.9.8a [11 Oct 2005] 10233 10234 * Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 10235 (part of SSL_OP_ALL). This option used to disable the 10236 countermeasure against man-in-the-middle protocol-version 10237 rollback in the SSL 2.0 server implementation, which is a bad 10238 idea. ([CVE-2005-2969]) 10239 10240 *Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 10241 for Information Security, National Institute of Advanced Industrial 10242 Science and Technology [AIST], Japan)* 10243 10244 * Add two function to clear and return the verify parameter flags. 10245 10246 *Steve Henson* 10247 10248 * Keep cipherlists sorted in the source instead of sorting them at 10249 runtime, thus removing the need for a lock. 10250 10251 *Nils Larsch* 10252 10253 * Avoid some small subgroup attacks in Diffie-Hellman. 10254 10255 *Nick Mathewson and Ben Laurie* 10256 10257 * Add functions for well-known primes. 10258 10259 *Nick Mathewson* 10260 10261 * Extended Windows CE support. 10262 10263 *Satoshi Nakamura and Andy Polyakov* 10264 10265 * Initialize SSL_METHOD structures at compile time instead of during 10266 runtime, thus removing the need for a lock. 10267 10268 *Steve Henson* 10269 10270 * Make PKCS7_decrypt() work even if no certificate is supplied by 10271 attempting to decrypt each encrypted key in turn. Add support to 10272 smime utility. 10273 10274 *Steve Henson* 10275 10276### Changes between 0.9.7h and 0.9.8 [05 Jul 2005] 10277 10278[NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after 10279OpenSSL 0.9.8.] 10280 10281 * Add libcrypto.pc and libssl.pc for those who feel they need them. 10282 10283 *Richard Levitte* 10284 10285 * Change CA.sh and CA.pl so they don't bundle the CSR and the private 10286 key into the same file any more. 10287 10288 *Richard Levitte* 10289 10290 * Add initial support for Win64, both IA64 and AMD64/x64 flavors. 10291 10292 *Andy Polyakov* 10293 10294 * Add -utf8 command line and config file option to 'ca'. 10295 10296 *Stefan <stf@udoma.org* 10297 10298 * Removed the macro des_crypt(), as it seems to conflict with some 10299 libraries. Use DES_crypt(). 10300 10301 *Richard Levitte* 10302 10303 * Correct naming of the 'chil' and '4758cca' ENGINEs. This 10304 involves renaming the source and generated shared-libs for 10305 both. The engines will accept the corrected or legacy ids 10306 ('ncipher' and '4758_cca' respectively) when binding. NB, 10307 this only applies when building 'shared'. 10308 10309 *Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe* 10310 10311 * Add attribute functions to EVP_PKEY structure. Modify 10312 PKCS12_create() to recognize a CSP name attribute and 10313 use it. Make -CSP option work again in pkcs12 utility. 10314 10315 *Steve Henson* 10316 10317 * Add new functionality to the bn blinding code: 10318 - automatic re-creation of the BN_BLINDING parameters after 10319 a fixed number of uses (currently 32) 10320 - add new function for parameter creation 10321 - introduce flags to control the update behaviour of the 10322 BN_BLINDING parameters 10323 - hide BN_BLINDING structure 10324 Add a second BN_BLINDING slot to the RSA structure to improve 10325 performance when a single RSA object is shared among several 10326 threads. 10327 10328 *Nils Larsch* 10329 10330 * Add support for DTLS. 10331 10332 *Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie* 10333 10334 * Add support for DER encoded private keys (SSL_FILETYPE_ASN1) 10335 to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file() 10336 10337 *Walter Goulet* 10338 10339 * Remove buggy and incomplete DH cert support from 10340 ssl/ssl_rsa.c and ssl/s3_both.c 10341 10342 *Nils Larsch* 10343 10344 * Use SHA-1 instead of MD5 as the default digest algorithm for 10345 the `apps/openssl` commands. 10346 10347 *Nils Larsch* 10348 10349 * Compile clean with "-Wall -Wmissing-prototypes 10350 -Wstrict-prototypes -Wmissing-declarations -Werror". Currently 10351 DEBUG_SAFESTACK must also be set. 10352 10353 *Ben Laurie* 10354 10355 * Change ./Configure so that certain algorithms can be disabled by default. 10356 The new counterpiece to "no-xxx" is "enable-xxx". 10357 10358 The patented RC5 and MDC2 algorithms will now be disabled unless 10359 "enable-rc5" and "enable-mdc2", respectively, are specified. 10360 10361 (IDEA remains enabled despite being patented. This is because IDEA 10362 is frequently required for interoperability, and there is no license 10363 fee for non-commercial use. As before, "no-idea" can be used to 10364 avoid this algorithm.) 10365 10366 *Bodo Moeller* 10367 10368 * Add processing of proxy certificates (see RFC 3820). This work was 10369 sponsored by KTH (The Royal Institute of Technology in Stockholm) and 10370 EGEE (Enabling Grids for E-science in Europe). 10371 10372 *Richard Levitte* 10373 10374 * RC4 performance overhaul on modern architectures/implementations, such 10375 as Intel P4, IA-64 and AMD64. 10376 10377 *Andy Polyakov* 10378 10379 * New utility extract-section.pl. This can be used specify an alternative 10380 section number in a pod file instead of having to treat each file as 10381 a separate case in Makefile. This can be done by adding two lines to the 10382 pod file: 10383 10384 =for comment openssl_section:XXX 10385 10386 The blank line is mandatory. 10387 10388 *Steve Henson* 10389 10390 * New arguments -certform, -keyform and -pass for s_client and s_server 10391 to allow alternative format key and certificate files and passphrase 10392 sources. 10393 10394 *Steve Henson* 10395 10396 * New structure X509_VERIFY_PARAM which combines current verify parameters, 10397 update associated structures and add various utility functions. 10398 10399 Add new policy related verify parameters, include policy checking in 10400 standard verify code. Enhance 'smime' application with extra parameters 10401 to support policy checking and print out. 10402 10403 *Steve Henson* 10404 10405 * Add a new engine to support VIA PadLock ACE extensions in the VIA C3 10406 Nehemiah processors. These extensions support AES encryption in hardware 10407 as well as RNG (though RNG support is currently disabled). 10408 10409 *Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov* 10410 10411 * Deprecate `BN_[get|set]_params()` functions (they were ignored internally). 10412 10413 *Geoff Thorpe* 10414 10415 * New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented. 10416 10417 *Andy Polyakov and a number of other people* 10418 10419 * Improved PowerPC platform support. Most notably BIGNUM assembler 10420 implementation contributed by IBM. 10421 10422 *Suresh Chari, Peter Waltenberg, Andy Polyakov* 10423 10424 * The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public 10425 exponent rather than 'unsigned long'. There is a corresponding change to 10426 the new 'rsa_keygen' element of the RSA_METHOD structure. 10427 10428 *Jelte Jansen, Geoff Thorpe* 10429 10430 * Functionality for creating the initial serial number file is now 10431 moved from CA.pl to the 'ca' utility with a new option -create_serial. 10432 10433 (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial 10434 number file to 1, which is bound to cause problems. To avoid 10435 the problems while respecting compatibility between different 0.9.7 10436 patchlevels, 0.9.7e employed 'openssl x509 -next_serial' in 10437 CA.pl for serial number initialization. With the new release 0.9.8, 10438 we can fix the problem directly in the 'ca' utility.) 10439 10440 *Steve Henson* 10441 10442 * Reduced header interdependencies by declaring more opaque objects in 10443 ossl_typ.h. As a consequence, including some headers (eg. engine.h) will 10444 give fewer recursive includes, which could break lazy source code - so 10445 this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always, 10446 developers should define this symbol when building and using openssl to 10447 ensure they track the recommended behaviour, interfaces, [etc], but 10448 backwards-compatible behaviour prevails when this isn't defined. 10449 10450 *Geoff Thorpe* 10451 10452 * New function X509_POLICY_NODE_print() which prints out policy nodes. 10453 10454 *Steve Henson* 10455 10456 * Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality. 10457 This will generate a random key of the appropriate length based on the 10458 cipher context. The EVP_CIPHER can provide its own random key generation 10459 routine to support keys of a specific form. This is used in the des and 10460 3des routines to generate a key of the correct parity. Update S/MIME 10461 code to use new functions and hence generate correct parity DES keys. 10462 Add EVP_CHECK_DES_KEY #define to return an error if the key is not 10463 valid (weak or incorrect parity). 10464 10465 *Steve Henson* 10466 10467 * Add a local set of CRLs that can be used by X509_verify_cert() as well 10468 as looking them up. This is useful when the verified structure may contain 10469 CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs 10470 present unless the new PKCS7_NO_CRL flag is asserted. 10471 10472 *Steve Henson* 10473 10474 * Extend ASN1 oid configuration module. It now additionally accepts the 10475 syntax: 10476 10477 shortName = some long name, 1.2.3.4 10478 10479 *Steve Henson* 10480 10481 * Reimplemented the BN_CTX implementation. There is now no more static 10482 limitation on the number of variables it can handle nor the depth of the 10483 "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack 10484 information can now expand as required, and rather than having a single 10485 static array of bignums, BN_CTX now uses a linked-list of such arrays 10486 allowing it to expand on demand whilst maintaining the usefulness of 10487 BN_CTX's "bundling". 10488 10489 *Geoff Thorpe* 10490 10491 * Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD 10492 to allow all RSA operations to function using a single BN_CTX. 10493 10494 *Geoff Thorpe* 10495 10496 * Preliminary support for certificate policy evaluation and checking. This 10497 is initially intended to pass the tests outlined in "Conformance Testing 10498 of Relying Party Client Certificate Path Processing Logic" v1.07. 10499 10500 *Steve Henson* 10501 10502 * bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and 10503 remained unused and not that useful. A variety of other little bignum 10504 tweaks and fixes have also been made continuing on from the audit (see 10505 below). 10506 10507 *Geoff Thorpe* 10508 10509 * Constify all or almost all d2i, c2i, s2i and r2i functions, along with 10510 associated ASN1, EVP and SSL functions and old ASN1 macros. 10511 10512 *Richard Levitte* 10513 10514 * BN_zero() only needs to set 'top' and 'neg' to zero for correct results, 10515 and this should never fail. So the return value from the use of 10516 BN_set_word() (which can fail due to needless expansion) is now deprecated; 10517 if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro. 10518 10519 *Geoff Thorpe* 10520 10521 * BN_CTX_get() should return zero-valued bignums, providing the same 10522 initialised value as BN_new(). 10523 10524 *Geoff Thorpe, suggested by Ulf Möller* 10525 10526 * Support for inhibitAnyPolicy certificate extension. 10527 10528 *Steve Henson* 10529 10530 * An audit of the BIGNUM code is underway, for which debugging code is 10531 enabled when BN_DEBUG is defined. This makes stricter enforcements on what 10532 is considered valid when processing BIGNUMs, and causes execution to 10533 assert() when a problem is discovered. If BN_DEBUG_RAND is defined, 10534 further steps are taken to deliberately pollute unused data in BIGNUM 10535 structures to try and expose faulty code further on. For now, openssl will 10536 (in its default mode of operation) continue to tolerate the inconsistent 10537 forms that it has tolerated in the past, but authors and packagers should 10538 consider trying openssl and their own applications when compiled with 10539 these debugging symbols defined. It will help highlight potential bugs in 10540 their own code, and will improve the test coverage for OpenSSL itself. At 10541 some point, these tighter rules will become openssl's default to improve 10542 maintainability, though the assert()s and other overheads will remain only 10543 in debugging configurations. See bn.h for more details. 10544 10545 *Geoff Thorpe, Nils Larsch, Ulf Möller* 10546 10547 * BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure 10548 that can only be obtained through BN_CTX_new() (which implicitly 10549 initialises it). The presence of this function only made it possible 10550 to overwrite an existing structure (and cause memory leaks). 10551 10552 *Geoff Thorpe* 10553 10554 * Because of the callback-based approach for implementing LHASH as a 10555 template type, lh_insert() adds opaque objects to hash-tables and 10556 lh_doall() or lh_doall_arg() are typically used with a destructor callback 10557 to clean up those corresponding objects before destroying the hash table 10558 (and losing the object pointers). So some over-zealous constifications in 10559 LHASH have been relaxed so that lh_insert() does not take (nor store) the 10560 objects as "const" and the `lh_doall[_arg]` callback wrappers are not 10561 prototyped to have "const" restrictions on the object pointers they are 10562 given (and so aren't required to cast them away any more). 10563 10564 *Geoff Thorpe* 10565 10566 * The tmdiff.h API was so ugly and minimal that our own timing utility 10567 (speed) prefers to use its own implementation. The two implementations 10568 haven't been consolidated as yet (volunteers?) but the tmdiff API has had 10569 its object type properly exposed (MS_TM) instead of casting to/from 10570 `char *`. This may still change yet if someone realises MS_TM and 10571 `ms_time_***` 10572 aren't necessarily the greatest nomenclatures - but this is what was used 10573 internally to the implementation so I've used that for now. 10574 10575 *Geoff Thorpe* 10576 10577 * Ensure that deprecated functions do not get compiled when 10578 OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of 10579 the self-tests were still using deprecated key-generation functions so 10580 these have been updated also. 10581 10582 *Geoff Thorpe* 10583 10584 * Reorganise PKCS#7 code to separate the digest location functionality 10585 into PKCS7_find_digest(), digest addition into PKCS7_bio_add_digest(). 10586 New function PKCS7_set_digest() to set the digest type for PKCS#7 10587 digestedData type. Add additional code to correctly generate the 10588 digestedData type and add support for this type in PKCS7 initialization 10589 functions. 10590 10591 *Steve Henson* 10592 10593 * New function PKCS7_set0_type_other() this initializes a PKCS7 10594 structure of type "other". 10595 10596 *Steve Henson* 10597 10598 * Fix prime generation loop in crypto/bn/bn_prime.pl by making 10599 sure the loop does correctly stop and breaking ("division by zero") 10600 modulus operations are not performed. The (pre-generated) prime 10601 table crypto/bn/bn_prime.h was already correct, but it could not be 10602 re-generated on some platforms because of the "division by zero" 10603 situation in the script. 10604 10605 *Ralf S. Engelschall* 10606 10607 * Update support for ECC-based TLS ciphersuites according to 10608 draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with 10609 SHA-1 now is only used for "small" curves (where the 10610 representation of a field element takes up to 24 bytes); for 10611 larger curves, the field element resulting from ECDH is directly 10612 used as premaster secret. 10613 10614 *Douglas Stebila (Sun Microsystems Laboratories)* 10615 10616 * Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2 10617 curve secp160r1 to the tests. 10618 10619 *Douglas Stebila (Sun Microsystems Laboratories)* 10620 10621 * Add the possibility to load symbols globally with DSO. 10622 10623 *Götz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte* 10624 10625 * Add the functions ERR_set_mark() and ERR_pop_to_mark() for better 10626 control of the error stack. 10627 10628 *Richard Levitte* 10629 10630 * Add support for STORE in ENGINE. 10631 10632 *Richard Levitte* 10633 10634 * Add the STORE type. The intention is to provide a common interface 10635 to certificate and key stores, be they simple file-based stores, or 10636 HSM-type store, or LDAP stores, or... 10637 NOTE: The code is currently UNTESTED and isn't really used anywhere. 10638 10639 *Richard Levitte* 10640 10641 * Add a generic structure called OPENSSL_ITEM. This can be used to 10642 pass a list of arguments to any function as well as provide a way 10643 for a function to pass data back to the caller. 10644 10645 *Richard Levitte* 10646 10647 * Add the functions BUF_strndup() and BUF_memdup(). BUF_strndup() 10648 works like BUF_strdup() but can be used to duplicate a portion of 10649 a string. The copy gets NUL-terminated. BUF_memdup() duplicates 10650 a memory area. 10651 10652 *Richard Levitte* 10653 10654 * Add the function sk_find_ex() which works like sk_find(), but will 10655 return an index to an element even if an exact match couldn't be 10656 found. The index is guaranteed to point at the element where the 10657 searched-for key would be inserted to preserve sorting order. 10658 10659 *Richard Levitte* 10660 10661 * Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but 10662 takes an extra flags argument for optional functionality. Currently, 10663 the following flags are defined: 10664 10665 OBJ_BSEARCH_VALUE_ON_NOMATCH 10666 This one gets OBJ_bsearch_ex() to return a pointer to the first 10667 element where the comparing function returns a negative or zero 10668 number. 10669 10670 OBJ_BSEARCH_FIRST_VALUE_ON_MATCH 10671 This one gets OBJ_bsearch_ex() to return a pointer to the first 10672 element where the comparing function returns zero. This is useful 10673 if there are more than one element where the comparing function 10674 returns zero. 10675 10676 *Richard Levitte* 10677 10678 * Make it possible to create self-signed certificates with 'openssl ca' 10679 in such a way that the self-signed certificate becomes part of the 10680 CA database and uses the same mechanisms for serial number generation 10681 as all other certificate signing. The new flag '-selfsign' enables 10682 this functionality. Adapt CA.sh and CA.pl.in. 10683 10684 *Richard Levitte* 10685 10686 * Add functionality to check the public key of a certificate request 10687 against a given private. This is useful to check that a certificate 10688 request can be signed by that key (self-signing). 10689 10690 *Richard Levitte* 10691 10692 * Make it possible to have multiple active certificates with the same 10693 subject in the CA index file. This is done only if the keyword 10694 'unique_subject' is set to 'no' in the main CA section (default 10695 if 'CA_default') of the configuration file. The value is saved 10696 with the database itself in a separate index attribute file, 10697 named like the index file with '.attr' appended to the name. 10698 10699 *Richard Levitte* 10700 10701 * Generate multi-valued AVAs using '+' notation in config files for 10702 req and dirName. 10703 10704 *Steve Henson* 10705 10706 * Support for nameConstraints certificate extension. 10707 10708 *Steve Henson* 10709 10710 * Support for policyConstraints certificate extension. 10711 10712 *Steve Henson* 10713 10714 * Support for policyMappings certificate extension. 10715 10716 *Steve Henson* 10717 10718 * Make sure the default DSA_METHOD implementation only uses its 10719 dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL, 10720 and change its own handlers to be NULL so as to remove unnecessary 10721 indirection. This lets alternative implementations fallback to the 10722 default implementation more easily. 10723 10724 *Geoff Thorpe* 10725 10726 * Support for directoryName in GeneralName related extensions 10727 in config files. 10728 10729 *Steve Henson* 10730 10731 * Make it possible to link applications using Makefile.shared. 10732 Make that possible even when linking against static libraries! 10733 10734 *Richard Levitte* 10735 10736 * Support for single pass processing for S/MIME signing. This now 10737 means that S/MIME signing can be done from a pipe, in addition 10738 cleartext signing (multipart/signed type) is effectively streaming 10739 and the signed data does not need to be all held in memory. 10740 10741 This is done with a new flag PKCS7_STREAM. When this flag is set 10742 PKCS7_sign() only initializes the PKCS7 structure and the actual signing 10743 is done after the data is output (and digests calculated) in 10744 SMIME_write_PKCS7(). 10745 10746 *Steve Henson* 10747 10748 * Add full support for -rpath/-R, both in shared libraries and 10749 applications, at least on the platforms where it's known how 10750 to do it. 10751 10752 *Richard Levitte* 10753 10754 * In crypto/ec/ec_mult.c, implement fast point multiplication with 10755 precomputation, based on wNAF splitting: EC_GROUP_precompute_mult() 10756 will now compute a table of multiples of the generator that 10757 makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul() 10758 faster (notably in the case of a single point multiplication, 10759 scalar * generator). 10760 10761 *Nils Larsch, Bodo Moeller* 10762 10763 * IPv6 support for certificate extensions. The various extensions 10764 which use the IP:a.b.c.d can now take IPv6 addresses using the 10765 formats of RFC1884 2.2 . IPv6 addresses are now also displayed 10766 correctly. 10767 10768 *Steve Henson* 10769 10770 * Added an ENGINE that implements RSA by performing private key 10771 exponentiations with the GMP library. The conversions to and from 10772 GMP's mpz_t format aren't optimised nor are any montgomery forms 10773 cached, and on x86 it appears OpenSSL's own performance has caught up. 10774 However there are likely to be other architectures where GMP could 10775 provide a boost. This ENGINE is not built in by default, but it can be 10776 specified at Configure time and should be accompanied by the necessary 10777 linker additions, eg; 10778 ./config -DOPENSSL_USE_GMP -lgmp 10779 10780 *Geoff Thorpe* 10781 10782 * "openssl engine" will not display ENGINE/DSO load failure errors when 10783 testing availability of engines with "-t" - the old behaviour is 10784 produced by increasing the feature's verbosity with "-tt". 10785 10786 *Geoff Thorpe* 10787 10788 * ECDSA routines: under certain error conditions uninitialized BN objects 10789 could be freed. Solution: make sure initialization is performed early 10790 enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de> 10791 via PR#459) 10792 10793 *Lutz Jaenicke* 10794 10795 * Key-generation can now be implemented in RSA_METHOD, DSA_METHOD 10796 and DH_METHOD (eg. by ENGINE implementations) to override the normal 10797 software implementations. For DSA and DH, parameter generation can 10798 also be overridden by providing the appropriate method callbacks. 10799 10800 *Geoff Thorpe* 10801 10802 * Change the "progress" mechanism used in key-generation and 10803 primality testing to functions that take a new BN_GENCB pointer in 10804 place of callback/argument pairs. The new API functions have `_ex` 10805 postfixes and the older functions are reimplemented as wrappers for 10806 the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide 10807 declarations of the old functions to help (graceful) attempts to 10808 migrate to the new functions. Also, the new key-generation API 10809 functions operate on a caller-supplied key-structure and return 10810 success/failure rather than returning a key or NULL - this is to 10811 help make "keygen" another member function of RSA_METHOD etc. 10812 10813 Example for using the new callback interface: 10814 10815 int (*my_callback)(int a, int b, BN_GENCB *cb) = ...; 10816 void *my_arg = ...; 10817 BN_GENCB my_cb; 10818 10819 BN_GENCB_set(&my_cb, my_callback, my_arg); 10820 10821 return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb); 10822 /* For the meaning of a, b in calls to my_callback(), see the 10823 * documentation of the function that calls the callback. 10824 * cb will point to my_cb; my_arg can be retrieved as cb->arg. 10825 * my_callback should return 1 if it wants BN_is_prime_ex() 10826 * to continue, or 0 to stop. 10827 */ 10828 10829 *Geoff Thorpe* 10830 10831 * Change the ZLIB compression method to be stateful, and make it 10832 available to TLS with the number defined in 10833 draft-ietf-tls-compression-04.txt. 10834 10835 *Richard Levitte* 10836 10837 * Add the ASN.1 structures and functions for CertificatePair, which 10838 is defined as follows (according to X.509_4thEditionDraftV6.pdf): 10839 10840 CertificatePair ::= SEQUENCE { 10841 forward [0] Certificate OPTIONAL, 10842 reverse [1] Certificate OPTIONAL, 10843 -- at least one of the pair shall be present -- } 10844 10845 Also implement the PEM functions to read and write certificate 10846 pairs, and defined the PEM tag as "CERTIFICATE PAIR". 10847 10848 This needed to be defined, mostly for the sake of the LDAP 10849 attribute crossCertificatePair, but may prove useful elsewhere as 10850 well. 10851 10852 *Richard Levitte* 10853 10854 * Make it possible to inhibit symlinking of shared libraries in 10855 Makefile.shared, for Cygwin's sake. 10856 10857 *Richard Levitte* 10858 10859 * Extend the BIGNUM API by creating a function 10860 void BN_set_negative(BIGNUM *a, int neg); 10861 and a macro that behave like 10862 int BN_is_negative(const BIGNUM *a); 10863 10864 to avoid the need to access 'a->neg' directly in applications. 10865 10866 *Nils Larsch* 10867 10868 * Implement fast modular reduction for pseudo-Mersenne primes 10869 used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c). 10870 EC_GROUP_new_curve_GFp() will now automatically use this 10871 if applicable. 10872 10873 *Nils Larsch <nla@trustcenter.de>* 10874 10875 * Add new lock type (CRYPTO_LOCK_BN). 10876 10877 *Bodo Moeller* 10878 10879 * Change the ENGINE framework to automatically load engines 10880 dynamically from specific directories unless they could be 10881 found to already be built in or loaded. Move all the 10882 current engines except for the cryptodev one to a new 10883 directory engines/. 10884 The engines in engines/ are built as shared libraries if 10885 the "shared" options was given to ./Configure or ./config. 10886 Otherwise, they are inserted in libcrypto.a. 10887 /usr/local/ssl/engines is the default directory for dynamic 10888 engines, but that can be overridden at configure time through 10889 the usual use of --prefix and/or --openssldir, and at run 10890 time with the environment variable OPENSSL_ENGINES. 10891 10892 *Geoff Thorpe and Richard Levitte* 10893 10894 * Add Makefile.shared, a helper makefile to build shared 10895 libraries. Adapt Makefile.org. 10896 10897 *Richard Levitte* 10898 10899 * Add version info to Win32 DLLs. 10900 10901 *Peter 'Luna' Runestig" <peter@runestig.com>* 10902 10903 * Add new 'medium level' PKCS#12 API. Certificates and keys 10904 can be added using this API to created arbitrary PKCS#12 10905 files while avoiding the low-level API. 10906 10907 New options to PKCS12_create(), key or cert can be NULL and 10908 will then be omitted from the output file. The encryption 10909 algorithm NIDs can be set to -1 for no encryption, the mac 10910 iteration count can be set to 0 to omit the mac. 10911 10912 Enhance pkcs12 utility by making the -nokeys and -nocerts 10913 options work when creating a PKCS#12 file. New option -nomac 10914 to omit the mac, NONE can be set for an encryption algorithm. 10915 New code is modified to use the enhanced PKCS12_create() 10916 instead of the low-level API. 10917 10918 *Steve Henson* 10919 10920 * Extend ASN1 encoder to support indefinite length constructed 10921 encoding. This can output sequences tags and octet strings in 10922 this form. Modify pk7_asn1.c to support indefinite length 10923 encoding. This is experimental and needs additional code to 10924 be useful, such as an ASN1 bio and some enhanced streaming 10925 PKCS#7 code. 10926 10927 Extend template encode functionality so that tagging is passed 10928 down to the template encoder. 10929 10930 *Steve Henson* 10931 10932 * Let 'openssl req' fail if an argument to '-newkey' is not 10933 recognized instead of using RSA as a default. 10934 10935 *Bodo Moeller* 10936 10937 * Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt. 10938 As these are not official, they are not included in "ALL"; 10939 the "ECCdraft" ciphersuite group alias can be used to select them. 10940 10941 *Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)* 10942 10943 * Add ECDH engine support. 10944 10945 *Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)* 10946 10947 * Add ECDH in new directory crypto/ecdh/. 10948 10949 *Douglas Stebila (Sun Microsystems Laboratories)* 10950 10951 * Let BN_rand_range() abort with an error after 100 iterations 10952 without success (which indicates a broken PRNG). 10953 10954 *Bodo Moeller* 10955 10956 * Change BN_mod_sqrt() so that it verifies that the input value 10957 is really the square of the return value. (Previously, 10958 BN_mod_sqrt would show GIGO behaviour.) 10959 10960 *Bodo Moeller* 10961 10962 * Add named elliptic curves over binary fields from X9.62, SECG, 10963 and WAP/WTLS; add OIDs that were still missing. 10964 10965 *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)* 10966 10967 * Extend the EC library for elliptic curves over binary fields 10968 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/). 10969 New EC_METHOD: 10970 10971 EC_GF2m_simple_method 10972 10973 New API functions: 10974 10975 EC_GROUP_new_curve_GF2m 10976 EC_GROUP_set_curve_GF2m 10977 EC_GROUP_get_curve_GF2m 10978 EC_POINT_set_affine_coordinates_GF2m 10979 EC_POINT_get_affine_coordinates_GF2m 10980 EC_POINT_set_compressed_coordinates_GF2m 10981 10982 Point compression for binary fields is disabled by default for 10983 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to 10984 enable it). 10985 10986 As binary polynomials are represented as BIGNUMs, various members 10987 of the EC_GROUP and EC_POINT data structures can be shared 10988 between the implementations for prime fields and binary fields; 10989 the above `..._GF2m functions` (except for EX_GROUP_new_curve_GF2m) 10990 are essentially identical to their `..._GFp` counterparts. 10991 (For simplicity, the `..._GFp` prefix has been dropped from 10992 various internal method names.) 10993 10994 An internal 'field_div' method (similar to 'field_mul' and 10995 'field_sqr') has been added; this is used only for binary fields. 10996 10997 *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)* 10998 10999 * Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult() 11000 through methods ('mul', 'precompute_mult'). 11001 11002 The generic implementations (now internally called 'ec_wNAF_mul' 11003 and 'ec_wNAF_precomputed_mult') remain the default if these 11004 methods are undefined. 11005 11006 *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)* 11007 11008 * New function EC_GROUP_get_degree, which is defined through 11009 EC_METHOD. For curves over prime fields, this returns the bit 11010 length of the modulus. 11011 11012 *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)* 11013 11014 * New functions EC_GROUP_dup, EC_POINT_dup. 11015 (These simply call ..._new and ..._copy). 11016 11017 *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)* 11018 11019 * Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c. 11020 Polynomials are represented as BIGNUMs (where the sign bit is not 11021 used) in the following functions [macros]: 11022 11023 BN_GF2m_add 11024 BN_GF2m_sub [= BN_GF2m_add] 11025 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr] 11026 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr] 11027 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr] 11028 BN_GF2m_mod_inv 11029 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr] 11030 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr] 11031 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr] 11032 BN_GF2m_cmp [= BN_ucmp] 11033 11034 (Note that only the 'mod' functions are actually for fields GF(2^m). 11035 BN_GF2m_add() is misnomer, but this is for the sake of consistency.) 11036 11037 For some functions, an the irreducible polynomial defining a 11038 field can be given as an 'unsigned int[]' with strictly 11039 decreasing elements giving the indices of those bits that are set; 11040 i.e., p[] represents the polynomial 11041 f(t) = t^p[0] + t^p[1] + ... + t^p[k] 11042 where 11043 p[0] > p[1] > ... > p[k] = 0. 11044 This applies to the following functions: 11045 11046 BN_GF2m_mod_arr 11047 BN_GF2m_mod_mul_arr 11048 BN_GF2m_mod_sqr_arr 11049 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv] 11050 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div] 11051 BN_GF2m_mod_exp_arr 11052 BN_GF2m_mod_sqrt_arr 11053 BN_GF2m_mod_solve_quad_arr 11054 BN_GF2m_poly2arr 11055 BN_GF2m_arr2poly 11056 11057 Conversion can be performed by the following functions: 11058 11059 BN_GF2m_poly2arr 11060 BN_GF2m_arr2poly 11061 11062 bntest.c has additional tests for binary polynomial arithmetic. 11063 11064 Two implementations for BN_GF2m_mod_div() are available. 11065 The default algorithm simply uses BN_GF2m_mod_inv() and 11066 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only 11067 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the 11068 copyright notice in crypto/bn/bn_gf2m.c before enabling it). 11069 11070 *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)* 11071 11072 * Add new error code 'ERR_R_DISABLED' that can be used when some 11073 functionality is disabled at compile-time. 11074 11075 *Douglas Stebila <douglas.stebila@sun.com>* 11076 11077 * Change default behaviour of 'openssl asn1parse' so that more 11078 information is visible when viewing, e.g., a certificate: 11079 11080 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump' 11081 mode the content of non-printable OCTET STRINGs is output in a 11082 style similar to INTEGERs, but with '[HEX DUMP]' prepended to 11083 avoid the appearance of a printable string. 11084 11085 *Nils Larsch <nla@trustcenter.de>* 11086 11087 * Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access 11088 functions 11089 EC_GROUP_set_asn1_flag() 11090 EC_GROUP_get_asn1_flag() 11091 EC_GROUP_set_point_conversion_form() 11092 EC_GROUP_get_point_conversion_form() 11093 These control ASN1 encoding details: 11094 - Curves (i.e., groups) are encoded explicitly unless asn1_flag 11095 has been set to OPENSSL_EC_NAMED_CURVE. 11096 - Points are encoded in uncompressed form by default; options for 11097 asn1_for are as for point2oct, namely 11098 POINT_CONVERSION_COMPRESSED 11099 POINT_CONVERSION_UNCOMPRESSED 11100 POINT_CONVERSION_HYBRID 11101 11102 Also add 'seed' and 'seed_len' members to EC_GROUP with access 11103 functions 11104 EC_GROUP_set_seed() 11105 EC_GROUP_get0_seed() 11106 EC_GROUP_get_seed_len() 11107 This is used only for ASN1 purposes (so far). 11108 11109 *Nils Larsch <nla@trustcenter.de>* 11110 11111 * Add 'field_type' member to EC_METHOD, which holds the NID 11112 of the appropriate field type OID. The new function 11113 EC_METHOD_get_field_type() returns this value. 11114 11115 *Nils Larsch <nla@trustcenter.de>* 11116 11117 * Add functions 11118 EC_POINT_point2bn() 11119 EC_POINT_bn2point() 11120 EC_POINT_point2hex() 11121 EC_POINT_hex2point() 11122 providing useful interfaces to EC_POINT_point2oct() and 11123 EC_POINT_oct2point(). 11124 11125 *Nils Larsch <nla@trustcenter.de>* 11126 11127 * Change internals of the EC library so that the functions 11128 EC_GROUP_set_generator() 11129 EC_GROUP_get_generator() 11130 EC_GROUP_get_order() 11131 EC_GROUP_get_cofactor() 11132 are implemented directly in crypto/ec/ec_lib.c and not dispatched 11133 to methods, which would lead to unnecessary code duplication when 11134 adding different types of curves. 11135 11136 *Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller* 11137 11138 * Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM 11139 arithmetic, and such that modified wNAFs are generated 11140 (which avoid length expansion in many cases). 11141 11142 *Bodo Moeller* 11143 11144 * Add a function EC_GROUP_check_discriminant() (defined via 11145 EC_METHOD) that verifies that the curve discriminant is non-zero. 11146 11147 Add a function EC_GROUP_check() that makes some sanity tests 11148 on a EC_GROUP, its generator and order. This includes 11149 EC_GROUP_check_discriminant(). 11150 11151 *Nils Larsch <nla@trustcenter.de>* 11152 11153 * Add ECDSA in new directory crypto/ecdsa/. 11154 11155 Add applications 'openssl ecparam' and 'openssl ecdsa' 11156 (these are based on 'openssl dsaparam' and 'openssl dsa'). 11157 11158 ECDSA support is also included in various other files across the 11159 library. Most notably, 11160 - 'openssl req' now has a '-newkey ecdsa:file' option; 11161 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA; 11162 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and 11163 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make 11164 them suitable for ECDSA where domain parameters must be 11165 extracted before the specific public key; 11166 - ECDSA engine support has been added. 11167 11168 *Nils Larsch <nla@trustcenter.de>* 11169 11170 * Include some named elliptic curves, and add OIDs from X9.62, 11171 SECG, and WAP/WTLS. Each curve can be obtained from the new 11172 function 11173 EC_GROUP_new_by_curve_name(), 11174 and the list of available named curves can be obtained with 11175 EC_get_builtin_curves(). 11176 Also add a 'curve_name' member to EC_GROUP objects, which can be 11177 accessed via 11178 EC_GROUP_set_curve_name() 11179 EC_GROUP_get_curve_name() 11180 11181 *Nils Larsch <larsch@trustcenter.de, Bodo Moeller* 11182 11183 * Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 11184 was actually never needed) and in BN_mul(). The removal in BN_mul() 11185 required a small change in bn_mul_part_recursive() and the addition 11186 of the functions bn_cmp_part_words(), bn_sub_part_words() and 11187 bn_add_part_words(), which do the same thing as bn_cmp_words(), 11188 bn_sub_words() and bn_add_words() except they take arrays with 11189 differing sizes. 11190 11191 *Richard Levitte* 11192 11193### Changes between 0.9.7l and 0.9.7m [23 Feb 2007] 11194 11195 * Cleanse PEM buffers before freeing them since they may contain 11196 sensitive data. 11197 11198 *Benjamin Bennett <ben@psc.edu>* 11199 11200 * Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 11201 a ciphersuite string such as "DEFAULT:RSA" cannot enable 11202 authentication-only ciphersuites. 11203 11204 *Bodo Moeller* 11205 11206 * Since AES128 and AES256 share a single mask bit in the logic of 11207 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 11208 kludge to work properly if AES128 is available and AES256 isn't. 11209 11210 *Victor Duchovni* 11211 11212 * Expand security boundary to match 1.1.1 module. 11213 11214 *Steve Henson* 11215 11216 * Remove redundant features: hash file source, editing of test vectors 11217 modify fipsld to use external fips_premain.c signature. 11218 11219 *Steve Henson* 11220 11221 * New perl script mkfipsscr.pl to create shell scripts or batch files to 11222 run algorithm test programs. 11223 11224 *Steve Henson* 11225 11226 * Make algorithm test programs more tolerant of whitespace. 11227 11228 *Steve Henson* 11229 11230 * Have SSL/TLS server implementation tolerate "mismatched" record 11231 protocol version while receiving ClientHello even if the 11232 ClientHello is fragmented. (The server can't insist on the 11233 particular protocol version it has chosen before the ServerHello 11234 message has informed the client about his choice.) 11235 11236 *Bodo Moeller* 11237 11238 * Load error codes if they are not already present instead of using a 11239 static variable. This allows them to be cleanly unloaded and reloaded. 11240 11241 *Steve Henson* 11242 11243### Changes between 0.9.7k and 0.9.7l [28 Sep 2006] 11244 11245 * Introduce limits to prevent malicious keys being able to 11246 cause a denial of service. ([CVE-2006-2940]) 11247 11248 *Steve Henson, Bodo Moeller* 11249 11250 * Fix ASN.1 parsing of certain invalid structures that can result 11251 in a denial of service. ([CVE-2006-2937]) [Steve Henson] 11252 11253 * Fix buffer overflow in SSL_get_shared_ciphers() function. 11254 ([CVE-2006-3738]) [Tavis Ormandy and Will Drewry, Google Security Team] 11255 11256 * Fix SSL client code which could crash if connecting to a 11257 malicious SSLv2 server. ([CVE-2006-4343]) 11258 11259 *Tavis Ormandy and Will Drewry, Google Security Team* 11260 11261 * Change ciphersuite string processing so that an explicit 11262 ciphersuite selects this one ciphersuite (so that "AES256-SHA" 11263 will no longer include "AES128-SHA"), and any other similar 11264 ciphersuite (same bitmap) from *other* protocol versions (so that 11265 "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the 11266 SSL 3.0/TLS 1.0 ciphersuite). This is a backport combining 11267 changes from 0.9.8b and 0.9.8d. 11268 11269 *Bodo Moeller* 11270 11271### Changes between 0.9.7j and 0.9.7k [05 Sep 2006] 11272 11273 * Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 11274 ([CVE-2006-4339]) [Ben Laurie and Google Security Team] 11275 11276 * Change the Unix randomness entropy gathering to use poll() when 11277 possible instead of select(), since the latter has some 11278 undesirable limitations. 11279 11280 *Darryl Miles via Richard Levitte and Bodo Moeller* 11281 11282 * Disable rogue ciphersuites: 11283 11284 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 11285 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 11286 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 11287 11288 The latter two were purportedly from 11289 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 11290 appear there. 11291 11292 Also deactivate the remaining ciphersuites from 11293 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 11294 unofficial, and the ID has long expired. 11295 11296 *Bodo Moeller* 11297 11298 * Fix RSA blinding Heisenbug (problems sometimes occurred on 11299 dual-core machines) and other potential thread-safety issues. 11300 11301 *Bodo Moeller* 11302 11303### Changes between 0.9.7i and 0.9.7j [04 May 2006] 11304 11305 * Adapt fipsld and the build system to link against the validated FIPS 11306 module in FIPS mode. 11307 11308 *Steve Henson* 11309 11310 * Fixes for VC++ 2005 build under Windows. 11311 11312 *Steve Henson* 11313 11314 * Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make 11315 from a Windows bash shell such as MSYS. It is autodetected from the 11316 "config" script when run from a VC++ environment. Modify standard VC++ 11317 build to use fipscanister.o from the GNU make build. 11318 11319 *Steve Henson* 11320 11321### Changes between 0.9.7h and 0.9.7i [14 Oct 2005] 11322 11323 * Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS. 11324 The value now differs depending on if you build for FIPS or not. 11325 BEWARE! A program linked with a shared FIPSed libcrypto can't be 11326 safely run with a non-FIPSed libcrypto, as it may crash because of 11327 the difference induced by this change. 11328 11329 *Andy Polyakov* 11330 11331### Changes between 0.9.7g and 0.9.7h [11 Oct 2005] 11332 11333 * Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 11334 (part of SSL_OP_ALL). This option used to disable the 11335 countermeasure against man-in-the-middle protocol-version 11336 rollback in the SSL 2.0 server implementation, which is a bad 11337 idea. ([CVE-2005-2969]) 11338 11339 *Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 11340 for Information Security, National Institute of Advanced Industrial 11341 Science and Technology [AIST, Japan)]* 11342 11343 * Minimal support for X9.31 signatures and PSS padding modes. This is 11344 mainly for FIPS compliance and not fully integrated at this stage. 11345 11346 *Steve Henson* 11347 11348 * For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform 11349 the exponentiation using a fixed-length exponent. (Otherwise, 11350 the information leaked through timing could expose the secret key 11351 after many signatures; cf. Bleichenbacher's attack on DSA with 11352 biased k.) 11353 11354 *Bodo Moeller* 11355 11356 * Make a new fixed-window mod_exp implementation the default for 11357 RSA, DSA, and DH private-key operations so that the sequence of 11358 squares and multiplies and the memory access pattern are 11359 independent of the particular secret key. This will mitigate 11360 cache-timing and potential related attacks. 11361 11362 BN_mod_exp_mont_consttime() is the new exponentiation implementation, 11363 and this is automatically used by BN_mod_exp_mont() if the new flag 11364 BN_FLG_EXP_CONSTTIME is set for the exponent. RSA, DSA, and DH 11365 will use this BN flag for private exponents unless the flag 11366 RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or 11367 DH_FLAG_NO_EXP_CONSTTIME, respectively, is set. 11368 11369 *Matthew D Wood (Intel Corp), with some changes by Bodo Moeller* 11370 11371 * Change the client implementation for SSLv23_method() and 11372 SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0 11373 Client Hello message format if the SSL_OP_NO_SSLv2 option is set. 11374 (Previously, the SSL 2.0 backwards compatible Client Hello 11375 message format would be used even with SSL_OP_NO_SSLv2.) 11376 11377 *Bodo Moeller* 11378 11379 * Add support for smime-type MIME parameter in S/MIME messages which some 11380 clients need. 11381 11382 *Steve Henson* 11383 11384 * New function BN_MONT_CTX_set_locked() to set montgomery parameters in 11385 a threadsafe manner. Modify rsa code to use new function and add calls 11386 to dsa and dh code (which had race conditions before). 11387 11388 *Steve Henson* 11389 11390 * Include the fixed error library code in the C error file definitions 11391 instead of fixing them up at runtime. This keeps the error code 11392 structures constant. 11393 11394 *Steve Henson* 11395 11396### Changes between 0.9.7f and 0.9.7g [11 Apr 2005] 11397 11398[NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after 11399OpenSSL 0.9.8.] 11400 11401 * Fixes for newer kerberos headers. NB: the casts are needed because 11402 the 'length' field is signed on one version and unsigned on another 11403 with no (?) obvious way to tell the difference, without these VC++ 11404 complains. Also the "definition" of FAR (blank) is no longer included 11405 nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up 11406 some needed definitions. 11407 11408 *Steve Henson* 11409 11410 * Undo Cygwin change. 11411 11412 *Ulf Möller* 11413 11414 * Added support for proxy certificates according to RFC 3820. 11415 Because they may be a security thread to unaware applications, 11416 they must be explicitly allowed in run-time. See 11417 docs/HOWTO/proxy_certificates.txt for further information. 11418 11419 *Richard Levitte* 11420 11421### Changes between 0.9.7e and 0.9.7f [22 Mar 2005] 11422 11423 * Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating 11424 server and client random values. Previously 11425 (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in 11426 less random data when sizeof(time_t) > 4 (some 64 bit platforms). 11427 11428 This change has negligible security impact because: 11429 11430 1. Server and client random values still have 24 bytes of pseudo random 11431 data. 11432 11433 2. Server and client random values are sent in the clear in the initial 11434 handshake. 11435 11436 3. The master secret is derived using the premaster secret (48 bytes in 11437 size for static RSA ciphersuites) as well as client server and random 11438 values. 11439 11440 The OpenSSL team would like to thank the UK NISCC for bringing this issue 11441 to our attention. 11442 11443 *Stephen Henson, reported by UK NISCC* 11444 11445 * Use Windows randomness collection on Cygwin. 11446 11447 *Ulf Möller* 11448 11449 * Fix hang in EGD/PRNGD query when communication socket is closed 11450 prematurely by EGD/PRNGD. 11451 11452 *Darren Tucker <dtucker@zip.com.au> via Lutz Jänicke, resolves #1014* 11453 11454 * Prompt for pass phrases when appropriate for PKCS12 input format. 11455 11456 *Steve Henson* 11457 11458 * Back-port of selected performance improvements from development 11459 branch, as well as improved support for PowerPC platforms. 11460 11461 *Andy Polyakov* 11462 11463 * Add lots of checks for memory allocation failure, error codes to indicate 11464 failure and freeing up memory if a failure occurs. 11465 11466 *Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson* 11467 11468 * Add new -passin argument to dgst. 11469 11470 *Steve Henson* 11471 11472 * Perform some character comparisons of different types in X509_NAME_cmp: 11473 this is needed for some certificates that re-encode DNs into UTF8Strings 11474 (in violation of RFC3280) and can't or won't issue name rollover 11475 certificates. 11476 11477 *Steve Henson* 11478 11479 * Make an explicit check during certificate validation to see that 11480 the CA setting in each certificate on the chain is correct. As a 11481 side effect always do the following basic checks on extensions, 11482 not just when there's an associated purpose to the check: 11483 11484 - if there is an unhandled critical extension (unless the user 11485 has chosen to ignore this fault) 11486 - if the path length has been exceeded (if one is set at all) 11487 - that certain extensions fit the associated purpose (if one has 11488 been given) 11489 11490 *Richard Levitte* 11491 11492### Changes between 0.9.7d and 0.9.7e [25 Oct 2004] 11493 11494 * Avoid a race condition when CRLs are checked in a multi threaded 11495 environment. This would happen due to the reordering of the revoked 11496 entries during signature checking and serial number lookup. Now the 11497 encoding is cached and the serial number sort performed under a lock. 11498 Add new STACK function sk_is_sorted(). 11499 11500 *Steve Henson* 11501 11502 * Add Delta CRL to the extension code. 11503 11504 *Steve Henson* 11505 11506 * Various fixes to s3_pkt.c so alerts are sent properly. 11507 11508 *David Holmes <d.holmes@f5.com>* 11509 11510 * Reduce the chances of duplicate issuer name and serial numbers (in 11511 violation of RFC3280) using the OpenSSL certificate creation utilities. 11512 This is done by creating a random 64 bit value for the initial serial 11513 number when a serial number file is created or when a self signed 11514 certificate is created using 'openssl req -x509'. The initial serial 11515 number file is created using 'openssl x509 -next_serial' in CA.pl 11516 rather than being initialized to 1. 11517 11518 *Steve Henson* 11519 11520### Changes between 0.9.7c and 0.9.7d [17 Mar 2004] 11521 11522 * Fix null-pointer assignment in do_change_cipher_spec() revealed 11523 by using the Codenomicon TLS Test Tool ([CVE-2004-0079]) 11524 11525 *Joe Orton, Steve Henson* 11526 11527 * Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites 11528 ([CVE-2004-0112]) 11529 11530 *Joe Orton, Steve Henson* 11531 11532 * Make it possible to have multiple active certificates with the same 11533 subject in the CA index file. This is done only if the keyword 11534 'unique_subject' is set to 'no' in the main CA section (default 11535 if 'CA_default') of the configuration file. The value is saved 11536 with the database itself in a separate index attribute file, 11537 named like the index file with '.attr' appended to the name. 11538 11539 *Richard Levitte* 11540 11541 * X509 verify fixes. Disable broken certificate workarounds when 11542 X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if 11543 keyUsage extension present. Don't accept CRLs with unhandled critical 11544 extensions: since verify currently doesn't process CRL extensions this 11545 rejects a CRL with *any* critical extensions. Add new verify error codes 11546 for these cases. 11547 11548 *Steve Henson* 11549 11550 * When creating an OCSP nonce use an OCTET STRING inside the extnValue. 11551 A clarification of RFC2560 will require the use of OCTET STRINGs and 11552 some implementations cannot handle the current raw format. Since OpenSSL 11553 copies and compares OCSP nonces as opaque blobs without any attempt at 11554 parsing them this should not create any compatibility issues. 11555 11556 *Steve Henson* 11557 11558 * New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when 11559 calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without 11560 this HMAC (and other) operations are several times slower than OpenSSL 11561 < 0.9.7. 11562 11563 *Steve Henson* 11564 11565 * Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex(). 11566 11567 *Peter Sylvester <Peter.Sylvester@EdelWeb.fr>* 11568 11569 * Use the correct content when signing type "other". 11570 11571 *Steve Henson* 11572 11573### Changes between 0.9.7b and 0.9.7c [30 Sep 2003] 11574 11575 * Fix various bugs revealed by running the NISCC test suite: 11576 11577 Stop out of bounds reads in the ASN1 code when presented with 11578 invalid tags (CVE-2003-0543 and CVE-2003-0544). 11579 11580 Free up ASN1_TYPE correctly if ANY type is invalid ([CVE-2003-0545]). 11581 11582 If verify callback ignores invalid public key errors don't try to check 11583 certificate signature with the NULL public key. 11584 11585 *Steve Henson* 11586 11587 * New -ignore_err option in ocsp application to stop the server 11588 exiting on the first error in a request. 11589 11590 *Steve Henson* 11591 11592 * In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 11593 if the server requested one: as stated in TLS 1.0 and SSL 3.0 11594 specifications. 11595 11596 *Steve Henson* 11597 11598 * In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 11599 extra data after the compression methods not only for TLS 1.0 11600 but also for SSL 3.0 (as required by the specification). 11601 11602 *Bodo Moeller; problem pointed out by Matthias Loepfe* 11603 11604 * Change X509_certificate_type() to mark the key as exported/exportable 11605 when it's 512 *bits* long, not 512 bytes. 11606 11607 *Richard Levitte* 11608 11609 * Change AES_cbc_encrypt() so it outputs exact multiple of 11610 blocks during encryption. 11611 11612 *Richard Levitte* 11613 11614 * Various fixes to base64 BIO and non blocking I/O. On write 11615 flushes were not handled properly if the BIO retried. On read 11616 data was not being buffered properly and had various logic bugs. 11617 This also affects blocking I/O when the data being decoded is a 11618 certain size. 11619 11620 *Steve Henson* 11621 11622 * Various S/MIME bugfixes and compatibility changes: 11623 output correct application/pkcs7 MIME type if 11624 PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures. 11625 Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening 11626 of files as .eml work). Correctly handle very long lines in MIME 11627 parser. 11628 11629 *Steve Henson* 11630 11631### Changes between 0.9.7a and 0.9.7b [10 Apr 2003] 11632 11633 * Countermeasure against the Klima-Pokorny-Rosa extension of 11634 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 11635 a protocol version number mismatch like a decryption error 11636 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 11637 11638 *Bodo Moeller* 11639 11640 * Turn on RSA blinding by default in the default implementation 11641 to avoid a timing attack. Applications that don't want it can call 11642 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 11643 They would be ill-advised to do so in most cases. 11644 11645 *Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller* 11646 11647 * Change RSA blinding code so that it works when the PRNG is not 11648 seeded (in this case, the secret RSA exponent is abused as 11649 an unpredictable seed -- if it is not unpredictable, there 11650 is no point in blinding anyway). Make RSA blinding thread-safe 11651 by remembering the creator's thread ID in rsa->blinding and 11652 having all other threads use local one-time blinding factors 11653 (this requires more computation than sharing rsa->blinding, but 11654 avoids excessive locking; and if an RSA object is not shared 11655 between threads, blinding will still be very fast). 11656 11657 *Bodo Moeller* 11658 11659 * Fixed a typo bug that would cause ENGINE_set_default() to set an 11660 ENGINE as defaults for all supported algorithms irrespective of 11661 the 'flags' parameter. 'flags' is now honoured, so applications 11662 should make sure they are passing it correctly. 11663 11664 *Geoff Thorpe* 11665 11666 * Target "mingw" now allows native Windows code to be generated in 11667 the Cygwin environment as well as with the MinGW compiler. 11668 11669 *Ulf Moeller* 11670 11671### Changes between 0.9.7 and 0.9.7a [19 Feb 2003] 11672 11673 * In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 11674 via timing by performing a MAC computation even if incorrect 11675 block cipher padding has been found. This is a countermeasure 11676 against active attacks where the attacker has to distinguish 11677 between bad padding and a MAC verification error. ([CVE-2003-0078]) 11678 11679 *Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 11680 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 11681 Martin Vuagnoux (EPFL, Ilion)* 11682 11683 * Make the no-err option work as intended. The intention with no-err 11684 is not to have the whole error stack handling routines removed from 11685 libcrypto, it's only intended to remove all the function name and 11686 reason texts, thereby removing some of the footprint that may not 11687 be interesting if those errors aren't displayed anyway. 11688 11689 NOTE: it's still possible for any application or module to have its 11690 own set of error texts inserted. The routines are there, just not 11691 used by default when no-err is given. 11692 11693 *Richard Levitte* 11694 11695 * Add support for FreeBSD on IA64. 11696 11697 *dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454* 11698 11699 * Adjust DES_cbc_cksum() so it returns the same value as the MIT 11700 Kerberos function mit_des_cbc_cksum(). Before this change, 11701 the value returned by DES_cbc_cksum() was like the one from 11702 mit_des_cbc_cksum(), except the bytes were swapped. 11703 11704 *Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte* 11705 11706 * Allow an application to disable the automatic SSL chain building. 11707 Before this a rather primitive chain build was always performed in 11708 ssl3_output_cert_chain(): an application had no way to send the 11709 correct chain if the automatic operation produced an incorrect result. 11710 11711 Now the chain builder is disabled if either: 11712 11713 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert(). 11714 11715 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set. 11716 11717 The reasoning behind this is that an application would not want the 11718 auto chain building to take place if extra chain certificates are 11719 present and it might also want a means of sending no additional 11720 certificates (for example the chain has two certificates and the 11721 root is omitted). 11722 11723 *Steve Henson* 11724 11725 * Add the possibility to build without the ENGINE framework. 11726 11727 *Steven Reddie <smr@essemer.com.au> via Richard Levitte* 11728 11729 * Under Win32 gmtime() can return NULL: check return value in 11730 OPENSSL_gmtime(). Add error code for case where gmtime() fails. 11731 11732 *Steve Henson* 11733 11734 * DSA routines: under certain error conditions uninitialized BN objects 11735 could be freed. Solution: make sure initialization is performed early 11736 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>, 11737 Nils Larsch <nla@trustcenter.de> via PR#459) 11738 11739 *Lutz Jaenicke* 11740 11741 * Another fix for SSLv2 session ID handling: the session ID was incorrectly 11742 checked on reconnect on the client side, therefore session resumption 11743 could still fail with a "ssl session id is different" error. This 11744 behaviour is masked when SSL_OP_ALL is used due to 11745 SSL_OP_MICROSOFT_SESS_ID_BUG being set. 11746 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 11747 followup to PR #377. 11748 11749 *Lutz Jaenicke* 11750 11751 * IA-32 assembler support enhancements: unified ELF targets, support 11752 for SCO/Caldera platforms, fix for Cygwin shared build. 11753 11754 *Andy Polyakov* 11755 11756 * Add support for FreeBSD on sparc64. As a consequence, support for 11757 FreeBSD on non-x86 processors is separate from x86 processors on 11758 the config script, much like the NetBSD support. 11759 11760 *Richard Levitte & Kris Kennaway <kris@obsecurity.org>* 11761 11762### Changes between 0.9.6h and 0.9.7 [31 Dec 2002] 11763 11764[NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after 11765OpenSSL 0.9.7.] 11766 11767 * Fix session ID handling in SSLv2 client code: the SERVER FINISHED 11768 code (06) was taken as the first octet of the session ID and the last 11769 octet was ignored consequently. As a result SSLv2 client side session 11770 caching could not have worked due to the session ID mismatch between 11771 client and server. 11772 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 11773 PR #377. 11774 11775 *Lutz Jaenicke* 11776 11777 * Change the declaration of needed Kerberos libraries to use EX_LIBS 11778 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is 11779 removed entirely. 11780 11781 *Richard Levitte* 11782 11783 * The hw_ncipher.c engine requires dynamic locks. Unfortunately, it 11784 seems that in spite of existing for more than a year, many application 11785 author have done nothing to provide the necessary callbacks, which 11786 means that this particular engine will not work properly anywhere. 11787 This is a very unfortunate situation which forces us, in the name 11788 of usability, to give the hw_ncipher.c a static lock, which is part 11789 of libcrypto. 11790 NOTE: This is for the 0.9.7 series ONLY. This hack will never 11791 appear in 0.9.8 or later. We EXPECT application authors to have 11792 dealt properly with this when 0.9.8 is released (unless we actually 11793 make such changes in the libcrypto locking code that changes will 11794 have to be made anyway). 11795 11796 *Richard Levitte* 11797 11798 * In asn1_d2i_read_bio() repeatedly call BIO_read() until all content 11799 octets have been read, EOF or an error occurs. Without this change 11800 some truncated ASN1 structures will not produce an error. 11801 11802 *Steve Henson* 11803 11804 * Disable Heimdal support, since it hasn't been fully implemented. 11805 Still give the possibility to force the use of Heimdal, but with 11806 warnings and a request that patches get sent to openssl-dev. 11807 11808 *Richard Levitte* 11809 11810 * Add the VC-CE target, introduce the WINCE sysname, and add 11811 INSTALL.WCE and appropriate conditionals to make it build. 11812 11813 *Steven Reddie <smr@essemer.com.au> via Richard Levitte* 11814 11815 * Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and 11816 cygssl-x.y.z.dll, where x, y and z are the major, minor and 11817 edit numbers of the version. 11818 11819 *Corinna Vinschen <vinschen@redhat.com> and Richard Levitte* 11820 11821 * Introduce safe string copy and catenation functions 11822 (BUF_strlcpy() and BUF_strlcat()). 11823 11824 *Ben Laurie (CHATS) and Richard Levitte* 11825 11826 * Avoid using fixed-size buffers for one-line DNs. 11827 11828 *Ben Laurie (CHATS)* 11829 11830 * Add BUF_MEM_grow_clean() to avoid information leakage when 11831 resizing buffers containing secrets, and use where appropriate. 11832 11833 *Ben Laurie (CHATS)* 11834 11835 * Avoid using fixed size buffers for configuration file location. 11836 11837 *Ben Laurie (CHATS)* 11838 11839 * Avoid filename truncation for various CA files. 11840 11841 *Ben Laurie (CHATS)* 11842 11843 * Use sizeof in preference to magic numbers. 11844 11845 *Ben Laurie (CHATS)* 11846 11847 * Avoid filename truncation in cert requests. 11848 11849 *Ben Laurie (CHATS)* 11850 11851 * Add assertions to check for (supposedly impossible) buffer 11852 overflows. 11853 11854 *Ben Laurie (CHATS)* 11855 11856 * Don't cache truncated DNS entries in the local cache (this could 11857 potentially lead to a spoofing attack). 11858 11859 *Ben Laurie (CHATS)* 11860 11861 * Fix various buffers to be large enough for hex/decimal 11862 representations in a platform independent manner. 11863 11864 *Ben Laurie (CHATS)* 11865 11866 * Add CRYPTO_realloc_clean() to avoid information leakage when 11867 resizing buffers containing secrets, and use where appropriate. 11868 11869 *Ben Laurie (CHATS)* 11870 11871 * Add BIO_indent() to avoid much slightly worrying code to do 11872 indents. 11873 11874 *Ben Laurie (CHATS)* 11875 11876 * Convert sprintf()/BIO_puts() to BIO_printf(). 11877 11878 *Ben Laurie (CHATS)* 11879 11880 * buffer_gets() could terminate with the buffer only half 11881 full. Fixed. 11882 11883 *Ben Laurie (CHATS)* 11884 11885 * Add assertions to prevent user-supplied crypto functions from 11886 overflowing internal buffers by having large block sizes, etc. 11887 11888 *Ben Laurie (CHATS)* 11889 11890 * New OPENSSL_assert() macro (similar to assert(), but enabled 11891 unconditionally). 11892 11893 *Ben Laurie (CHATS)* 11894 11895 * Eliminate unused copy of key in RC4. 11896 11897 *Ben Laurie (CHATS)* 11898 11899 * Eliminate unused and incorrectly sized buffers for IV in pem.h. 11900 11901 *Ben Laurie (CHATS)* 11902 11903 * Fix off-by-one error in EGD path. 11904 11905 *Ben Laurie (CHATS)* 11906 11907 * If RANDFILE path is too long, ignore instead of truncating. 11908 11909 *Ben Laurie (CHATS)* 11910 11911 * Eliminate unused and incorrectly sized X.509 structure 11912 CBCParameter. 11913 11914 *Ben Laurie (CHATS)* 11915 11916 * Eliminate unused and dangerous function knumber(). 11917 11918 *Ben Laurie (CHATS)* 11919 11920 * Eliminate unused and dangerous structure, KSSL_ERR. 11921 11922 *Ben Laurie (CHATS)* 11923 11924 * Protect against overlong session ID context length in an encoded 11925 session object. Since these are local, this does not appear to be 11926 exploitable. 11927 11928 *Ben Laurie (CHATS)* 11929 11930 * Change from security patch (see 0.9.6e below) that did not affect 11931 the 0.9.6 release series: 11932 11933 Remote buffer overflow in SSL3 protocol - an attacker could 11934 supply an oversized master key in Kerberos-enabled versions. 11935 ([CVE-2002-0657]) 11936 11937 *Ben Laurie (CHATS)* 11938 11939 * Change the SSL kerb5 codes to match RFC 2712. 11940 11941 *Richard Levitte* 11942 11943 * Make -nameopt work fully for req and add -reqopt switch. 11944 11945 *Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson* 11946 11947 * The "block size" for block ciphers in CFB and OFB mode should be 1. 11948 11949 *Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>* 11950 11951 * Make sure tests can be performed even if the corresponding algorithms 11952 have been removed entirely. This was also the last step to make 11953 OpenSSL compilable with DJGPP under all reasonable conditions. 11954 11955 *Richard Levitte, Doug Kaufman <dkaufman@rahul.net>* 11956 11957 * Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT 11958 to allow version independent disabling of normally unselected ciphers, 11959 which may be activated as a side-effect of selecting a single cipher. 11960 11961 (E.g., cipher list string "RSA" enables ciphersuites that are left 11962 out of "ALL" because they do not provide symmetric encryption. 11963 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.) 11964 11965 *Lutz Jaenicke, Bodo Moeller* 11966 11967 * Add appropriate support for separate platform-dependent build 11968 directories. The recommended way to make a platform-dependent 11969 build directory is the following (tested on Linux), maybe with 11970 some local tweaks: 11971 11972 # Place yourself outside of the OpenSSL source tree. In 11973 # this example, the environment variable OPENSSL_SOURCE 11974 # is assumed to contain the absolute OpenSSL source directory. 11975 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`" 11976 cd objtree/"`uname -s`-`uname -r`-`uname -m`" 11977 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do 11978 mkdir -p `dirname $F` 11979 ln -s $OPENSSL_SOURCE/$F $F 11980 done 11981 11982 To be absolutely sure not to disturb the source tree, a "make clean" 11983 is a good thing. If it isn't successful, don't worry about it, 11984 it probably means the source directory is very clean. 11985 11986 *Richard Levitte* 11987 11988 * Make sure any ENGINE control commands make local copies of string 11989 pointers passed to them whenever necessary. Otherwise it is possible 11990 the caller may have overwritten (or deallocated) the original string 11991 data when a later ENGINE operation tries to use the stored values. 11992 11993 *Götz Babin-Ebell <babinebell@trustcenter.de>* 11994 11995 * Improve diagnostics in file reading and command-line digests. 11996 11997 *Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>* 11998 11999 * Add AES modes CFB and OFB to the object database. Correct an 12000 error in AES-CFB decryption. 12001 12002 *Richard Levitte* 12003 12004 * Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this 12005 allows existing EVP_CIPHER_CTX structures to be reused after 12006 calling `EVP_*Final()`. This behaviour is used by encryption 12007 BIOs and some applications. This has the side effect that 12008 applications must explicitly clean up cipher contexts with 12009 EVP_CIPHER_CTX_cleanup() or they will leak memory. 12010 12011 *Steve Henson* 12012 12013 * Check the values of dna and dnb in bn_mul_recursive before calling 12014 bn_mul_comba (a non zero value means the a or b arrays do not contain 12015 n2 elements) and fallback to bn_mul_normal if either is not zero. 12016 12017 *Steve Henson* 12018 12019 * Fix escaping of non-ASCII characters when using the -subj option 12020 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>) 12021 12022 *Lutz Jaenicke* 12023 12024 * Make object definitions compliant to LDAP (RFC2256): SN is the short 12025 form for "surname", serialNumber has no short form. 12026 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798; 12027 therefore remove "mail" short name for "internet 7". 12028 The OID for unique identifiers in X509 certificates is 12029 x500UniqueIdentifier, not uniqueIdentifier. 12030 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>) 12031 12032 *Lutz Jaenicke* 12033 12034 * Add an "init" command to the ENGINE config module and auto initialize 12035 ENGINEs. Without any "init" command the ENGINE will be initialized 12036 after all ctrl commands have been executed on it. If init=1 the 12037 ENGINE is initialized at that point (ctrls before that point are run 12038 on the uninitialized ENGINE and after on the initialized one). If 12039 init=0 then the ENGINE will not be initialized at all. 12040 12041 *Steve Henson* 12042 12043 * Fix the 'app_verify_callback' interface so that the user-defined 12044 argument is actually passed to the callback: In the 12045 SSL_CTX_set_cert_verify_callback() prototype, the callback 12046 declaration has been changed from 12047 int (*cb)() 12048 into 12049 int (*cb)(X509_STORE_CTX *,void *); 12050 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call 12051 i=s->ctx->app_verify_callback(&ctx) 12052 has been changed into 12053 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg). 12054 12055 To update applications using SSL_CTX_set_cert_verify_callback(), 12056 a dummy argument can be added to their callback functions. 12057 12058 *D. K. Smetters <smetters@parc.xerox.com>* 12059 12060 * Added the '4758cca' ENGINE to support IBM 4758 cards. 12061 12062 *Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe* 12063 12064 * Add and OPENSSL_LOAD_CONF define which will cause 12065 OpenSSL_add_all_algorithms() to load the openssl.cnf config file. 12066 This allows older applications to transparently support certain 12067 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading. 12068 Two new functions OPENSSL_add_all_algorithms_noconf() which will never 12069 load the config file and OPENSSL_add_all_algorithms_conf() which will 12070 always load it have also been added. 12071 12072 *Steve Henson* 12073 12074 * Add the OFB, CFB and CTR (all with 128 bit feedback) to AES. 12075 Adjust NIDs and EVP layer. 12076 12077 *Stephen Sprunk <stephen@sprunk.org> and Richard Levitte* 12078 12079 * Config modules support in openssl utility. 12080 12081 Most commands now load modules from the config file, 12082 though in a few (such as version) this isn't done 12083 because it couldn't be used for anything. 12084 12085 In the case of ca and req the config file used is 12086 the same as the utility itself: that is the -config 12087 command line option can be used to specify an 12088 alternative file. 12089 12090 *Steve Henson* 12091 12092 * Move default behaviour from OPENSSL_config(). If appname is NULL 12093 use "openssl_conf" if filename is NULL use default openssl config file. 12094 12095 *Steve Henson* 12096 12097 * Add an argument to OPENSSL_config() to allow the use of an alternative 12098 config section name. Add a new flag to tolerate a missing config file 12099 and move code to CONF_modules_load_file(). 12100 12101 *Steve Henson* 12102 12103 * Support for crypto accelerator cards from Accelerated Encryption 12104 Processing, www.aep.ie. (Use engine 'aep') 12105 The support was copied from 0.9.6c [engine] and adapted/corrected 12106 to work with the new engine framework. 12107 12108 *AEP Inc. and Richard Levitte* 12109 12110 * Support for SureWare crypto accelerator cards from Baltimore 12111 Technologies. (Use engine 'sureware') 12112 The support was copied from 0.9.6c [engine] and adapted 12113 to work with the new engine framework. 12114 12115 *Richard Levitte* 12116 12117 * Have the CHIL engine fork-safe (as defined by nCipher) and actually 12118 make the newer ENGINE framework commands for the CHIL engine work. 12119 12120 *Toomas Kiisk <vix@cyber.ee> and Richard Levitte* 12121 12122 * Make it possible to produce shared libraries on ReliantUNIX. 12123 12124 *Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte* 12125 12126 * Add the configuration target debug-linux-ppro. 12127 Make 'openssl rsa' use the general key loading routines 12128 implemented in `apps.c`, and make those routines able to 12129 handle the key format FORMAT_NETSCAPE and the variant 12130 FORMAT_IISSGC. 12131 12132 *Toomas Kiisk <vix@cyber.ee> via Richard Levitte* 12133 12134 * Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 12135 12136 *Toomas Kiisk <vix@cyber.ee> via Richard Levitte* 12137 12138 * Add -keyform to rsautl, and document -engine. 12139 12140 *Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>* 12141 12142 * Change BIO_new_file (crypto/bio/bss_file.c) to use new 12143 BIO_R_NO_SUCH_FILE error code rather than the generic 12144 ERR_R_SYS_LIB error code if fopen() fails with ENOENT. 12145 12146 *Ben Laurie* 12147 12148 * Add new functions 12149 ERR_peek_last_error 12150 ERR_peek_last_error_line 12151 ERR_peek_last_error_line_data. 12152 These are similar to 12153 ERR_peek_error 12154 ERR_peek_error_line 12155 ERR_peek_error_line_data, 12156 but report on the latest error recorded rather than the first one 12157 still in the error queue. 12158 12159 *Ben Laurie, Bodo Moeller* 12160 12161 * default_algorithms option in ENGINE config module. This allows things 12162 like: 12163 default_algorithms = ALL 12164 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS 12165 12166 *Steve Henson* 12167 12168 * Preliminary ENGINE config module. 12169 12170 *Steve Henson* 12171 12172 * New experimental application configuration code. 12173 12174 *Steve Henson* 12175 12176 * Change the AES code to follow the same name structure as all other 12177 symmetric ciphers, and behave the same way. Move everything to 12178 the directory crypto/aes, thereby obsoleting crypto/rijndael. 12179 12180 *Stephen Sprunk <stephen@sprunk.org> and Richard Levitte* 12181 12182 * SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c. 12183 12184 *Ben Laurie and Theo de Raadt* 12185 12186 * Add option to output public keys in req command. 12187 12188 *Massimiliano Pala madwolf@openca.org* 12189 12190 * Use wNAFs in EC_POINTs_mul() for improved efficiency 12191 (up to about 10% better than before for P-192 and P-224). 12192 12193 *Bodo Moeller* 12194 12195 * New functions/macros 12196 12197 SSL_CTX_set_msg_callback(ctx, cb) 12198 SSL_CTX_set_msg_callback_arg(ctx, arg) 12199 SSL_set_msg_callback(ssl, cb) 12200 SSL_set_msg_callback_arg(ssl, arg) 12201 12202 to request calling a callback function 12203 12204 void cb(int write_p, int version, int content_type, 12205 const void *buf, size_t len, SSL *ssl, void *arg) 12206 12207 whenever a protocol message has been completely received 12208 (write_p == 0) or sent (write_p == 1). Here 'version' is the 12209 protocol version according to which the SSL library interprets 12210 the current protocol message (SSL2_VERSION, SSL3_VERSION, or 12211 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or 12212 the content type as defined in the SSL 3.0/TLS 1.0 protocol 12213 specification (change_cipher_spec(20), alert(21), handshake(22)). 12214 'buf' and 'len' point to the actual message, 'ssl' to the 12215 SSL object, and 'arg' is the application-defined value set by 12216 SSL[_CTX]_set_msg_callback_arg(). 12217 12218 'openssl s_client' and 'openssl s_server' have new '-msg' options 12219 to enable a callback that displays all protocol messages. 12220 12221 *Bodo Moeller* 12222 12223 * Change the shared library support so shared libraries are built as 12224 soon as the corresponding static library is finished, and thereby get 12225 openssl and the test programs linked against the shared library. 12226 This still only happens when the keyword "shard" has been given to 12227 the configuration scripts. 12228 12229 NOTE: shared library support is still an experimental thing, and 12230 backward binary compatibility is still not guaranteed. 12231 12232 *"Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte* 12233 12234 * Add support for Subject Information Access extension. 12235 12236 *Peter Sylvester <Peter.Sylvester@EdelWeb.fr>* 12237 12238 * Make BUF_MEM_grow() behaviour more consistent: Initialise to zero 12239 additional bytes when new memory had to be allocated, not just 12240 when reusing an existing buffer. 12241 12242 *Bodo Moeller* 12243 12244 * New command line and configuration option 'utf8' for the req command. 12245 This allows field values to be specified as UTF8 strings. 12246 12247 *Steve Henson* 12248 12249 * Add -multi and -mr options to "openssl speed" - giving multiple parallel 12250 runs for the former and machine-readable output for the latter. 12251 12252 *Ben Laurie* 12253 12254 * Add '-noemailDN' option to 'openssl ca'. This prevents inclusion 12255 of the e-mail address in the DN (i.e., it will go into a certificate 12256 extension only). The new configuration file option 'email_in_dn = no' 12257 has the same effect. 12258 12259 *Massimiliano Pala madwolf@openca.org* 12260 12261 * Change all functions with names starting with `des_` to be starting 12262 with `DES_` instead. Add wrappers that are compatible with libdes, 12263 but are named `_ossl_old_des_*`. Finally, add macros that map the 12264 `des_*` symbols to the corresponding `_ossl_old_des_*` if libdes 12265 compatibility is desired. If OpenSSL 0.9.6c compatibility is 12266 desired, the `des_*` symbols will be mapped to `DES_*`, with one 12267 exception. 12268 12269 Since we provide two compatibility mappings, the user needs to 12270 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes 12271 compatibility is desired. The default (i.e., when that macro 12272 isn't defined) is OpenSSL 0.9.6c compatibility. 12273 12274 There are also macros that enable and disable the support of old 12275 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT 12276 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those 12277 are defined, the default will apply: to support the old des routines. 12278 12279 In either case, one must include openssl/des.h to get the correct 12280 definitions. Do not try to just include openssl/des_old.h, that 12281 won't work. 12282 12283 NOTE: This is a major break of an old API into a new one. Software 12284 authors are encouraged to switch to the `DES_` style functions. Some 12285 time in the future, des_old.h and the libdes compatibility functions 12286 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the 12287 default), and then completely removed. 12288 12289 *Richard Levitte* 12290 12291 * Test for certificates which contain unsupported critical extensions. 12292 If such a certificate is found during a verify operation it is 12293 rejected by default: this behaviour can be overridden by either 12294 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or 12295 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function 12296 X509_supported_extension() has also been added which returns 1 if a 12297 particular extension is supported. 12298 12299 *Steve Henson* 12300 12301 * Modify the behaviour of EVP cipher functions in similar way to digests 12302 to retain compatibility with existing code. 12303 12304 *Steve Henson* 12305 12306 * Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain 12307 compatibility with existing code. In particular the 'ctx' parameter does 12308 not have to be to be initialized before the call to EVP_DigestInit() and 12309 it is tidied up after a call to EVP_DigestFinal(). New function 12310 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function 12311 EVP_MD_CTX_copy() changed to not require the destination to be 12312 initialized valid and new function EVP_MD_CTX_copy_ex() added which 12313 requires the destination to be valid. 12314 12315 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(), 12316 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex(). 12317 12318 *Steve Henson* 12319 12320 * Change ssl3_get_message (ssl/s3_both.c) and the functions using it 12321 so that complete 'Handshake' protocol structures are kept in memory 12322 instead of overwriting 'msg_type' and 'length' with 'body' data. 12323 12324 *Bodo Moeller* 12325 12326 * Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32. 12327 12328 *Massimo Santin via Richard Levitte* 12329 12330 * Major restructuring to the underlying ENGINE code. This includes 12331 reduction of linker bloat, separation of pure "ENGINE" manipulation 12332 (initialisation, etc) from functionality dealing with implementations 12333 of specific crypto interfaces. This change also introduces integrated 12334 support for symmetric ciphers and digest implementations - so ENGINEs 12335 can now accelerate these by providing EVP_CIPHER and EVP_MD 12336 implementations of their own. This is detailed in 12337 [crypto/engine/README.md](crypto/engine/README.md) 12338 as it couldn't be adequately described here. However, there are a few 12339 API changes worth noting - some RSA, DSA, DH, and RAND functions that 12340 were changed in the original introduction of ENGINE code have now 12341 reverted back - the hooking from this code to ENGINE is now a good 12342 deal more passive and at run-time, operations deal directly with 12343 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than 12344 dereferencing through an ENGINE pointer any more. Also, the ENGINE 12345 functions dealing with `BN_MOD_EXP[_CRT]` handlers have been removed - 12346 they were not being used by the framework as there is no concept of a 12347 BIGNUM_METHOD and they could not be generalised to the new 12348 'ENGINE_TABLE' mechanism that underlies the new code. Similarly, 12349 ENGINE_cpy() has been removed as it cannot be consistently defined in 12350 the new code. 12351 12352 *Geoff Thorpe* 12353 12354 * Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds. 12355 12356 *Steve Henson* 12357 12358 * Change mkdef.pl to sort symbols that get the same entry number, 12359 and make sure the automatically generated functions `ERR_load_*` 12360 become part of libeay.num as well. 12361 12362 *Richard Levitte* 12363 12364 * New function SSL_renegotiate_pending(). This returns true once 12365 renegotiation has been requested (either SSL_renegotiate() call 12366 or HelloRequest/ClientHello received from the peer) and becomes 12367 false once a handshake has been completed. 12368 (For servers, SSL_renegotiate() followed by SSL_do_handshake() 12369 sends a HelloRequest, but does not ensure that a handshake takes 12370 place. SSL_renegotiate_pending() is useful for checking if the 12371 client has followed the request.) 12372 12373 *Bodo Moeller* 12374 12375 * New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION. 12376 By default, clients may request session resumption even during 12377 renegotiation (if session ID contexts permit); with this option, 12378 session resumption is possible only in the first handshake. 12379 12380 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes 12381 more bits available for options that should not be part of 12382 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION). 12383 12384 *Bodo Moeller* 12385 12386 * Add some demos for certificate and certificate request creation. 12387 12388 *Steve Henson* 12389 12390 * Make maximum certificate chain size accepted from the peer application 12391 settable (`SSL*_get/set_max_cert_list()`), as proposed by 12392 "Douglas E. Engert" <deengert@anl.gov>. 12393 12394 *Lutz Jaenicke* 12395 12396 * Add support for shared libraries for Unixware-7 12397 (Boyd Lynn Gerber <gerberb@zenez.com>). 12398 12399 *Lutz Jaenicke* 12400 12401 * Add a "destroy" handler to ENGINEs that allows structural cleanup to 12402 be done prior to destruction. Use this to unload error strings from 12403 ENGINEs that load their own error strings. NB: This adds two new API 12404 functions to "get" and "set" this destroy handler in an ENGINE. 12405 12406 *Geoff Thorpe* 12407 12408 * Alter all existing ENGINE implementations (except "openssl" and 12409 "openbsd") to dynamically instantiate their own error strings. This 12410 makes them more flexible to be built both as statically-linked ENGINEs 12411 and self-contained shared-libraries loadable via the "dynamic" ENGINE. 12412 Also, add stub code to each that makes building them as self-contained 12413 shared-libraries easier (see [README-Engine.md](README-Engine.md)). 12414 12415 *Geoff Thorpe* 12416 12417 * Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE 12418 implementations into applications that are completely implemented in 12419 self-contained shared-libraries. The "dynamic" ENGINE exposes control 12420 commands that can be used to configure what shared-library to load and 12421 to control aspects of the way it is handled. Also, made an update to 12422 the [README-Engine.md](README-Engine.md) file 12423 that brings its information up-to-date and 12424 provides some information and instructions on the "dynamic" ENGINE 12425 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc). 12426 12427 *Geoff Thorpe* 12428 12429 * Make it possible to unload ranges of ERR strings with a new 12430 "ERR_unload_strings" function. 12431 12432 *Geoff Thorpe* 12433 12434 * Add a copy() function to EVP_MD. 12435 12436 *Ben Laurie* 12437 12438 * Make EVP_MD routines take a context pointer instead of just the 12439 md_data void pointer. 12440 12441 *Ben Laurie* 12442 12443 * Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates 12444 that the digest can only process a single chunk of data 12445 (typically because it is provided by a piece of 12446 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application 12447 is only going to provide a single chunk of data, and hence the 12448 framework needn't accumulate the data for oneshot drivers. 12449 12450 *Ben Laurie* 12451 12452 * As with "ERR", make it possible to replace the underlying "ex_data" 12453 functions. This change also alters the storage and management of global 12454 ex_data state - it's now all inside ex_data.c and all "class" code (eg. 12455 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class 12456 index counters. The API functions that use this state have been changed 12457 to take a "class_index" rather than pointers to the class's local STACK 12458 and counter, and there is now an API function to dynamically create new 12459 classes. This centralisation allows us to (a) plug a lot of the 12460 thread-safety problems that existed, and (b) makes it possible to clean 12461 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b) 12462 such data would previously have always leaked in application code and 12463 workarounds were in place to make the memory debugging turn a blind eye 12464 to it. Application code that doesn't use this new function will still 12465 leak as before, but their memory debugging output will announce it now 12466 rather than letting it slide. 12467 12468 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change 12469 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now 12470 has a return value to indicate success or failure. 12471 12472 *Geoff Thorpe* 12473 12474 * Make it possible to replace the underlying "ERR" functions such that the 12475 global state (2 LHASH tables and 2 locks) is only used by the "default" 12476 implementation. This change also adds two functions to "get" and "set" 12477 the implementation prior to it being automatically set the first time 12478 any other ERR function takes place. Ie. an application can call "get", 12479 pass the return value to a module it has just loaded, and that module 12480 can call its own "set" function using that value. This means the 12481 module's "ERR" operations will use (and modify) the error state in the 12482 application and not in its own statically linked copy of OpenSSL code. 12483 12484 *Geoff Thorpe* 12485 12486 * Give DH, DSA, and RSA types their own `*_up_ref()` function to increment 12487 reference counts. This performs normal REF_PRINT/REF_CHECK macros on 12488 the operation, and provides a more encapsulated way for external code 12489 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code 12490 to use these functions rather than manually incrementing the counts. 12491 12492 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()". 12493 12494 *Geoff Thorpe* 12495 12496 * Add EVP test program. 12497 12498 *Ben Laurie* 12499 12500 * Add symmetric cipher support to ENGINE. Expect the API to change! 12501 12502 *Ben Laurie* 12503 12504 * New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name() 12505 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(), 12506 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate(). 12507 These allow a CRL to be built without having to access X509_CRL fields 12508 directly. Modify 'ca' application to use new functions. 12509 12510 *Steve Henson* 12511 12512 * Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended 12513 bug workarounds. Rollback attack detection is a security feature. 12514 The problem will only arise on OpenSSL servers when TLSv1 is not 12515 available (sslv3_server_method() or SSL_OP_NO_TLSv1). 12516 Software authors not wanting to support TLSv1 will have special reasons 12517 for their choice and can explicitly enable this option. 12518 12519 *Bodo Moeller, Lutz Jaenicke* 12520 12521 * Rationalise EVP so it can be extended: don't include a union of 12522 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX 12523 (similar to those existing for EVP_CIPHER_CTX). 12524 Usage example: 12525 12526 EVP_MD_CTX md; 12527 12528 EVP_MD_CTX_init(&md); /* new function call */ 12529 EVP_DigestInit(&md, EVP_sha1()); 12530 EVP_DigestUpdate(&md, in, len); 12531 EVP_DigestFinal(&md, out, NULL); 12532 EVP_MD_CTX_cleanup(&md); /* new function call */ 12533 12534 *Ben Laurie* 12535 12536 * Make DES key schedule conform to the usual scheme, as well as 12537 correcting its structure. This means that calls to DES functions 12538 now have to pass a pointer to a des_key_schedule instead of a 12539 plain des_key_schedule (which was actually always a pointer 12540 anyway): E.g., 12541 12542 des_key_schedule ks; 12543 12544 des_set_key_checked(..., &ks); 12545 des_ncbc_encrypt(..., &ks, ...); 12546 12547 (Note that a later change renames 'des_...' into 'DES_...'.) 12548 12549 *Ben Laurie* 12550 12551 * Initial reduction of linker bloat: the use of some functions, such as 12552 PEM causes large amounts of unused functions to be linked in due to 12553 poor organisation. For example pem_all.c contains every PEM function 12554 which has a knock on effect of linking in large amounts of (unused) 12555 ASN1 code. Grouping together similar functions and splitting unrelated 12556 functions prevents this. 12557 12558 *Steve Henson* 12559 12560 * Cleanup of EVP macros. 12561 12562 *Ben Laurie* 12563 12564 * Change historical references to `{NID,SN,LN}_des_ede` and ede3 to add the 12565 correct `_ecb suffix`. 12566 12567 *Ben Laurie* 12568 12569 * Add initial OCSP responder support to ocsp application. The 12570 revocation information is handled using the text based index 12571 use by the ca application. The responder can either handle 12572 requests generated internally, supplied in files (for example 12573 via a CGI script) or using an internal minimal server. 12574 12575 *Steve Henson* 12576 12577 * Add configuration choices to get zlib compression for TLS. 12578 12579 *Richard Levitte* 12580 12581 * Changes to Kerberos SSL for RFC 2712 compliance: 12582 1. Implemented real KerberosWrapper, instead of just using 12583 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>] 12584 2. Implemented optional authenticator field of KerberosWrapper. 12585 12586 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req, 12587 and authenticator structs; see crypto/krb5/. 12588 12589 Generalized Kerberos calls to support multiple Kerberos libraries. 12590 *Vern Staats <staatsvr@asc.hpc.mil>, Jeffrey Altman <jaltman@columbia.edu> 12591 via Richard Levitte* 12592 12593 * Cause 'openssl speed' to use fully hard-coded DSA keys as it 12594 already does with RSA. testdsa.h now has 'priv_key/pub_key' 12595 values for each of the key sizes rather than having just 12596 parameters (and 'speed' generating keys each time). 12597 12598 *Geoff Thorpe* 12599 12600 * Speed up EVP routines. 12601 Before: 12602crypt 12603pe 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes 12604s-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k 12605s-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k 12606s-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k 12607crypt 12608s-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k 12609s-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k 12610s-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k 12611 After: 12612crypt 12613s-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k 12614crypt 12615s-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k 12616 12617 *Ben Laurie* 12618 12619 * Added the OS2-EMX target. 12620 12621 *"Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte* 12622 12623 * Rewrite commands to use `NCONF` routines instead of the old `CONF`. 12624 New functions to support `NCONF` routines in extension code. 12625 New function `CONF_set_nconf()` 12626 to allow functions which take an `NCONF` to also handle the old `LHASH` 12627 structure: this means that the old `CONF` compatible routines can be 12628 retained (in particular w.rt. extensions) without having to duplicate the 12629 code. New function `X509V3_add_ext_nconf_sk()` to add extensions to a stack. 12630 12631 *Steve Henson* 12632 12633 * Enhance the general user interface with mechanisms for inner control 12634 and with possibilities to have yes/no kind of prompts. 12635 12636 *Richard Levitte* 12637 12638 * Change all calls to low-level digest routines in the library and 12639 applications to use EVP. Add missing calls to HMAC_cleanup() and 12640 don't assume HMAC_CTX can be copied using memcpy(). 12641 12642 *Verdon Walker <VWalker@novell.com>, Steve Henson* 12643 12644 * Add the possibility to control engines through control names but with 12645 arbitrary arguments instead of just a string. 12646 Change the key loaders to take a UI_METHOD instead of a callback 12647 function pointer. NOTE: this breaks binary compatibility with earlier 12648 versions of OpenSSL [engine]. 12649 Adapt the nCipher code for these new conditions and add a card insertion 12650 callback. 12651 12652 *Richard Levitte* 12653 12654 * Enhance the general user interface with mechanisms to better support 12655 dialog box interfaces, application-defined prompts, the possibility 12656 to use defaults (for example default passwords from somewhere else) 12657 and interrupts/cancellations. 12658 12659 *Richard Levitte* 12660 12661 * Tidy up PKCS#12 attribute handling. Add support for the CSP name 12662 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility. 12663 12664 *Steve Henson* 12665 12666 * Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also 12667 tidy up some unnecessarily weird code in 'sk_new()'). 12668 12669 *Geoff, reported by Diego Tartara <dtartara@novamens.com>* 12670 12671 * Change the key loading routines for ENGINEs to use the same kind 12672 callback (pem_password_cb) as all other routines that need this 12673 kind of callback. 12674 12675 *Richard Levitte* 12676 12677 * Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with 12678 256 bit (=32 byte) keys. Of course seeding with more entropy bytes 12679 than this minimum value is recommended. 12680 12681 *Lutz Jaenicke* 12682 12683 * New random seeder for OpenVMS, using the system process statistics 12684 that are easily reachable. 12685 12686 *Richard Levitte* 12687 12688 * Windows apparently can't transparently handle global 12689 variables defined in DLLs. Initialisations such as: 12690 12691 const ASN1_ITEM *it = &ASN1_INTEGER_it; 12692 12693 won't compile. This is used by the any applications that need to 12694 declare their own ASN1 modules. This was fixed by adding the option 12695 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly 12696 needed for static libraries under Win32. 12697 12698 *Steve Henson* 12699 12700 * New functions X509_PURPOSE_set() and X509_TRUST_set() to handle 12701 setting of purpose and trust fields. New X509_STORE trust and 12702 purpose functions and tidy up setting in other SSL functions. 12703 12704 *Steve Henson* 12705 12706 * Add copies of X509_STORE_CTX fields and callbacks to X509_STORE 12707 structure. These are inherited by X509_STORE_CTX when it is 12708 initialised. This allows various defaults to be set in the 12709 X509_STORE structure (such as flags for CRL checking and custom 12710 purpose or trust settings) for functions which only use X509_STORE_CTX 12711 internally such as S/MIME. 12712 12713 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and 12714 trust settings if they are not set in X509_STORE. This allows X509_STORE 12715 purposes and trust (in S/MIME for example) to override any set by default. 12716 12717 Add command line options for CRL checking to smime, s_client and s_server 12718 applications. 12719 12720 *Steve Henson* 12721 12722 * Initial CRL based revocation checking. If the CRL checking flag(s) 12723 are set then the CRL is looked up in the X509_STORE structure and 12724 its validity and signature checked, then if the certificate is found 12725 in the CRL the verify fails with a revoked error. 12726 12727 Various new CRL related callbacks added to X509_STORE_CTX structure. 12728 12729 Command line options added to 'verify' application to support this. 12730 12731 This needs some additional work, such as being able to handle multiple 12732 CRLs with different times, extension based lookup (rather than just 12733 by subject name) and ultimately more complete V2 CRL extension 12734 handling. 12735 12736 *Steve Henson* 12737 12738 * Add a general user interface API (crypto/ui/). This is designed 12739 to replace things like des_read_password and friends (backward 12740 compatibility functions using this new API are provided). 12741 The purpose is to remove prompting functions from the DES code 12742 section as well as provide for prompting through dialog boxes in 12743 a window system and the like. 12744 12745 *Richard Levitte* 12746 12747 * Add "ex_data" support to ENGINE so implementations can add state at a 12748 per-structure level rather than having to store it globally. 12749 12750 *Geoff* 12751 12752 * Make it possible for ENGINE structures to be copied when retrieved by 12753 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY. 12754 This causes the "original" ENGINE structure to act like a template, 12755 analogous to the RSA vs. RSA_METHOD type of separation. Because of this 12756 operational state can be localised to each ENGINE structure, despite the 12757 fact they all share the same "methods". New ENGINE structures returned in 12758 this case have no functional references and the return value is the single 12759 structural reference. This matches the single structural reference returned 12760 by ENGINE_by_id() normally, when it is incremented on the pre-existing 12761 ENGINE structure. 12762 12763 *Geoff* 12764 12765 * Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this 12766 needs to match any other type at all we need to manually clear the 12767 tag cache. 12768 12769 *Steve Henson* 12770 12771 * Changes to the "openssl engine" utility to include; 12772 - verbosity levels ('-v', '-vv', and '-vvv') that provide information 12773 about an ENGINE's available control commands. 12774 - executing control commands from command line arguments using the 12775 '-pre' and '-post' switches. '-post' is only used if '-t' is 12776 specified and the ENGINE is successfully initialised. The syntax for 12777 the individual commands are colon-separated, for example; 12778 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so 12779 12780 *Geoff* 12781 12782 * New dynamic control command support for ENGINEs. ENGINEs can now 12783 declare their own commands (numbers), names (strings), descriptions, 12784 and input types for run-time discovery by calling applications. A 12785 subset of these commands are implicitly classed as "executable" 12786 depending on their input type, and only these can be invoked through 12787 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this 12788 can be based on user input, config files, etc). The distinction is 12789 that "executable" commands cannot return anything other than a boolean 12790 result and can only support numeric or string input, whereas some 12791 discoverable commands may only be for direct use through 12792 ENGINE_ctrl(), eg. supporting the exchange of binary data, function 12793 pointers, or other custom uses. The "executable" commands are to 12794 support parameterisations of ENGINE behaviour that can be 12795 unambiguously defined by ENGINEs and used consistently across any 12796 OpenSSL-based application. Commands have been added to all the 12797 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow 12798 control over shared-library paths without source code alterations. 12799 12800 *Geoff* 12801 12802 * Changed all ENGINE implementations to dynamically allocate their 12803 ENGINEs rather than declaring them statically. Apart from this being 12804 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction, 12805 this also allows the implementations to compile without using the 12806 internal engine_int.h header. 12807 12808 *Geoff* 12809 12810 * Minor adjustment to "rand" code. RAND_get_rand_method() now returns a 12811 'const' value. Any code that should be able to modify a RAND_METHOD 12812 should already have non-const pointers to it (ie. they should only 12813 modify their own ones). 12814 12815 *Geoff* 12816 12817 * Made a variety of little tweaks to the ENGINE code. 12818 - "atalla" and "ubsec" string definitions were moved from header files 12819 to C code. "nuron" string definitions were placed in variables 12820 rather than hard-coded - allowing parameterisation of these values 12821 later on via ctrl() commands. 12822 - Removed unused "#if 0"'d code. 12823 - Fixed engine list iteration code so it uses ENGINE_free() to release 12824 structural references. 12825 - Constified the RAND_METHOD element of ENGINE structures. 12826 - Constified various get/set functions as appropriate and added 12827 missing functions (including a catch-all ENGINE_cpy that duplicates 12828 all ENGINE values onto a new ENGINE except reference counts/state). 12829 - Removed NULL parameter checks in get/set functions. Setting a method 12830 or function to NULL is a way of cancelling out a previously set 12831 value. Passing a NULL ENGINE parameter is just plain stupid anyway 12832 and doesn't justify the extra error symbols and code. 12833 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for 12834 flags from engine_int.h to engine.h. 12835 - Changed prototypes for ENGINE handler functions (init(), finish(), 12836 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter. 12837 12838 *Geoff* 12839 12840 * Implement binary inversion algorithm for BN_mod_inverse in addition 12841 to the algorithm using long division. The binary algorithm can be 12842 used only if the modulus is odd. On 32-bit systems, it is faster 12843 only for relatively small moduli (roughly 20-30% for 128-bit moduli, 12844 roughly 5-15% for 256-bit moduli), so we use it only for moduli 12845 up to 450 bits. In 64-bit environments, the binary algorithm 12846 appears to be advantageous for much longer moduli; here we use it 12847 for moduli up to 2048 bits. 12848 12849 *Bodo Moeller* 12850 12851 * Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code 12852 could not support the combine flag in choice fields. 12853 12854 *Steve Henson* 12855 12856 * Add a 'copy_extensions' option to the 'ca' utility. This copies 12857 extensions from a certificate request to the certificate. 12858 12859 *Steve Henson* 12860 12861 * Allow multiple 'certopt' and 'nameopt' options to be separated 12862 by commas. Add 'namopt' and 'certopt' options to the 'ca' config 12863 file: this allows the display of the certificate about to be 12864 signed to be customised, to allow certain fields to be included 12865 or excluded and extension details. The old system didn't display 12866 multicharacter strings properly, omitted fields not in the policy 12867 and couldn't display additional details such as extensions. 12868 12869 *Steve Henson* 12870 12871 * Function EC_POINTs_mul for multiple scalar multiplication 12872 of an arbitrary number of elliptic curve points 12873 \sum scalars[i]*points[i], 12874 optionally including the generator defined for the EC_GROUP: 12875 scalar*generator + \sum scalars[i]*points[i]. 12876 12877 EC_POINT_mul is a simple wrapper function for the typical case 12878 that the point list has just one item (besides the optional 12879 generator). 12880 12881 *Bodo Moeller* 12882 12883 * First EC_METHODs for curves over GF(p): 12884 12885 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr 12886 operations and provides various method functions that can also 12887 operate with faster implementations of modular arithmetic. 12888 12889 EC_GFp_mont_method() reuses most functions that are part of 12890 EC_GFp_simple_method, but uses Montgomery arithmetic. 12891 12892 *Bodo Moeller; point addition and point doubling 12893 implementation directly derived from source code provided by 12894 Lenka Fibikova <fibikova@exp-math.uni-essen.de>* 12895 12896 * Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h, 12897 crypto/ec/ec_lib.c): 12898 12899 Curves are EC_GROUP objects (with an optional group generator) 12900 based on EC_METHODs that are built into the library. 12901 12902 Points are EC_POINT objects based on EC_GROUP objects. 12903 12904 Most of the framework would be able to handle curves over arbitrary 12905 finite fields, but as there are no obvious types for fields other 12906 than GF(p), some functions are limited to that for now. 12907 12908 *Bodo Moeller* 12909 12910 * Add the -HTTP option to s_server. It is similar to -WWW, but requires 12911 that the file contains a complete HTTP response. 12912 12913 *Richard Levitte* 12914 12915 * Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl 12916 change the def and num file printf format specifier from "%-40sXXX" 12917 to "%-39s XXX". The latter will always guarantee a space after the 12918 field while the former will cause them to run together if the field 12919 is 40 of more characters long. 12920 12921 *Steve Henson* 12922 12923 * Constify the cipher and digest 'method' functions and structures 12924 and modify related functions to take constant EVP_MD and EVP_CIPHER 12925 pointers. 12926 12927 *Steve Henson* 12928 12929 * Hide BN_CTX structure details in bn_lcl.h instead of publishing them 12930 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32. 12931 12932 *Bodo Moeller* 12933 12934 * Modify `EVP_Digest*()` routines so they now return values. Although the 12935 internal software routines can never fail additional hardware versions 12936 might. 12937 12938 *Steve Henson* 12939 12940 * Clean up crypto/err/err.h and change some error codes to avoid conflicts: 12941 12942 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7 12943 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32. 12944 12945 ASN1 error codes 12946 ERR_R_NESTED_ASN1_ERROR 12947 ... 12948 ERR_R_MISSING_ASN1_EOS 12949 were 4 .. 9, conflicting with 12950 ERR_LIB_RSA (= ERR_R_RSA_LIB) 12951 ... 12952 ERR_LIB_PEM (= ERR_R_PEM_LIB). 12953 They are now 58 .. 63 (i.e., just below ERR_R_FATAL). 12954 12955 Add new error code 'ERR_R_INTERNAL_ERROR'. 12956 12957 *Bodo Moeller* 12958 12959 * Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock 12960 suffices. 12961 12962 *Bodo Moeller* 12963 12964 * New option '-subj arg' for 'openssl req' and 'openssl ca'. This 12965 sets the subject name for a new request or supersedes the 12966 subject name in a given request. Formats that can be parsed are 12967 'CN=Some Name, OU=myOU, C=IT' 12968 and 12969 'CN=Some Name/OU=myOU/C=IT'. 12970 12971 Add options '-batch' and '-verbose' to 'openssl req'. 12972 12973 *Massimiliano Pala <madwolf@hackmasters.net>* 12974 12975 * Introduce the possibility to access global variables through 12976 functions on platform were that's the best way to handle exporting 12977 global variables in shared libraries. To enable this functionality, 12978 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro 12979 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter 12980 is normally done by Configure or something similar). 12981 12982 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL 12983 in the source file (foo.c) like this: 12984 12985 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1; 12986 OPENSSL_IMPLEMENT_GLOBAL(double,bar); 12987 12988 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL 12989 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this: 12990 12991 OPENSSL_DECLARE_GLOBAL(int,foo); 12992 #define foo OPENSSL_GLOBAL_REF(foo) 12993 OPENSSL_DECLARE_GLOBAL(double,bar); 12994 #define bar OPENSSL_GLOBAL_REF(bar) 12995 12996 The #defines are very important, and therefore so is including the 12997 header file everywhere where the defined globals are used. 12998 12999 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition 13000 of ASN.1 items, but that structure is a bit different. 13001 13002 The largest change is in util/mkdef.pl which has been enhanced with 13003 better and easier to understand logic to choose which symbols should 13004 go into the Windows .def files as well as a number of fixes and code 13005 cleanup (among others, algorithm keywords are now sorted 13006 lexicographically to avoid constant rewrites). 13007 13008 *Richard Levitte* 13009 13010 * In BN_div() keep a copy of the sign of 'num' before writing the 13011 result to 'rm' because if rm==num the value will be overwritten 13012 and produce the wrong result if 'num' is negative: this caused 13013 problems with BN_mod() and BN_nnmod(). 13014 13015 *Steve Henson* 13016 13017 * Function OCSP_request_verify(). This checks the signature on an 13018 OCSP request and verifies the signer certificate. The signer 13019 certificate is just checked for a generic purpose and OCSP request 13020 trust settings. 13021 13022 *Steve Henson* 13023 13024 * Add OCSP_check_validity() function to check the validity of OCSP 13025 responses. OCSP responses are prepared in real time and may only 13026 be a few seconds old. Simply checking that the current time lies 13027 between thisUpdate and nextUpdate max reject otherwise valid responses 13028 caused by either OCSP responder or client clock inaccuracy. Instead 13029 we allow thisUpdate and nextUpdate to fall within a certain period of 13030 the current time. The age of the response can also optionally be 13031 checked. Two new options -validity_period and -status_age added to 13032 ocsp utility. 13033 13034 *Steve Henson* 13035 13036 * If signature or public key algorithm is unrecognized print out its 13037 OID rather that just UNKNOWN. 13038 13039 *Steve Henson* 13040 13041 * Change OCSP_cert_to_id() to tolerate a NULL subject certificate and 13042 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate 13043 ID to be generated from the issuer certificate alone which can then be 13044 passed to OCSP_id_issuer_cmp(). 13045 13046 *Steve Henson* 13047 13048 * New compilation option ASN1_ITEM_FUNCTIONS. This causes the new 13049 ASN1 modules to export functions returning ASN1_ITEM pointers 13050 instead of the ASN1_ITEM structures themselves. This adds several 13051 new macros which allow the underlying ASN1 function/structure to 13052 be accessed transparently. As a result code should not use ASN1_ITEM 13053 references directly (such as &X509_it) but instead use the relevant 13054 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow 13055 use of the new ASN1 code on platforms where exporting structures 13056 is problematical (for example in shared libraries) but exporting 13057 functions returning pointers to structures is not. 13058 13059 *Steve Henson* 13060 13061 * Add support for overriding the generation of SSL/TLS session IDs. 13062 These callbacks can be registered either in an SSL_CTX or per SSL. 13063 The purpose of this is to allow applications to control, if they wish, 13064 the arbitrary values chosen for use as session IDs, particularly as it 13065 can be useful for session caching in multiple-server environments. A 13066 command-line switch for testing this (and any client code that wishes 13067 to use such a feature) has been added to "s_server". 13068 13069 *Geoff Thorpe, Lutz Jaenicke* 13070 13071 * Modify mkdef.pl to recognise and parse preprocessor conditionals 13072 of the form `#if defined(...) || defined(...) || ...` and 13073 `#if !defined(...) && !defined(...) && ...`. This also avoids 13074 the growing number of special cases it was previously handling. 13075 13076 *Richard Levitte* 13077 13078 * Make all configuration macros available for application by making 13079 sure they are available in opensslconf.h, by giving them names starting 13080 with `OPENSSL_` to avoid conflicts with other packages and by making 13081 sure e_os2.h will cover all platform-specific cases together with 13082 opensslconf.h. 13083 Additionally, it is now possible to define configuration/platform- 13084 specific names (called "system identities"). In the C code, these 13085 are prefixed with `OPENSSL_SYSNAME_`. e_os2.h will create another 13086 macro with the name beginning with `OPENSSL_SYS_`, which is determined 13087 from `OPENSSL_SYSNAME_*` or compiler-specific macros depending on 13088 what is available. 13089 13090 *Richard Levitte* 13091 13092 * New option -set_serial to 'req' and 'x509' this allows the serial 13093 number to use to be specified on the command line. Previously self 13094 signed certificates were hard coded with serial number 0 and the 13095 CA options of 'x509' had to use a serial number in a file which was 13096 auto incremented. 13097 13098 *Steve Henson* 13099 13100 * New options to 'ca' utility to support V2 CRL entry extensions. 13101 Currently CRL reason, invalidity date and hold instruction are 13102 supported. Add new CRL extensions to V3 code and some new objects. 13103 13104 *Steve Henson* 13105 13106 * New function EVP_CIPHER_CTX_set_padding() this is used to 13107 disable standard block padding (aka PKCS#5 padding) in the EVP 13108 API, which was previously mandatory. This means that the data is 13109 not padded in any way and so the total length much be a multiple 13110 of the block size, otherwise an error occurs. 13111 13112 *Steve Henson* 13113 13114 * Initial (incomplete) OCSP SSL support. 13115 13116 *Steve Henson* 13117 13118 * New function OCSP_parse_url(). This splits up a URL into its host, 13119 port and path components: primarily to parse OCSP URLs. New -url 13120 option to ocsp utility. 13121 13122 *Steve Henson* 13123 13124 * New nonce behavior. The return value of OCSP_check_nonce() now 13125 reflects the various checks performed. Applications can decide 13126 whether to tolerate certain situations such as an absent nonce 13127 in a response when one was present in a request: the ocsp application 13128 just prints out a warning. New function OCSP_add1_basic_nonce() 13129 this is to allow responders to include a nonce in a response even if 13130 the request is nonce-less. 13131 13132 *Steve Henson* 13133 13134 * Disable stdin buffering in `load_cert()` (`apps/apps.c`) so that no certs are 13135 skipped when using openssl x509 multiple times on a single input file, 13136 e.g. `(openssl x509 -out cert1; openssl x509 -out cert2) <certs`. 13137 13138 *Bodo Moeller* 13139 13140 * Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string() 13141 set string type: to handle setting ASN1_TIME structures. Fix ca 13142 utility to correctly initialize revocation date of CRLs. 13143 13144 *Steve Henson* 13145 13146 * New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override 13147 the clients preferred ciphersuites and rather use its own preferences. 13148 Should help to work around M$ SGC (Server Gated Cryptography) bug in 13149 Internet Explorer by ensuring unchanged hash method during stepup. 13150 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.) 13151 13152 *Lutz Jaenicke* 13153 13154 * Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael 13155 to aes and add a new 'exist' option to print out symbols that don't 13156 appear to exist. 13157 13158 *Steve Henson* 13159 13160 * Additional options to ocsp utility to allow flags to be set and 13161 additional certificates supplied. 13162 13163 *Steve Henson* 13164 13165 * Add the option -VAfile to 'openssl ocsp', so the user can give the 13166 OCSP client a number of certificate to only verify the response 13167 signature against. 13168 13169 *Richard Levitte* 13170 13171 * Update Rijndael code to version 3.0 and change EVP AES ciphers to 13172 handle the new API. Currently only ECB, CBC modes supported. Add new 13173 AES OIDs. 13174 13175 Add TLS AES ciphersuites as described in RFC3268, "Advanced 13176 Encryption Standard (AES) Ciphersuites for Transport Layer 13177 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were 13178 not enabled by default and were not part of the "ALL" ciphersuite 13179 alias because they were not yet official; they could be 13180 explicitly requested by specifying the "AESdraft" ciphersuite 13181 group alias. In the final release of OpenSSL 0.9.7, the group 13182 alias is called "AES" and is part of "ALL".) 13183 13184 *Ben Laurie, Steve Henson, Bodo Moeller* 13185 13186 * New function OCSP_copy_nonce() to copy nonce value (if present) from 13187 request to response. 13188 13189 *Steve Henson* 13190 13191 * Functions for OCSP responders. OCSP_request_onereq_count(), 13192 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info() 13193 extract information from a certificate request. OCSP_response_create() 13194 creates a response and optionally adds a basic response structure. 13195 OCSP_basic_add1_status() adds a complete single response to a basic 13196 response and returns the OCSP_SINGLERESP structure just added (to allow 13197 extensions to be included for example). OCSP_basic_add1_cert() adds a 13198 certificate to a basic response and OCSP_basic_sign() signs a basic 13199 response with various flags. New helper functions ASN1_TIME_check() 13200 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime() 13201 (converts ASN1_TIME to GeneralizedTime). 13202 13203 *Steve Henson* 13204 13205 * Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}() 13206 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key 13207 structure from a certificate. X509_pubkey_digest() digests the public_key 13208 contents: this is used in various key identifiers. 13209 13210 *Steve Henson* 13211 13212 * Make sk_sort() tolerate a NULL argument. 13213 13214 *Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>* 13215 13216 * New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates 13217 passed by the function are trusted implicitly. If any of them signed the 13218 response then it is assumed to be valid and is not verified. 13219 13220 *Steve Henson* 13221 13222 * In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT 13223 to data. This was previously part of the PKCS7 ASN1 code. This 13224 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures. 13225 *Steve Henson, reported by Kenneth R. Robinette 13226 <support@securenetterm.com>* 13227 13228 * Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1 13229 routines: without these tracing memory leaks is very painful. 13230 Fix leaks in PKCS12 and PKCS7 routines. 13231 13232 *Steve Henson* 13233 13234 * Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new(). 13235 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which 13236 effectively meant GeneralizedTime would never be used. Now it 13237 is initialised to -1 but X509_time_adj() now has to check the value 13238 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or 13239 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime. 13240 *Steve Henson, reported by Kenneth R. Robinette 13241 <support@securenetterm.com>* 13242 13243 * Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously 13244 result in a zero length in the ASN1_INTEGER structure which was 13245 not consistent with the structure when d2i_ASN1_INTEGER() was used 13246 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER() 13247 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER() 13248 where it did not print out a minus for negative ASN1_INTEGER. 13249 13250 *Steve Henson* 13251 13252 * Add summary printout to ocsp utility. The various functions which 13253 convert status values to strings have been renamed to: 13254 OCSP_response_status_str(), OCSP_cert_status_str() and 13255 OCSP_crl_reason_str() and are no longer static. New options 13256 to verify nonce values and to disable verification. OCSP response 13257 printout format cleaned up. 13258 13259 *Steve Henson* 13260 13261 * Add additional OCSP certificate checks. These are those specified 13262 in RFC2560. This consists of two separate checks: the CA of the 13263 certificate being checked must either be the OCSP signer certificate 13264 or the issuer of the OCSP signer certificate. In the latter case the 13265 OCSP signer certificate must contain the OCSP signing extended key 13266 usage. This check is performed by attempting to match the OCSP 13267 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash 13268 in the OCSP_CERTID structures of the response. 13269 13270 *Steve Henson* 13271 13272 * Initial OCSP certificate verification added to OCSP_basic_verify() 13273 and related routines. This uses the standard OpenSSL certificate 13274 verify routines to perform initial checks (just CA validity) and 13275 to obtain the certificate chain. Then additional checks will be 13276 performed on the chain. Currently the root CA is checked to see 13277 if it is explicitly trusted for OCSP signing. This is used to set 13278 a root CA as a global signing root: that is any certificate that 13279 chains to that CA is an acceptable OCSP signing certificate. 13280 13281 *Steve Henson* 13282 13283 * New '-extfile ...' option to 'openssl ca' for reading X.509v3 13284 extensions from a separate configuration file. 13285 As when reading extensions from the main configuration file, 13286 the '-extensions ...' option may be used for specifying the 13287 section to use. 13288 13289 *Massimiliano Pala <madwolf@comune.modena.it>* 13290 13291 * New OCSP utility. Allows OCSP requests to be generated or 13292 read. The request can be sent to a responder and the output 13293 parsed, outputted or printed in text form. Not complete yet: 13294 still needs to check the OCSP response validity. 13295 13296 *Steve Henson* 13297 13298 * New subcommands for 'openssl ca': 13299 `openssl ca -status <serial>` prints the status of the cert with 13300 the given serial number (according to the index file). 13301 `openssl ca -updatedb` updates the expiry status of certificates 13302 in the index file. 13303 13304 *Massimiliano Pala <madwolf@comune.modena.it>* 13305 13306 * New '-newreq-nodes' command option to CA.pl. This is like 13307 '-newreq', but calls 'openssl req' with the '-nodes' option 13308 so that the resulting key is not encrypted. 13309 13310 *Damien Miller <djm@mindrot.org>* 13311 13312 * New configuration for the GNU Hurd. 13313 13314 *Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte* 13315 13316 * Initial code to implement OCSP basic response verify. This 13317 is currently incomplete. Currently just finds the signer's 13318 certificate and verifies the signature on the response. 13319 13320 *Steve Henson* 13321 13322 * New SSLeay_version code SSLEAY_DIR to determine the compiled-in 13323 value of OPENSSLDIR. This is available via the new '-d' option 13324 to 'openssl version', and is also included in 'openssl version -a'. 13325 13326 *Bodo Moeller* 13327 13328 * Allowing defining memory allocation callbacks that will be given 13329 file name and line number information in additional arguments 13330 (a `const char*` and an int). The basic functionality remains, as 13331 well as the original possibility to just replace malloc(), 13332 realloc() and free() by functions that do not know about these 13333 additional arguments. To register and find out the current 13334 settings for extended allocation functions, the following 13335 functions are provided: 13336 13337 CRYPTO_set_mem_ex_functions 13338 CRYPTO_set_locked_mem_ex_functions 13339 CRYPTO_get_mem_ex_functions 13340 CRYPTO_get_locked_mem_ex_functions 13341 13342 These work the same way as CRYPTO_set_mem_functions and friends. 13343 `CRYPTO_get_[locked_]mem_functions` now writes 0 where such an 13344 extended allocation function is enabled. 13345 Similarly, `CRYPTO_get_[locked_]mem_ex_functions` writes 0 where 13346 a conventional allocation function is enabled. 13347 13348 *Richard Levitte, Bodo Moeller* 13349 13350 * Finish off removing the remaining LHASH function pointer casts. 13351 There should no longer be any prototype-casting required when using 13352 the LHASH abstraction, and any casts that remain are "bugs". See 13353 the callback types and macros at the head of lhash.h for details 13354 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example). 13355 13356 *Geoff Thorpe* 13357 13358 * Add automatic query of EGD sockets in RAND_poll() for the unix variant. 13359 If /dev/[u]random devices are not available or do not return enough 13360 entropy, EGD style sockets (served by EGD or PRNGD) will automatically 13361 be queried. 13362 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and 13363 /etc/entropy will be queried once each in this sequence, querying stops 13364 when enough entropy was collected without querying more sockets. 13365 13366 *Lutz Jaenicke* 13367 13368 * Change the Unix RAND_poll() variant to be able to poll several 13369 random devices, as specified by DEVRANDOM, until a sufficient amount 13370 of data has been collected. We spend at most 10 ms on each file 13371 (select timeout) and read in non-blocking mode. DEVRANDOM now 13372 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom" 13373 (previously it was just the string "/dev/urandom"), so on typical 13374 platforms the 10 ms delay will never occur. 13375 Also separate out the Unix variant to its own file, rand_unix.c. 13376 For VMS, there's a currently-empty rand_vms.c. 13377 13378 *Richard Levitte* 13379 13380 * Move OCSP client related routines to ocsp_cl.c. These 13381 provide utility functions which an application needing 13382 to issue a request to an OCSP responder and analyse the 13383 response will typically need: as opposed to those which an 13384 OCSP responder itself would need which will be added later. 13385 13386 OCSP_request_sign() signs an OCSP request with an API similar 13387 to PKCS7_sign(). OCSP_response_status() returns status of OCSP 13388 response. OCSP_response_get1_basic() extracts basic response 13389 from response. OCSP_resp_find_status(): finds and extracts status 13390 information from an OCSP_CERTID structure (which will be created 13391 when the request structure is built). These are built from lower 13392 level functions which work on OCSP_SINGLERESP structures but 13393 won't normally be used unless the application wishes to examine 13394 extensions in the OCSP response for example. 13395 13396 Replace nonce routines with a pair of functions. 13397 OCSP_request_add1_nonce() adds a nonce value and optionally 13398 generates a random value. OCSP_check_nonce() checks the 13399 validity of the nonce in an OCSP response. 13400 13401 *Steve Henson* 13402 13403 * Change function OCSP_request_add() to OCSP_request_add0_id(). 13404 This doesn't copy the supplied OCSP_CERTID and avoids the 13405 need to free up the newly created id. Change return type 13406 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure. 13407 This can then be used to add extensions to the request. 13408 Deleted OCSP_request_new(), since most of its functionality 13409 is now in OCSP_REQUEST_new() (and the case insensitive name 13410 clash) apart from the ability to set the request name which 13411 will be added elsewhere. 13412 13413 *Steve Henson* 13414 13415 * Update OCSP API. Remove obsolete extensions argument from 13416 various functions. Extensions are now handled using the new 13417 OCSP extension code. New simple OCSP HTTP function which 13418 can be used to send requests and parse the response. 13419 13420 *Steve Henson* 13421 13422 * Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new 13423 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN 13424 uses the special reorder version of SET OF to sort the attributes 13425 and reorder them to match the encoded order. This resolves a long 13426 standing problem: a verify on a PKCS7 structure just after signing 13427 it used to fail because the attribute order did not match the 13428 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes: 13429 it uses the received order. This is necessary to tolerate some broken 13430 software that does not order SET OF. This is handled by encoding 13431 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class) 13432 to produce the required SET OF. 13433 13434 *Steve Henson* 13435 13436 * Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and 13437 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header 13438 files to get correct declarations of the ASN.1 item variables. 13439 13440 *Richard Levitte* 13441 13442 * Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many 13443 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs: 13444 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was 13445 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i(). 13446 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant 13447 ASN1_ITEM and no wrapper functions. 13448 13449 *Steve Henson* 13450 13451 * New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These 13452 replace the old function pointer based I/O routines. Change most of 13453 the `*_d2i_bio()` and `*_d2i_fp()` functions to use these. 13454 13455 *Steve Henson* 13456 13457 * Enhance mkdef.pl to be more accepting about spacing in C preprocessor 13458 lines, recognize more "algorithms" that can be deselected, and make 13459 it complain about algorithm deselection that isn't recognised. 13460 13461 *Richard Levitte* 13462 13463 * New ASN1 functions to handle dup, sign, verify, digest, pack and 13464 unpack operations in terms of ASN1_ITEM. Modify existing wrappers 13465 to use new functions. Add NO_ASN1_OLD which can be set to remove 13466 some old style ASN1 functions: this can be used to determine if old 13467 code will still work when these eventually go away. 13468 13469 *Steve Henson* 13470 13471 * New extension functions for OCSP structures, these follow the 13472 same conventions as certificates and CRLs. 13473 13474 *Steve Henson* 13475 13476 * New function X509V3_add1_i2d(). This automatically encodes and 13477 adds an extension. Its behaviour can be customised with various 13478 flags to append, replace or delete. Various wrappers added for 13479 certificates and CRLs. 13480 13481 *Steve Henson* 13482 13483 * Fix to avoid calling the underlying ASN1 print routine when 13484 an extension cannot be parsed. Correct a typo in the 13485 OCSP_SERVICELOC extension. Tidy up print OCSP format. 13486 13487 *Steve Henson* 13488 13489 * Make mkdef.pl parse some of the ASN1 macros and add appropriate 13490 entries for variables. 13491 13492 *Steve Henson* 13493 13494 * Add functionality to `apps/openssl.c` for detecting locking 13495 problems: As the program is single-threaded, all we have 13496 to do is register a locking callback using an array for 13497 storing which locks are currently held by the program. 13498 13499 *Bodo Moeller* 13500 13501 * Use a lock around the call to CRYPTO_get_ex_new_index() in 13502 SSL_get_ex_data_X509_STORE_idx(), which is used in 13503 ssl_verify_cert_chain() and thus can be called at any time 13504 during TLS/SSL handshakes so that thread-safety is essential. 13505 Unfortunately, the ex_data design is not at all suited 13506 for multi-threaded use, so it probably should be abolished. 13507 13508 *Bodo Moeller* 13509 13510 * Added Broadcom "ubsec" ENGINE to OpenSSL. 13511 13512 *Broadcom, tweaked and integrated by Geoff Thorpe* 13513 13514 * Move common extension printing code to new function 13515 X509V3_print_extensions(). Reorganise OCSP print routines and 13516 implement some needed OCSP ASN1 functions. Add OCSP extensions. 13517 13518 *Steve Henson* 13519 13520 * New function X509_signature_print() to remove duplication in some 13521 print routines. 13522 13523 *Steve Henson* 13524 13525 * Add a special meaning when SET OF and SEQUENCE OF flags are both 13526 set (this was treated exactly the same as SET OF previously). This 13527 is used to reorder the STACK representing the structure to match the 13528 encoding. This will be used to get round a problem where a PKCS7 13529 structure which was signed could not be verified because the STACK 13530 order did not reflect the encoded order. 13531 13532 *Steve Henson* 13533 13534 * Reimplement the OCSP ASN1 module using the new code. 13535 13536 *Steve Henson* 13537 13538 * Update the X509V3 code to permit the use of an ASN1_ITEM structure 13539 for its ASN1 operations. The old style function pointers still exist 13540 for now but they will eventually go away. 13541 13542 *Steve Henson* 13543 13544 * Merge in replacement ASN1 code from the ASN1 branch. This almost 13545 completely replaces the old ASN1 functionality with a table driven 13546 encoder and decoder which interprets an ASN1_ITEM structure describing 13547 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is 13548 largely maintained. Almost all of the old asn1_mac.h macro based ASN1 13549 has also been converted to the new form. 13550 13551 *Steve Henson* 13552 13553 * Change BN_mod_exp_recp so that negative moduli are tolerated 13554 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set 13555 so that BN_mod_exp_mont and BN_mod_exp_mont_word work 13556 for negative moduli. 13557 13558 *Bodo Moeller* 13559 13560 * Fix BN_uadd and BN_usub: Always return non-negative results instead 13561 of not touching the result's sign bit. 13562 13563 *Bodo Moeller* 13564 13565 * BN_div bugfix: If the result is 0, the sign (res->neg) must not be 13566 set. 13567 13568 *Bodo Moeller* 13569 13570 * Changed the LHASH code to use prototypes for callbacks, and created 13571 macros to declare and implement thin (optionally static) functions 13572 that provide type-safety and avoid function pointer casting for the 13573 type-specific callbacks. 13574 13575 *Geoff Thorpe* 13576 13577 * Added Kerberos Cipher Suites to be used with TLS, as written in 13578 RFC 2712. 13579 *Veers Staats <staatsvr@asc.hpc.mil>, 13580 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte* 13581 13582 * Reformat the FAQ so the different questions and answers can be divided 13583 in sections depending on the subject. 13584 13585 *Richard Levitte* 13586 13587 * Have the zlib compression code load ZLIB.DLL dynamically under 13588 Windows. 13589 13590 *Richard Levitte* 13591 13592 * New function BN_mod_sqrt for computing square roots modulo a prime 13593 (using the probabilistic Tonelli-Shanks algorithm unless 13594 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can 13595 be handled deterministically). 13596 13597 *Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller* 13598 13599 * Make BN_mod_inverse faster by explicitly handling small quotients 13600 in the Euclid loop. (Speed gain about 20% for small moduli [256 or 13601 512 bits], about 30% for larger ones [1024 or 2048 bits].) 13602 13603 *Bodo Moeller* 13604 13605 * New function BN_kronecker. 13606 13607 *Bodo Moeller* 13608 13609 * Fix BN_gcd so that it works on negative inputs; the result is 13610 positive unless both parameters are zero. 13611 Previously something reasonably close to an infinite loop was 13612 possible because numbers could be growing instead of shrinking 13613 in the implementation of Euclid's algorithm. 13614 13615 *Bodo Moeller* 13616 13617 * Fix BN_is_word() and BN_is_one() macros to take into account the 13618 sign of the number in question. 13619 13620 Fix BN_is_word(a,w) to work correctly for w == 0. 13621 13622 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w) 13623 because its test if the absolute value of 'a' equals 'w'. 13624 Note that BN_abs_is_word does *not* handle w == 0 reliably; 13625 it exists mostly for use in the implementations of BN_is_zero(), 13626 BN_is_one(), and BN_is_word(). 13627 13628 *Bodo Moeller* 13629 13630 * New function BN_swap. 13631 13632 *Bodo Moeller* 13633 13634 * Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that 13635 the exponentiation functions are more likely to produce reasonable 13636 results on negative inputs. 13637 13638 *Bodo Moeller* 13639 13640 * Change BN_mod_mul so that the result is always non-negative. 13641 Previously, it could be negative if one of the factors was negative; 13642 I don't think anyone really wanted that behaviour. 13643 13644 *Bodo Moeller* 13645 13646 * Move `BN_mod_...` functions into new file `crypto/bn/bn_mod.c` 13647 (except for exponentiation, which stays in `crypto/bn/bn_exp.c`, 13648 and `BN_mod_mul_reciprocal`, which stays in `crypto/bn/bn_recp.c`) 13649 and add new functions: 13650 13651 BN_nnmod 13652 BN_mod_sqr 13653 BN_mod_add 13654 BN_mod_add_quick 13655 BN_mod_sub 13656 BN_mod_sub_quick 13657 BN_mod_lshift1 13658 BN_mod_lshift1_quick 13659 BN_mod_lshift 13660 BN_mod_lshift_quick 13661 13662 These functions always generate non-negative results. 13663 13664 `BN_nnmod` otherwise is `like BN_mod` (if `BN_mod` computes a remainder `r` 13665 such that `|m| < r < 0`, `BN_nnmod` will output `rem + |m|` instead). 13666 13667 `BN_mod_XXX_quick(r, a, [b,] m)` generates the same result as 13668 `BN_mod_XXX(r, a, [b,] m, ctx)`, but requires that `a` [and `b`] 13669 be reduced modulo `m`. 13670 13671 *Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller* 13672 13673<!-- 13674 The following entry accidentally appeared in the CHANGES file 13675 distributed with OpenSSL 0.9.7. The modifications described in 13676 it do *not* apply to OpenSSL 0.9.7. 13677 13678 * Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 13679 was actually never needed) and in BN_mul(). The removal in BN_mul() 13680 required a small change in bn_mul_part_recursive() and the addition 13681 of the functions bn_cmp_part_words(), bn_sub_part_words() and 13682 bn_add_part_words(), which do the same thing as bn_cmp_words(), 13683 bn_sub_words() and bn_add_words() except they take arrays with 13684 differing sizes. 13685 13686 *Richard Levitte* 13687--> 13688 13689 * In 'openssl passwd', verify passwords read from the terminal 13690 unless the '-salt' option is used (which usually means that 13691 verification would just waste user's time since the resulting 13692 hash is going to be compared with some given password hash) 13693 or the new '-noverify' option is used. 13694 13695 This is an incompatible change, but it does not affect 13696 non-interactive use of 'openssl passwd' (passwords on the command 13697 line, '-stdin' option, '-in ...' option) and thus should not 13698 cause any problems. 13699 13700 *Bodo Moeller* 13701 13702 * Remove all references to RSAref, since there's no more need for it. 13703 13704 *Richard Levitte* 13705 13706 * Make DSO load along a path given through an environment variable 13707 (SHLIB_PATH) with shl_load(). 13708 13709 *Richard Levitte* 13710 13711 * Constify the ENGINE code as a result of BIGNUM constification. 13712 Also constify the RSA code and most things related to it. In a 13713 few places, most notable in the depth of the ASN.1 code, ugly 13714 casts back to non-const were required (to be solved at a later 13715 time) 13716 13717 *Richard Levitte* 13718 13719 * Make it so the openssl application has all engines loaded by default. 13720 13721 *Richard Levitte* 13722 13723 * Constify the BIGNUM routines a little more. 13724 13725 *Richard Levitte* 13726 13727 * Add the following functions: 13728 13729 ENGINE_load_cswift() 13730 ENGINE_load_chil() 13731 ENGINE_load_atalla() 13732 ENGINE_load_nuron() 13733 ENGINE_load_builtin_engines() 13734 13735 That way, an application can itself choose if external engines that 13736 are built-in in OpenSSL shall ever be used or not. The benefit is 13737 that applications won't have to be linked with libdl or other dso 13738 libraries unless it's really needed. 13739 13740 Changed 'openssl engine' to load all engines on demand. 13741 Changed the engine header files to avoid the duplication of some 13742 declarations (they differed!). 13743 13744 *Richard Levitte* 13745 13746 * 'openssl engine' can now list capabilities. 13747 13748 *Richard Levitte* 13749 13750 * Better error reporting in 'openssl engine'. 13751 13752 *Richard Levitte* 13753 13754 * Never call load_dh_param(NULL) in s_server. 13755 13756 *Bodo Moeller* 13757 13758 * Add engine application. It can currently list engines by name and 13759 identity, and test if they are actually available. 13760 13761 *Richard Levitte* 13762 13763 * Improve RPM specification file by forcing symbolic linking and making 13764 sure the installed documentation is also owned by root.root. 13765 13766 *Damien Miller <djm@mindrot.org>* 13767 13768 * Give the OpenSSL applications more possibilities to make use of 13769 keys (public as well as private) handled by engines. 13770 13771 *Richard Levitte* 13772 13773 * Add OCSP code that comes from CertCo. 13774 13775 *Richard Levitte* 13776 13777 * Add VMS support for the Rijndael code. 13778 13779 *Richard Levitte* 13780 13781 * Added untested support for Nuron crypto accelerator. 13782 13783 *Ben Laurie* 13784 13785 * Add support for external cryptographic devices. This code was 13786 previously distributed separately as the "engine" branch. 13787 13788 *Geoff Thorpe, Richard Levitte* 13789 13790 * Rework the filename-translation in the DSO code. It is now possible to 13791 have far greater control over how a "name" is turned into a filename 13792 depending on the operating environment and any oddities about the 13793 different shared library filenames on each system. 13794 13795 *Geoff Thorpe* 13796 13797 * Support threads on FreeBSD-elf in Configure. 13798 13799 *Richard Levitte* 13800 13801 * Fix for SHA1 assembly problem with MASM: it produces 13802 warnings about corrupt line number information when assembling 13803 with debugging information. This is caused by the overlapping 13804 of two sections. 13805 13806 *Bernd Matthes <mainbug@celocom.de>, Steve Henson* 13807 13808 * NCONF changes. 13809 NCONF_get_number() has no error checking at all. As a replacement, 13810 NCONF_get_number_e() is defined (`_e` for "error checking") and is 13811 promoted strongly. The old NCONF_get_number is kept around for 13812 binary backward compatibility. 13813 Make it possible for methods to load from something other than a BIO, 13814 by providing a function pointer that is given a name instead of a BIO. 13815 For example, this could be used to load configuration data from an 13816 LDAP server. 13817 13818 *Richard Levitte* 13819 13820 * Fix for non blocking accept BIOs. Added new I/O special reason 13821 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs 13822 with non blocking I/O was not possible because no retry code was 13823 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover 13824 this case. 13825 13826 *Steve Henson* 13827 13828 * Added the beginnings of Rijndael support. 13829 13830 *Ben Laurie* 13831 13832 * Fix for bug in DirectoryString mask setting. Add support for 13833 X509_NAME_print_ex() in 'req' and X509_print_ex() function 13834 to allow certificate printing to more controllable, additional 13835 'certopt' option to 'x509' to allow new printing options to be 13836 set. 13837 13838 *Steve Henson* 13839 13840 * Clean old EAY MD5 hack from e_os.h. 13841 13842 *Richard Levitte* 13843 13844### Changes between 0.9.6l and 0.9.6m [17 Mar 2004] 13845 13846 * Fix null-pointer assignment in do_change_cipher_spec() revealed 13847 by using the Codenomicon TLS Test Tool ([CVE-2004-0079]) 13848 13849 *Joe Orton, Steve Henson* 13850 13851### Changes between 0.9.6k and 0.9.6l [04 Nov 2003] 13852 13853 * Fix additional bug revealed by the NISCC test suite: 13854 13855 Stop bug triggering large recursion when presented with 13856 certain ASN.1 tags ([CVE-2003-0851]) 13857 13858 *Steve Henson* 13859 13860### Changes between 0.9.6j and 0.9.6k [30 Sep 2003] 13861 13862 * Fix various bugs revealed by running the NISCC test suite: 13863 13864 Stop out of bounds reads in the ASN1 code when presented with 13865 invalid tags (CVE-2003-0543 and CVE-2003-0544). 13866 13867 If verify callback ignores invalid public key errors don't try to check 13868 certificate signature with the NULL public key. 13869 13870 *Steve Henson* 13871 13872 * In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 13873 if the server requested one: as stated in TLS 1.0 and SSL 3.0 13874 specifications. 13875 13876 *Steve Henson* 13877 13878 * In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 13879 extra data after the compression methods not only for TLS 1.0 13880 but also for SSL 3.0 (as required by the specification). 13881 13882 *Bodo Moeller; problem pointed out by Matthias Loepfe* 13883 13884 * Change X509_certificate_type() to mark the key as exported/exportable 13885 when it's 512 *bits* long, not 512 bytes. 13886 13887 *Richard Levitte* 13888 13889### Changes between 0.9.6i and 0.9.6j [10 Apr 2003] 13890 13891 * Countermeasure against the Klima-Pokorny-Rosa extension of 13892 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 13893 a protocol version number mismatch like a decryption error 13894 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 13895 13896 *Bodo Moeller* 13897 13898 * Turn on RSA blinding by default in the default implementation 13899 to avoid a timing attack. Applications that don't want it can call 13900 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 13901 They would be ill-advised to do so in most cases. 13902 13903 *Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller* 13904 13905 * Change RSA blinding code so that it works when the PRNG is not 13906 seeded (in this case, the secret RSA exponent is abused as 13907 an unpredictable seed -- if it is not unpredictable, there 13908 is no point in blinding anyway). Make RSA blinding thread-safe 13909 by remembering the creator's thread ID in rsa->blinding and 13910 having all other threads use local one-time blinding factors 13911 (this requires more computation than sharing rsa->blinding, but 13912 avoids excessive locking; and if an RSA object is not shared 13913 between threads, blinding will still be very fast). 13914 13915 *Bodo Moeller* 13916 13917### Changes between 0.9.6h and 0.9.6i [19 Feb 2003] 13918 13919 * In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 13920 via timing by performing a MAC computation even if incorrect 13921 block cipher padding has been found. This is a countermeasure 13922 against active attacks where the attacker has to distinguish 13923 between bad padding and a MAC verification error. ([CVE-2003-0078]) 13924 13925 *Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 13926 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 13927 Martin Vuagnoux (EPFL, Ilion)* 13928 13929### Changes between 0.9.6g and 0.9.6h [5 Dec 2002] 13930 13931 * New function OPENSSL_cleanse(), which is used to cleanse a section of 13932 memory from its contents. This is done with a counter that will 13933 place alternating values in each byte. This can be used to solve 13934 two issues: 1) the removal of calls to memset() by highly optimizing 13935 compilers, and 2) cleansing with other values than 0, since those can 13936 be read through on certain media, for example a swap space on disk. 13937 13938 *Geoff Thorpe* 13939 13940 * Bugfix: client side session caching did not work with external caching, 13941 because the session->cipher setting was not restored when reloading 13942 from the external cache. This problem was masked, when 13943 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set. 13944 (Found by Steve Haslam <steve@araqnid.ddts.net>.) 13945 13946 *Lutz Jaenicke* 13947 13948 * Fix client_certificate (ssl/s2_clnt.c): The permissible total 13949 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33. 13950 13951 *Zeev Lieber <zeev-l@yahoo.com>* 13952 13953 * Undo an undocumented change introduced in 0.9.6e which caused 13954 repeated calls to OpenSSL_add_all_ciphers() and 13955 OpenSSL_add_all_digests() to be ignored, even after calling 13956 EVP_cleanup(). 13957 13958 *Richard Levitte* 13959 13960 * Change the default configuration reader to deal with last line not 13961 being properly terminated. 13962 13963 *Richard Levitte* 13964 13965 * Change X509_NAME_cmp() so it applies the special rules on handling 13966 DN values that are of type PrintableString, as well as RDNs of type 13967 emailAddress where the value has the type ia5String. 13968 13969 *stefank@valicert.com via Richard Levitte* 13970 13971 * Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half 13972 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently 13973 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be 13974 the bitwise-OR of the two for use by the majority of applications 13975 wanting this behaviour, and update the docs. The documented 13976 behaviour and actual behaviour were inconsistent and had been 13977 changing anyway, so this is more a bug-fix than a behavioural 13978 change. 13979 13980 *Geoff Thorpe, diagnosed by Nadav Har'El* 13981 13982 * Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c 13983 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes). 13984 13985 *Bodo Moeller* 13986 13987 * Fix initialization code race conditions in 13988 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(), 13989 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(), 13990 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(), 13991 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(), 13992 ssl2_get_cipher_by_char(), 13993 ssl3_get_cipher_by_char(). 13994 13995 *Patrick McCormick <patrick@tellme.com>, Bodo Moeller* 13996 13997 * Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after 13998 the cached sessions are flushed, as the remove_cb() might use ex_data 13999 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com> 14000 (see [openssl.org #212]). 14001 14002 *Geoff Thorpe, Lutz Jaenicke* 14003 14004 * Fix typo in OBJ_txt2obj which incorrectly passed the content 14005 length, instead of the encoding length to d2i_ASN1_OBJECT. 14006 14007 *Steve Henson* 14008 14009### Changes between 0.9.6f and 0.9.6g [9 Aug 2002] 14010 14011 * [In 0.9.6g-engine release:] 14012 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use `_stdcall`). 14013 14014 *Lynn Gazis <lgazis@rainbow.com>* 14015 14016### Changes between 0.9.6e and 0.9.6f [8 Aug 2002] 14017 14018 * Fix ASN1 checks. Check for overflow by comparing with LONG_MAX 14019 and get fix the header length calculation. 14020 *Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>, 14021 Alon Kantor <alonk@checkpoint.com> (and others), Steve Henson* 14022 14023 * Use proper error handling instead of 'assertions' in buffer 14024 overflow checks added in 0.9.6e. This prevents DoS (the 14025 assertions could call abort()). 14026 14027 *Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller* 14028 14029### Changes between 0.9.6d and 0.9.6e [30 Jul 2002] 14030 14031 * Add various sanity checks to asn1_get_length() to reject 14032 the ASN1 length bytes if they exceed sizeof(long), will appear 14033 negative or the content length exceeds the length of the 14034 supplied buffer. 14035 14036 *Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>* 14037 14038 * Fix cipher selection routines: ciphers without encryption had no flags 14039 for the cipher strength set and where therefore not handled correctly 14040 by the selection routines (PR #130). 14041 14042 *Lutz Jaenicke* 14043 14044 * Fix EVP_dsa_sha macro. 14045 14046 *Nils Larsch* 14047 14048 * New option 14049 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS 14050 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure 14051 that was added in OpenSSL 0.9.6d. 14052 14053 As the countermeasure turned out to be incompatible with some 14054 broken SSL implementations, the new option is part of SSL_OP_ALL. 14055 SSL_OP_ALL is usually employed when compatibility with weird SSL 14056 implementations is desired (e.g. '-bugs' option to 's_client' and 14057 's_server'), so the new option is automatically set in many 14058 applications. 14059 14060 *Bodo Moeller* 14061 14062 * Changes in security patch: 14063 14064 Changes marked "(CHATS)" were sponsored by the Defense Advanced 14065 Research Projects Agency (DARPA) and Air Force Research Laboratory, 14066 Air Force Materiel Command, USAF, under agreement number 14067 F30602-01-2-0537. 14068 14069 * Add various sanity checks to asn1_get_length() to reject 14070 the ASN1 length bytes if they exceed sizeof(long), will appear 14071 negative or the content length exceeds the length of the 14072 supplied buffer. ([CVE-2002-0659]) 14073 14074 *Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>* 14075 14076 * Assertions for various potential buffer overflows, not known to 14077 happen in practice. 14078 14079 *Ben Laurie (CHATS)* 14080 14081 * Various temporary buffers to hold ASCII versions of integers were 14082 too small for 64 bit platforms. ([CVE-2002-0655]) 14083 *Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)>* 14084 14085 * Remote buffer overflow in SSL3 protocol - an attacker could 14086 supply an oversized session ID to a client. ([CVE-2002-0656]) 14087 14088 *Ben Laurie (CHATS)* 14089 14090 * Remote buffer overflow in SSL2 protocol - an attacker could 14091 supply an oversized client master key. ([CVE-2002-0656]) 14092 14093 *Ben Laurie (CHATS)* 14094 14095### Changes between 0.9.6c and 0.9.6d [9 May 2002] 14096 14097 * Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not 14098 encoded as NULL) with id-dsa-with-sha1. 14099 14100 *Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller* 14101 14102 * Check various `X509_...()` return values in `apps/req.c`. 14103 14104 *Nils Larsch <nla@trustcenter.de>* 14105 14106 * Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines: 14107 an end-of-file condition would erroneously be flagged, when the CRLF 14108 was just at the end of a processed block. The bug was discovered when 14109 processing data through a buffering memory BIO handing the data to a 14110 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov 14111 <ptsekov@syntrex.com> and Nedelcho Stanev. 14112 14113 *Lutz Jaenicke* 14114 14115 * Implement a countermeasure against a vulnerability recently found 14116 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment 14117 before application data chunks to avoid the use of known IVs 14118 with data potentially chosen by the attacker. 14119 14120 *Bodo Moeller* 14121 14122 * Fix length checks in ssl3_get_client_hello(). 14123 14124 *Bodo Moeller* 14125 14126 * TLS/SSL library bugfix: use s->s3->in_read_app_data differently 14127 to prevent ssl3_read_internal() from incorrectly assuming that 14128 ssl3_read_bytes() found application data while handshake 14129 processing was enabled when in fact s->s3->in_read_app_data was 14130 merely automatically cleared during the initial handshake. 14131 14132 *Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>* 14133 14134 * Fix object definitions for Private and Enterprise: they were not 14135 recognized in their shortname (=lowercase) representation. Extend 14136 obj_dat.pl to issue an error when using undefined keywords instead 14137 of silently ignoring the problem (Svenning Sorensen 14138 <sss@sss.dnsalias.net>). 14139 14140 *Lutz Jaenicke* 14141 14142 * Fix DH_generate_parameters() so that it works for 'non-standard' 14143 generators, i.e. generators other than 2 and 5. (Previously, the 14144 code did not properly initialise the 'add' and 'rem' values to 14145 BN_generate_prime().) 14146 14147 In the new general case, we do not insist that 'generator' is 14148 actually a primitive root: This requirement is rather pointless; 14149 a generator of the order-q subgroup is just as good, if not 14150 better. 14151 14152 *Bodo Moeller* 14153 14154 * Map new X509 verification errors to alerts. Discovered and submitted by 14155 Tom Wu <tom@arcot.com>. 14156 14157 *Lutz Jaenicke* 14158 14159 * Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from 14160 returning non-zero before the data has been completely received 14161 when using non-blocking I/O. 14162 14163 *Bodo Moeller; problem pointed out by John Hughes* 14164 14165 * Some of the ciphers missed the strength entry (SSL_LOW etc). 14166 14167 *Ben Laurie, Lutz Jaenicke* 14168 14169 * Fix bug in SSL_clear(): bad sessions were not removed (found by 14170 Yoram Zahavi <YoramZ@gilian.com>). 14171 14172 *Lutz Jaenicke* 14173 14174 * Add information about CygWin 1.3 and on, and preserve proper 14175 configuration for the versions before that. 14176 14177 *Corinna Vinschen <vinschen@redhat.com> and Richard Levitte* 14178 14179 * Make removal from session cache (SSL_CTX_remove_session()) more robust: 14180 check whether we deal with a copy of a session and do not delete from 14181 the cache in this case. Problem reported by "Izhar Shoshani Levi" 14182 <izhar@checkpoint.com>. 14183 14184 *Lutz Jaenicke* 14185 14186 * Do not store session data into the internal session cache, if it 14187 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP 14188 flag is set). Proposed by Aslam <aslam@funk.com>. 14189 14190 *Lutz Jaenicke* 14191 14192 * Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested 14193 value is 0. 14194 14195 *Richard Levitte* 14196 14197 * [In 0.9.6d-engine release:] 14198 Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 14199 14200 *Toomas Kiisk <vix@cyber.ee> via Richard Levitte* 14201 14202 * Add the configuration target linux-s390x. 14203 14204 *Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte* 14205 14206 * The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of 14207 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag 14208 variable as an indication that a ClientHello message has been 14209 received. As the flag value will be lost between multiple 14210 invocations of ssl3_accept when using non-blocking I/O, the 14211 function may not be aware that a handshake has actually taken 14212 place, thus preventing a new session from being added to the 14213 session cache. 14214 14215 To avoid this problem, we now set s->new_session to 2 instead of 14216 using a local variable. 14217 14218 *Lutz Jaenicke, Bodo Moeller* 14219 14220 * Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c) 14221 if the SSL_R_LENGTH_MISMATCH error is detected. 14222 14223 *Geoff Thorpe, Bodo Moeller* 14224 14225 * New 'shared_ldflag' column in Configure platform table. 14226 14227 *Richard Levitte* 14228 14229 * Fix EVP_CIPHER_mode macro. 14230 14231 *"Dan S. Camper" <dan@bti.net>* 14232 14233 * Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown 14234 type, we must throw them away by setting rr->length to 0. 14235 14236 *D P Chang <dpc@qualys.com>* 14237 14238### Changes between 0.9.6b and 0.9.6c [21 dec 2001] 14239 14240 * Fix BN_rand_range bug pointed out by Dominikus Scherkl 14241 <Dominikus.Scherkl@biodata.com>. (The previous implementation 14242 worked incorrectly for those cases where range = `10..._2` and 14243 `3*range` is two bits longer than range.) 14244 14245 *Bodo Moeller* 14246 14247 * Only add signing time to PKCS7 structures if it is not already 14248 present. 14249 14250 *Steve Henson* 14251 14252 * Fix crypto/objects/objects.h: "ld-ce" should be "id-ce", 14253 OBJ_ld_ce should be OBJ_id_ce. 14254 Also some ip-pda OIDs in crypto/objects/objects.txt were 14255 incorrect (cf. RFC 3039). 14256 14257 *Matt Cooper, Frederic Giudicelli, Bodo Moeller* 14258 14259 * Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid() 14260 returns early because it has nothing to do. 14261 14262 *Andy Schneider <andy.schneider@bjss.co.uk>* 14263 14264 * [In 0.9.6c-engine release:] 14265 Fix mutex callback return values in crypto/engine/hw_ncipher.c. 14266 14267 *Andy Schneider <andy.schneider@bjss.co.uk>* 14268 14269 * [In 0.9.6c-engine release:] 14270 Add support for Cryptographic Appliance's keyserver technology. 14271 (Use engine 'keyclient') 14272 14273 *Cryptographic Appliances and Geoff Thorpe* 14274 14275 * Add a configuration entry for OS/390 Unix. The C compiler 'c89' 14276 is called via tools/c89.sh because arguments have to be 14277 rearranged (all '-L' options must appear before the first object 14278 modules). 14279 14280 *Richard Shapiro <rshapiro@abinitio.com>* 14281 14282 * [In 0.9.6c-engine release:] 14283 Add support for Broadcom crypto accelerator cards, backported 14284 from 0.9.7. 14285 14286 *Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox* 14287 14288 * [In 0.9.6c-engine release:] 14289 Add support for SureWare crypto accelerator cards from 14290 Baltimore Technologies. (Use engine 'sureware') 14291 14292 *Baltimore Technologies and Mark Cox* 14293 14294 * [In 0.9.6c-engine release:] 14295 Add support for crypto accelerator cards from Accelerated 14296 Encryption Processing, www.aep.ie. (Use engine 'aep') 14297 14298 *AEP Inc. and Mark Cox* 14299 14300 * Add a configuration entry for gcc on UnixWare. 14301 14302 *Gary Benson <gbenson@redhat.com>* 14303 14304 * Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake 14305 messages are stored in a single piece (fixed-length part and 14306 variable-length part combined) and fix various bugs found on the way. 14307 14308 *Bodo Moeller* 14309 14310 * Disable caching in BIO_gethostbyname(), directly use gethostbyname() 14311 instead. BIO_gethostbyname() does not know what timeouts are 14312 appropriate, so entries would stay in cache even when they have 14313 become invalid. 14314 *Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com>* 14315 14316 * Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when 14317 faced with a pathologically small ClientHello fragment that does 14318 not contain client_version: Instead of aborting with an error, 14319 simply choose the highest available protocol version (i.e., 14320 TLS 1.0 unless it is disabled). In practice, ClientHello 14321 messages are never sent like this, but this change gives us 14322 strictly correct behaviour at least for TLS. 14323 14324 *Bodo Moeller* 14325 14326 * Fix SSL handshake functions and SSL_clear() such that SSL_clear() 14327 never resets s->method to s->ctx->method when called from within 14328 one of the SSL handshake functions. 14329 14330 *Bodo Moeller; problem pointed out by Niko Baric* 14331 14332 * In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert 14333 (sent using the client's version number) if client_version is 14334 smaller than the protocol version in use. Also change 14335 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if 14336 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then 14337 the client will at least see that alert. 14338 14339 *Bodo Moeller* 14340 14341 * Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation 14342 correctly. 14343 14344 *Bodo Moeller* 14345 14346 * Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a 14347 client receives HelloRequest while in a handshake. 14348 14349 *Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>* 14350 14351 * Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C 14352 should end in 'break', not 'goto end' which circumvents various 14353 cleanups done in state SSL_ST_OK. But session related stuff 14354 must be disabled for SSL_ST_OK in the case that we just sent a 14355 HelloRequest. 14356 14357 Also avoid some overhead by not calling ssl_init_wbio_buffer() 14358 before just sending a HelloRequest. 14359 14360 *Bodo Moeller, Eric Rescorla <ekr@rtfm.com>* 14361 14362 * Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't 14363 reveal whether illegal block cipher padding was found or a MAC 14364 verification error occurred. (Neither SSLerr() codes nor alerts 14365 are directly visible to potential attackers, but the information 14366 may leak via logfiles.) 14367 14368 Similar changes are not required for the SSL 2.0 implementation 14369 because the number of padding bytes is sent in clear for SSL 2.0, 14370 and the extra bytes are just ignored. However ssl/s2_pkt.c 14371 failed to verify that the purported number of padding bytes is in 14372 the legal range. 14373 14374 *Bodo Moeller* 14375 14376 * Add OpenUNIX-8 support including shared libraries 14377 (Boyd Lynn Gerber <gerberb@zenez.com>). 14378 14379 *Lutz Jaenicke* 14380 14381 * Improve RSA_padding_check_PKCS1_OAEP() check again to avoid 14382 'wristwatch attack' using huge encoding parameters (cf. 14383 James H. Manger's CRYPTO 2001 paper). Note that the 14384 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use 14385 encoding parameters and hence was not vulnerable. 14386 14387 *Bodo Moeller* 14388 14389 * BN_sqr() bug fix. 14390 14391 *Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>* 14392 14393 * Rabin-Miller test analyses assume uniformly distributed witnesses, 14394 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand() 14395 followed by modular reduction. 14396 14397 *Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>* 14398 14399 * Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range() 14400 equivalent based on BN_pseudo_rand() instead of BN_rand(). 14401 14402 *Bodo Moeller* 14403 14404 * s3_srvr.c: allow sending of large client certificate lists (> 16 kB). 14405 This function was broken, as the check for a new client hello message 14406 to handle SGC did not allow these large messages. 14407 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.) 14408 14409 *Lutz Jaenicke* 14410 14411 * Add alert descriptions for TLSv1 to `SSL_alert_desc_string[_long]()`. 14412 14413 *Lutz Jaenicke* 14414 14415 * Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl() 14416 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>). 14417 14418 *Lutz Jaenicke* 14419 14420 * Rework the configuration and shared library support for Tru64 Unix. 14421 The configuration part makes use of modern compiler features and 14422 still retains old compiler behavior for those that run older versions 14423 of the OS. The shared library support part includes a variant that 14424 uses the RPATH feature, and is available through the special 14425 configuration target "alpha-cc-rpath", which will never be selected 14426 automatically. 14427 14428 *Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte* 14429 14430 * In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message() 14431 with the same message size as in ssl3_get_certificate_request(). 14432 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest 14433 messages might inadvertently be reject as too long. 14434 14435 *Petr Lampa <lampa@fee.vutbr.cz>* 14436 14437 * Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX). 14438 14439 *Andy Polyakov* 14440 14441 * Modified SSL library such that the verify_callback that has been set 14442 specifically for an SSL object with SSL_set_verify() is actually being 14443 used. Before the change, a verify_callback set with this function was 14444 ignored and the verify_callback() set in the SSL_CTX at the time of 14445 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced 14446 to allow the necessary settings. 14447 14448 *Lutz Jaenicke* 14449 14450 * Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c 14451 explicitly to NULL, as at least on Solaris 8 this seems not always to be 14452 done automatically (in contradiction to the requirements of the C 14453 standard). This made problems when used from OpenSSH. 14454 14455 *Lutz Jaenicke* 14456 14457 * In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored 14458 dh->length and always used 14459 14460 BN_rand_range(priv_key, dh->p). 14461 14462 BN_rand_range() is not necessary for Diffie-Hellman, and this 14463 specific range makes Diffie-Hellman unnecessarily inefficient if 14464 dh->length (recommended exponent length) is much smaller than the 14465 length of dh->p. We could use BN_rand_range() if the order of 14466 the subgroup was stored in the DH structure, but we only have 14467 dh->length. 14468 14469 So switch back to 14470 14471 BN_rand(priv_key, l, ...) 14472 14473 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1 14474 otherwise. 14475 14476 *Bodo Moeller* 14477 14478 * In 14479 14480 RSA_eay_public_encrypt 14481 RSA_eay_private_decrypt 14482 RSA_eay_private_encrypt (signing) 14483 RSA_eay_public_decrypt (signature verification) 14484 14485 (default implementations for RSA_public_encrypt, 14486 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt), 14487 always reject numbers >= n. 14488 14489 *Bodo Moeller* 14490 14491 * In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2 14492 to synchronize access to 'locking_thread'. This is necessary on 14493 systems where access to 'locking_thread' (an 'unsigned long' 14494 variable) is not atomic. 14495 14496 *Bodo Moeller* 14497 14498 * In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID 14499 *before* setting the 'crypto_lock_rand' flag. The previous code had 14500 a race condition if 0 is a valid thread ID. 14501 14502 *Travis Vitek <vitek@roguewave.com>* 14503 14504 * Add support for shared libraries under Irix. 14505 14506 *Albert Chin-A-Young <china@thewrittenword.com>* 14507 14508 * Add configuration option to build on Linux on both big-endian and 14509 little-endian MIPS. 14510 14511 *Ralf Baechle <ralf@uni-koblenz.de>* 14512 14513 * Add the possibility to create shared libraries on HP-UX. 14514 14515 *Richard Levitte* 14516 14517### Changes between 0.9.6a and 0.9.6b [9 Jul 2001] 14518 14519 * Change ssleay_rand_bytes (crypto/rand/md_rand.c) 14520 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by 14521 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>: 14522 PRNG state recovery was possible based on the output of 14523 one PRNG request appropriately sized to gain knowledge on 14524 'md' followed by enough consecutive 1-byte PRNG requests 14525 to traverse all of 'state'. 14526 14527 1. When updating 'md_local' (the current thread's copy of 'md') 14528 during PRNG output generation, hash all of the previous 14529 'md_local' value, not just the half used for PRNG output. 14530 14531 2. Make the number of bytes from 'state' included into the hash 14532 independent from the number of PRNG bytes requested. 14533 14534 The first measure alone would be sufficient to avoid 14535 Markku-Juhani's attack. (Actually it had never occurred 14536 to me that the half of 'md_local' used for chaining was the 14537 half from which PRNG output bytes were taken -- I had always 14538 assumed that the secret half would be used.) The second 14539 measure makes sure that additional data from 'state' is never 14540 mixed into 'md_local' in small portions; this heuristically 14541 further strengthens the PRNG. 14542 14543 *Bodo Moeller* 14544 14545 * Fix crypto/bn/asm/mips3.s. 14546 14547 *Andy Polyakov* 14548 14549 * When only the key is given to "enc", the IV is undefined. Print out 14550 an error message in this case. 14551 14552 *Lutz Jaenicke* 14553 14554 * Handle special case when X509_NAME is empty in X509 printing routines. 14555 14556 *Steve Henson* 14557 14558 * In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are 14559 positive and less than q. 14560 14561 *Bodo Moeller* 14562 14563 * Don't change `*pointer` in CRYPTO_add_lock() is add_lock_callback is 14564 used: it isn't thread safe and the add_lock_callback should handle 14565 that itself. 14566 14567 *Paul Rose <Paul.Rose@bridge.com>* 14568 14569 * Verify that incoming data obeys the block size in 14570 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c). 14571 14572 *Bodo Moeller* 14573 14574 * Fix OAEP check. 14575 14576 *Ulf Möller, Bodo Möller* 14577 14578 * The countermeasure against Bleichbacher's attack on PKCS #1 v1.5 14579 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5 14580 when fixing the server behaviour for backwards-compatible 'client 14581 hello' messages. (Note that the attack is impractical against 14582 SSL 3.0 and TLS 1.0 anyway because length and version checking 14583 means that the probability of guessing a valid ciphertext is 14584 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98 14585 paper.) 14586 14587 Before 0.9.5, the countermeasure (hide the error by generating a 14588 random 'decryption result') did not work properly because 14589 ERR_clear_error() was missing, meaning that SSL_get_error() would 14590 detect the supposedly ignored error. 14591 14592 Both problems are now fixed. 14593 14594 *Bodo Moeller* 14595 14596 * In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096 14597 (previously it was 1024). 14598 14599 *Bodo Moeller* 14600 14601 * Fix for compatibility mode trust settings: ignore trust settings 14602 unless some valid trust or reject settings are present. 14603 14604 *Steve Henson* 14605 14606 * Fix for blowfish EVP: its a variable length cipher. 14607 14608 *Steve Henson* 14609 14610 * Fix various bugs related to DSA S/MIME verification. Handle missing 14611 parameters in DSA public key structures and return an error in the 14612 DSA routines if parameters are absent. 14613 14614 *Steve Henson* 14615 14616 * In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd" 14617 in the current directory if neither $RANDFILE nor $HOME was set. 14618 RAND_file_name() in 0.9.6a returned NULL in this case. This has 14619 caused some confusion to Windows users who haven't defined $HOME. 14620 Thus RAND_file_name() is changed again: e_os.h can define a 14621 DEFAULT_HOME, which will be used if $HOME is not set. 14622 For Windows, we use "C:"; on other platforms, we still require 14623 environment variables. 14624 14625 * Move 'if (!initialized) RAND_poll()' into regions protected by 14626 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids 14627 having multiple threads call RAND_poll() concurrently. 14628 14629 *Bodo Moeller* 14630 14631 * In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a 14632 combination of a flag and a thread ID variable. 14633 Otherwise while one thread is in ssleay_rand_bytes (which sets the 14634 flag), *other* threads can enter ssleay_add_bytes without obeying 14635 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock 14636 that they do not hold after the first thread unsets add_do_not_lock). 14637 14638 *Bodo Moeller* 14639 14640 * Change bctest again: '-x' expressions are not available in all 14641 versions of 'test'. 14642 14643 *Bodo Moeller* 14644 14645### Changes between 0.9.6 and 0.9.6a [5 Apr 2001] 14646 14647 * Fix a couple of memory leaks in PKCS7_dataDecode() 14648 14649 *Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>* 14650 14651 * Change Configure and Makefiles to provide EXE_EXT, which will contain 14652 the default extension for executables, if any. Also, make the perl 14653 scripts that use symlink() to test if it really exists and use "cp" 14654 if it doesn't. All this made OpenSSL compilable and installable in 14655 CygWin. 14656 14657 *Richard Levitte* 14658 14659 * Fix for asn1_GetSequence() for indefinite length constructed data. 14660 If SEQUENCE is length is indefinite just set c->slen to the total 14661 amount of data available. 14662 14663 *Steve Henson, reported by shige@FreeBSD.org* 14664 14665 *This change does not apply to 0.9.7.* 14666 14667 * Change bctest to avoid here-documents inside command substitution 14668 (workaround for FreeBSD /bin/sh bug). 14669 For compatibility with Ultrix, avoid shell functions (introduced 14670 in the bctest version that searches along $PATH). 14671 14672 *Bodo Moeller* 14673 14674 * Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes 14675 with des_encrypt() defined on some operating systems, like Solaris 14676 and UnixWare. 14677 14678 *Richard Levitte* 14679 14680 * Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton: 14681 On the Importance of Eliminating Errors in Cryptographic 14682 Computations, J. Cryptology 14 (2001) 2, 101-119, 14683 <http://theory.stanford.edu/~dabo/papers/faults.ps.gz>). 14684 14685 *Ulf Moeller* 14686 14687 * MIPS assembler BIGNUM division bug fix. 14688 14689 *Andy Polyakov* 14690 14691 * Disabled incorrect Alpha assembler code. 14692 14693 *Richard Levitte* 14694 14695 * Fix PKCS#7 decode routines so they correctly update the length 14696 after reading an EOC for the EXPLICIT tag. 14697 14698 *Steve Henson* 14699 14700 *This change does not apply to 0.9.7.* 14701 14702 * Fix bug in PKCS#12 key generation routines. This was triggered 14703 if a 3DES key was generated with a 0 initial byte. Include 14704 PKCS12_BROKEN_KEYGEN compilation option to retain the old 14705 (but broken) behaviour. 14706 14707 *Steve Henson* 14708 14709 * Enhance bctest to search for a working bc along $PATH and print 14710 it when found. 14711 14712 *Tim Rice <tim@multitalents.net> via Richard Levitte* 14713 14714 * Fix memory leaks in err.c: free err_data string if necessary; 14715 don't write to the wrong index in ERR_set_error_data. 14716 14717 *Bodo Moeller* 14718 14719 * Implement ssl23_peek (analogous to ssl23_read), which previously 14720 did not exist. 14721 14722 *Bodo Moeller* 14723 14724 * Replace rdtsc with `_emit` statements for VC++ version 5. 14725 14726 *Jeremy Cooper <jeremy@baymoo.org>* 14727 14728 * Make it possible to reuse SSLv2 sessions. 14729 14730 *Richard Levitte* 14731 14732 * In copy_email() check for >= 0 as a return value for 14733 X509_NAME_get_index_by_NID() since 0 is a valid index. 14734 14735 *Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>* 14736 14737 * Avoid coredump with unsupported or invalid public keys by checking if 14738 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when 14739 PKCS7_verify() fails with non detached data. 14740 14741 *Steve Henson* 14742 14743 * Don't use getenv in library functions when run as setuid/setgid. 14744 New function OPENSSL_issetugid(). 14745 14746 *Ulf Moeller* 14747 14748 * Avoid false positives in memory leak detection code (crypto/mem_dbg.c) 14749 due to incorrect handling of multi-threading: 14750 14751 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl(). 14752 14753 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on(). 14754 14755 3. Count how many times MemCheck_off() has been called so that 14756 nested use can be treated correctly. This also avoids 14757 inband-signalling in the previous code (which relied on the 14758 assumption that thread ID 0 is impossible). 14759 14760 *Bodo Moeller* 14761 14762 * Add "-rand" option also to s_client and s_server. 14763 14764 *Lutz Jaenicke* 14765 14766 * Fix CPU detection on Irix 6.x. 14767 *Kurt Hockenbury <khockenb@stevens-tech.edu> and 14768 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>* 14769 14770 * Fix X509_NAME bug which produced incorrect encoding if X509_NAME 14771 was empty. 14772 14773 *Steve Henson* 14774 14775 *This change does not apply to 0.9.7.* 14776 14777 * Use the cached encoding of an X509_NAME structure rather than 14778 copying it. This is apparently the reason for the libsafe "errors" 14779 but the code is actually correct. 14780 14781 *Steve Henson* 14782 14783 * Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent 14784 Bleichenbacher's DSA attack. 14785 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits 14786 to be set and top=0 forces the highest bit to be set; top=-1 is new 14787 and leaves the highest bit random. 14788 14789 *Ulf Moeller, Bodo Moeller* 14790 14791 * In the `NCONF_...`-based implementations for `CONF_...` queries 14792 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using 14793 a temporary CONF structure with the data component set to NULL 14794 (which gives segmentation faults in lh_retrieve). 14795 Instead, use NULL for the CONF pointer in CONF_get_string and 14796 CONF_get_number (which may use environment variables) and directly 14797 return NULL from CONF_get_section. 14798 14799 *Bodo Moeller* 14800 14801 * Fix potential buffer overrun for EBCDIC. 14802 14803 *Ulf Moeller* 14804 14805 * Tolerate nonRepudiation as being valid for S/MIME signing and certSign 14806 keyUsage if basicConstraints absent for a CA. 14807 14808 *Steve Henson* 14809 14810 * Make SMIME_write_PKCS7() write mail header values with a format that 14811 is more generally accepted (no spaces before the semicolon), since 14812 some programs can't parse those values properly otherwise. Also make 14813 sure BIO's that break lines after each write do not create invalid 14814 headers. 14815 14816 *Richard Levitte* 14817 14818 * Make the CRL encoding routines work with empty SEQUENCE OF. The 14819 macros previously used would not encode an empty SEQUENCE OF 14820 and break the signature. 14821 14822 *Steve Henson* 14823 14824 *This change does not apply to 0.9.7.* 14825 14826 * Zero the premaster secret after deriving the master secret in 14827 DH ciphersuites. 14828 14829 *Steve Henson* 14830 14831 * Add some EVP_add_digest_alias registrations (as found in 14832 OpenSSL_add_all_digests()) to SSL_library_init() 14833 aka OpenSSL_add_ssl_algorithms(). This provides improved 14834 compatibility with peers using X.509 certificates 14835 with unconventional AlgorithmIdentifier OIDs. 14836 14837 *Bodo Moeller* 14838 14839 * Fix for Irix with NO_ASM. 14840 14841 *"Bruce W. Forsberg" <bruce.forsberg@baesystems.com>* 14842 14843 * ./config script fixes. 14844 14845 *Ulf Moeller, Richard Levitte* 14846 14847 * Fix 'openssl passwd -1'. 14848 14849 *Bodo Moeller* 14850 14851 * Change PKCS12_key_gen_asc() so it can cope with non null 14852 terminated strings whose length is passed in the passlen 14853 parameter, for example from PEM callbacks. This was done 14854 by adding an extra length parameter to asc2uni(). 14855 14856 *Steve Henson, reported by <oddissey@samsung.co.kr>* 14857 14858 * Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn 14859 call failed, free the DSA structure. 14860 14861 *Bodo Moeller* 14862 14863 * Fix to uni2asc() to cope with zero length Unicode strings. 14864 These are present in some PKCS#12 files. 14865 14866 *Steve Henson* 14867 14868 * Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c). 14869 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits 14870 when writing a 32767 byte record. 14871 14872 *Bodo Moeller; problem reported by Eric Day <eday@concentric.net>* 14873 14874 * In `RSA_eay_public_{en,ed}crypt` and RSA_eay_mod_exp (rsa_eay.c), 14875 obtain lock CRYPTO_LOCK_RSA before setting `rsa->_method_mod_{n,p,q}`. 14876 14877 (RSA objects have a reference count access to which is protected 14878 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c], 14879 so they are meant to be shared between threads.) 14880 *Bodo Moeller, Geoff Thorpe; original patch submitted by 14881 "Reddie, Steven" <Steven.Reddie@ca.com>* 14882 14883 * Fix a deadlock in CRYPTO_mem_leaks(). 14884 14885 *Bodo Moeller* 14886 14887 * Use better test patterns in bntest. 14888 14889 *Ulf Möller* 14890 14891 * rand_win.c fix for Borland C. 14892 14893 *Ulf Möller* 14894 14895 * BN_rshift bugfix for n == 0. 14896 14897 *Bodo Moeller* 14898 14899 * Add a 'bctest' script that checks for some known 'bc' bugs 14900 so that 'make test' does not abort just because 'bc' is broken. 14901 14902 *Bodo Moeller* 14903 14904 * Store verify_result within SSL_SESSION also for client side to 14905 avoid potential security hole. (Re-used sessions on the client side 14906 always resulted in verify_result==X509_V_OK, not using the original 14907 result of the server certificate verification.) 14908 14909 *Lutz Jaenicke* 14910 14911 * Fix ssl3_pending: If the record in s->s3->rrec is not of type 14912 SSL3_RT_APPLICATION_DATA, return 0. 14913 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true. 14914 14915 *Bodo Moeller* 14916 14917 * Fix SSL_peek: 14918 Both ssl2_peek and ssl3_peek, which were totally broken in earlier 14919 releases, have been re-implemented by renaming the previous 14920 implementations of ssl2_read and ssl3_read to ssl2_read_internal 14921 and ssl3_read_internal, respectively, and adding 'peek' parameters 14922 to them. The new ssl[23]_{read,peek} functions are calls to 14923 ssl[23]_read_internal with the 'peek' flag set appropriately. 14924 A 'peek' parameter has also been added to ssl3_read_bytes, which 14925 does the actual work for ssl3_read_internal. 14926 14927 *Bodo Moeller* 14928 14929 * Initialise "ex_data" member of RSA/DSA/DH structures prior to calling 14930 the method-specific "init()" handler. Also clean up ex_data after 14931 calling the method-specific "finish()" handler. Previously, this was 14932 happening the other way round. 14933 14934 *Geoff Thorpe* 14935 14936 * Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16. 14937 The previous value, 12, was not always sufficient for BN_mod_exp(). 14938 14939 *Bodo Moeller* 14940 14941 * Make sure that shared libraries get the internal name engine with 14942 the full version number and not just 0. This should mark the 14943 shared libraries as not backward compatible. Of course, this should 14944 be changed again when we can guarantee backward binary compatibility. 14945 14946 *Richard Levitte* 14947 14948 * Fix typo in get_cert_by_subject() in by_dir.c 14949 14950 *Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>* 14951 14952 * Rework the system to generate shared libraries: 14953 14954 - Make note of the expected extension for the shared libraries and 14955 if there is a need for symbolic links from for example libcrypto.so.0 14956 to libcrypto.so.0.9.7. There is extended info in Configure for 14957 that. 14958 14959 - Make as few rebuilds of the shared libraries as possible. 14960 14961 - Still avoid linking the OpenSSL programs with the shared libraries. 14962 14963 - When installing, install the shared libraries separately from the 14964 static ones. 14965 14966 *Richard Levitte* 14967 14968 * Fix SSL_CTX_set_read_ahead macro to actually use its argument. 14969 14970 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new 14971 and not in SSL_clear because the latter is also used by the 14972 accept/connect functions; previously, the settings made by 14973 SSL_set_read_ahead would be lost during the handshake. 14974 14975 *Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>* 14976 14977 * Correct util/mkdef.pl to be selective about disabled algorithms. 14978 Previously, it would create entries for disabled algorithms no 14979 matter what. 14980 14981 *Richard Levitte* 14982 14983 * Added several new manual pages for SSL_* function. 14984 14985 *Lutz Jaenicke* 14986 14987### Changes between 0.9.5a and 0.9.6 [24 Sep 2000] 14988 14989 * In ssl23_get_client_hello, generate an error message when faced 14990 with an initial SSL 3.0/TLS record that is too small to contain the 14991 first two bytes of the ClientHello message, i.e. client_version. 14992 (Note that this is a pathologic case that probably has never happened 14993 in real life.) The previous approach was to use the version number 14994 from the record header as a substitute; but our protocol choice 14995 should not depend on that one because it is not authenticated 14996 by the Finished messages. 14997 14998 *Bodo Moeller* 14999 15000 * More robust randomness gathering functions for Windows. 15001 15002 *Jeffrey Altman <jaltman@columbia.edu>* 15003 15004 * For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is 15005 not set then we don't setup the error code for issuer check errors 15006 to avoid possibly overwriting other errors which the callback does 15007 handle. If an application does set the flag then we assume it knows 15008 what it is doing and can handle the new informational codes 15009 appropriately. 15010 15011 *Steve Henson* 15012 15013 * Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for 15014 a general "ANY" type, as such it should be able to decode anything 15015 including tagged types. However it didn't check the class so it would 15016 wrongly interpret tagged types in the same way as their universal 15017 counterpart and unknown types were just rejected. Changed so that the 15018 tagged and unknown types are handled in the same way as a SEQUENCE: 15019 that is the encoding is stored intact. There is also a new type 15020 "V_ASN1_OTHER" which is used when the class is not universal, in this 15021 case we have no idea what the actual type is so we just lump them all 15022 together. 15023 15024 *Steve Henson* 15025 15026 * On VMS, stdout may very well lead to a file that is written to 15027 in a record-oriented fashion. That means that every write() will 15028 write a separate record, which will be read separately by the 15029 programs trying to read from it. This can be very confusing. 15030 15031 The solution is to put a BIO filter in the way that will buffer 15032 text until a linefeed is reached, and then write everything a 15033 line at a time, so every record written will be an actual line, 15034 not chunks of lines and not (usually doesn't happen, but I've 15035 seen it once) several lines in one record. BIO_f_linebuffer() is 15036 the answer. 15037 15038 Currently, it's a VMS-only method, because that's where it has 15039 been tested well enough. 15040 15041 *Richard Levitte* 15042 15043 * Remove 'optimized' squaring variant in BN_mod_mul_montgomery, 15044 it can return incorrect results. 15045 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a, 15046 but it was in 0.9.6-beta[12].) 15047 15048 *Bodo Moeller* 15049 15050 * Disable the check for content being present when verifying detached 15051 signatures in pk7_smime.c. Some versions of Netscape (wrongly) 15052 include zero length content when signing messages. 15053 15054 *Steve Henson* 15055 15056 * New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR 15057 BIO_ctrl (for BIO pairs). 15058 15059 *Bodo Möller* 15060 15061 * Add DSO method for VMS. 15062 15063 *Richard Levitte* 15064 15065 * Bug fix: Montgomery multiplication could produce results with the 15066 wrong sign. 15067 15068 *Ulf Möller* 15069 15070 * Add RPM specification openssl.spec and modify it to build three 15071 packages. The default package contains applications, application 15072 documentation and run-time libraries. The devel package contains 15073 include files, static libraries and function documentation. The 15074 doc package contains the contents of the doc directory. The original 15075 openssl.spec was provided by Damien Miller <djm@mindrot.org>. 15076 15077 *Richard Levitte* 15078 15079 * Add a large number of documentation files for many SSL routines. 15080 15081 *Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>* 15082 15083 * Add a configuration entry for Sony News 4. 15084 15085 *NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>* 15086 15087 * Don't set the two most significant bits to one when generating a 15088 random number < q in the DSA library. 15089 15090 *Ulf Möller* 15091 15092 * New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default 15093 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if 15094 the underlying transport is blocking) if a handshake took place. 15095 (The default behaviour is needed by applications such as s_client 15096 and s_server that use select() to determine when to use SSL_read; 15097 but for applications that know in advance when to expect data, it 15098 just makes things more complicated.) 15099 15100 *Bodo Moeller* 15101 15102 * Add RAND_egd_bytes(), which gives control over the number of bytes read 15103 from EGD. 15104 15105 *Ben Laurie* 15106 15107 * Add a few more EBCDIC conditionals that make `req` and `x509` 15108 work better on such systems. 15109 15110 *Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>* 15111 15112 * Add two demo programs for PKCS12_parse() and PKCS12_create(). 15113 Update PKCS12_parse() so it copies the friendlyName and the 15114 keyid to the certificates aux info. 15115 15116 *Steve Henson* 15117 15118 * Fix bug in PKCS7_verify() which caused an infinite loop 15119 if there was more than one signature. 15120 15121 *Sven Uszpelkat <su@celocom.de>* 15122 15123 * Major change in util/mkdef.pl to include extra information 15124 about each symbol, as well as presenting variables as well 15125 as functions. This change means that there's n more need 15126 to rebuild the .num files when some algorithms are excluded. 15127 15128 *Richard Levitte* 15129 15130 * Allow the verify time to be set by an application, 15131 rather than always using the current time. 15132 15133 *Steve Henson* 15134 15135 * Phase 2 verify code reorganisation. The certificate 15136 verify code now looks up an issuer certificate by a 15137 number of criteria: subject name, authority key id 15138 and key usage. It also verifies self signed certificates 15139 by the same criteria. The main comparison function is 15140 X509_check_issued() which performs these checks. 15141 15142 Lot of changes were necessary in order to support this 15143 without completely rewriting the lookup code. 15144 15145 Authority and subject key identifier are now cached. 15146 15147 The LHASH 'certs' is X509_STORE has now been replaced 15148 by a STACK_OF(X509_OBJECT). This is mainly because an 15149 LHASH can't store or retrieve multiple objects with 15150 the same hash value. 15151 15152 As a result various functions (which were all internal 15153 use only) have changed to handle the new X509_STORE 15154 structure. This will break anything that messed round 15155 with X509_STORE internally. 15156 15157 The functions X509_STORE_add_cert() now checks for an 15158 exact match, rather than just subject name. 15159 15160 The X509_STORE API doesn't directly support the retrieval 15161 of multiple certificates matching a given criteria, however 15162 this can be worked round by performing a lookup first 15163 (which will fill the cache with candidate certificates) 15164 and then examining the cache for matches. This is probably 15165 the best we can do without throwing out X509_LOOKUP 15166 entirely (maybe later...). 15167 15168 The X509_VERIFY_CTX structure has been enhanced considerably. 15169 15170 All certificate lookup operations now go via a get_issuer() 15171 callback. Although this currently uses an X509_STORE it 15172 can be replaced by custom lookups. This is a simple way 15173 to bypass the X509_STORE hackery necessary to make this 15174 work and makes it possible to use more efficient techniques 15175 in future. A very simple version which uses a simple 15176 STACK for its trusted certificate store is also provided 15177 using X509_STORE_CTX_trusted_stack(). 15178 15179 The verify_cb() and verify() callbacks now have equivalents 15180 in the X509_STORE_CTX structure. 15181 15182 X509_STORE_CTX also has a 'flags' field which can be used 15183 to customise the verify behaviour. 15184 15185 *Steve Henson* 15186 15187 * Add new PKCS#7 signing option PKCS7_NOSMIMECAP which 15188 excludes S/MIME capabilities. 15189 15190 *Steve Henson* 15191 15192 * When a certificate request is read in keep a copy of the 15193 original encoding of the signed data and use it when outputting 15194 again. Signatures then use the original encoding rather than 15195 a decoded, encoded version which may cause problems if the 15196 request is improperly encoded. 15197 15198 *Steve Henson* 15199 15200 * For consistency with other BIO_puts implementations, call 15201 buffer_write(b, ...) directly in buffer_puts instead of calling 15202 BIO_write(b, ...). 15203 15204 In BIO_puts, increment b->num_write as in BIO_write. 15205 15206 *Peter.Sylvester@EdelWeb.fr* 15207 15208 * Fix BN_mul_word for the case where the word is 0. (We have to use 15209 BN_zero, we may not return a BIGNUM with an array consisting of 15210 words set to zero.) 15211 15212 *Bodo Moeller* 15213 15214 * Avoid calling abort() from within the library when problems are 15215 detected, except if preprocessor symbols have been defined 15216 (such as REF_CHECK, BN_DEBUG etc.). 15217 15218 *Bodo Moeller* 15219 15220 * New openssl application 'rsautl'. This utility can be 15221 used for low-level RSA operations. DER public key 15222 BIO/fp routines also added. 15223 15224 *Steve Henson* 15225 15226 * New Configure entry and patches for compiling on QNX 4. 15227 15228 *Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>* 15229 15230 * A demo state-machine implementation was sponsored by 15231 Nuron (<http://www.nuron.com/>) and is now available in 15232 demos/state_machine. 15233 15234 *Ben Laurie* 15235 15236 * New options added to the 'dgst' utility for signature 15237 generation and verification. 15238 15239 *Steve Henson* 15240 15241 * Unrecognized PKCS#7 content types are now handled via a 15242 catch all ASN1_TYPE structure. This allows unsupported 15243 types to be stored as a "blob" and an application can 15244 encode and decode it manually. 15245 15246 *Steve Henson* 15247 15248 * Fix various signed/unsigned issues to make a_strex.c 15249 compile under VC++. 15250 15251 *Oscar Jacobsson <oscar.jacobsson@celocom.com>* 15252 15253 * ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct 15254 length if passed a buffer. ASN1_INTEGER_to_BN failed 15255 if passed a NULL BN and its argument was negative. 15256 15257 *Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>* 15258 15259 * Modification to PKCS#7 encoding routines to output definite 15260 length encoding. Since currently the whole structures are in 15261 memory there's not real point in using indefinite length 15262 constructed encoding. However if OpenSSL is compiled with 15263 the flag PKCS7_INDEFINITE_ENCODING the old form is used. 15264 15265 *Steve Henson* 15266 15267 * Added BIO_vprintf() and BIO_vsnprintf(). 15268 15269 *Richard Levitte* 15270 15271 * Added more prefixes to parse for in the strings written 15272 through a logging bio, to cover all the levels that are available 15273 through syslog. The prefixes are now: 15274 15275 PANIC, EMERG, EMR => LOG_EMERG 15276 ALERT, ALR => LOG_ALERT 15277 CRIT, CRI => LOG_CRIT 15278 ERROR, ERR => LOG_ERR 15279 WARNING, WARN, WAR => LOG_WARNING 15280 NOTICE, NOTE, NOT => LOG_NOTICE 15281 INFO, INF => LOG_INFO 15282 DEBUG, DBG => LOG_DEBUG 15283 15284 and as before, if none of those prefixes are present at the 15285 beginning of the string, LOG_ERR is chosen. 15286 15287 On Win32, the `LOG_*` levels are mapped according to this: 15288 15289 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE 15290 LOG_WARNING => EVENTLOG_WARNING_TYPE 15291 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE 15292 15293 *Richard Levitte* 15294 15295 * Made it possible to reconfigure with just the configuration 15296 argument "reconf" or "reconfigure". The command line arguments 15297 are stored in Makefile.ssl in the variable CONFIGURE_ARGS, 15298 and are retrieved from there when reconfiguring. 15299 15300 *Richard Levitte* 15301 15302 * MD4 implemented. 15303 15304 *Assar Westerlund <assar@sics.se>, Richard Levitte* 15305 15306 * Add the arguments -CAfile and -CApath to the pkcs12 utility. 15307 15308 *Richard Levitte* 15309 15310 * The obj_dat.pl script was messing up the sorting of object 15311 names. The reason was that it compared the quoted version 15312 of strings as a result "OCSP" > "OCSP Signing" because 15313 " > SPACE. Changed script to store unquoted versions of 15314 names and add quotes on output. It was also omitting some 15315 names from the lookup table if they were given a default 15316 value (that is if SN is missing it is given the same 15317 value as LN and vice versa), these are now added on the 15318 grounds that if an object has a name we should be able to 15319 look it up. Finally added warning output when duplicate 15320 short or long names are found. 15321 15322 *Steve Henson* 15323 15324 * Changes needed for Tandem NSK. 15325 15326 *Scott Uroff <scott@xypro.com>* 15327 15328 * Fix SSL 2.0 rollback checking: Due to an off-by-one error in 15329 RSA_padding_check_SSLv23(), special padding was never detected 15330 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol 15331 version rollback attacks was not effective. 15332 15333 In s23_clnt.c, don't use special rollback-attack detection padding 15334 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the 15335 client; similarly, in s23_srvr.c, don't do the rollback check if 15336 SSL 2.0 is the only protocol enabled in the server. 15337 15338 *Bodo Moeller* 15339 15340 * Make it possible to get hexdumps of unprintable data with 'openssl 15341 asn1parse'. By implication, the functions ASN1_parse_dump() and 15342 BIO_dump_indent() are added. 15343 15344 *Richard Levitte* 15345 15346 * New functions ASN1_STRING_print_ex() and X509_NAME_print_ex() 15347 these print out strings and name structures based on various 15348 flags including RFC2253 support and proper handling of 15349 multibyte characters. Added options to the 'x509' utility 15350 to allow the various flags to be set. 15351 15352 *Steve Henson* 15353 15354 * Various fixes to use ASN1_TIME instead of ASN1_UTCTIME. 15355 Also change the functions X509_cmp_current_time() and 15356 X509_gmtime_adj() work with an ASN1_TIME structure, 15357 this will enable certificates using GeneralizedTime in validity 15358 dates to be checked. 15359 15360 *Steve Henson* 15361 15362 * Make the NEG_PUBKEY_BUG code (which tolerates invalid 15363 negative public key encodings) on by default, 15364 NO_NEG_PUBKEY_BUG can be set to disable it. 15365 15366 *Steve Henson* 15367 15368 * New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT 15369 content octets. An i2c_ASN1_OBJECT is unnecessary because 15370 the encoding can be trivially obtained from the structure. 15371 15372 *Steve Henson* 15373 15374 * crypto/err.c locking bugfix: Use write locks (`CRYPTO_w_[un]lock`), 15375 not read locks (`CRYPTO_r_[un]lock`). 15376 15377 *Bodo Moeller* 15378 15379 * A first attempt at creating official support for shared 15380 libraries through configuration. I've kept it so the 15381 default is static libraries only, and the OpenSSL programs 15382 are always statically linked for now, but there are 15383 preparations for dynamic linking in place. 15384 This has been tested on Linux and Tru64. 15385 15386 *Richard Levitte* 15387 15388 * Randomness polling function for Win9x, as described in: 15389 Peter Gutmann, Software Generation of Practically Strong 15390 Random Numbers. 15391 15392 *Ulf Möller* 15393 15394 * Fix so PRNG is seeded in req if using an already existing 15395 DSA key. 15396 15397 *Steve Henson* 15398 15399 * New options to smime application. -inform and -outform 15400 allow alternative formats for the S/MIME message including 15401 PEM and DER. The -content option allows the content to be 15402 specified separately. This should allow things like Netscape 15403 form signing output easier to verify. 15404 15405 *Steve Henson* 15406 15407 * Fix the ASN1 encoding of tags using the 'long form'. 15408 15409 *Steve Henson* 15410 15411 * New ASN1 functions, `i2c_*` and `c2i_*` for INTEGER and BIT 15412 STRING types. These convert content octets to and from the 15413 underlying type. The actual tag and length octets are 15414 already assumed to have been read in and checked. These 15415 are needed because all other string types have virtually 15416 identical handling apart from the tag. By having versions 15417 of the ASN1 functions that just operate on content octets 15418 IMPLICIT tagging can be handled properly. It also allows 15419 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED 15420 and ASN1_INTEGER are identical apart from the tag. 15421 15422 *Steve Henson* 15423 15424 * Change the handling of OID objects as follows: 15425 15426 - New object identifiers are inserted in objects.txt, following 15427 the syntax given in [crypto/objects/README.md](crypto/objects/README.md). 15428 - objects.pl is used to process obj_mac.num and create a new 15429 obj_mac.h. 15430 - obj_dat.pl is used to create a new obj_dat.h, using the data in 15431 obj_mac.h. 15432 15433 This is currently kind of a hack, and the perl code in objects.pl 15434 isn't very elegant, but it works as I intended. The simplest way 15435 to check that it worked correctly is to look in obj_dat.h and 15436 check the array nid_objs and make sure the objects haven't moved 15437 around (this is important!). Additions are OK, as well as 15438 consistent name changes. 15439 15440 *Richard Levitte* 15441 15442 * Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1'). 15443 15444 *Bodo Moeller* 15445 15446 * Addition of the command line parameter '-rand file' to 'openssl req'. 15447 The given file adds to whatever has already been seeded into the 15448 random pool through the RANDFILE configuration file option or 15449 environment variable, or the default random state file. 15450 15451 *Richard Levitte* 15452 15453 * mkstack.pl now sorts each macro group into lexical order. 15454 Previously the output order depended on the order the files 15455 appeared in the directory, resulting in needless rewriting 15456 of safestack.h . 15457 15458 *Steve Henson* 15459 15460 * Patches to make OpenSSL compile under Win32 again. Mostly 15461 work arounds for the VC++ problem that it treats func() as 15462 func(void). Also stripped out the parts of mkdef.pl that 15463 added extra typesafe functions: these no longer exist. 15464 15465 *Steve Henson* 15466 15467 * Reorganisation of the stack code. The macros are now all 15468 collected in safestack.h . Each macro is defined in terms of 15469 a "stack macro" of the form `SKM_<name>(type, a, b)`. The 15470 DEBUG_SAFESTACK is now handled in terms of function casts, 15471 this has the advantage of retaining type safety without the 15472 use of additional functions. If DEBUG_SAFESTACK is not defined 15473 then the non typesafe macros are used instead. Also modified the 15474 mkstack.pl script to handle the new form. Needs testing to see 15475 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK 15476 the default if no major problems. Similar behaviour for ASN1_SET_OF 15477 and PKCS12_STACK_OF. 15478 15479 *Steve Henson* 15480 15481 * When some versions of IIS use the 'NET' form of private key the 15482 key derivation algorithm is different. Normally MD5(password) is 15483 used as a 128 bit RC4 key. In the modified case 15484 MD5(MD5(password) + "SGCKEYSALT") is used instead. Added some 15485 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same 15486 as the old Netscape_RSA functions except they have an additional 15487 'sgckey' parameter which uses the modified algorithm. Also added 15488 an -sgckey command line option to the rsa utility. Thanks to 15489 Adrian Peck <bertie@ncipher.com> for posting details of the modified 15490 algorithm to openssl-dev. 15491 15492 *Steve Henson* 15493 15494 * The evp_local.h macros were using 'c.##kname' which resulted in 15495 invalid expansion on some systems (SCO 5.0.5 for example). 15496 Corrected to 'c.kname'. 15497 15498 *Phillip Porch <root@theporch.com>* 15499 15500 * New X509_get1_email() and X509_REQ_get1_email() functions that return 15501 a STACK of email addresses from a certificate or request, these look 15502 in the subject name and the subject alternative name extensions and 15503 omit any duplicate addresses. 15504 15505 *Steve Henson* 15506 15507 * Re-implement BN_mod_exp2_mont using independent (and larger) windows. 15508 This makes DSA verification about 2 % faster. 15509 15510 *Bodo Moeller* 15511 15512 * Increase maximum window size in `BN_mod_exp_...` to 6 bits instead of 5 15513 (meaning that now 2^5 values will be precomputed, which is only 4 KB 15514 plus overhead for 1024 bit moduli). 15515 This makes exponentiations about 0.5 % faster for 1024 bit 15516 exponents (as measured by "openssl speed rsa2048"). 15517 15518 *Bodo Moeller* 15519 15520 * Rename memory handling macros to avoid conflicts with other 15521 software: 15522 Malloc => OPENSSL_malloc 15523 Malloc_locked => OPENSSL_malloc_locked 15524 Realloc => OPENSSL_realloc 15525 Free => OPENSSL_free 15526 15527 *Richard Levitte* 15528 15529 * New function BN_mod_exp_mont_word for small bases (roughly 15% 15530 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange). 15531 15532 *Bodo Moeller* 15533 15534 * CygWin32 support. 15535 15536 *John Jarvie <jjarvie@newsguy.com>* 15537 15538 * The type-safe stack code has been rejigged. It is now only compiled 15539 in when OpenSSL is configured with the DEBUG_SAFESTACK option and 15540 by default all type-specific stack functions are "#define"d back to 15541 standard stack functions. This results in more streamlined output 15542 but retains the type-safety checking possibilities of the original 15543 approach. 15544 15545 *Geoff Thorpe* 15546 15547 * The STACK code has been cleaned up, and certain type declarations 15548 that didn't make a lot of sense have been brought in line. This has 15549 also involved a cleanup of sorts in safestack.h to more correctly 15550 map type-safe stack functions onto their plain stack counterparts. 15551 This work has also resulted in a variety of "const"ifications of 15552 lots of the code, especially `_cmp` operations which should normally 15553 be prototyped with "const" parameters anyway. 15554 15555 *Geoff Thorpe* 15556 15557 * When generating bytes for the first time in md_rand.c, 'stir the pool' 15558 by seeding with STATE_SIZE dummy bytes (with zero entropy count). 15559 (The PRNG state consists of two parts, the large pool 'state' and 'md', 15560 where all of 'md' is used each time the PRNG is used, but 'state' 15561 is used only indexed by a cyclic counter. As entropy may not be 15562 well distributed from the beginning, 'md' is important as a 15563 chaining variable. However, the output function chains only half 15564 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains 15565 all of 'md', and seeding with STATE_SIZE dummy bytes will result 15566 in all of 'state' being rewritten, with the new values depending 15567 on virtually all of 'md'. This overcomes the 80 bit limitation.) 15568 15569 *Bodo Moeller* 15570 15571 * In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when 15572 the handshake is continued after ssl_verify_cert_chain(); 15573 otherwise, if SSL_VERIFY_NONE is set, remaining error codes 15574 can lead to 'unexplainable' connection aborts later. 15575 15576 *Bodo Moeller; problem tracked down by Lutz Jaenicke* 15577 15578 * Major EVP API cipher revision. 15579 Add hooks for extra EVP features. This allows various cipher 15580 parameters to be set in the EVP interface. Support added for variable 15581 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and 15582 setting of RC2 and RC5 parameters. 15583 15584 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length 15585 ciphers. 15586 15587 Remove lots of duplicated code from the EVP library. For example *every* 15588 cipher init() function handles the 'iv' in the same way according to the 15589 cipher mode. They also all do nothing if the 'key' parameter is NULL and 15590 for CFB and OFB modes they zero ctx->num. 15591 15592 New functionality allows removal of S/MIME code RC2 hack. 15593 15594 Most of the routines have the same form and so can be declared in terms 15595 of macros. 15596 15597 By shifting this to the top level EVP_CipherInit() it can be removed from 15598 all individual ciphers. If the cipher wants to handle IVs or keys 15599 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT 15600 flags. 15601 15602 Change lots of functions like EVP_EncryptUpdate() to now return a 15603 value: although software versions of the algorithms cannot fail 15604 any installed hardware versions can. 15605 15606 *Steve Henson* 15607 15608 * Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if 15609 this option is set, tolerate broken clients that send the negotiated 15610 protocol version number instead of the requested protocol version 15611 number. 15612 15613 *Bodo Moeller* 15614 15615 * Call dh_tmp_cb (set by `..._TMP_DH_CB`) with correct 'is_export' flag; 15616 i.e. non-zero for export ciphersuites, zero otherwise. 15617 Previous versions had this flag inverted, inconsistent with 15618 rsa_tmp_cb (..._TMP_RSA_CB). 15619 15620 *Bodo Moeller; problem reported by Amit Chopra* 15621 15622 * Add missing DSA library text string. Work around for some IIS 15623 key files with invalid SEQUENCE encoding. 15624 15625 *Steve Henson* 15626 15627 * Add a document (doc/standards.txt) that list all kinds of standards 15628 and so on that are implemented in OpenSSL. 15629 15630 *Richard Levitte* 15631 15632 * Enhance c_rehash script. Old version would mishandle certificates 15633 with the same subject name hash and wouldn't handle CRLs at all. 15634 Added -fingerprint option to crl utility, to support new c_rehash 15635 features. 15636 15637 *Steve Henson* 15638 15639 * Eliminate non-ANSI declarations in crypto.h and stack.h. 15640 15641 *Ulf Möller* 15642 15643 * Fix for SSL server purpose checking. Server checking was 15644 rejecting certificates which had extended key usage present 15645 but no ssl client purpose. 15646 15647 *Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>* 15648 15649 * Make PKCS#12 code work with no password. The PKCS#12 spec 15650 is a little unclear about how a blank password is handled. 15651 Since the password in encoded as a BMPString with terminating 15652 double NULL a zero length password would end up as just the 15653 double NULL. However no password at all is different and is 15654 handled differently in the PKCS#12 key generation code. NS 15655 treats a blank password as zero length. MSIE treats it as no 15656 password on export: but it will try both on import. We now do 15657 the same: PKCS12_parse() tries zero length and no password if 15658 the password is set to "" or NULL (NULL is now a valid password: 15659 it wasn't before) as does the pkcs12 application. 15660 15661 *Steve Henson* 15662 15663 * Bugfixes in `apps/x509.c`: Avoid a memory leak; and don't use 15664 perror when PEM_read_bio_X509_REQ fails, the error message must 15665 be obtained from the error queue. 15666 15667 *Bodo Moeller* 15668 15669 * Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing 15670 it in ERR_remove_state if appropriate, and change ERR_get_state 15671 accordingly to avoid race conditions (this is necessary because 15672 thread_hash is no longer constant once set). 15673 15674 *Bodo Moeller* 15675 15676 * Bugfix for linux-elf makefile.one. 15677 15678 *Ulf Möller* 15679 15680 * RSA_get_default_method() will now cause a default 15681 RSA_METHOD to be chosen if one doesn't exist already. 15682 Previously this was only set during a call to RSA_new() 15683 or RSA_new_method(NULL) meaning it was possible for 15684 RSA_get_default_method() to return NULL. 15685 15686 *Geoff Thorpe* 15687 15688 * Added native name translation to the existing DSO code 15689 that will convert (if the flag to do so is set) filenames 15690 that are sufficiently small and have no path information 15691 into a canonical native form. Eg. "blah" converted to 15692 "libblah.so" or "blah.dll" etc. 15693 15694 *Geoff Thorpe* 15695 15696 * New function ERR_error_string_n(e, buf, len) which is like 15697 ERR_error_string(e, buf), but writes at most 'len' bytes 15698 including the 0 terminator. For ERR_error_string_n, 'buf' 15699 may not be NULL. 15700 15701 *Damien Miller <djm@mindrot.org>, Bodo Moeller* 15702 15703 * CONF library reworked to become more general. A new CONF 15704 configuration file reader "class" is implemented as well as a 15705 new functions (`NCONF_*`, for "New CONF") to handle it. The now 15706 old `CONF_*` functions are still there, but are reimplemented to 15707 work in terms of the new functions. Also, a set of functions 15708 to handle the internal storage of the configuration data is 15709 provided to make it easier to write new configuration file 15710 reader "classes" (I can definitely see something reading a 15711 configuration file in XML format, for example), called `_CONF_*`, 15712 or "the configuration storage API"... 15713 15714 The new configuration file reading functions are: 15715 15716 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio, 15717 NCONF_get_section, NCONF_get_string, NCONF_get_numbre 15718 15719 NCONF_default, NCONF_WIN32 15720 15721 NCONF_dump_fp, NCONF_dump_bio 15722 15723 NCONF_default and NCONF_WIN32 are method (or "class") choosers, 15724 NCONF_new creates a new CONF object. This works in the same way 15725 as other interfaces in OpenSSL, like the BIO interface. 15726 `NCONF_dump_*` dump the internal storage of the configuration file, 15727 which is useful for debugging. All other functions take the same 15728 arguments as the old `CONF_*` functions with the exception of the 15729 first that must be a `CONF *` instead of a `LHASH *`. 15730 15731 To make it easier to use the new classes with the old `CONF_*` functions, 15732 the function CONF_set_default_method is provided. 15733 15734 *Richard Levitte* 15735 15736 * Add '-tls1' option to 'openssl ciphers', which was already 15737 mentioned in the documentation but had not been implemented. 15738 (This option is not yet really useful because even the additional 15739 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.) 15740 15741 *Bodo Moeller* 15742 15743 * Initial DSO code added into libcrypto for letting OpenSSL (and 15744 OpenSSL-based applications) load shared libraries and bind to 15745 them in a portable way. 15746 15747 *Geoff Thorpe, with contributions from Richard Levitte* 15748 15749### Changes between 0.9.5 and 0.9.5a [1 Apr 2000] 15750 15751 * Make sure _lrotl and _lrotr are only used with MSVC. 15752 15753 * Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status 15754 (the default implementation of RAND_status). 15755 15756 * Rename openssl x509 option '-crlext', which was added in 0.9.5, 15757 to '-clrext' (= clear extensions), as intended and documented. 15758 *Bodo Moeller; inconsistency pointed out by Michael Attili 15759 <attili@amaxo.com>* 15760 15761 * Fix for HMAC. It wasn't zeroing the rest of the block if the key length 15762 was larger than the MD block size. 15763 15764 *Steve Henson, pointed out by Yost William <YostW@tce.com>* 15765 15766 * Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument 15767 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set() 15768 using the passed key: if the passed key was a private key the result 15769 of X509_print(), for example, would be to print out all the private key 15770 components. 15771 15772 *Steve Henson* 15773 15774 * des_quad_cksum() byte order bug fix. 15775 *Ulf Möller, using the problem description in krb4-0.9.7, where 15776 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>* 15777 15778 * Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly 15779 discouraged. 15780 15781 *Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>* 15782 15783 * For easily testing in shell scripts whether some command 15784 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX' 15785 returns with exit code 0 iff no command of the given name is available. 15786 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases, 15787 the output goes to stdout and nothing is printed to stderr. 15788 Additional arguments are always ignored. 15789 15790 Since for each cipher there is a command of the same name, 15791 the 'no-cipher' compilation switches can be tested this way. 15792 15793 ('openssl no-XXX' is not able to detect pseudo-commands such 15794 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.) 15795 15796 *Bodo Moeller* 15797 15798 * Update test suite so that 'make test' succeeds in 'no-rsa' configuration. 15799 15800 *Bodo Moeller* 15801 15802 * For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE 15803 is set; it will be thrown away anyway because each handshake creates 15804 its own key. 15805 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition 15806 to parameters -- in previous versions (since OpenSSL 0.9.3) the 15807 'default key' from SSL_CTX_set_tmp_dh would always be lost, meaning 15808 you effectively got SSL_OP_SINGLE_DH_USE when using this macro. 15809 15810 *Bodo Moeller* 15811 15812 * New s_client option -ign_eof: EOF at stdin is ignored, and 15813 'Q' and 'R' lose their special meanings (quit/renegotiate). 15814 This is part of what -quiet does; unlike -quiet, -ign_eof 15815 does not suppress any output. 15816 15817 *Richard Levitte* 15818 15819 * Add compatibility options to the purpose and trust code. The 15820 purpose X509_PURPOSE_ANY is "any purpose" which automatically 15821 accepts a certificate or CA, this was the previous behaviour, 15822 with all the associated security issues. 15823 15824 X509_TRUST_COMPAT is the old trust behaviour: only and 15825 automatically trust self signed roots in certificate store. A 15826 new trust setting X509_TRUST_DEFAULT is used to specify that 15827 a purpose has no associated trust setting and it should instead 15828 use the value in the default purpose. 15829 15830 *Steve Henson* 15831 15832 * Fix the PKCS#8 DSA private key code so it decodes keys again 15833 and fix a memory leak. 15834 15835 *Steve Henson* 15836 15837 * In util/mkerr.pl (which implements 'make errors'), preserve 15838 reason strings from the previous version of the .c file, as 15839 the default to have only downcase letters (and digits) in 15840 automatically generated reasons codes is not always appropriate. 15841 15842 *Bodo Moeller* 15843 15844 * In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table 15845 using strerror. Previously, ERR_reason_error_string() returned 15846 library names as reason strings for SYSerr; but SYSerr is a special 15847 case where small numbers are errno values, not library numbers. 15848 15849 *Bodo Moeller* 15850 15851 * Add '-dsaparam' option to 'openssl dhparam' application. This 15852 converts DSA parameters into DH parameters. (When creating parameters, 15853 DSA_generate_parameters is used.) 15854 15855 *Bodo Moeller* 15856 15857 * Include 'length' (recommended exponent length) in C code generated 15858 by 'openssl dhparam -C'. 15859 15860 *Bodo Moeller* 15861 15862 * The second argument to set_label in perlasm was already being used 15863 so couldn't be used as a "file scope" flag. Moved to third argument 15864 which was free. 15865 15866 *Steve Henson* 15867 15868 * In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes 15869 instead of RAND_bytes for encryption IVs and salts. 15870 15871 *Bodo Moeller* 15872 15873 * Include RAND_status() into RAND_METHOD instead of implementing 15874 it only for md_rand.c Otherwise replacing the PRNG by calling 15875 RAND_set_rand_method would be impossible. 15876 15877 *Bodo Moeller* 15878 15879 * Don't let DSA_generate_key() enter an infinite loop if the random 15880 number generation fails. 15881 15882 *Bodo Moeller* 15883 15884 * New 'rand' application for creating pseudo-random output. 15885 15886 *Bodo Moeller* 15887 15888 * Added configuration support for Linux/IA64 15889 15890 *Rolf Haberrecker <rolf@suse.de>* 15891 15892 * Assembler module support for Mingw32. 15893 15894 *Ulf Möller* 15895 15896 * Shared library support for HPUX (in shlib/). 15897 15898 *Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous* 15899 15900 * Shared library support for Solaris gcc. 15901 15902 *Lutz Behnke <behnke@trustcenter.de>* 15903 15904### Changes between 0.9.4 and 0.9.5 [28 Feb 2000] 15905 15906 * PKCS7_encrypt() was adding text MIME headers twice because they 15907 were added manually and by SMIME_crlf_copy(). 15908 15909 *Steve Henson* 15910 15911 * In bntest.c don't call BN_rand with zero bits argument. 15912 15913 *Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>* 15914 15915 * BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n] 15916 case was implemented. This caused BN_div_recp() to fail occasionally. 15917 15918 *Ulf Möller* 15919 15920 * Add an optional second argument to the set_label() in the perl 15921 assembly language builder. If this argument exists and is set 15922 to 1 it signals that the assembler should use a symbol whose 15923 scope is the entire file, not just the current function. This 15924 is needed with MASM which uses the format label:: for this scope. 15925 15926 *Steve Henson, pointed out by Peter Runestig <peter@runestig.com>* 15927 15928 * Change the ASN1 types so they are typedefs by default. Before 15929 almost all types were #define'd to ASN1_STRING which was causing 15930 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING) 15931 for example. 15932 15933 *Steve Henson* 15934 15935 * Change names of new functions to the new get1/get0 naming 15936 convention: After 'get1', the caller owns a reference count 15937 and has to call `..._free`; 'get0' returns a pointer to some 15938 data structure without incrementing reference counters. 15939 (Some of the existing 'get' functions increment a reference 15940 counter, some don't.) 15941 Similarly, 'set1' and 'add1' functions increase reference 15942 counters or duplicate objects. 15943 15944 *Steve Henson* 15945 15946 * Allow for the possibility of temp RSA key generation failure: 15947 the code used to assume it always worked and crashed on failure. 15948 15949 *Steve Henson* 15950 15951 * Fix potential buffer overrun problem in BIO_printf(). 15952 *Ulf Möller, using public domain code by Patrick Powell; problem 15953 pointed out by David Sacerdote <das33@cornell.edu>* 15954 15955 * Support EGD <http://www.lothar.com/tech/crypto/>. New functions 15956 RAND_egd() and RAND_status(). In the command line application, 15957 the EGD socket can be specified like a seed file using RANDFILE 15958 or -rand. 15959 15960 *Ulf Möller* 15961 15962 * Allow the string CERTIFICATE to be tolerated in PKCS#7 structures. 15963 Some CAs (e.g. Verisign) distribute certificates in this form. 15964 15965 *Steve Henson* 15966 15967 * Remove the SSL_ALLOW_ADH compile option and set the default cipher 15968 list to exclude them. This means that no special compilation option 15969 is needed to use anonymous DH: it just needs to be included in the 15970 cipher list. 15971 15972 *Steve Henson* 15973 15974 * Change the EVP_MD_CTX_type macro so its meaning consistent with 15975 EVP_MD_type. The old functionality is available in a new macro called 15976 EVP_MD_md(). Change code that uses it and update docs. 15977 15978 *Steve Henson* 15979 15980 * `..._ctrl` functions now have corresponding `..._callback_ctrl` functions 15981 where the `void *` argument is replaced by a function pointer argument. 15982 Previously `void *` was abused to point to functions, which works on 15983 many platforms, but is not correct. As these functions are usually 15984 called by macros defined in OpenSSL header files, most source code 15985 should work without changes. 15986 15987 *Richard Levitte* 15988 15989 * `<openssl/opensslconf.h>` (which is created by Configure) now contains 15990 sections with information on -D... compiler switches used for 15991 compiling the library so that applications can see them. To enable 15992 one of these sections, a pre-processor symbol `OPENSSL_..._DEFINES` 15993 must be defined. E.g., 15994 #define OPENSSL_ALGORITHM_DEFINES 15995 #include <openssl/opensslconf.h> 15996 defines all pertinent `NO_<algo>` symbols, such as NO_IDEA, NO_RSA, etc. 15997 15998 *Richard Levitte, Ulf and Bodo Möller* 15999 16000 * Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS 16001 record layer. 16002 16003 *Bodo Moeller* 16004 16005 * Change the 'other' type in certificate aux info to a STACK_OF 16006 X509_ALGOR. Although not an AlgorithmIdentifier as such it has 16007 the required ASN1 format: arbitrary types determined by an OID. 16008 16009 *Steve Henson* 16010 16011 * Add some PEM_write_X509_REQ_NEW() functions and a command line 16012 argument to 'req'. This is not because the function is newer or 16013 better than others it just uses the work 'NEW' in the certificate 16014 request header lines. Some software needs this. 16015 16016 *Steve Henson* 16017 16018 * Reorganise password command line arguments: now passwords can be 16019 obtained from various sources. Delete the PEM_cb function and make 16020 it the default behaviour: i.e. if the callback is NULL and the 16021 usrdata argument is not NULL interpret it as a null terminated pass 16022 phrase. If usrdata and the callback are NULL then the pass phrase 16023 is prompted for as usual. 16024 16025 *Steve Henson* 16026 16027 * Add support for the Compaq Atalla crypto accelerator. If it is installed, 16028 the support is automatically enabled. The resulting binaries will 16029 autodetect the card and use it if present. 16030 16031 *Ben Laurie and Compaq Inc.* 16032 16033 * Work around for Netscape hang bug. This sends certificate request 16034 and server done in one record. Since this is perfectly legal in the 16035 SSL/TLS protocol it isn't a "bug" option and is on by default. See 16036 the bugs/SSLv3 entry for more info. 16037 16038 *Steve Henson* 16039 16040 * HP-UX tune-up: new unified configs, HP C compiler bug workaround. 16041 16042 *Andy Polyakov* 16043 16044 * Add -rand argument to smime and pkcs12 applications and read/write 16045 of seed file. 16046 16047 *Steve Henson* 16048 16049 * New 'passwd' tool for crypt(3) and apr1 password hashes. 16050 16051 *Bodo Moeller* 16052 16053 * Add command line password options to the remaining applications. 16054 16055 *Steve Henson* 16056 16057 * Bug fix for BN_div_recp() for numerators with an even number of 16058 bits. 16059 16060 *Ulf Möller* 16061 16062 * More tests in bntest.c, and changed test_bn output. 16063 16064 *Ulf Möller* 16065 16066 * ./config recognizes MacOS X now. 16067 16068 *Andy Polyakov* 16069 16070 * Bug fix for BN_div() when the first words of num and divisor are 16071 equal (it gave wrong results if `(rem=(n1-q*d0)&BN_MASK2) < d0)`. 16072 16073 *Ulf Möller* 16074 16075 * Add support for various broken PKCS#8 formats, and command line 16076 options to produce them. 16077 16078 *Steve Henson* 16079 16080 * New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to 16081 get temporary BIGNUMs from a BN_CTX. 16082 16083 *Ulf Möller* 16084 16085 * Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont() 16086 for p == 0. 16087 16088 *Ulf Möller* 16089 16090 * Change the `SSLeay_add_all_*()` functions to `OpenSSL_add_all_*()` and 16091 include a #define from the old name to the new. The original intent 16092 was that statically linked binaries could for example just call 16093 SSLeay_add_all_ciphers() to just add ciphers to the table and not 16094 link with digests. This never worked because SSLeay_add_all_digests() 16095 and SSLeay_add_all_ciphers() were in the same source file so calling 16096 one would link with the other. They are now in separate source files. 16097 16098 *Steve Henson* 16099 16100 * Add a new -notext option to 'ca' and a -pubkey option to 'spkac'. 16101 16102 *Steve Henson* 16103 16104 * Use a less unusual form of the Miller-Rabin primality test (it used 16105 a binary algorithm for exponentiation integrated into the Miller-Rabin 16106 loop, our standard modexp algorithms are faster). 16107 16108 *Bodo Moeller* 16109 16110 * Support for the EBCDIC character set completed. 16111 16112 *Martin Kraemer <Martin.Kraemer@Mch.SNI.De>* 16113 16114 * Source code cleanups: use const where appropriate, eliminate casts, 16115 use `void *` instead of `char *` in lhash. 16116 16117 *Ulf Möller* 16118 16119 * Bugfix: ssl3_send_server_key_exchange was not restartable 16120 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of 16121 this the server could overwrite ephemeral keys that the client 16122 has already seen). 16123 16124 *Bodo Moeller* 16125 16126 * Turn DSA_is_prime into a macro that calls BN_is_prime, 16127 using 50 iterations of the Rabin-Miller test. 16128 16129 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50 16130 iterations of the Rabin-Miller test as required by the appendix 16131 to FIPS PUB 186[-1]) instead of DSA_is_prime. 16132 As BN_is_prime_fasttest includes trial division, DSA parameter 16133 generation becomes much faster. 16134 16135 This implies a change for the callback functions in DSA_is_prime 16136 and DSA_generate_parameters: The callback function is called once 16137 for each positive witness in the Rabin-Miller test, not just 16138 occasionally in the inner loop; and the parameters to the 16139 callback function now provide an iteration count for the outer 16140 loop rather than for the current invocation of the inner loop. 16141 DSA_generate_parameters additionally can call the callback 16142 function with an 'iteration count' of -1, meaning that a 16143 candidate has passed the trial division test (when q is generated 16144 from an application-provided seed, trial division is skipped). 16145 16146 *Bodo Moeller* 16147 16148 * New function BN_is_prime_fasttest that optionally does trial 16149 division before starting the Rabin-Miller test and has 16150 an additional BN_CTX * argument (whereas BN_is_prime always 16151 has to allocate at least one BN_CTX). 16152 'callback(1, -1, cb_arg)' is called when a number has passed the 16153 trial division stage. 16154 16155 *Bodo Moeller* 16156 16157 * Fix for bug in CRL encoding. The validity dates weren't being handled 16158 as ASN1_TIME. 16159 16160 *Steve Henson* 16161 16162 * New -pkcs12 option to CA.pl script to write out a PKCS#12 file. 16163 16164 *Steve Henson* 16165 16166 * New function BN_pseudo_rand(). 16167 16168 *Ulf Möller* 16169 16170 * Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable) 16171 bignum version of BN_from_montgomery() with the working code from 16172 SSLeay 0.9.0 (the word based version is faster anyway), and clean up 16173 the comments. 16174 16175 *Ulf Möller* 16176 16177 * Avoid a race condition in s2_clnt.c (function get_server_hello) that 16178 made it impossible to use the same SSL_SESSION data structure in 16179 SSL2 clients in multiple threads. 16180 16181 *Bodo Moeller* 16182 16183 * The return value of RAND_load_file() no longer counts bytes obtained 16184 by stat(). RAND_load_file(..., -1) is new and uses the complete file 16185 to seed the PRNG (previously an explicit byte count was required). 16186 16187 *Ulf Möller, Bodo Möller* 16188 16189 * Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes 16190 used `char *` instead of `void *` and had casts all over the place. 16191 16192 *Steve Henson* 16193 16194 * Make BN_generate_prime() return NULL on error if ret!=NULL. 16195 16196 *Ulf Möller* 16197 16198 * Retain source code compatibility for BN_prime_checks macro: 16199 BN_is_prime(..., BN_prime_checks, ...) now uses 16200 BN_prime_checks_for_size to determine the appropriate number of 16201 Rabin-Miller iterations. 16202 16203 *Ulf Möller* 16204 16205 * Diffie-Hellman uses "safe" primes: DH_check() return code renamed to 16206 DH_CHECK_P_NOT_SAFE_PRIME. 16207 (Check if this is true? OpenPGP calls them "strong".) 16208 16209 *Ulf Möller* 16210 16211 * Merge the functionality of "dh" and "gendh" programs into a new program 16212 "dhparam". The old programs are retained for now but will handle DH keys 16213 (instead of parameters) in future. 16214 16215 *Steve Henson* 16216 16217 * Make the ciphers, s_server and s_client programs check the return values 16218 when a new cipher list is set. 16219 16220 *Steve Henson* 16221 16222 * Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit 16223 ciphers. Before when the 56bit ciphers were enabled the sorting was 16224 wrong. 16225 16226 The syntax for the cipher sorting has been extended to support sorting by 16227 cipher-strength (using the strength_bits hard coded in the tables). 16228 The new command is `@STRENGTH` (see also `doc/apps/ciphers.pod`). 16229 16230 Fix a bug in the cipher-command parser: when supplying a cipher command 16231 string with an "undefined" symbol (neither command nor alphanumeric 16232 *A-Za-z0-9*, ssl_set_cipher_list used to hang in an endless loop. Now 16233 an error is flagged. 16234 16235 Due to the strength-sorting extension, the code of the 16236 ssl_create_cipher_list() function was completely rearranged. I hope that 16237 the readability was also increased :-) 16238 16239 *Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>* 16240 16241 * Minor change to 'x509' utility. The -CAcreateserial option now uses 1 16242 for the first serial number and places 2 in the serial number file. This 16243 avoids problems when the root CA is created with serial number zero and 16244 the first user certificate has the same issuer name and serial number 16245 as the root CA. 16246 16247 *Steve Henson* 16248 16249 * Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses 16250 the new code. Add documentation for this stuff. 16251 16252 *Steve Henson* 16253 16254 * Changes to X509_ATTRIBUTE utilities. These have been renamed from 16255 `X509_*()` to `X509at_*()` on the grounds that they don't handle X509 16256 structures and behave in an analogous way to the X509v3 functions: 16257 they shouldn't be called directly but wrapper functions should be used 16258 instead. 16259 16260 So we also now have some wrapper functions that call the X509at functions 16261 when passed certificate requests. (TO DO: similar things can be done with 16262 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other 16263 things. Some of these need some d2i or i2d and print functionality 16264 because they handle more complex structures.) 16265 16266 *Steve Henson* 16267 16268 * Add missing #ifndefs that caused missing symbols when building libssl 16269 as a shared library without RSA. Use #ifndef NO_SSL2 instead of 16270 NO_RSA in `ssl/s2*.c`. 16271 16272 *Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf Möller* 16273 16274 * Precautions against using the PRNG uninitialized: RAND_bytes() now 16275 has a return value which indicates the quality of the random data 16276 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's 16277 error queue. New function RAND_pseudo_bytes() generates output that is 16278 guaranteed to be unique but not unpredictable. RAND_add is like 16279 RAND_seed, but takes an extra argument for an entropy estimate 16280 (RAND_seed always assumes full entropy). 16281 16282 *Ulf Möller* 16283 16284 * Do more iterations of Rabin-Miller probable prime test (specifically, 16285 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes 16286 instead of only 2 for all lengths; see BN_prime_checks_for_size definition 16287 in crypto/bn/bn_prime.c for the complete table). This guarantees a 16288 false-positive rate of at most 2^-80 for random input. 16289 16290 *Bodo Moeller* 16291 16292 * Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs. 16293 16294 *Bodo Moeller* 16295 16296 * New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain 16297 in the 0.9.5 release), this returns the chain 16298 from an X509_CTX structure with a dup of the stack and all 16299 the X509 reference counts upped: so the stack will exist 16300 after X509_CTX_cleanup() has been called. Modify pkcs12.c 16301 to use this. 16302 16303 Also make SSL_SESSION_print() print out the verify return 16304 code. 16305 16306 *Steve Henson* 16307 16308 * Add manpage for the pkcs12 command. Also change the default 16309 behaviour so MAC iteration counts are used unless the new 16310 -nomaciter option is used. This improves file security and 16311 only older versions of MSIE (4.0 for example) need it. 16312 16313 *Steve Henson* 16314 16315 * Honor the no-xxx Configure options when creating .DEF files. 16316 16317 *Ulf Möller* 16318 16319 * Add PKCS#10 attributes to field table: challengePassword, 16320 unstructuredName and unstructuredAddress. These are taken from 16321 draft PKCS#9 v2.0 but are compatible with v1.2 provided no 16322 international characters are used. 16323 16324 More changes to X509_ATTRIBUTE code: allow the setting of types 16325 based on strings. Remove the 'loc' parameter when adding 16326 attributes because these will be a SET OF encoding which is sorted 16327 in ASN1 order. 16328 16329 *Steve Henson* 16330 16331 * Initial changes to the 'req' utility to allow request generation 16332 automation. This will allow an application to just generate a template 16333 file containing all the field values and have req construct the 16334 request. 16335 16336 Initial support for X509_ATTRIBUTE handling. Stacks of these are 16337 used all over the place including certificate requests and PKCS#7 16338 structures. They are currently handled manually where necessary with 16339 some primitive wrappers for PKCS#7. The new functions behave in a 16340 manner analogous to the X509 extension functions: they allow 16341 attributes to be looked up by NID and added. 16342 16343 Later something similar to the X509V3 code would be desirable to 16344 automatically handle the encoding, decoding and printing of the 16345 more complex types. The string types like challengePassword can 16346 be handled by the string table functions. 16347 16348 Also modified the multi byte string table handling. Now there is 16349 a 'global mask' which masks out certain types. The table itself 16350 can use the flag STABLE_NO_MASK to ignore the mask setting: this 16351 is useful when for example there is only one permissible type 16352 (as in countryName) and using the mask might result in no valid 16353 types at all. 16354 16355 *Steve Henson* 16356 16357 * Clean up 'Finished' handling, and add functions SSL_get_finished and 16358 SSL_get_peer_finished to allow applications to obtain the latest 16359 Finished messages sent to the peer or expected from the peer, 16360 respectively. (SSL_get_peer_finished is usually the Finished message 16361 actually received from the peer, otherwise the protocol will be aborted.) 16362 16363 As the Finished message are message digests of the complete handshake 16364 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can 16365 be used for external authentication procedures when the authentication 16366 provided by SSL/TLS is not desired or is not enough. 16367 16368 *Bodo Moeller* 16369 16370 * Enhanced support for Alpha Linux is added. Now ./config checks if 16371 the host supports BWX extension and if Compaq C is present on the 16372 $PATH. Just exploiting of the BWX extension results in 20-30% 16373 performance kick for some algorithms, e.g. DES and RC4 to mention 16374 a couple. Compaq C in turn generates ~20% faster code for MD5 and 16375 SHA1. 16376 16377 *Andy Polyakov* 16378 16379 * Add support for MS "fast SGC". This is arguably a violation of the 16380 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with 16381 weak crypto and after checking the certificate is SGC a second one 16382 with strong crypto. MS SGC stops the first handshake after receiving 16383 the server certificate message and sends a second client hello. Since 16384 a server will typically do all the time consuming operations before 16385 expecting any further messages from the client (server key exchange 16386 is the most expensive) there is little difference between the two. 16387 16388 To get OpenSSL to support MS SGC we have to permit a second client 16389 hello message after we have sent server done. In addition we have to 16390 reset the MAC if we do get this second client hello. 16391 16392 *Steve Henson* 16393 16394 * Add a function 'd2i_AutoPrivateKey()' this will automatically decide 16395 if a DER encoded private key is RSA or DSA traditional format. Changed 16396 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional" 16397 format DER encoded private key. Newer code should use PKCS#8 format which 16398 has the key type encoded in the ASN1 structure. Added DER private key 16399 support to pkcs8 application. 16400 16401 *Steve Henson* 16402 16403 * SSL 3/TLS 1 servers now don't request certificates when an anonymous 16404 ciphersuites has been selected (as required by the SSL 3/TLS 1 16405 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT 16406 is set, we interpret this as a request to violate the specification 16407 (the worst that can happen is a handshake failure, and 'correct' 16408 behaviour would result in a handshake failure anyway). 16409 16410 *Bodo Moeller* 16411 16412 * In SSL_CTX_add_session, take into account that there might be multiple 16413 SSL_SESSION structures with the same session ID (e.g. when two threads 16414 concurrently obtain them from an external cache). 16415 The internal cache can handle only one SSL_SESSION with a given ID, 16416 so if there's a conflict, we now throw out the old one to achieve 16417 consistency. 16418 16419 *Bodo Moeller* 16420 16421 * Add OIDs for idea and blowfish in CBC mode. This will allow both 16422 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to 16423 some routines that use cipher OIDs: some ciphers do not have OIDs 16424 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for 16425 example. 16426 16427 *Steve Henson* 16428 16429 * Simplify the trust setting structure and code. Now we just have 16430 two sequences of OIDs for trusted and rejected settings. These will 16431 typically have values the same as the extended key usage extension 16432 and any application specific purposes. 16433 16434 The trust checking code now has a default behaviour: it will just 16435 check for an object with the same NID as the passed id. Functions can 16436 be provided to override either the default behaviour or the behaviour 16437 for a given id. SSL client, server and email already have functions 16438 in place for compatibility: they check the NID and also return "trusted" 16439 if the certificate is self signed. 16440 16441 *Steve Henson* 16442 16443 * Add d2i,i2d bio/fp functions for PrivateKey: these convert the 16444 traditional format into an EVP_PKEY structure. 16445 16446 *Steve Henson* 16447 16448 * Add a password callback function PEM_cb() which either prompts for 16449 a password if usr_data is NULL or otherwise assumes it is a null 16450 terminated password. Allow passwords to be passed on command line 16451 environment or config files in a few more utilities. 16452 16453 *Steve Henson* 16454 16455 * Add a bunch of DER and PEM functions to handle PKCS#8 format private 16456 keys. Add some short names for PKCS#8 PBE algorithms and allow them 16457 to be specified on the command line for the pkcs8 and pkcs12 utilities. 16458 Update documentation. 16459 16460 *Steve Henson* 16461 16462 * Support for ASN1 "NULL" type. This could be handled before by using 16463 ASN1_TYPE but there wasn't any function that would try to read a NULL 16464 and produce an error if it couldn't. For compatibility we also have 16465 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and 16466 don't allocate anything because they don't need to. 16467 16468 *Steve Henson* 16469 16470 * Initial support for MacOS is now provided. Examine INSTALL.MacOS 16471 for details. 16472 16473 *Andy Polyakov, Roy Woods <roy@centicsystems.ca>* 16474 16475 * Rebuild of the memory allocation routines used by OpenSSL code and 16476 possibly others as well. The purpose is to make an interface that 16477 provide hooks so anyone can build a separate set of allocation and 16478 deallocation routines to be used by OpenSSL, for example memory 16479 pool implementations, or something else, which was previously hard 16480 since Malloc(), Realloc() and Free() were defined as macros having 16481 the values malloc, realloc and free, respectively (except for Win32 16482 compilations). The same is provided for memory debugging code. 16483 OpenSSL already comes with functionality to find memory leaks, but 16484 this gives people a chance to debug other memory problems. 16485 16486 With these changes, a new set of functions and macros have appeared: 16487 16488 CRYPTO_set_mem_debug_functions() [F] 16489 CRYPTO_get_mem_debug_functions() [F] 16490 CRYPTO_dbg_set_options() [F] 16491 CRYPTO_dbg_get_options() [F] 16492 CRYPTO_malloc_debug_init() [M] 16493 16494 The memory debug functions are NULL by default, unless the library 16495 is compiled with CRYPTO_MDEBUG or friends is defined. If someone 16496 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which 16497 gives the standard debugging functions that come with OpenSSL) or 16498 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions 16499 provided by the library user) must be used. When the standard 16500 debugging functions are used, CRYPTO_dbg_set_options can be used to 16501 request additional information: 16502 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting 16503 the CRYPTO_MDEBUG_xxx macro when compiling the library. 16504 16505 Also, things like CRYPTO_set_mem_functions will always give the 16506 expected result (the new set of functions is used for allocation 16507 and deallocation) at all times, regardless of platform and compiler 16508 options. 16509 16510 To finish it up, some functions that were never use in any other 16511 way than through macros have a new API and new semantic: 16512 16513 CRYPTO_dbg_malloc() 16514 CRYPTO_dbg_realloc() 16515 CRYPTO_dbg_free() 16516 16517 All macros of value have retained their old syntax. 16518 16519 *Richard Levitte and Bodo Moeller* 16520 16521 * Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the 16522 ordering of SMIMECapabilities wasn't in "strength order" and there 16523 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature 16524 algorithm. 16525 16526 *Steve Henson* 16527 16528 * Some ASN1 types with illegal zero length encoding (INTEGER, 16529 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines. 16530 16531 *Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson* 16532 16533 * Merge in my S/MIME library for OpenSSL. This provides a simple 16534 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough 16535 functionality to handle multipart/signed properly) and a utility 16536 called 'smime' to call all this stuff. This is based on code I 16537 originally wrote for Celo who have kindly allowed it to be 16538 included in OpenSSL. 16539 16540 *Steve Henson* 16541 16542 * Add variants des_set_key_checked and des_set_key_unchecked of 16543 des_set_key (aka des_key_sched). Global variable des_check_key 16544 decides which of these is called by des_set_key; this way 16545 des_check_key behaves as it always did, but applications and 16546 the library itself, which was buggy for des_check_key == 1, 16547 have a cleaner way to pick the version they need. 16548 16549 *Bodo Moeller* 16550 16551 * New function PKCS12_newpass() which changes the password of a 16552 PKCS12 structure. 16553 16554 *Steve Henson* 16555 16556 * Modify X509_TRUST and X509_PURPOSE so it also uses a static and 16557 dynamic mix. In both cases the ids can be used as an index into the 16558 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add() 16559 functions so they accept a list of the field values and the 16560 application doesn't need to directly manipulate the X509_TRUST 16561 structure. 16562 16563 *Steve Henson* 16564 16565 * Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't 16566 need initialising. 16567 16568 *Steve Henson* 16569 16570 * Modify the way the V3 extension code looks up extensions. This now 16571 works in a similar way to the object code: we have some "standard" 16572 extensions in a static table which is searched with OBJ_bsearch() 16573 and the application can add dynamic ones if needed. The file 16574 crypto/x509v3/ext_dat.h now has the info: this file needs to be 16575 updated whenever a new extension is added to the core code and kept 16576 in ext_nid order. There is a simple program 'tabtest.c' which checks 16577 this. New extensions are not added too often so this file can readily 16578 be maintained manually. 16579 16580 There are two big advantages in doing things this way. The extensions 16581 can be looked up immediately and no longer need to be "added" using 16582 X509V3_add_standard_extensions(): this function now does nothing. 16583 Side note: I get *lots* of email saying the extension code doesn't 16584 work because people forget to call this function. 16585 Also no dynamic allocation is done unless new extensions are added: 16586 so if we don't add custom extensions there is no need to call 16587 X509V3_EXT_cleanup(). 16588 16589 *Steve Henson* 16590 16591 * Modify enc utility's salting as follows: make salting the default. Add a 16592 magic header, so unsalted files fail gracefully instead of just decrypting 16593 to garbage. This is because not salting is a big security hole, so people 16594 should be discouraged from doing it. 16595 16596 *Ben Laurie* 16597 16598 * Fixes and enhancements to the 'x509' utility. It allowed a message 16599 digest to be passed on the command line but it only used this 16600 parameter when signing a certificate. Modified so all relevant 16601 operations are affected by the digest parameter including the 16602 -fingerprint and -x509toreq options. Also -x509toreq choked if a 16603 DSA key was used because it didn't fix the digest. 16604 16605 *Steve Henson* 16606 16607 * Initial certificate chain verify code. Currently tests the untrusted 16608 certificates for consistency with the verify purpose (which is set 16609 when the X509_STORE_CTX structure is set up) and checks the pathlength. 16610 16611 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour: 16612 this is because it will reject chains with invalid extensions whereas 16613 every previous version of OpenSSL and SSLeay made no checks at all. 16614 16615 Trust code: checks the root CA for the relevant trust settings. Trust 16616 settings have an initial value consistent with the verify purpose: e.g. 16617 if the verify purpose is for SSL client use it expects the CA to be 16618 trusted for SSL client use. However the default value can be changed to 16619 permit custom trust settings: one example of this would be to only trust 16620 certificates from a specific "secure" set of CAs. 16621 16622 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions 16623 which should be used for version portability: especially since the 16624 verify structure is likely to change more often now. 16625 16626 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions 16627 to set them. If not set then assume SSL clients will verify SSL servers 16628 and vice versa. 16629 16630 Two new options to the verify program: -untrusted allows a set of 16631 untrusted certificates to be passed in and -purpose which sets the 16632 intended purpose of the certificate. If a purpose is set then the 16633 new chain verify code is used to check extension consistency. 16634 16635 *Steve Henson* 16636 16637 * Support for the authority information access extension. 16638 16639 *Steve Henson* 16640 16641 * Modify RSA and DSA PEM read routines to transparently handle 16642 PKCS#8 format private keys. New *_PUBKEY_* functions that handle 16643 public keys in a format compatible with certificate 16644 SubjectPublicKeyInfo structures. Unfortunately there were already 16645 functions called *_PublicKey_* which used various odd formats so 16646 these are retained for compatibility: however the DSA variants were 16647 never in a public release so they have been deleted. Changed dsa/rsa 16648 utilities to handle the new format: note no releases ever handled public 16649 keys so we should be OK. 16650 16651 The primary motivation for this change is to avoid the same fiasco 16652 that dogs private keys: there are several incompatible private key 16653 formats some of which are standard and some OpenSSL specific and 16654 require various evil hacks to allow partial transparent handling and 16655 even then it doesn't work with DER formats. Given the option anything 16656 other than PKCS#8 should be dumped: but the other formats have to 16657 stay in the name of compatibility. 16658 16659 With public keys and the benefit of hindsight one standard format 16660 is used which works with EVP_PKEY, RSA or DSA structures: though 16661 it clearly returns an error if you try to read the wrong kind of key. 16662 16663 Added a -pubkey option to the 'x509' utility to output the public key. 16664 Also rename the `EVP_PKEY_get_*()` to `EVP_PKEY_rget_*()` 16665 (renamed to `EVP_PKEY_get1_*()` in the OpenSSL 0.9.5 release) and add 16666 `EVP_PKEY_rset_*()` functions (renamed to `EVP_PKEY_set1_*()`) 16667 that do the same as the `EVP_PKEY_assign_*()` except they up the 16668 reference count of the added key (they don't "swallow" the 16669 supplied key). 16670 16671 *Steve Henson* 16672 16673 * Fixes to crypto/x509/by_file.c the code to read in certificates and 16674 CRLs would fail if the file contained no certificates or no CRLs: 16675 added a new function to read in both types and return the number 16676 read: this means that if none are read it will be an error. The 16677 DER versions of the certificate and CRL reader would always fail 16678 because it isn't possible to mix certificates and CRLs in DER format 16679 without choking one or the other routine. Changed this to just read 16680 a certificate: this is the best we can do. Also modified the code 16681 in `apps/verify.c` to take notice of return codes: it was previously 16682 attempting to read in certificates from NULL pointers and ignoring 16683 any errors: this is one reason why the cert and CRL reader seemed 16684 to work. It doesn't check return codes from the default certificate 16685 routines: these may well fail if the certificates aren't installed. 16686 16687 *Steve Henson* 16688 16689 * Code to support otherName option in GeneralName. 16690 16691 *Steve Henson* 16692 16693 * First update to verify code. Change the verify utility 16694 so it warns if it is passed a self signed certificate: 16695 for consistency with the normal behaviour. X509_verify 16696 has been modified to it will now verify a self signed 16697 certificate if *exactly* the same certificate appears 16698 in the store: it was previously impossible to trust a 16699 single self signed certificate. This means that: 16700 openssl verify ss.pem 16701 now gives a warning about a self signed certificate but 16702 openssl verify -CAfile ss.pem ss.pem 16703 is OK. 16704 16705 *Steve Henson* 16706 16707 * For servers, store verify_result in SSL_SESSION data structure 16708 (and add it to external session representation). 16709 This is needed when client certificate verifications fails, 16710 but an application-provided verification callback (set by 16711 SSL_CTX_set_cert_verify_callback) allows accepting the session 16712 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK 16713 but returns 1): When the session is reused, we have to set 16714 ssl->verify_result to the appropriate error code to avoid 16715 security holes. 16716 16717 *Bodo Moeller, problem pointed out by Lutz Jaenicke* 16718 16719 * Fix a bug in the new PKCS#7 code: it didn't consider the 16720 case in PKCS7_dataInit() where the signed PKCS7 structure 16721 didn't contain any existing data because it was being created. 16722 16723 *Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson* 16724 16725 * Add a salt to the key derivation routines in enc.c. This 16726 forms the first 8 bytes of the encrypted file. Also add a 16727 -S option to allow a salt to be input on the command line. 16728 16729 *Steve Henson* 16730 16731 * New function X509_cmp(). Oddly enough there wasn't a function 16732 to compare two certificates. We do this by working out the SHA1 16733 hash and comparing that. X509_cmp() will be needed by the trust 16734 code. 16735 16736 *Steve Henson* 16737 16738 * SSL_get1_session() is like SSL_get_session(), but increments 16739 the reference count in the SSL_SESSION returned. 16740 16741 *Geoff Thorpe <geoff@eu.c2.net>* 16742 16743 * Fix for 'req': it was adding a null to request attributes. 16744 Also change the X509_LOOKUP and X509_INFO code to handle 16745 certificate auxiliary information. 16746 16747 *Steve Henson* 16748 16749 * Add support for 40 and 64 bit RC2 and RC4 algorithms: document 16750 the 'enc' command. 16751 16752 *Steve Henson* 16753 16754 * Add the possibility to add extra information to the memory leak 16755 detecting output, to form tracebacks, showing from where each 16756 allocation was originated: CRYPTO_push_info("constant string") adds 16757 the string plus current file name and line number to a per-thread 16758 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info() 16759 is like calling CYRPTO_pop_info() until the stack is empty. 16760 Also updated memory leak detection code to be multi-thread-safe. 16761 16762 *Richard Levitte* 16763 16764 * Add options -text and -noout to pkcs7 utility and delete the 16765 encryption options which never did anything. Update docs. 16766 16767 *Steve Henson* 16768 16769 * Add options to some of the utilities to allow the pass phrase 16770 to be included on either the command line (not recommended on 16771 OSes like Unix) or read from the environment. Update the 16772 manpages and fix a few bugs. 16773 16774 *Steve Henson* 16775 16776 * Add a few manpages for some of the openssl commands. 16777 16778 *Steve Henson* 16779 16780 * Fix the -revoke option in ca. It was freeing up memory twice, 16781 leaking and not finding already revoked certificates. 16782 16783 *Steve Henson* 16784 16785 * Extensive changes to support certificate auxiliary information. 16786 This involves the use of X509_CERT_AUX structure and X509_AUX 16787 functions. An X509_AUX function such as PEM_read_X509_AUX() 16788 can still read in a certificate file in the usual way but it 16789 will also read in any additional "auxiliary information". By 16790 doing things this way a fair degree of compatibility can be 16791 retained: existing certificates can have this information added 16792 using the new 'x509' options. 16793 16794 Current auxiliary information includes an "alias" and some trust 16795 settings. The trust settings will ultimately be used in enhanced 16796 certificate chain verification routines: currently a certificate 16797 can only be trusted if it is self signed and then it is trusted 16798 for all purposes. 16799 16800 *Steve Henson* 16801 16802 * Fix assembler for Alpha (tested only on DEC OSF not Linux or `*BSD`). 16803 The problem was that one of the replacement routines had not been working 16804 since SSLeay releases. For now the offending routine has been replaced 16805 with non-optimised assembler. Even so, this now gives around 95% 16806 performance improvement for 1024 bit RSA signs. 16807 16808 *Mark Cox* 16809 16810 * Hack to fix PKCS#7 decryption when used with some unorthodox RC2 16811 handling. Most clients have the effective key size in bits equal to 16812 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key. 16813 A few however don't do this and instead use the size of the decrypted key 16814 to determine the RC2 key length and the AlgorithmIdentifier to determine 16815 the effective key length. In this case the effective key length can still 16816 be 40 bits but the key length can be 168 bits for example. This is fixed 16817 by manually forcing an RC2 key into the EVP_PKEY structure because the 16818 EVP code can't currently handle unusual RC2 key sizes: it always assumes 16819 the key length and effective key length are equal. 16820 16821 *Steve Henson* 16822 16823 * Add a bunch of functions that should simplify the creation of 16824 X509_NAME structures. Now you should be able to do: 16825 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0); 16826 and have it automatically work out the correct field type and fill in 16827 the structures. The more adventurous can try: 16828 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0); 16829 and it will (hopefully) work out the correct multibyte encoding. 16830 16831 *Steve Henson* 16832 16833 * Change the 'req' utility to use the new field handling and multibyte 16834 copy routines. Before the DN field creation was handled in an ad hoc 16835 way in req, ca, and x509 which was rather broken and didn't support 16836 BMPStrings or UTF8Strings. Since some software doesn't implement 16837 BMPStrings or UTF8Strings yet, they can be enabled using the config file 16838 using the dirstring_type option. See the new comment in the default 16839 openssl.cnf for more info. 16840 16841 *Steve Henson* 16842 16843 * Make crypto/rand/md_rand.c more robust: 16844 - Assure unique random numbers after fork(). 16845 - Make sure that concurrent threads access the global counter and 16846 md serializably so that we never lose entropy in them 16847 or use exactly the same state in multiple threads. 16848 Access to the large state is not always serializable because 16849 the additional locking could be a performance killer, and 16850 md should be large enough anyway. 16851 16852 *Bodo Moeller* 16853 16854 * New file `apps/app_rand.c` with commonly needed functionality 16855 for handling the random seed file. 16856 16857 Use the random seed file in some applications that previously did not: 16858 ca, 16859 dsaparam -genkey (which also ignored its '-rand' option), 16860 s_client, 16861 s_server, 16862 x509 (when signing). 16863 Except on systems with /dev/urandom, it is crucial to have a random 16864 seed file at least for key creation, DSA signing, and for DH exchanges; 16865 for RSA signatures we could do without one. 16866 16867 gendh and gendsa (unlike genrsa) used to read only the first byte 16868 of each file listed in the '-rand' option. The function as previously 16869 found in genrsa is now in app_rand.c and is used by all programs 16870 that support '-rand'. 16871 16872 *Bodo Moeller* 16873 16874 * In RAND_write_file, use mode 0600 for creating files; 16875 don't just chmod when it may be too late. 16876 16877 *Bodo Moeller* 16878 16879 * Report an error from X509_STORE_load_locations 16880 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed. 16881 16882 *Bill Perry* 16883 16884 * New function ASN1_mbstring_copy() this copies a string in either 16885 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format 16886 into an ASN1_STRING type. A mask of permissible types is passed 16887 and it chooses the "minimal" type to use or an error if not type 16888 is suitable. 16889 16890 *Steve Henson* 16891 16892 * Add function equivalents to the various macros in asn1.h. The old 16893 macros are retained with an `M_` prefix. Code inside the library can 16894 use the `M_` macros. External code (including the openssl utility) 16895 should *NOT* in order to be "shared library friendly". 16896 16897 *Steve Henson* 16898 16899 * Add various functions that can check a certificate's extensions 16900 to see if it usable for various purposes such as SSL client, 16901 server or S/MIME and CAs of these types. This is currently 16902 VERY EXPERIMENTAL but will ultimately be used for certificate chain 16903 verification. Also added a -purpose flag to x509 utility to 16904 print out all the purposes. 16905 16906 *Steve Henson* 16907 16908 * Add a CRYPTO_EX_DATA to X509 certificate structure and associated 16909 functions. 16910 16911 *Steve Henson* 16912 16913 * New `X509V3_{X509,CRL,REVOKED}_get_d2i()` functions. These will search 16914 for, obtain and decode and extension and obtain its critical flag. 16915 This allows all the necessary extension code to be handled in a 16916 single function call. 16917 16918 *Steve Henson* 16919 16920 * RC4 tune-up featuring 30-40% performance improvement on most RISC 16921 platforms. See crypto/rc4/rc4_enc.c for further details. 16922 16923 *Andy Polyakov* 16924 16925 * New -noout option to asn1parse. This causes no output to be produced 16926 its main use is when combined with -strparse and -out to extract data 16927 from a file (which may not be in ASN.1 format). 16928 16929 *Steve Henson* 16930 16931 * Fix for pkcs12 program. It was hashing an invalid certificate pointer 16932 when producing the local key id. 16933 16934 *Richard Levitte <levitte@stacken.kth.se>* 16935 16936 * New option -dhparam in s_server. This allows a DH parameter file to be 16937 stated explicitly. If it is not stated then it tries the first server 16938 certificate file. The previous behaviour hard coded the filename 16939 "server.pem". 16940 16941 *Steve Henson* 16942 16943 * Add -pubin and -pubout options to the rsa and dsa commands. These allow 16944 a public key to be input or output. For example: 16945 openssl rsa -in key.pem -pubout -out pubkey.pem 16946 Also added necessary DSA public key functions to handle this. 16947 16948 *Steve Henson* 16949 16950 * Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained 16951 in the message. This was handled by allowing 16952 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it. 16953 16954 *Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>* 16955 16956 * Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null 16957 to the end of the strings whereas this didn't. This would cause problems 16958 if strings read with d2i_ASN1_bytes() were later modified. 16959 16960 *Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>* 16961 16962 * Fix for base64 decode bug. When a base64 bio reads only one line of 16963 data and it contains EOF it will end up returning an error. This is 16964 caused by input 46 bytes long. The cause is due to the way base64 16965 BIOs find the start of base64 encoded data. They do this by trying a 16966 trial decode on each line until they find one that works. When they 16967 do a flag is set and it starts again knowing it can pass all the 16968 data directly through the decoder. Unfortunately it doesn't reset 16969 the context it uses. This means that if EOF is reached an attempt 16970 is made to pass two EOFs through the context and this causes the 16971 resulting error. This can also cause other problems as well. As is 16972 usual with these problems it takes *ages* to find and the fix is 16973 trivial: move one line. 16974 16975 *Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer)* 16976 16977 * Ugly workaround to get s_client and s_server working under Windows. The 16978 old code wouldn't work because it needed to select() on sockets and the 16979 tty (for keypresses and to see if data could be written). Win32 only 16980 supports select() on sockets so we select() with a 1s timeout on the 16981 sockets and then see if any characters are waiting to be read, if none 16982 are present then we retry, we also assume we can always write data to 16983 the tty. This isn't nice because the code then blocks until we've 16984 received a complete line of data and it is effectively polling the 16985 keyboard at 1s intervals: however it's quite a bit better than not 16986 working at all :-) A dedicated Windows application might handle this 16987 with an event loop for example. 16988 16989 *Steve Henson* 16990 16991 * Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign 16992 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions 16993 will be called when RSA_sign() and RSA_verify() are used. This is useful 16994 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available. 16995 For this to work properly RSA_public_decrypt() and RSA_private_encrypt() 16996 should *not* be used: RSA_sign() and RSA_verify() must be used instead. 16997 This necessitated the support of an extra signature type NID_md5_sha1 16998 for SSL signatures and modifications to the SSL library to use it instead 16999 of calling RSA_public_decrypt() and RSA_private_encrypt(). 17000 17001 *Steve Henson* 17002 17003 * Add new -verify -CAfile and -CApath options to the crl program, these 17004 will lookup a CRL issuers certificate and verify the signature in a 17005 similar way to the verify program. Tidy up the crl program so it 17006 no longer accesses structures directly. Make the ASN1 CRL parsing a bit 17007 less strict. It will now permit CRL extensions even if it is not 17008 a V2 CRL: this will allow it to tolerate some broken CRLs. 17009 17010 *Steve Henson* 17011 17012 * Initialize all non-automatic variables each time one of the openssl 17013 sub-programs is started (this is necessary as they may be started 17014 multiple times from the "OpenSSL>" prompt). 17015 17016 *Lennart Bang, Bodo Moeller* 17017 17018 * Preliminary compilation option RSA_NULL which disables RSA crypto without 17019 removing all other RSA functionality (this is what NO_RSA does). This 17020 is so (for example) those in the US can disable those operations covered 17021 by the RSA patent while allowing storage and parsing of RSA keys and RSA 17022 key generation. 17023 17024 *Steve Henson* 17025 17026 * Non-copying interface to BIO pairs. 17027 (still largely untested) 17028 17029 *Bodo Moeller* 17030 17031 * New function ASN1_tag2str() to convert an ASN1 tag to a descriptive 17032 ASCII string. This was handled independently in various places before. 17033 17034 *Steve Henson* 17035 17036 * New functions UTF8_getc() and UTF8_putc() that parse and generate 17037 UTF8 strings a character at a time. 17038 17039 *Steve Henson* 17040 17041 * Use client_version from client hello to select the protocol 17042 (s23_srvr.c) and for RSA client key exchange verification 17043 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications. 17044 17045 *Bodo Moeller* 17046 17047 * Add various utility functions to handle SPKACs, these were previously 17048 handled by poking round in the structure internals. Added new function 17049 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to 17050 print, verify and generate SPKACs. Based on an original idea from 17051 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified. 17052 17053 *Steve Henson* 17054 17055 * RIPEMD160 is operational on all platforms and is back in 'make test'. 17056 17057 *Andy Polyakov* 17058 17059 * Allow the config file extension section to be overwritten on the 17060 command line. Based on an original idea from Massimiliano Pala 17061 <madwolf@comune.modena.it>. The new option is called -extensions 17062 and can be applied to ca, req and x509. Also -reqexts to override 17063 the request extensions in req and -crlexts to override the crl extensions 17064 in ca. 17065 17066 *Steve Henson* 17067 17068 * Add new feature to the SPKAC handling in ca. Now you can include 17069 the same field multiple times by preceding it by "XXXX." for example: 17070 1.OU="Unit name 1" 17071 2.OU="Unit name 2" 17072 this is the same syntax as used in the req config file. 17073 17074 *Steve Henson* 17075 17076 * Allow certificate extensions to be added to certificate requests. These 17077 are specified in a 'req_extensions' option of the req section of the 17078 config file. They can be printed out with the -text option to req but 17079 are otherwise ignored at present. 17080 17081 *Steve Henson* 17082 17083 * Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first 17084 data read consists of only the final block it would not decrypted because 17085 EVP_CipherUpdate() would correctly report zero bytes had been decrypted. 17086 A misplaced 'break' also meant the decrypted final block might not be 17087 copied until the next read. 17088 17089 *Steve Henson* 17090 17091 * Initial support for DH_METHOD. Again based on RSA_METHOD. Also added 17092 a few extra parameters to the DH structure: these will be useful if 17093 for example we want the value of 'q' or implement X9.42 DH. 17094 17095 *Steve Henson* 17096 17097 * Initial support for DSA_METHOD. This is based on the RSA_METHOD and 17098 provides hooks that allow the default DSA functions or functions on a 17099 "per key" basis to be replaced. This allows hardware acceleration and 17100 hardware key storage to be handled without major modification to the 17101 library. Also added low-level modexp hooks and CRYPTO_EX structure and 17102 associated functions. 17103 17104 *Steve Henson* 17105 17106 * Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO 17107 as "read only": it can't be written to and the buffer it points to will 17108 not be freed. Reading from a read only BIO is much more efficient than 17109 a normal memory BIO. This was added because there are several times when 17110 an area of memory needs to be read from a BIO. The previous method was 17111 to create a memory BIO and write the data to it, this results in two 17112 copies of the data and an O(n^2) reading algorithm. There is a new 17113 function BIO_new_mem_buf() which creates a read only memory BIO from 17114 an area of memory. Also modified the PKCS#7 routines to use read only 17115 memory BIOs. 17116 17117 *Steve Henson* 17118 17119 * Bugfix: ssl23_get_client_hello did not work properly when called in 17120 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of 17121 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read, 17122 but a retry condition occurred while trying to read the rest. 17123 17124 *Bodo Moeller* 17125 17126 * The PKCS7_ENC_CONTENT_new() function was setting the content type as 17127 NID_pkcs7_encrypted by default: this was wrong since this should almost 17128 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle 17129 the encrypted data type: this is a more sensible place to put it and it 17130 allows the PKCS#12 code to be tidied up that duplicated this 17131 functionality. 17132 17133 *Steve Henson* 17134 17135 * Changed obj_dat.pl script so it takes its input and output files on 17136 the command line. This should avoid shell escape redirection problems 17137 under Win32. 17138 17139 *Steve Henson* 17140 17141 * Initial support for certificate extension requests, these are included 17142 in things like Xenroll certificate requests. Included functions to allow 17143 extensions to be obtained and added. 17144 17145 *Steve Henson* 17146 17147 * -crlf option to s_client and s_server for sending newlines as 17148 CRLF (as required by many protocols). 17149 17150 *Bodo Moeller* 17151 17152### Changes between 0.9.3a and 0.9.4 [09 Aug 1999] 17153 17154 * Install libRSAglue.a when OpenSSL is built with RSAref. 17155 17156 *Ralf S. Engelschall* 17157 17158 * A few more `#ifndef NO_FP_API / #endif` pairs for consistency. 17159 17160 *Andrija Antonijevic <TheAntony2@bigfoot.com>* 17161 17162 * Fix -startdate and -enddate (which was missing) arguments to 'ca' 17163 program. 17164 17165 *Steve Henson* 17166 17167 * New function DSA_dup_DH, which duplicates DSA parameters/keys as 17168 DH parameters/keys (q is lost during that conversion, but the resulting 17169 DH parameters contain its length). 17170 17171 For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is 17172 much faster than DH_generate_parameters (which creates parameters 17173 where `p = 2*q + 1`), and also the smaller q makes DH computations 17174 much more efficient (160-bit exponentiation instead of 1024-bit 17175 exponentiation); so this provides a convenient way to support DHE 17176 ciphersuites in SSL/TLS servers (see ssl/ssltest.c). It is of 17177 utter importance to use 17178 SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 17179 or 17180 SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 17181 when such DH parameters are used, because otherwise small subgroup 17182 attacks may become possible! 17183 17184 *Bodo Moeller* 17185 17186 * Avoid memory leak in i2d_DHparams. 17187 17188 *Bodo Moeller* 17189 17190 * Allow the -k option to be used more than once in the enc program: 17191 this allows the same encrypted message to be read by multiple recipients. 17192 17193 *Steve Henson* 17194 17195 * New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts 17196 an ASN1_OBJECT to a text string. If the "no_name" parameter is set then 17197 it will always use the numerical form of the OID, even if it has a short 17198 or long name. 17199 17200 *Steve Henson* 17201 17202 * Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp 17203 method only got called if p,q,dmp1,dmq1,iqmp components were present, 17204 otherwise bn_mod_exp was called. In the case of hardware keys for example 17205 no private key components need be present and it might store extra data 17206 in the RSA structure, which cannot be accessed from bn_mod_exp. 17207 By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for 17208 private key operations. 17209 17210 *Steve Henson* 17211 17212 * Added support for SPARC Linux. 17213 17214 *Andy Polyakov* 17215 17216 * pem_password_cb function type incompatibly changed from 17217 typedef int pem_password_cb(char *buf, int size, int rwflag); 17218 to 17219 ....(char *buf, int size, int rwflag, void *userdata); 17220 so that applications can pass data to their callbacks: 17221 The `PEM[_ASN1]_{read,write}...` functions and macros now take an 17222 additional void * argument, which is just handed through whenever 17223 the password callback is called. 17224 17225 *Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller* 17226 17227 New function SSL_CTX_set_default_passwd_cb_userdata. 17228 17229 Compatibility note: As many C implementations push function arguments 17230 onto the stack in reverse order, the new library version is likely to 17231 interoperate with programs that have been compiled with the old 17232 pem_password_cb definition (PEM_whatever takes some data that 17233 happens to be on the stack as its last argument, and the callback 17234 just ignores this garbage); but there is no guarantee whatsoever that 17235 this will work. 17236 17237 * The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=... 17238 (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused 17239 problems not only on Windows, but also on some Unix platforms. 17240 To avoid problematic command lines, these definitions are now in an 17241 auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl 17242 for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds). 17243 17244 *Bodo Moeller* 17245 17246 * MIPS III/IV assembler module is reimplemented. 17247 17248 *Andy Polyakov* 17249 17250 * More DES library cleanups: remove references to srand/rand and 17251 delete an unused file. 17252 17253 *Ulf Möller* 17254 17255 * Add support for the free Netwide assembler (NASM) under Win32, 17256 since not many people have MASM (ml) and it can be hard to obtain. 17257 This is currently experimental but it seems to work OK and pass all 17258 the tests. Check out INSTALL.W32 for info. 17259 17260 *Steve Henson* 17261 17262 * Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections 17263 without temporary keys kept an extra copy of the server key, 17264 and connections with temporary keys did not free everything in case 17265 of an error. 17266 17267 *Bodo Moeller* 17268 17269 * New function RSA_check_key and new openssl rsa option -check 17270 for verifying the consistency of RSA keys. 17271 17272 *Ulf Moeller, Bodo Moeller* 17273 17274 * Various changes to make Win32 compile work: 17275 1. Casts to avoid "loss of data" warnings in p5_crpt2.c 17276 2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned 17277 comparison" warnings. 17278 3. Add `sk_<TYPE>_sort` to DEF file generator and do make update. 17279 17280 *Steve Henson* 17281 17282 * Add a debugging option to PKCS#5 v2 key generation function: when 17283 you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and 17284 derived keys are printed to stderr. 17285 17286 *Steve Henson* 17287 17288 * Copy the flags in ASN1_STRING_dup(). 17289 17290 *Roman E. Pavlov <pre@mo.msk.ru>* 17291 17292 * The x509 application mishandled signing requests containing DSA 17293 keys when the signing key was also DSA and the parameters didn't match. 17294 17295 It was supposed to omit the parameters when they matched the signing key: 17296 the verifying software was then supposed to automatically use the CA's 17297 parameters if they were absent from the end user certificate. 17298 17299 Omitting parameters is no longer recommended. The test was also 17300 the wrong way round! This was probably due to unusual behaviour in 17301 EVP_cmp_parameters() which returns 1 if the parameters match. 17302 This meant that parameters were omitted when they *didn't* match and 17303 the certificate was useless. Certificates signed with 'ca' didn't have 17304 this bug. 17305 17306 *Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>* 17307 17308 * Memory leak checking (-DCRYPTO_MDEBUG) had some problems. 17309 The interface is as follows: 17310 Applications can use 17311 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(), 17312 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop(); 17313 "off" is now the default. 17314 The library internally uses 17315 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(), 17316 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on() 17317 to disable memory-checking temporarily. 17318 17319 Some inconsistent states that previously were possible (and were 17320 even the default) are now avoided. 17321 17322 -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time 17323 with each memory chunk allocated; this is occasionally more helpful 17324 than just having a counter. 17325 17326 -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID. 17327 17328 -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future 17329 extensions. 17330 17331 *Bodo Moeller* 17332 17333 * Introduce "mode" for SSL structures (with defaults in SSL_CTX), 17334 which largely parallels "options", but is for changing API behaviour, 17335 whereas "options" are about protocol behaviour. 17336 Initial "mode" flags are: 17337 17338 SSL_MODE_ENABLE_PARTIAL_WRITE Allow SSL_write to report success when 17339 a single record has been written. 17340 SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER Don't insist that SSL_write 17341 retries use the same buffer location. 17342 (But all of the contents must be 17343 copied!) 17344 17345 *Bodo Moeller* 17346 17347 * Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options 17348 worked. 17349 17350 * Fix problems with no-hmac etc. 17351 17352 *Ulf Möller, pointed out by Brian Wellington <bwelling@tislabs.com>* 17353 17354 * New functions RSA_get_default_method(), RSA_set_method() and 17355 RSA_get_method(). These allows replacement of RSA_METHODs without having 17356 to mess around with the internals of an RSA structure. 17357 17358 *Steve Henson* 17359 17360 * Fix memory leaks in DSA_do_sign and DSA_is_prime. 17361 Also really enable memory leak checks in openssl.c and in some 17362 test programs. 17363 17364 *Chad C. Mulligan, Bodo Moeller* 17365 17366 * Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess 17367 up the length of negative integers. This has now been simplified to just 17368 store the length when it is first determined and use it later, rather 17369 than trying to keep track of where data is copied and updating it to 17370 point to the end. 17371 *Steve Henson, reported by Brien Wheeler <bwheeler@authentica-security.com>* 17372 17373 * Add a new function PKCS7_signatureVerify. This allows the verification 17374 of a PKCS#7 signature but with the signing certificate passed to the 17375 function itself. This contrasts with PKCS7_dataVerify which assumes the 17376 certificate is present in the PKCS#7 structure. This isn't always the 17377 case: certificates can be omitted from a PKCS#7 structure and be 17378 distributed by "out of band" means (such as a certificate database). 17379 17380 *Steve Henson* 17381 17382 * Complete the `PEM_*` macros with DECLARE_PEM versions to replace the 17383 function prototypes in pem.h, also change util/mkdef.pl to add the 17384 necessary function names. 17385 17386 *Steve Henson* 17387 17388 * mk1mf.pl (used by Windows builds) did not properly read the 17389 options set by Configure in the top level Makefile, and Configure 17390 was not even able to write more than one option correctly. 17391 Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended. 17392 17393 *Bodo Moeller* 17394 17395 * New functions CONF_load_bio() and CONF_load_fp() to allow a config 17396 file to be loaded from a BIO or FILE pointer. The BIO version will 17397 for example allow memory BIOs to contain config info. 17398 17399 *Steve Henson* 17400 17401 * New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS. 17402 Whoever hopes to achieve shared-library compatibility across versions 17403 must use this, not the compile-time macro. 17404 (Exercise 0.9.4: Which is the minimum library version required by 17405 such programs?) 17406 Note: All this applies only to multi-threaded programs, others don't 17407 need locks. 17408 17409 *Bodo Moeller* 17410 17411 * Add missing case to s3_clnt.c state machine -- one of the new SSL tests 17412 through a BIO pair triggered the default case, i.e. 17413 SSLerr(...,SSL_R_UNKNOWN_STATE). 17414 17415 *Bodo Moeller* 17416 17417 * New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications 17418 can use the SSL library even if none of the specific BIOs is 17419 appropriate. 17420 17421 *Bodo Moeller* 17422 17423 * Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value 17424 for the encoded length. 17425 17426 *Jeon KyoungHo <khjeon@sds.samsung.co.kr>* 17427 17428 * Add initial documentation of the X509V3 functions. 17429 17430 *Steve Henson* 17431 17432 * Add a new pair of functions PEM_write_PKCS8PrivateKey() and 17433 PEM_write_bio_PKCS8PrivateKey() that are equivalent to 17434 PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more 17435 secure PKCS#8 private key format with a high iteration count. 17436 17437 *Steve Henson* 17438 17439 * Fix determination of Perl interpreter: A perl or perl5 17440 *directory* in $PATH was also accepted as the interpreter. 17441 17442 *Ralf S. Engelschall* 17443 17444 * Fix demos/sign/sign.c: well there wasn't anything strictly speaking 17445 wrong with it but it was very old and did things like calling 17446 PEM_ASN1_read() directly and used MD5 for the hash not to mention some 17447 unusual formatting. 17448 17449 *Steve Henson* 17450 17451 * Fix demos/selfsign.c: it used obsolete and deleted functions, changed 17452 to use the new extension code. 17453 17454 *Steve Henson* 17455 17456 * Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c 17457 with macros. This should make it easier to change their form, add extra 17458 arguments etc. Fix a few PEM prototypes which didn't have cipher as a 17459 constant. 17460 17461 *Steve Henson* 17462 17463 * Add to configuration table a new entry that can specify an alternative 17464 name for unistd.h (for pre-POSIX systems); we need this for NeXTstep, 17465 according to Mark Crispin <MRC@Panda.COM>. 17466 17467 *Bodo Moeller* 17468 17469 * DES CBC did not update the IV. Weird. 17470 17471 *Ben Laurie* 17472lse 17473 des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does. 17474 Changing the behaviour of the former might break existing programs -- 17475 where IV updating is needed, des_ncbc_encrypt can be used. 17476ndif 17477 17478 * When bntest is run from "make test" it drives bc to check its 17479 calculations, as well as internally checking them. If an internal check 17480 fails, it needs to cause bc to give a non-zero result or make test carries 17481 on without noticing the failure. Fixed. 17482 17483 *Ben Laurie* 17484 17485 * DES library cleanups. 17486 17487 *Ulf Möller* 17488 17489 * Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be 17490 used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit 17491 ciphers. NOTE: although the key derivation function has been verified 17492 against some published test vectors it has not been extensively tested 17493 yet. Added a -v2 "cipher" option to pkcs8 application to allow the use 17494 of v2.0. 17495 17496 *Steve Henson* 17497 17498 * Instead of "mkdir -p", which is not fully portable, use new 17499 Perl script "util/mkdir-p.pl". 17500 17501 *Bodo Moeller* 17502 17503 * Rewrite the way password based encryption (PBE) is handled. It used to 17504 assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter 17505 structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms 17506 but doesn't apply to PKCS#5 v2.0 where it can be something else. Now 17507 the 'parameter' field of the AlgorithmIdentifier is passed to the 17508 underlying key generation function so it must do its own ASN1 parsing. 17509 This has also changed the EVP_PBE_CipherInit() function which now has a 17510 'parameter' argument instead of literal salt and iteration count values 17511 and the function EVP_PBE_ALGOR_CipherInit() has been deleted. 17512 17513 *Steve Henson* 17514 17515 * Support for PKCS#5 v1.5 compatible password based encryption algorithms 17516 and PKCS#8 functionality. New 'pkcs8' application linked to openssl. 17517 Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE 17518 KEY" because this clashed with PKCS#8 unencrypted string. Since this 17519 value was just used as a "magic string" and not used directly its 17520 value doesn't matter. 17521 17522 *Steve Henson* 17523 17524 * Introduce some semblance of const correctness to BN. Shame C doesn't 17525 support mutable. 17526 17527 *Ben Laurie* 17528 17529 * "linux-sparc64" configuration (ultrapenguin). 17530 17531 *Ray Miller <ray.miller@oucs.ox.ac.uk>* 17532 "linux-sparc" configuration. 17533 17534 *Christian Forster <fo@hawo.stw.uni-erlangen.de>* 17535 17536 * config now generates no-xxx options for missing ciphers. 17537 17538 *Ulf Möller* 17539 17540 * Support the EBCDIC character set (work in progress). 17541 File ebcdic.c not yet included because it has a different license. 17542 17543 *Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>* 17544 17545 * Support BS2000/OSD-POSIX. 17546 17547 *Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>* 17548 17549 * Make callbacks for key generation use `void *` instead of `char *`. 17550 17551 *Ben Laurie* 17552 17553 * Make S/MIME samples compile (not yet tested). 17554 17555 *Ben Laurie* 17556 17557 * Additional typesafe stacks. 17558 17559 *Ben Laurie* 17560 17561 * New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x). 17562 17563 *Bodo Moeller* 17564 17565### Changes between 0.9.3 and 0.9.3a [29 May 1999] 17566 17567 * New configuration variant "sco5-gcc". 17568 17569 * Updated some demos. 17570 17571 *Sean O Riordain, Wade Scholine* 17572 17573 * Add missing BIO_free at exit of pkcs12 application. 17574 17575 *Wu Zhigang* 17576 17577 * Fix memory leak in conf.c. 17578 17579 *Steve Henson* 17580 17581 * Updates for Win32 to assembler version of MD5. 17582 17583 *Steve Henson* 17584 17585 * Set #! path to perl in `apps/der_chop` to where we found it 17586 instead of using a fixed path. 17587 17588 *Bodo Moeller* 17589 17590 * SHA library changes for irix64-mips4-cc. 17591 17592 *Andy Polyakov* 17593 17594 * Improvements for VMS support. 17595 17596 *Richard Levitte* 17597 17598### Changes between 0.9.2b and 0.9.3 [24 May 1999] 17599 17600 * Bignum library bug fix. IRIX 6 passes "make test" now! 17601 This also avoids the problems with SC4.2 and unpatched SC5. 17602 17603 *Andy Polyakov <appro@fy.chalmers.se>* 17604 17605 * New functions sk_num, sk_value and sk_set to replace the previous macros. 17606 These are required because of the typesafe stack would otherwise break 17607 existing code. If old code used a structure member which used to be STACK 17608 and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with 17609 sk_num or sk_value it would produce an error because the num, data members 17610 are not present in STACK_OF. Now it just produces a warning. sk_set 17611 replaces the old method of assigning a value to sk_value 17612 (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code 17613 that does this will no longer work (and should use sk_set instead) but 17614 this could be regarded as a "questionable" behaviour anyway. 17615 17616 *Steve Henson* 17617 17618 * Fix most of the other PKCS#7 bugs. The "experimental" code can now 17619 correctly handle encrypted S/MIME data. 17620 17621 *Steve Henson* 17622 17623 * Change type of various DES function arguments from des_cblock 17624 (which means, in function argument declarations, pointer to char) 17625 to des_cblock * (meaning pointer to array with 8 char elements), 17626 which allows the compiler to do more typechecking; it was like 17627 that back in SSLeay, but with lots of ugly casts. 17628 17629 Introduce new type const_des_cblock. 17630 17631 *Bodo Moeller* 17632 17633 * Reorganise the PKCS#7 library and get rid of some of the more obvious 17634 problems: find RecipientInfo structure that matches recipient certificate 17635 and initialise the ASN1 structures properly based on passed cipher. 17636 17637 *Steve Henson* 17638 17639 * Belatedly make the BN tests actually check the results. 17640 17641 *Ben Laurie* 17642 17643 * Fix the encoding and decoding of negative ASN1 INTEGERS and conversion 17644 to and from BNs: it was completely broken. New compilation option 17645 NEG_PUBKEY_BUG to allow for some broken certificates that encode public 17646 key elements as negative integers. 17647 17648 *Steve Henson* 17649 17650 * Reorganize and speed up MD5. 17651 17652 *Andy Polyakov <appro@fy.chalmers.se>* 17653 17654 * VMS support. 17655 17656 *Richard Levitte <richard@levitte.org>* 17657 17658 * New option -out to asn1parse to allow the parsed structure to be 17659 output to a file. This is most useful when combined with the -strparse 17660 option to examine the output of things like OCTET STRINGS. 17661 17662 *Steve Henson* 17663 17664 * Make SSL library a little more fool-proof by not requiring any longer 17665 that `SSL_set_{accept,connect}_state` be called before 17666 `SSL_{accept,connect}` may be used (`SSL_set_..._state` is omitted 17667 in many applications because usually everything *appeared* to work as 17668 intended anyway -- now it really works as intended). 17669 17670 *Bodo Moeller* 17671 17672 * Move openssl.cnf out of lib/. 17673 17674 *Ulf Möller* 17675 17676 * Fix various things to let OpenSSL even pass "egcc -pipe -O2 -Wall 17677 -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes 17678 -Wmissing-declarations -Wnested-externs -Winline" with EGCS 1.1.2+ 17679 17680 *Ralf S. Engelschall* 17681 17682 * Various fixes to the EVP and PKCS#7 code. It may now be able to 17683 handle PKCS#7 enveloped data properly. 17684 17685 *Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve* 17686 17687 * Create a duplicate of the SSL_CTX's CERT in SSL_new instead of 17688 copying pointers. The cert_st handling is changed by this in 17689 various ways (and thus what used to be known as ctx->default_cert 17690 is now called ctx->cert, since we don't resort to `s->ctx->[default_]cert` 17691 any longer when s->cert does not give us what we need). 17692 ssl_cert_instantiate becomes obsolete by this change. 17693 As soon as we've got the new code right (possibly it already is?), 17694 we have solved a couple of bugs of the earlier code where s->cert 17695 was used as if it could not have been shared with other SSL structures. 17696 17697 Note that using the SSL API in certain dirty ways now will result 17698 in different behaviour than observed with earlier library versions: 17699 Changing settings for an `SSL_CTX *ctx` after having done s = SSL_new(ctx) 17700 does not influence s as it used to. 17701 17702 In order to clean up things more thoroughly, inside SSL_SESSION 17703 we don't use CERT any longer, but a new structure SESS_CERT 17704 that holds per-session data (if available); currently, this is 17705 the peer's certificate chain and, for clients, the server's certificate 17706 and temporary key. CERT holds only those values that can have 17707 meaningful defaults in an SSL_CTX. 17708 17709 *Bodo Moeller* 17710 17711 * New function X509V3_EXT_i2d() to create an X509_EXTENSION structure 17712 from the internal representation. Various PKCS#7 fixes: remove some 17713 evil casts and set the enc_dig_alg field properly based on the signing 17714 key type. 17715 17716 *Steve Henson* 17717 17718 * Allow PKCS#12 password to be set from the command line or the 17719 environment. Let 'ca' get its config file name from the environment 17720 variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req' 17721 and 'x509'). 17722 17723 *Steve Henson* 17724 17725 * Allow certificate policies extension to use an IA5STRING for the 17726 organization field. This is contrary to the PKIX definition but 17727 VeriSign uses it and IE5 only recognises this form. Document 'x509' 17728 extension option. 17729 17730 *Steve Henson* 17731 17732 * Add PEDANTIC compiler flag to allow compilation with gcc -pedantic, 17733 without disallowing inline assembler and the like for non-pedantic builds. 17734 17735 *Ben Laurie* 17736 17737 * Support Borland C++ builder. 17738 17739 *Janez Jere <jj@void.si>, modified by Ulf Möller* 17740 17741 * Support Mingw32. 17742 17743 *Ulf Möller* 17744 17745 * SHA-1 cleanups and performance enhancements. 17746 17747 *Andy Polyakov <appro@fy.chalmers.se>* 17748 17749 * Sparc v8plus assembler for the bignum library. 17750 17751 *Andy Polyakov <appro@fy.chalmers.se>* 17752 17753 * Accept any -xxx and +xxx compiler options in Configure. 17754 17755 *Ulf Möller* 17756 17757 * Update HPUX configuration. 17758 17759 *Anonymous* 17760 17761 * Add missing `sk_<type>_unshift()` function to safestack.h 17762 17763 *Ralf S. Engelschall* 17764 17765 * New function SSL_CTX_use_certificate_chain_file that sets the 17766 "extra_cert"s in addition to the certificate. (This makes sense 17767 only for "PEM" format files, as chains as a whole are not 17768 DER-encoded.) 17769 17770 *Bodo Moeller* 17771 17772 * Support verify_depth from the SSL API. 17773 x509_vfy.c had what can be considered an off-by-one-error: 17774 Its depth (which was not part of the external interface) 17775 was actually counting the number of certificates in a chain; 17776 now it really counts the depth. 17777 17778 *Bodo Moeller* 17779 17780 * Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used 17781 instead of X509err, which often resulted in confusing error 17782 messages since the error codes are not globally unique 17783 (e.g. an alleged error in ssl3_accept when a certificate 17784 didn't match the private key). 17785 17786 * New function SSL_CTX_set_session_id_context that allows to set a default 17787 value (so that you don't need SSL_set_session_id_context for each 17788 connection using the SSL_CTX). 17789 17790 *Bodo Moeller* 17791 17792 * OAEP decoding bug fix. 17793 17794 *Ulf Möller* 17795 17796 * Support INSTALL_PREFIX for package builders, as proposed by 17797 David Harris. 17798 17799 *Bodo Moeller* 17800 17801 * New Configure options "threads" and "no-threads". For systems 17802 where the proper compiler options are known (currently Solaris 17803 and Linux), "threads" is the default. 17804 17805 *Bodo Moeller* 17806 17807 * New script util/mklink.pl as a faster substitute for util/mklink.sh. 17808 17809 *Bodo Moeller* 17810 17811 * Install various scripts to $(OPENSSLDIR)/misc, not to 17812 $(INSTALLTOP)/bin -- they shouldn't clutter directories 17813 such as /usr/local/bin. 17814 17815 *Bodo Moeller* 17816 17817 * "make linux-shared" to build shared libraries. 17818 17819 *Niels Poppe <niels@netbox.org>* 17820 17821 * New Configure option `no-<cipher>` (rsa, idea, rc5, ...). 17822 17823 *Ulf Möller* 17824 17825 * Add the PKCS#12 API documentation to openssl.txt. Preliminary support for 17826 extension adding in x509 utility. 17827 17828 *Steve Henson* 17829 17830 * Remove NOPROTO sections and error code comments. 17831 17832 *Ulf Möller* 17833 17834 * Partial rewrite of the DEF file generator to now parse the ANSI 17835 prototypes. 17836 17837 *Steve Henson* 17838 17839 * New Configure options --prefix=DIR and --openssldir=DIR. 17840 17841 *Ulf Möller* 17842 17843 * Complete rewrite of the error code script(s). It is all now handled 17844 by one script at the top level which handles error code gathering, 17845 header rewriting and C source file generation. It should be much better 17846 than the old method: it now uses a modified version of Ulf's parser to 17847 read the ANSI prototypes in all header files (thus the old K&R definitions 17848 aren't needed for error creation any more) and do a better job of 17849 translating function codes into names. The old 'ASN1 error code embedded 17850 in a comment' is no longer necessary and it doesn't use .err files which 17851 have now been deleted. Also the error code call doesn't have to appear all 17852 on one line (which resulted in some large lines...). 17853 17854 *Steve Henson* 17855 17856 * Change #include filenames from `<foo.h>` to `<openssl/foo.h>`. 17857 17858 *Bodo Moeller* 17859 17860 * Change behaviour of ssl2_read when facing length-0 packets: Don't return 17861 0 (which usually indicates a closed connection), but continue reading. 17862 17863 *Bodo Moeller* 17864 17865 * Fix some race conditions. 17866 17867 *Bodo Moeller* 17868 17869 * Add support for CRL distribution points extension. Add Certificate 17870 Policies and CRL distribution points documentation. 17871 17872 *Steve Henson* 17873 17874 * Move the autogenerated header file parts to crypto/opensslconf.h. 17875 17876 *Ulf Möller* 17877 17878 * Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of 17879 8 of keying material. Merlin has also confirmed interop with this fix 17880 between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0. 17881 17882 *Merlin Hughes <merlin@baltimore.ie>* 17883 17884 * Fix lots of warnings. 17885 17886 *Richard Levitte <levitte@stacken.kth.se>* 17887 17888 * In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if 17889 the directory spec didn't end with a LIST_SEPARATOR_CHAR. 17890 17891 *Richard Levitte <levitte@stacken.kth.se>* 17892 17893 * Fix problems with sizeof(long) == 8. 17894 17895 *Andy Polyakov <appro@fy.chalmers.se>* 17896 17897 * Change functions to ANSI C. 17898 17899 *Ulf Möller* 17900 17901 * Fix typos in error codes. 17902 17903 *Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf Möller* 17904 17905 * Remove defunct assembler files from Configure. 17906 17907 *Ulf Möller* 17908 17909 * SPARC v8 assembler BIGNUM implementation. 17910 17911 *Andy Polyakov <appro@fy.chalmers.se>* 17912 17913 * Support for Certificate Policies extension: both print and set. 17914 Various additions to support the r2i method this uses. 17915 17916 *Steve Henson* 17917 17918 * A lot of constification, and fix a bug in X509_NAME_oneline() that could 17919 return a const string when you are expecting an allocated buffer. 17920 17921 *Ben Laurie* 17922 17923 * Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE 17924 types DirectoryString and DisplayText. 17925 17926 *Steve Henson* 17927 17928 * Add code to allow r2i extensions to access the configuration database, 17929 add an LHASH database driver and add several ctx helper functions. 17930 17931 *Steve Henson* 17932 17933 * Fix an evil bug in bn_expand2() which caused various BN functions to 17934 fail when they extended the size of a BIGNUM. 17935 17936 *Steve Henson* 17937 17938 * Various utility functions to handle SXNet extension. Modify mkdef.pl to 17939 support typesafe stack. 17940 17941 *Steve Henson* 17942 17943 * Fix typo in SSL_[gs]et_options(). 17944 17945 *Nils Frostberg <nils@medcom.se>* 17946 17947 * Delete various functions and files that belonged to the (now obsolete) 17948 old X509V3 handling code. 17949 17950 *Steve Henson* 17951 17952 * New Configure option "rsaref". 17953 17954 *Ulf Möller* 17955 17956 * Don't auto-generate pem.h. 17957 17958 *Bodo Moeller* 17959 17960 * Introduce type-safe ASN.1 SETs. 17961 17962 *Ben Laurie* 17963 17964 * Convert various additional casted stacks to type-safe STACK_OF() variants. 17965 17966 *Ben Laurie, Ralf S. Engelschall, Steve Henson* 17967 17968 * Introduce type-safe STACKs. This will almost certainly break lots of code 17969 that links with OpenSSL (well at least cause lots of warnings), but fear 17970 not: the conversion is trivial, and it eliminates loads of evil casts. A 17971 few STACKed things have been converted already. Feel free to convert more. 17972 In the fullness of time, I'll do away with the STACK type altogether. 17973 17974 *Ben Laurie* 17975 17976 * Add `openssl ca -revoke <certfile>` facility which revokes a certificate 17977 specified in `<certfile>` by updating the entry in the index.txt file. 17978 This way one no longer has to edit the index.txt file manually for 17979 revoking a certificate. The -revoke option does the gory details now. 17980 17981 *Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall* 17982 17983 * Fix `openssl crl -noout -text` combination where `-noout` killed the 17984 `-text` option at all and this way the `-noout -text` combination was 17985 inconsistent in `openssl crl` with the friends in `openssl x509|rsa|dsa`. 17986 17987 *Ralf S. Engelschall* 17988 17989 * Make sure a corresponding plain text error message exists for the 17990 X509_V_ERR_CERT_REVOKED/23 error number which can occur when a 17991 verify callback function determined that a certificate was revoked. 17992 17993 *Ralf S. Engelschall* 17994 17995 * Bugfix: In test/testenc, don't test `openssl <cipher>` for 17996 ciphers that were excluded, e.g. by -DNO_IDEA. Also, test 17997 all available ciphers including rc5, which was forgotten until now. 17998 In order to let the testing shell script know which algorithms 17999 are available, a new (up to now undocumented) command 18000 `openssl list-cipher-commands` is used. 18001 18002 *Bodo Moeller* 18003 18004 * Bugfix: s_client occasionally would sleep in select() when 18005 it should have checked SSL_pending() first. 18006 18007 *Bodo Moeller* 18008 18009 * New functions DSA_do_sign and DSA_do_verify to provide access to 18010 the raw DSA values prior to ASN.1 encoding. 18011 18012 *Ulf Möller* 18013 18014 * Tweaks to Configure 18015 18016 *Niels Poppe <niels@netbox.org>* 18017 18018 * Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support, 18019 yet... 18020 18021 *Steve Henson* 18022 18023 * New variables $(RANLIB) and $(PERL) in the Makefiles. 18024 18025 *Ulf Möller* 18026 18027 * New config option to avoid instructions that are illegal on the 80386. 18028 The default code is faster, but requires at least a 486. 18029 18030 *Ulf Möller* 18031 18032 * Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and 18033 SSL2_SERVER_VERSION (not used at all) macros, which are now the 18034 same as SSL2_VERSION anyway. 18035 18036 *Bodo Moeller* 18037 18038 * New "-showcerts" option for s_client. 18039 18040 *Bodo Moeller* 18041 18042 * Still more PKCS#12 integration. Add pkcs12 application to openssl 18043 application. Various cleanups and fixes. 18044 18045 *Steve Henson* 18046 18047 * More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and 18048 modify error routines to work internally. Add error codes and PBE init 18049 to library startup routines. 18050 18051 *Steve Henson* 18052 18053 * Further PKCS#12 integration. Added password based encryption, PKCS#8 and 18054 packing functions to asn1 and evp. Changed function names and error 18055 codes along the way. 18056 18057 *Steve Henson* 18058 18059 * PKCS12 integration: and so it begins... First of several patches to 18060 slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12 18061 objects to objects.h 18062 18063 *Steve Henson* 18064 18065 * Add a new 'indent' option to some X509V3 extension code. Initial ASN1 18066 and display support for Thawte strong extranet extension. 18067 18068 *Steve Henson* 18069 18070 * Add LinuxPPC support. 18071 18072 *Jeff Dubrule <igor@pobox.org>* 18073 18074 * Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to 18075 bn_div_words in alpha.s. 18076 18077 *Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie* 18078 18079 * Make sure the RSA OAEP test is skipped under -DRSAref because 18080 OAEP isn't supported when OpenSSL is built with RSAref. 18081 18082 *Ulf Moeller <ulf@fitug.de>* 18083 18084 * Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h 18085 so they no longer are missing under -DNOPROTO. 18086 18087 *Soren S. Jorvang <soren@t.dk>* 18088 18089### Changes between 0.9.1c and 0.9.2b [22 Mar 1999] 18090 18091 * Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still 18092 doesn't work when the session is reused. Coming soon! 18093 18094 *Ben Laurie* 18095 18096 * Fix a security hole, that allows sessions to be reused in the wrong 18097 context thus bypassing client cert protection! All software that uses 18098 client certs and session caches in multiple contexts NEEDS PATCHING to 18099 allow session reuse! A fuller solution is in the works. 18100 18101 *Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)* 18102 18103 * Some more source tree cleanups (removed obsolete files 18104 crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed 18105 permission on "config" script to be executable) and a fix for the INSTALL 18106 document. 18107 18108 *Ulf Moeller <ulf@fitug.de>* 18109 18110 * Remove some legacy and erroneous uses of malloc, free instead of 18111 Malloc, Free. 18112 18113 *Lennart Bang <lob@netstream.se>, with minor changes by Steve* 18114 18115 * Make rsa_oaep_test return non-zero on error. 18116 18117 *Ulf Moeller <ulf@fitug.de>* 18118 18119 * Add support for native Solaris shared libraries. Configure 18120 solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice 18121 if someone would make that last step automatic. 18122 18123 *Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>* 18124 18125 * ctx_size was not built with the right compiler during "make links". Fixed. 18126 18127 *Ben Laurie* 18128 18129 * Change the meaning of 'ALL' in the cipher list. It now means "everything 18130 except NULL ciphers". This means the default cipher list will no longer 18131 enable NULL ciphers. They need to be specifically enabled e.g. with 18132 the string "DEFAULT:eNULL". 18133 18134 *Steve Henson* 18135 18136 * Fix to RSA private encryption routines: if p < q then it would 18137 occasionally produce an invalid result. This will only happen with 18138 externally generated keys because OpenSSL (and SSLeay) ensure p > q. 18139 18140 *Steve Henson* 18141 18142 * Be less restrictive and allow also `perl util/perlpath.pl 18143 /path/to/bin/perl` in addition to `perl util/perlpath.pl /path/to/bin`, 18144 because this way one can also use an interpreter named `perl5` (which is 18145 usually the name of Perl 5.xxx on platforms where an Perl 4.x is still 18146 installed as `perl`). 18147 18148 *Matthias Loepfe <Matthias.Loepfe@adnovum.ch>* 18149 18150 * Let util/clean-depend.pl work also with older Perl 5.00x versions. 18151 18152 *Matthias Loepfe <Matthias.Loepfe@adnovum.ch>* 18153 18154 * Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add 18155 advapi32.lib to Win32 build and change the pem test comparison 18156 to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the 18157 suggestion). Fix misplaced ASNI prototypes and declarations in evp.h 18158 and crypto/des/ede_cbcm_enc.c. 18159 18160 *Steve Henson* 18161 18162 * DES quad checksum was broken on big-endian architectures. Fixed. 18163 18164 *Ben Laurie* 18165 18166 * Comment out two functions in bio.h that aren't implemented. Fix up the 18167 Win32 test batch file so it (might) work again. The Win32 test batch file 18168 is horrible: I feel ill.... 18169 18170 *Steve Henson* 18171 18172 * Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected 18173 in e_os.h. Audit of header files to check ANSI and non ANSI 18174 sections: 10 functions were absent from non ANSI section and not exported 18175 from Windows DLLs. Fixed up libeay.num for new functions. 18176 18177 *Steve Henson* 18178 18179 * Make `openssl version` output lines consistent. 18180 18181 *Ralf S. Engelschall* 18182 18183 * Fix Win32 symbol export lists for BIO functions: Added 18184 BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data 18185 to ms/libeay{16,32}.def. 18186 18187 *Ralf S. Engelschall* 18188 18189 * Second round of fixing the OpenSSL perl/ stuff. It now at least compiled 18190 fine under Unix and passes some trivial tests I've now added. But the 18191 whole stuff is horribly incomplete, so a README.1ST with a disclaimer was 18192 added to make sure no one expects that this stuff really works in the 18193 OpenSSL 0.9.2 release. Additionally I've started to clean the XS sources 18194 up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and 18195 openssl_bio.xs. 18196 18197 *Ralf S. Engelschall* 18198 18199 * Fix the generation of two part addresses in perl. 18200 18201 *Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie* 18202 18203 * Add config entry for Linux on MIPS. 18204 18205 *John Tobey <jtobey@channel1.com>* 18206 18207 * Make links whenever Configure is run, unless we are on Windoze. 18208 18209 *Ben Laurie* 18210 18211 * Permit extensions to be added to CRLs using crl_section in openssl.cnf. 18212 Currently only issuerAltName and AuthorityKeyIdentifier make any sense 18213 in CRLs. 18214 18215 *Steve Henson* 18216 18217 * Add a useful kludge to allow package maintainers to specify compiler and 18218 other platforms details on the command line without having to patch the 18219 Configure script every time: One now can use 18220 `perl Configure <id>:<details>`, 18221 i.e. platform ids are allowed to have details appended 18222 to them (separated by colons). This is treated as there would be a static 18223 pre-configured entry in Configure's %table under key `<id>` with value 18224 `<details>` and `perl Configure <id>` is called. So, when you want to 18225 perform a quick test-compile under FreeBSD 3.1 with pgcc and without 18226 assembler stuff you can use `perl Configure "FreeBSD-elf:pgcc:-O6:::"` 18227 now, which overrides the FreeBSD-elf entry on-the-fly. 18228 18229 *Ralf S. Engelschall* 18230 18231 * Disable new TLS1 ciphersuites by default: they aren't official yet. 18232 18233 *Ben Laurie* 18234 18235 * Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified 18236 on the `perl Configure ...` command line. This way one can compile 18237 OpenSSL libraries with Position Independent Code (PIC) which is needed 18238 for linking it into DSOs. 18239 18240 *Ralf S. Engelschall* 18241 18242 * Remarkably, export ciphers were totally broken and no-one had noticed! 18243 Fixed. 18244 18245 *Ben Laurie* 18246 18247 * Cleaned up the LICENSE document: The official contact for any license 18248 questions now is the OpenSSL core team under openssl-core@openssl.org. 18249 And add a paragraph about the dual-license situation to make sure people 18250 recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply 18251 to the OpenSSL toolkit. 18252 18253 *Ralf S. Engelschall* 18254 18255 * General source tree makefile cleanups: Made `making xxx in yyy...` 18256 display consistent in the source tree and replaced `/bin/rm` by `rm`. 18257 Additionally cleaned up the `make links` target: Remove unnecessary 18258 semicolons, subsequent redundant removes, inline point.sh into mklink.sh 18259 to speed processing and no longer clutter the display with confusing 18260 stuff. Instead only the actually done links are displayed. 18261 18262 *Ralf S. Engelschall* 18263 18264 * Permit null encryption ciphersuites, used for authentication only. It used 18265 to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this. 18266 It is now necessary to set SSL_FORBID_ENULL to prevent the use of null 18267 encryption. 18268 18269 *Ben Laurie* 18270 18271 * Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder 18272 signed attributes when verifying signatures (this would break them), 18273 the detached data encoding was wrong and public keys obtained using 18274 X509_get_pubkey() weren't freed. 18275 18276 *Steve Henson* 18277 18278 * Add text documentation for the BUFFER functions. Also added a work around 18279 to a Win95 console bug. This was triggered by the password read stuff: the 18280 last character typed gets carried over to the next fread(). If you were 18281 generating a new cert request using 'req' for example then the last 18282 character of the passphrase would be CR which would then enter the first 18283 field as blank. 18284 18285 *Steve Henson* 18286 18287 * Added the new 'Includes OpenSSL Cryptography Software' button as 18288 doc/openssl_button.{gif,html} which is similar in style to the old SSLeay 18289 button and can be used by applications based on OpenSSL to show the 18290 relationship to the OpenSSL project. 18291 18292 *Ralf S. Engelschall* 18293 18294 * Remove confusing variables in function signatures in files 18295 ssl/ssl_lib.c and ssl/ssl.h. 18296 18297 *Lennart Bong <lob@kulthea.stacken.kth.se>* 18298 18299 * Don't install bss_file.c under PREFIX/include/ 18300 18301 *Lennart Bong <lob@kulthea.stacken.kth.se>* 18302 18303 * Get the Win32 compile working again. Modify mkdef.pl so it can handle 18304 functions that return function pointers and has support for NT specific 18305 stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various 18306 #ifdef WIN32 and WINNTs sprinkled about the place and some changes from 18307 unsigned to signed types: this was killing the Win32 compile. 18308 18309 *Steve Henson* 18310 18311 * Add new certificate file to stack functions, 18312 SSL_add_dir_cert_subjects_to_stack() and 18313 SSL_add_file_cert_subjects_to_stack(). These largely supplant 18314 SSL_load_client_CA_file(), and can be used to add multiple certs easily 18315 to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()). 18316 This means that Apache-SSL and similar packages don't have to mess around 18317 to add as many CAs as they want to the preferred list. 18318 18319 *Ben Laurie* 18320 18321 * Experiment with doxygen documentation. Currently only partially applied to 18322 ssl/ssl_lib.c. 18323 See <http://www.stack.nl/~dimitri/doxygen/index.html>, and run doxygen with 18324 openssl.doxy as the configuration file. 18325 18326 *Ben Laurie* 18327 18328 * Get rid of remaining C++-style comments which strict C compilers hate. 18329 18330 *Ralf S. Engelschall, pointed out by Carlos Amengual* 18331 18332 * Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not 18333 compiled in by default: it has problems with large keys. 18334 18335 *Steve Henson* 18336 18337 * Add a bunch of SSL_xxx() functions for configuring the temporary RSA and 18338 DH private keys and/or callback functions which directly correspond to 18339 their SSL_CTX_xxx() counterparts but work on a per-connection basis. This 18340 is needed for applications which have to configure certificates on a 18341 per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis 18342 (e.g. s_server). 18343 For the RSA certificate situation is makes no difference, but 18344 for the DSA certificate situation this fixes the "no shared cipher" 18345 problem where the OpenSSL cipher selection procedure failed because the 18346 temporary keys were not overtaken from the context and the API provided 18347 no way to reconfigure them. 18348 The new functions now let applications reconfigure the stuff and they 18349 are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh, 18350 SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback. Additionally a new 18351 non-public-API function ssl_cert_instantiate() is used as a helper 18352 function and also to reduce code redundancy inside ssl_rsa.c. 18353 18354 *Ralf S. Engelschall* 18355 18356 * Move s_server -dcert and -dkey options out of the undocumented feature 18357 area because they are useful for the DSA situation and should be 18358 recognized by the users. 18359 18360 *Ralf S. Engelschall* 18361 18362 * Fix the cipher decision scheme for export ciphers: the export bits are 18363 *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within 18364 SSL_EXP_MASK. So, the original variable has to be used instead of the 18365 already masked variable. 18366 18367 *Richard Levitte <levitte@stacken.kth.se>* 18368 18369 * Fix `port` variable from `int` to `unsigned int` in crypto/bio/b_sock.c 18370 18371 *Richard Levitte <levitte@stacken.kth.se>* 18372 18373 * Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal() 18374 from `int` to `unsigned int` because it is a length and initialized by 18375 EVP_DigestFinal() which expects an `unsigned int *`. 18376 18377 *Richard Levitte <levitte@stacken.kth.se>* 18378 18379 * Don't hard-code path to Perl interpreter on shebang line of Configure 18380 script. Instead use the usual Shell->Perl transition trick. 18381 18382 *Ralf S. Engelschall* 18383 18384 * Make `openssl x509 -noout -modulus`' functional also for DSA certificates 18385 (in addition to RSA certificates) to match the behaviour of `openssl dsa 18386 -noout -modulus` as it's already the case for `openssl rsa -noout 18387 -modulus`. For RSA the -modulus is the real "modulus" while for DSA 18388 currently the public key is printed (a decision which was already done by 18389 `openssl dsa -modulus` in the past) which serves a similar purpose. 18390 Additionally the NO_RSA no longer completely removes the whole -modulus 18391 option; it now only avoids using the RSA stuff. Same applies to NO_DSA 18392 now, too. 18393 18394 *Ralf S. Engelschall* 18395 18396 * Add Arne Ansper's reliable BIO - this is an encrypted, block-digested 18397 BIO. See the source (crypto/evp/bio_ok.c) for more info. 18398 18399 *Arne Ansper <arne@ats.cyber.ee>* 18400 18401 * Dump the old yucky req code that tried (and failed) to allow raw OIDs 18402 to be added. Now both 'req' and 'ca' can use new objects defined in the 18403 config file. 18404 18405 *Steve Henson* 18406 18407 * Add cool BIO that does syslog (or event log on NT). 18408 18409 *Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie* 18410 18411 * Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5, 18412 TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and 18413 TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher 18414 Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt. 18415 18416 *Ben Laurie* 18417 18418 * Add preliminary config info for new extension code. 18419 18420 *Steve Henson* 18421 18422 * Make RSA_NO_PADDING really use no padding. 18423 18424 *Ulf Moeller <ulf@fitug.de>* 18425 18426 * Generate errors when private/public key check is done. 18427 18428 *Ben Laurie* 18429 18430 * Overhaul for 'crl' utility. New function X509_CRL_print. Partial support 18431 for some CRL extensions and new objects added. 18432 18433 *Steve Henson* 18434 18435 * Really fix the ASN1 IMPLICIT bug this time... Partial support for private 18436 key usage extension and fuller support for authority key id. 18437 18438 *Steve Henson* 18439 18440 * Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved 18441 padding method for RSA, which is recommended for new applications in PKCS 18442 #1 v2.0 (RFC 2437, October 1998). 18443 OAEP (Optimal Asymmetric Encryption Padding) has better theoretical 18444 foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure 18445 against Bleichbacher's attack on RSA. 18446 *Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by 18447 Ben Laurie* 18448 18449 * Updates to the new SSL compression code 18450 18451 *Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)* 18452 18453 * Fix so that the version number in the master secret, when passed 18454 via RSA, checks that if TLS was proposed, but we roll back to SSLv3 18455 (because the server will not accept higher), that the version number 18456 is 0x03,0x01, not 0x03,0x00 18457 18458 *Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)* 18459 18460 * Run extensive memory leak checks on SSL commands. Fixed *lots* of memory 18461 leaks in `ssl/` relating to new `X509_get_pubkey()` behaviour. Also fixes 18462 in `apps/` and an unrelated leak in `crypto/dsa/dsa_vrf.c`. 18463 18464 *Steve Henson* 18465 18466 * Support for RAW extensions where an arbitrary extension can be 18467 created by including its DER encoding. See `apps/openssl.cnf` for 18468 an example. 18469 18470 *Steve Henson* 18471 18472 * Make sure latest Perl versions don't interpret some generated C array 18473 code as Perl array code in the crypto/err/err_genc.pl script. 18474 18475 *Lars Weber <3weber@informatik.uni-hamburg.de>* 18476 18477 * Modify ms/do_ms.bat to not generate assembly language makefiles since 18478 not many people have the assembler. Various Win32 compilation fixes and 18479 update to the INSTALL.W32 file with (hopefully) more accurate Win32 18480 build instructions. 18481 18482 *Steve Henson* 18483 18484 * Modify configure script 'Configure' to automatically create crypto/date.h 18485 file under Win32 and also build pem.h from pem.org. New script 18486 util/mkfiles.pl to create the MINFO file on environments that can't do a 18487 'make files': perl util/mkfiles.pl >MINFO should work. 18488 18489 *Steve Henson* 18490 18491 * Major rework of DES function declarations, in the pursuit of correctness 18492 and purity. As a result, many evil casts evaporated, and some weirdness, 18493 too. You may find this causes warnings in your code. Zapping your evil 18494 casts will probably fix them. Mostly. 18495 18496 *Ben Laurie* 18497 18498 * Fix for a typo in asn1.h. Bug fix to object creation script 18499 obj_dat.pl. It considered a zero in an object definition to mean 18500 "end of object": none of the objects in objects.h have any zeros 18501 so it wasn't spotted. 18502 18503 *Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>* 18504 18505 * Add support for Triple DES Cipher Block Chaining with Output Feedback 18506 Masking (CBCM). In the absence of test vectors, the best I have been able 18507 to do is check that the decrypt undoes the encrypt, so far. Send me test 18508 vectors if you have them. 18509 18510 *Ben Laurie* 18511 18512 * Correct calculation of key length for export ciphers (too much space was 18513 allocated for null ciphers). This has not been tested! 18514 18515 *Ben Laurie* 18516 18517 * Modifications to the mkdef.pl for Win32 DEF file creation. The usage 18518 message is now correct (it understands "crypto" and "ssl" on its 18519 command line). There is also now an "update" option. This will update 18520 the util/ssleay.num and util/libeay.num files with any new functions. 18521 If you do a: 18522 perl util/mkdef.pl crypto ssl update 18523 it will update them. 18524 18525 *Steve Henson* 18526 18527 * Overhauled the Perl interface: 18528 - ported BN stuff to OpenSSL's different BN library 18529 - made the perl/ source tree CVS-aware 18530 - renamed the package from SSLeay to OpenSSL (the files still contain 18531 their history because I've copied them in the repository) 18532 - removed obsolete files (the test scripts will be replaced 18533 by better Test::Harness variants in the future) 18534 18535 *Ralf S. Engelschall* 18536 18537 * First cut for a very conservative source tree cleanup: 18538 1. merge various obsolete readme texts into doc/ssleay.txt 18539 where we collect the old documents and readme texts. 18540 2. remove the first part of files where I'm already sure that we no 18541 longer need them because of three reasons: either they are just temporary 18542 files which were left by Eric or they are preserved original files where 18543 I've verified that the diff is also available in the CVS via "cvs diff 18544 -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for 18545 the crypto/md/ stuff). 18546 18547 *Ralf S. Engelschall* 18548 18549 * More extension code. Incomplete support for subject and issuer alt 18550 name, issuer and authority key id. Change the i2v function parameters 18551 and add an extra 'crl' parameter in the X509V3_CTX structure: guess 18552 what that's for :-) Fix to ASN1 macro which messed up 18553 IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED. 18554 18555 *Steve Henson* 18556 18557 * Preliminary support for ENUMERATED type. This is largely copied from the 18558 INTEGER code. 18559 18560 *Steve Henson* 18561 18562 * Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy. 18563 18564 *Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)* 18565 18566 * Make sure `make rehash` target really finds the `openssl` program. 18567 18568 *Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>* 18569 18570 * Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd 18571 like to hear about it if this slows down other processors. 18572 18573 *Ben Laurie* 18574 18575 * Add CygWin32 platform information to Configure script. 18576 18577 *Alan Batie <batie@aahz.jf.intel.com>* 18578 18579 * Fixed ms/32all.bat script: `no_asm` -> `no-asm` 18580 18581 *Rainer W. Gerling <gerling@mpg-gv.mpg.de>* 18582 18583 * New program nseq to manipulate netscape certificate sequences 18584 18585 *Steve Henson* 18586 18587 * Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a 18588 few typos. 18589 18590 *Steve Henson* 18591 18592 * Fixes to BN code. Previously the default was to define BN_RECURSION 18593 but the BN code had some problems that would cause failures when 18594 doing certificate verification and some other functions. 18595 18596 *Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)* 18597 18598 * Add ASN1 and PEM code to support netscape certificate sequences. 18599 18600 *Steve Henson* 18601 18602 * Add ASN1 and PEM code to support netscape certificate sequences. 18603 18604 *Steve Henson* 18605 18606 * Add several PKIX and private extended key usage OIDs. 18607 18608 *Steve Henson* 18609 18610 * Modify the 'ca' program to handle the new extension code. Modify 18611 openssl.cnf for new extension format, add comments. 18612 18613 *Steve Henson* 18614 18615 * More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req' 18616 and add a sample to openssl.cnf so req -x509 now adds appropriate 18617 CA extensions. 18618 18619 *Steve Henson* 18620 18621 * Continued X509 V3 changes. Add to other makefiles, integrate with the 18622 error code, add initial support to X509_print() and x509 application. 18623 18624 *Steve Henson* 18625 18626 * Takes a deep breath and start adding X509 V3 extension support code. Add 18627 files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this 18628 stuff is currently isolated and isn't even compiled yet. 18629 18630 *Steve Henson* 18631 18632 * Continuing patches for GeneralizedTime. Fix up certificate and CRL 18633 ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print. 18634 Removed the versions check from X509 routines when loading extensions: 18635 this allows certain broken certificates that don't set the version 18636 properly to be processed. 18637 18638 *Steve Henson* 18639 18640 * Deal with irritating shit to do with dependencies, in YAAHW (Yet Another 18641 Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which 18642 can still be regenerated with "make depend". 18643 18644 *Ben Laurie* 18645 18646 * Spelling mistake in C version of CAST-128. 18647 18648 *Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>* 18649 18650 * Changes to the error generation code. The perl script err-code.pl 18651 now reads in the old error codes and retains the old numbers, only 18652 adding new ones if necessary. It also only changes the .err files if new 18653 codes are added. The makefiles have been modified to only insert errors 18654 when needed (to avoid needlessly modifying header files). This is done 18655 by only inserting errors if the .err file is newer than the auto generated 18656 C file. To rebuild all the error codes from scratch (the old behaviour) 18657 either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl 18658 or delete all the .err files. 18659 18660 *Steve Henson* 18661 18662 * CAST-128 was incorrectly implemented for short keys. The C version has 18663 been fixed, but is untested. The assembler versions are also fixed, but 18664 new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing 18665 to regenerate it if needed. 18666 *Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun 18667 Hagino <itojun@kame.net>* 18668 18669 * File was opened incorrectly in randfile.c. 18670 18671 *Ulf Möller <ulf@fitug.de>* 18672 18673 * Beginning of support for GeneralizedTime. d2i, i2d, check and print 18674 functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or 18675 GeneralizedTime. ASN1_TIME is the proper type used in certificates et 18676 al: it's just almost always a UTCTime. Note this patch adds new error 18677 codes so do a "make errors" if there are problems. 18678 18679 *Steve Henson* 18680 18681 * Correct Linux 1 recognition in config. 18682 18683 *Ulf Möller <ulf@fitug.de>* 18684 18685 * Remove pointless MD5 hash when using DSA keys in ca. 18686 18687 *Anonymous <nobody@replay.com>* 18688 18689 * Generate an error if given an empty string as a cert directory. Also 18690 generate an error if handed NULL (previously returned 0 to indicate an 18691 error, but didn't set one). 18692 18693 *Ben Laurie, reported by Anonymous <nobody@replay.com>* 18694 18695 * Add prototypes to SSL methods. Make SSL_write's buffer const, at last. 18696 18697 *Ben Laurie* 18698 18699 * Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct 18700 parameters. This was causing a warning which killed off the Win32 compile. 18701 18702 *Steve Henson* 18703 18704 * Remove C++ style comments from crypto/bn/bn_local.h. 18705 18706 *Neil Costigan <neil.costigan@celocom.com>* 18707 18708 * The function OBJ_txt2nid was broken. It was supposed to return a nid 18709 based on a text string, looking up short and long names and finally 18710 "dot" format. The "dot" format stuff didn't work. Added new function 18711 OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote 18712 OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the 18713 OID is not part of the table. 18714 18715 *Steve Henson* 18716 18717 * Add prototypes to X509 lookup/verify methods, fixing a bug in 18718 X509_LOOKUP_by_alias(). 18719 18720 *Ben Laurie* 18721 18722 * Sort openssl functions by name. 18723 18724 *Ben Laurie* 18725 18726 * Get the `gendsa` command working and add it to the `list` command. Remove 18727 encryption from sample DSA keys (in case anyone is interested the password 18728 was "1234"). 18729 18730 *Steve Henson* 18731 18732 * Make *all* `*_free` functions accept a NULL pointer. 18733 18734 *Frans Heymans <fheymans@isaserver.be>* 18735 18736 * If a DH key is generated in s3_srvr.c, don't blow it by trying to use 18737 NULL pointers. 18738 18739 *Anonymous <nobody@replay.com>* 18740 18741 * s_server should send the CAfile as acceptable CAs, not its own cert. 18742 18743 *Bodo Moeller <3moeller@informatik.uni-hamburg.de>* 18744 18745 * Don't blow it for numeric `-newkey` arguments to `apps/req`. 18746 18747 *Bodo Moeller <3moeller@informatik.uni-hamburg.de>* 18748 18749 * Temp key "for export" tests were wrong in s3_srvr.c. 18750 18751 *Anonymous <nobody@replay.com>* 18752 18753 * Add prototype for temp key callback functions 18754 SSL_CTX_set_tmp_{rsa,dh}_callback(). 18755 18756 *Ben Laurie* 18757 18758 * Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and 18759 DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey(). 18760 18761 *Steve Henson* 18762 18763 * X509_name_add_entry() freed the wrong thing after an error. 18764 18765 *Arne Ansper <arne@ats.cyber.ee>* 18766 18767 * rsa_eay.c would attempt to free a NULL context. 18768 18769 *Arne Ansper <arne@ats.cyber.ee>* 18770 18771 * BIO_s_socket() had a broken should_retry() on Windoze. 18772 18773 *Arne Ansper <arne@ats.cyber.ee>* 18774 18775 * BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH. 18776 18777 *Arne Ansper <arne@ats.cyber.ee>* 18778 18779 * Make sure the already existing X509_STORE->depth variable is initialized 18780 in X509_STORE_new(), but document the fact that this variable is still 18781 unused in the certificate verification process. 18782 18783 *Ralf S. Engelschall* 18784 18785 * Fix the various library and `apps/` files to free up pkeys obtained from 18786 X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions. 18787 18788 *Steve Henson* 18789 18790 * Fix reference counting in X509_PUBKEY_get(). This makes 18791 demos/maurice/example2.c work, amongst others, probably. 18792 18793 *Steve Henson and Ben Laurie* 18794 18795 * First cut of a cleanup for `apps/`. First the `ssleay` program is now named 18796 `openssl` and second, the shortcut symlinks for the `openssl <command>` 18797 are no longer created. This way we have a single and consistent command 18798 line interface `openssl <command>`, similar to `cvs <command>`. 18799 18800 *Ralf S. Engelschall, Paul Sutton and Ben Laurie* 18801 18802 * ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey 18803 BIT STRING wrapper always have zero unused bits. 18804 18805 *Steve Henson* 18806 18807 * Add CA.pl, perl version of CA.sh, add extended key usage OID. 18808 18809 *Steve Henson* 18810 18811 * Make the top-level INSTALL documentation easier to understand. 18812 18813 *Paul Sutton* 18814 18815 * Makefiles updated to exit if an error occurs in a sub-directory 18816 make (including if user presses ^C) [Paul Sutton] 18817 18818 * Make Montgomery context stuff explicit in RSA data structure. 18819 18820 *Ben Laurie* 18821 18822 * Fix build order of pem and err to allow for generated pem.h. 18823 18824 *Ben Laurie* 18825 18826 * Fix renumbering bug in X509_NAME_delete_entry(). 18827 18828 *Ben Laurie* 18829 18830 * Enhanced the err-ins.pl script so it makes the error library number 18831 global and can add a library name. This is needed for external ASN1 and 18832 other error libraries. 18833 18834 *Steve Henson* 18835 18836 * Fixed sk_insert which never worked properly. 18837 18838 *Steve Henson* 18839 18840 * Fix ASN1 macros so they can handle indefinite length constructed 18841 EXPLICIT tags. Some non standard certificates use these: they can now 18842 be read in. 18843 18844 *Steve Henson* 18845 18846 * Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc) 18847 into a single doc/ssleay.txt bundle. This way the information is still 18848 preserved but no longer messes up this directory. Now it's new room for 18849 the new set of documentation files. 18850 18851 *Ralf S. Engelschall* 18852 18853 * SETs were incorrectly DER encoded. This was a major pain, because they 18854 shared code with SEQUENCEs, which aren't coded the same. This means that 18855 almost everything to do with SETs or SEQUENCEs has either changed name or 18856 number of arguments. 18857 18858 *Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>* 18859 18860 * Fix test data to work with the above. 18861 18862 *Ben Laurie* 18863 18864 * Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but 18865 was already fixed by Eric for 0.9.1 it seems. 18866 18867 *Ben Laurie - pointed out by Ulf Möller <ulf@fitug.de>* 18868 18869 * Autodetect FreeBSD3. 18870 18871 *Ben Laurie* 18872 18873 * Fix various bugs in Configure. This affects the following platforms: 18874 nextstep 18875 ncr-scde 18876 unixware-2.0 18877 unixware-2.0-pentium 18878 sco5-cc. 18879 18880 *Ben Laurie* 18881 18882 * Eliminate generated files from CVS. Reorder tests to regenerate files 18883 before they are needed. 18884 18885 *Ben Laurie* 18886 18887 * Generate Makefile.ssl from Makefile.org (to keep CVS happy). 18888 18889 *Ben Laurie* 18890 18891### Changes between 0.9.1b and 0.9.1c [23-Dec-1998] 18892 18893 * Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and 18894 changed SSLeay to OpenSSL in version strings. 18895 18896 *Ralf S. Engelschall* 18897 18898 * Some fixups to the top-level documents. 18899 18900 *Paul Sutton* 18901 18902 * Fixed the nasty bug where rsaref.h was not found under compile-time 18903 because the symlink to include/ was missing. 18904 18905 *Ralf S. Engelschall* 18906 18907 * Incorporated the popular no-RSA/DSA-only patches 18908 which allow to compile a RSA-free SSLeay. 18909 18910 *Andrew Cooke / Interrader Ldt., Ralf S. Engelschall* 18911 18912 * Fixed nasty rehash problem under `make -f Makefile.ssl links` 18913 when "ssleay" is still not found. 18914 18915 *Ralf S. Engelschall* 18916 18917 * Added more platforms to Configure: Cray T3E, HPUX 11, 18918 18919 *Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>* 18920 18921 * Updated the README file. 18922 18923 *Ralf S. Engelschall* 18924 18925 * Added various .cvsignore files in the CVS repository subdirs 18926 to make a "cvs update" really silent. 18927 18928 *Ralf S. Engelschall* 18929 18930 * Recompiled the error-definition header files and added 18931 missing symbols to the Win32 linker tables. 18932 18933 *Ralf S. Engelschall* 18934 18935 * Cleaned up the top-level documents; 18936 o new files: CHANGES and LICENSE 18937 o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay 18938 o merged COPYRIGHT into LICENSE 18939 o removed obsolete TODO file 18940 o renamed MICROSOFT to INSTALL.W32 18941 18942 *Ralf S. Engelschall* 18943 18944 * Removed dummy files from the 0.9.1b source tree: 18945 crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi 18946 crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f 18947 crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f 18948 crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f 18949 util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f 18950 18951 *Ralf S. Engelschall* 18952 18953 * Added various platform portability fixes. 18954 18955 *Mark J. Cox* 18956 18957 * The Genesis of the OpenSSL rpject: 18958 We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A. 18959 Young and Tim J. Hudson created while they were working for C2Net until 18960 summer 1998. 18961 18962 *The OpenSSL Project* 18963 18964### Changes between 0.9.0b and 0.9.1b [not released] 18965 18966 * Updated a few CA certificates under certs/ 18967 18968 *Eric A. Young* 18969 18970 * Changed some BIGNUM api stuff. 18971 18972 *Eric A. Young* 18973 18974 * Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD, 18975 DGUX x86, Linux Alpha, etc. 18976 18977 *Eric A. Young* 18978 18979 * New COMP library [crypto/comp/] for SSL Record Layer Compression: 18980 RLE (dummy implemented) and ZLIB (really implemented when ZLIB is 18981 available). 18982 18983 *Eric A. Young* 18984 18985 * Add -strparse option to asn1pars program which parses nested 18986 binary structures 18987 18988 *Dr Stephen Henson <shenson@bigfoot.com>* 18989 18990 * Added "oid_file" to ssleay.cnf for "ca" and "req" programs. 18991 18992 *Eric A. Young* 18993 18994 * DSA fix for "ca" program. 18995 18996 *Eric A. Young* 18997 18998 * Added "-genkey" option to "dsaparam" program. 18999 19000 *Eric A. Young* 19001 19002 * Added RIPE MD160 (rmd160) message digest. 19003 19004 *Eric A. Young* 19005 19006 * Added -a (all) option to "ssleay version" command. 19007 19008 *Eric A. Young* 19009 19010 * Added PLATFORM define which is the id given to Configure. 19011 19012 *Eric A. Young* 19013 19014 * Added MemCheck_XXXX functions to crypto/mem.c for memory checking. 19015 19016 *Eric A. Young* 19017 19018 * Extended the ASN.1 parser routines. 19019 19020 *Eric A. Young* 19021 19022 * Extended BIO routines to support REUSEADDR, seek, tell, etc. 19023 19024 *Eric A. Young* 19025 19026 * Added a BN_CTX to the BN library. 19027 19028 *Eric A. Young* 19029 19030 * Fixed the weak key values in DES library 19031 19032 *Eric A. Young* 19033 19034 * Changed API in EVP library for cipher aliases. 19035 19036 *Eric A. Young* 19037 19038 * Added support for RC2/64bit cipher. 19039 19040 *Eric A. Young* 19041 19042 * Converted the lhash library to the crypto/mem.c functions. 19043 19044 *Eric A. Young* 19045 19046 * Added more recognized ASN.1 object ids. 19047 19048 *Eric A. Young* 19049 19050 * Added more RSA padding checks for SSL/TLS. 19051 19052 *Eric A. Young* 19053 19054 * Added BIO proxy/filter functionality. 19055 19056 *Eric A. Young* 19057 19058 * Added extra_certs to SSL_CTX which can be used 19059 send extra CA certificates to the client in the CA cert chain sending 19060 process. It can be configured with SSL_CTX_add_extra_chain_cert(). 19061 19062 *Eric A. Young* 19063 19064 * Now Fortezza is denied in the authentication phase because 19065 this is key exchange mechanism is not supported by SSLeay at all. 19066 19067 *Eric A. Young* 19068 19069 * Additional PKCS1 checks. 19070 19071 *Eric A. Young* 19072 19073 * Support the string "TLSv1" for all TLS v1 ciphers. 19074 19075 *Eric A. Young* 19076 19077 * Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the 19078 ex_data index of the SSL context in the X509_STORE_CTX ex_data. 19079 19080 *Eric A. Young* 19081 19082 * Fixed a few memory leaks. 19083 19084 *Eric A. Young* 19085 19086 * Fixed various code and comment typos. 19087 19088 *Eric A. Young* 19089 19090 * A minor bug in ssl/s3_clnt.c where there would always be 4 0 19091 bytes sent in the client random. 19092 19093 *Edward Bishop <ebishop@spyglass.com>* 19094 19095<!-- Links --> 19096 19097[CVE-2020-1971]: https://www.openssl.org/news/vulnerabilities.html#CVE-2020-1971 19098[CVE-2020-1967]: https://www.openssl.org/news/vulnerabilities.html#CVE-2020-1967 19099[CVE-2019-1563]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1563 19100[CVE-2019-1559]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1559 19101[CVE-2019-1552]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1552 19102[CVE-2019-1551]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1551 19103[CVE-2019-1549]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1549 19104[CVE-2019-1547]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1547 19105[CVE-2019-1543]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1543 19106[CVE-2018-5407]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-5407 19107[CVE-2018-0739]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0739 19108[CVE-2018-0737]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0737 19109[CVE-2018-0735]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0735 19110[CVE-2018-0734]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0734 19111[CVE-2018-0733]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0733 19112[CVE-2018-0732]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0732 19113[CVE-2017-3738]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3738 19114[CVE-2017-3737]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3737 19115[CVE-2017-3736]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3736 19116[CVE-2017-3735]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3735 19117[CVE-2017-3733]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3733 19118[CVE-2017-3732]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3732 19119[CVE-2017-3731]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3731 19120[CVE-2017-3730]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3730 19121[CVE-2016-7055]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-7055 19122[CVE-2016-7054]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-7054 19123[CVE-2016-7053]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-7053 19124[CVE-2016-7052]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-7052 19125[CVE-2016-6309]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6309 19126[CVE-2016-6308]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6308 19127[CVE-2016-6307]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6307 19128[CVE-2016-6306]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6306 19129[CVE-2016-6305]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6305 19130[CVE-2016-6304]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6304 19131[CVE-2016-6303]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6303 19132[CVE-2016-6302]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6302 19133[CVE-2016-2183]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2183 19134[CVE-2016-2182]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2182 19135[CVE-2016-2181]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2181 19136[CVE-2016-2180]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2180 19137[CVE-2016-2179]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2179 19138[CVE-2016-2178]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2178 19139[CVE-2016-2177]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2177 19140[CVE-2016-2176]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2176 19141[CVE-2016-2109]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2109 19142[CVE-2016-2107]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2107 19143[CVE-2016-2106]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2106 19144[CVE-2016-2105]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2105 19145[CVE-2016-0800]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0800 19146[CVE-2016-0799]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0799 19147[CVE-2016-0798]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0798 19148[CVE-2016-0797]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0797 19149[CVE-2016-0705]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0705 19150[CVE-2016-0702]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0702 19151[CVE-2016-0701]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0701 19152[CVE-2015-3197]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3197 19153[CVE-2015-3196]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3196 19154[CVE-2015-3195]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3195 19155[CVE-2015-3194]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3194 19156[CVE-2015-3193]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3193 19157[CVE-2015-1793]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1793 19158[CVE-2015-1792]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1792 19159[CVE-2015-1791]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1791 19160[CVE-2015-1790]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1790 19161[CVE-2015-1789]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1789 19162[CVE-2015-1788]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1788 19163[CVE-2015-1787]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1787 19164[CVE-2015-0293]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0293 19165[CVE-2015-0291]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0291 19166[CVE-2015-0290]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0290 19167[CVE-2015-0289]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0289 19168[CVE-2015-0288]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0288 19169[CVE-2015-0287]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0287 19170[CVE-2015-0286]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0286 19171[CVE-2015-0285]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0285 19172[CVE-2015-0209]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0209 19173[CVE-2015-0208]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0208 19174[CVE-2015-0207]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0207 19175[CVE-2015-0206]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0206 19176[CVE-2015-0205]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0205 19177[CVE-2015-0204]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0204 19178[CVE-2014-8275]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-8275 19179[CVE-2014-5139]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-5139 19180[CVE-2014-3572]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3572 19181[CVE-2014-3571]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3571 19182[CVE-2014-3570]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3570 19183[CVE-2014-3569]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3569 19184[CVE-2014-3568]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3568 19185[CVE-2014-3567]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3567 19186[CVE-2014-3566]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3566 19187[CVE-2014-3513]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3513 19188[CVE-2014-3512]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3512 19189[CVE-2014-3511]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3511 19190[CVE-2014-3510]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3510 19191[CVE-2014-3509]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3509 19192[CVE-2014-3508]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3508 19193[CVE-2014-3507]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3507 19194[CVE-2014-3506]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3506 19195[CVE-2014-3505]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3505 19196[CVE-2014-3470]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3470 19197[CVE-2014-0224]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0224 19198[CVE-2014-0221]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0221 19199[CVE-2014-0195]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0195 19200[CVE-2014-0160]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0160 19201[CVE-2014-0076]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0076 19202[CVE-2013-6450]: https://www.openssl.org/news/vulnerabilities.html#CVE-2013-6450 19203[CVE-2013-4353]: https://www.openssl.org/news/vulnerabilities.html#CVE-2013-4353 19204[CVE-2013-0169]: https://www.openssl.org/news/vulnerabilities.html#CVE-2013-0169 19205[CVE-2013-0166]: https://www.openssl.org/news/vulnerabilities.html#CVE-2013-0166 19206[CVE-2012-2686]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-2686 19207[CVE-2012-2333]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-2333 19208[CVE-2012-2110]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-2110 19209[CVE-2012-0884]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-0884 19210[CVE-2012-0050]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-0050 19211[CVE-2012-0027]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-0027 19212[CVE-2011-4619]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4619 19213[CVE-2011-4577]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4577 19214[CVE-2011-4576]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4576 19215[CVE-2011-4109]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4109 19216[CVE-2011-4108]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4108 19217[CVE-2011-3210]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-3210 19218[CVE-2011-3207]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-3207 19219[CVE-2011-0014]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-0014 19220[CVE-2010-4252]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-4252 19221[CVE-2010-4180]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-4180 19222[CVE-2010-3864]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-3864 19223[CVE-2010-1633]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-1633 19224[CVE-2010-0740]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-0740 19225[CVE-2010-0433]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-0433 19226[CVE-2009-4355]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-4355 19227[CVE-2009-3555]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-3555 19228[CVE-2009-3245]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-3245 19229[CVE-2009-1386]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-1386 19230[CVE-2009-1379]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-1379 19231[CVE-2009-1378]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-1378 19232[CVE-2009-1377]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-1377 19233[CVE-2009-0789]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-0789 19234[CVE-2009-0591]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-0591 19235[CVE-2009-0590]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-0590 19236[CVE-2008-5077]: https://www.openssl.org/news/vulnerabilities.html#CVE-2008-5077 19237[CVE-2008-1678]: https://www.openssl.org/news/vulnerabilities.html#CVE-2008-1678 19238[CVE-2008-1672]: https://www.openssl.org/news/vulnerabilities.html#CVE-2008-1672 19239[CVE-2008-0891]: https://www.openssl.org/news/vulnerabilities.html#CVE-2008-0891 19240[CVE-2007-5135]: https://www.openssl.org/news/vulnerabilities.html#CVE-2007-5135 19241[CVE-2007-4995]: https://www.openssl.org/news/vulnerabilities.html#CVE-2007-4995 19242[CVE-2006-4343]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-4343 19243[CVE-2006-4339]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-4339 19244[CVE-2006-3738]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-3738 19245[CVE-2006-2940]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-2940 19246[CVE-2006-2937]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-2937 19247[CVE-2005-2969]: https://www.openssl.org/news/vulnerabilities.html#CVE-2005-2969 19248[CVE-2004-0112]: https://www.openssl.org/news/vulnerabilities.html#CVE-2004-0112 19249[CVE-2004-0079]: https://www.openssl.org/news/vulnerabilities.html#CVE-2004-0079 19250[CVE-2003-0851]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0851 19251[CVE-2003-0545]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0545 19252[CVE-2003-0544]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0544 19253[CVE-2003-0543]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0543 19254[CVE-2003-0078]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0078 19255[CVE-2002-0659]: https://www.openssl.org/news/vulnerabilities.html#CVE-2002-0659 19256[CVE-2002-0657]: https://www.openssl.org/news/vulnerabilities.html#CVE-2002-0657 19257[CVE-2002-0656]: https://www.openssl.org/news/vulnerabilities.html#CVE-2002-0656 19258[CVE-2002-0655]: https://www.openssl.org/news/vulnerabilities.html#CVE-2002-0655 19259