CHANGES revision 352193
1 2 OpenSSL CHANGES 3 _______________ 4 5 This is a high-level summary of the most important changes. 6 For a full list of changes, see the git commit log; for example, 7 https://github.com/openssl/openssl/commits/ and pick the appropriate 8 release branch. 9 10 Changes between 1.0.2s and 1.0.2t [10 Sep 2019] 11 12 *) For built-in EC curves, ensure an EC_GROUP built from the curve name is 13 used even when parsing explicit parameters, when loading a serialized key 14 or calling `EC_GROUP_new_from_ecpkparameters()`/ 15 `EC_GROUP_new_from_ecparameters()`. 16 This prevents bypass of security hardening and performance gains, 17 especially for curves with specialized EC_METHODs. 18 By default, if a key encoded with explicit parameters is loaded and later 19 serialized, the output is still encoded with explicit parameters, even if 20 internally a "named" EC_GROUP is used for computation. 21 [Nicola Tuveri] 22 23 *) Compute ECC cofactors if not provided during EC_GROUP construction. Before 24 this change, EC_GROUP_set_generator would accept order and/or cofactor as 25 NULL. After this change, only the cofactor parameter can be NULL. It also 26 does some minimal sanity checks on the passed order. 27 (CVE-2019-1547) 28 [Billy Bob Brumley] 29 30 *) Fixed a padding oracle in PKCS7_dataDecode and CMS_decrypt_set1_pkey. 31 An attack is simple, if the first CMS_recipientInfo is valid but the 32 second CMS_recipientInfo is chosen ciphertext. If the second 33 recipientInfo decodes to PKCS #1 v1.5 form plaintext, the correct 34 encryption key will be replaced by garbage, and the message cannot be 35 decoded, but if the RSA decryption fails, the correct encryption key is 36 used and the recipient will not notice the attack. 37 As a work around for this potential attack the length of the decrypted 38 key must be equal to the cipher default key length, in case the 39 certifiate is not given and all recipientInfo are tried out. 40 The old behaviour can be re-enabled in the CMS code by setting the 41 CMS_DEBUG_DECRYPT flag. 42 (CVE-2019-1563) 43 [Bernd Edlinger] 44 45 *) Document issue with installation paths in diverse Windows builds 46 47 '/usr/local/ssl' is an unsafe prefix for location to install OpenSSL 48 binaries and run-time config file. 49 (CVE-2019-1552) 50 [Richard Levitte] 51 52 Changes between 1.0.2r and 1.0.2s [28 May 2019] 53 54 *) Change the default RSA, DSA and DH size to 2048 bit instead of 1024. 55 This changes the size when using the genpkey app when no size is given. It 56 fixes an omission in earlier changes that changed all RSA, DSA and DH 57 generation apps to use 2048 bits by default. 58 [Kurt Roeckx] 59 60 *) Add FIPS support for Android Arm 64-bit 61 62 Support for Android Arm 64-bit was added to the OpenSSL FIPS Object 63 Module in Version 2.0.10. For some reason, the corresponding target 64 'android64-aarch64' was missing OpenSSL 1.0.2, whence it could not be 65 built with FIPS support on Android Arm 64-bit. This omission has been 66 fixed. 67 [Matthias St. Pierre] 68 69 Changes between 1.0.2q and 1.0.2r [26 Feb 2019] 70 71 *) 0-byte record padding oracle 72 73 If an application encounters a fatal protocol error and then calls 74 SSL_shutdown() twice (once to send a close_notify, and once to receive one) 75 then OpenSSL can respond differently to the calling application if a 0 byte 76 record is received with invalid padding compared to if a 0 byte record is 77 received with an invalid MAC. If the application then behaves differently 78 based on that in a way that is detectable to the remote peer, then this 79 amounts to a padding oracle that could be used to decrypt data. 80 81 In order for this to be exploitable "non-stitched" ciphersuites must be in 82 use. Stitched ciphersuites are optimised implementations of certain 83 commonly used ciphersuites. Also the application must call SSL_shutdown() 84 twice even if a protocol error has occurred (applications should not do 85 this but some do anyway). 86 87 This issue was discovered by Juraj Somorovsky, Robert Merget and Nimrod 88 Aviram, with additional investigation by Steven Collison and Andrew 89 Hourselt. It was reported to OpenSSL on 10th December 2018. 90 (CVE-2019-1559) 91 [Matt Caswell] 92 93 *) Move strictness check from EVP_PKEY_asn1_new() to EVP_PKEY_asn1_add0(). 94 [Richard Levitte] 95 96 Changes between 1.0.2p and 1.0.2q [20 Nov 2018] 97 98 *) Microarchitecture timing vulnerability in ECC scalar multiplication 99 100 OpenSSL ECC scalar multiplication, used in e.g. ECDSA and ECDH, has been 101 shown to be vulnerable to a microarchitecture timing side channel attack. 102 An attacker with sufficient access to mount local timing attacks during 103 ECDSA signature generation could recover the private key. 104 105 This issue was reported to OpenSSL on 26th October 2018 by Alejandro 106 Cabrera Aldaya, Billy Brumley, Sohaib ul Hassan, Cesar Pereida Garcia and 107 Nicola Tuveri. 108 (CVE-2018-5407) 109 [Billy Brumley] 110 111 *) Timing vulnerability in DSA signature generation 112 113 The OpenSSL DSA signature algorithm has been shown to be vulnerable to a 114 timing side channel attack. An attacker could use variations in the signing 115 algorithm to recover the private key. 116 117 This issue was reported to OpenSSL on 16th October 2018 by Samuel Weiser. 118 (CVE-2018-0734) 119 [Paul Dale] 120 121 *) Resolve a compatibility issue in EC_GROUP handling with the FIPS Object 122 Module, accidentally introduced while backporting security fixes from the 123 development branch and hindering the use of ECC in FIPS mode. 124 [Nicola Tuveri] 125 126 Changes between 1.0.2o and 1.0.2p [14 Aug 2018] 127 128 *) Client DoS due to large DH parameter 129 130 During key agreement in a TLS handshake using a DH(E) based ciphersuite a 131 malicious server can send a very large prime value to the client. This will 132 cause the client to spend an unreasonably long period of time generating a 133 key for this prime resulting in a hang until the client has finished. This 134 could be exploited in a Denial Of Service attack. 135 136 This issue was reported to OpenSSL on 5th June 2018 by Guido Vranken 137 (CVE-2018-0732) 138 [Guido Vranken] 139 140 *) Cache timing vulnerability in RSA Key Generation 141 142 The OpenSSL RSA Key generation algorithm has been shown to be vulnerable to 143 a cache timing side channel attack. An attacker with sufficient access to 144 mount cache timing attacks during the RSA key generation process could 145 recover the private key. 146 147 This issue was reported to OpenSSL on 4th April 2018 by Alejandro Cabrera 148 Aldaya, Billy Brumley, Cesar Pereida Garcia and Luis Manuel Alvarez Tapia. 149 (CVE-2018-0737) 150 [Billy Brumley] 151 152 *) Make EVP_PKEY_asn1_new() a bit stricter about its input. A NULL pem_str 153 parameter is no longer accepted, as it leads to a corrupt table. NULL 154 pem_str is reserved for alias entries only. 155 [Richard Levitte] 156 157 *) Revert blinding in ECDSA sign and instead make problematic addition 158 length-invariant. Switch even to fixed-length Montgomery multiplication. 159 [Andy Polyakov] 160 161 *) Change generating and checking of primes so that the error rate of not 162 being prime depends on the intended use based on the size of the input. 163 For larger primes this will result in more rounds of Miller-Rabin. 164 The maximal error rate for primes with more than 1080 bits is lowered 165 to 2^-128. 166 [Kurt Roeckx, Annie Yousar] 167 168 *) Increase the number of Miller-Rabin rounds for DSA key generating to 64. 169 [Kurt Roeckx] 170 171 *) Add blinding to ECDSA and DSA signatures to protect against side channel 172 attacks discovered by Keegan Ryan (NCC Group). 173 [Matt Caswell] 174 175 *) When unlocking a pass phrase protected PEM file or PKCS#8 container, we 176 now allow empty (zero character) pass phrases. 177 [Richard Levitte] 178 179 *) Certificate time validation (X509_cmp_time) enforces stricter 180 compliance with RFC 5280. Fractional seconds and timezone offsets 181 are no longer allowed. 182 [Emilia K��sper] 183 184 Changes between 1.0.2n and 1.0.2o [27 Mar 2018] 185 186 *) Constructed ASN.1 types with a recursive definition could exceed the stack 187 188 Constructed ASN.1 types with a recursive definition (such as can be found 189 in PKCS7) could eventually exceed the stack given malicious input with 190 excessive recursion. This could result in a Denial Of Service attack. There 191 are no such structures used within SSL/TLS that come from untrusted sources 192 so this is considered safe. 193 194 This issue was reported to OpenSSL on 4th January 2018 by the OSS-fuzz 195 project. 196 (CVE-2018-0739) 197 [Matt Caswell] 198 199 Changes between 1.0.2m and 1.0.2n [7 Dec 2017] 200 201 *) Read/write after SSL object in error state 202 203 OpenSSL 1.0.2 (starting from version 1.0.2b) introduced an "error state" 204 mechanism. The intent was that if a fatal error occurred during a handshake 205 then OpenSSL would move into the error state and would immediately fail if 206 you attempted to continue the handshake. This works as designed for the 207 explicit handshake functions (SSL_do_handshake(), SSL_accept() and 208 SSL_connect()), however due to a bug it does not work correctly if 209 SSL_read() or SSL_write() is called directly. In that scenario, if the 210 handshake fails then a fatal error will be returned in the initial function 211 call. If SSL_read()/SSL_write() is subsequently called by the application 212 for the same SSL object then it will succeed and the data is passed without 213 being decrypted/encrypted directly from the SSL/TLS record layer. 214 215 In order to exploit this issue an application bug would have to be present 216 that resulted in a call to SSL_read()/SSL_write() being issued after having 217 already received a fatal error. 218 219 This issue was reported to OpenSSL by David Benjamin (Google). 220 (CVE-2017-3737) 221 [Matt Caswell] 222 223 *) rsaz_1024_mul_avx2 overflow bug on x86_64 224 225 There is an overflow bug in the AVX2 Montgomery multiplication procedure 226 used in exponentiation with 1024-bit moduli. No EC algorithms are affected. 227 Analysis suggests that attacks against RSA and DSA as a result of this 228 defect would be very difficult to perform and are not believed likely. 229 Attacks against DH1024 are considered just feasible, because most of the 230 work necessary to deduce information about a private key may be performed 231 offline. The amount of resources required for such an attack would be 232 significant. However, for an attack on TLS to be meaningful, the server 233 would have to share the DH1024 private key among multiple clients, which is 234 no longer an option since CVE-2016-0701. 235 236 This only affects processors that support the AVX2 but not ADX extensions 237 like Intel Haswell (4th generation). 238 239 This issue was reported to OpenSSL by David Benjamin (Google). The issue 240 was originally found via the OSS-Fuzz project. 241 (CVE-2017-3738) 242 [Andy Polyakov] 243 244 Changes between 1.0.2l and 1.0.2m [2 Nov 2017] 245 246 *) bn_sqrx8x_internal carry bug on x86_64 247 248 There is a carry propagating bug in the x86_64 Montgomery squaring 249 procedure. No EC algorithms are affected. Analysis suggests that attacks 250 against RSA and DSA as a result of this defect would be very difficult to 251 perform and are not believed likely. Attacks against DH are considered just 252 feasible (although very difficult) because most of the work necessary to 253 deduce information about a private key may be performed offline. The amount 254 of resources required for such an attack would be very significant and 255 likely only accessible to a limited number of attackers. An attacker would 256 additionally need online access to an unpatched system using the target 257 private key in a scenario with persistent DH parameters and a private 258 key that is shared between multiple clients. 259 260 This only affects processors that support the BMI1, BMI2 and ADX extensions 261 like Intel Broadwell (5th generation) and later or AMD Ryzen. 262 263 This issue was reported to OpenSSL by the OSS-Fuzz project. 264 (CVE-2017-3736) 265 [Andy Polyakov] 266 267 *) Malformed X.509 IPAddressFamily could cause OOB read 268 269 If an X.509 certificate has a malformed IPAddressFamily extension, 270 OpenSSL could do a one-byte buffer overread. The most likely result 271 would be an erroneous display of the certificate in text format. 272 273 This issue was reported to OpenSSL by the OSS-Fuzz project. 274 (CVE-2017-3735) 275 [Rich Salz] 276 277 Changes between 1.0.2k and 1.0.2l [25 May 2017] 278 279 *) Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target 280 platform rather than 'mingw'. 281 [Richard Levitte] 282 283 Changes between 1.0.2j and 1.0.2k [26 Jan 2017] 284 285 *) Truncated packet could crash via OOB read 286 287 If one side of an SSL/TLS path is running on a 32-bit host and a specific 288 cipher is being used, then a truncated packet can cause that host to 289 perform an out-of-bounds read, usually resulting in a crash. 290 291 This issue was reported to OpenSSL by Robert ��wi��cki of Google. 292 (CVE-2017-3731) 293 [Andy Polyakov] 294 295 *) BN_mod_exp may produce incorrect results on x86_64 296 297 There is a carry propagating bug in the x86_64 Montgomery squaring 298 procedure. No EC algorithms are affected. Analysis suggests that attacks 299 against RSA and DSA as a result of this defect would be very difficult to 300 perform and are not believed likely. Attacks against DH are considered just 301 feasible (although very difficult) because most of the work necessary to 302 deduce information about a private key may be performed offline. The amount 303 of resources required for such an attack would be very significant and 304 likely only accessible to a limited number of attackers. An attacker would 305 additionally need online access to an unpatched system using the target 306 private key in a scenario with persistent DH parameters and a private 307 key that is shared between multiple clients. For example this can occur by 308 default in OpenSSL DHE based SSL/TLS ciphersuites. Note: This issue is very 309 similar to CVE-2015-3193 but must be treated as a separate problem. 310 311 This issue was reported to OpenSSL by the OSS-Fuzz project. 312 (CVE-2017-3732) 313 [Andy Polyakov] 314 315 *) Montgomery multiplication may produce incorrect results 316 317 There is a carry propagating bug in the Broadwell-specific Montgomery 318 multiplication procedure that handles input lengths divisible by, but 319 longer than 256 bits. Analysis suggests that attacks against RSA, DSA 320 and DH private keys are impossible. This is because the subroutine in 321 question is not used in operations with the private key itself and an input 322 of the attacker's direct choice. Otherwise the bug can manifest itself as 323 transient authentication and key negotiation failures or reproducible 324 erroneous outcome of public-key operations with specially crafted input. 325 Among EC algorithms only Brainpool P-512 curves are affected and one 326 presumably can attack ECDH key negotiation. Impact was not analyzed in 327 detail, because pre-requisites for attack are considered unlikely. Namely 328 multiple clients have to choose the curve in question and the server has to 329 share the private key among them, neither of which is default behaviour. 330 Even then only clients that chose the curve will be affected. 331 332 This issue was publicly reported as transient failures and was not 333 initially recognized as a security issue. Thanks to Richard Morgan for 334 providing reproducible case. 335 (CVE-2016-7055) 336 [Andy Polyakov] 337 338 *) OpenSSL now fails if it receives an unrecognised record type in TLS1.0 339 or TLS1.1. Previously this only happened in SSLv3 and TLS1.2. This is to 340 prevent issues where no progress is being made and the peer continually 341 sends unrecognised record types, using up resources processing them. 342 [Matt Caswell] 343 344 Changes between 1.0.2i and 1.0.2j [26 Sep 2016] 345 346 *) Missing CRL sanity check 347 348 A bug fix which included a CRL sanity check was added to OpenSSL 1.1.0 349 but was omitted from OpenSSL 1.0.2i. As a result any attempt to use 350 CRLs in OpenSSL 1.0.2i will crash with a null pointer exception. 351 352 This issue only affects the OpenSSL 1.0.2i 353 (CVE-2016-7052) 354 [Matt Caswell] 355 356 Changes between 1.0.2h and 1.0.2i [22 Sep 2016] 357 358 *) OCSP Status Request extension unbounded memory growth 359 360 A malicious client can send an excessively large OCSP Status Request 361 extension. If that client continually requests renegotiation, sending a 362 large OCSP Status Request extension each time, then there will be unbounded 363 memory growth on the server. This will eventually lead to a Denial Of 364 Service attack through memory exhaustion. Servers with a default 365 configuration are vulnerable even if they do not support OCSP. Builds using 366 the "no-ocsp" build time option are not affected. 367 368 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 369 (CVE-2016-6304) 370 [Matt Caswell] 371 372 *) In order to mitigate the SWEET32 attack, the DES ciphers were moved from 373 HIGH to MEDIUM. 374 375 This issue was reported to OpenSSL Karthikeyan Bhargavan and Gaetan 376 Leurent (INRIA) 377 (CVE-2016-2183) 378 [Rich Salz] 379 380 *) OOB write in MDC2_Update() 381 382 An overflow can occur in MDC2_Update() either if called directly or 383 through the EVP_DigestUpdate() function using MDC2. If an attacker 384 is able to supply very large amounts of input data after a previous 385 call to EVP_EncryptUpdate() with a partial block then a length check 386 can overflow resulting in a heap corruption. 387 388 The amount of data needed is comparable to SIZE_MAX which is impractical 389 on most platforms. 390 391 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 392 (CVE-2016-6303) 393 [Stephen Henson] 394 395 *) Malformed SHA512 ticket DoS 396 397 If a server uses SHA512 for TLS session ticket HMAC it is vulnerable to a 398 DoS attack where a malformed ticket will result in an OOB read which will 399 ultimately crash. 400 401 The use of SHA512 in TLS session tickets is comparatively rare as it requires 402 a custom server callback and ticket lookup mechanism. 403 404 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 405 (CVE-2016-6302) 406 [Stephen Henson] 407 408 *) OOB write in BN_bn2dec() 409 410 The function BN_bn2dec() does not check the return value of BN_div_word(). 411 This can cause an OOB write if an application uses this function with an 412 overly large BIGNUM. This could be a problem if an overly large certificate 413 or CRL is printed out from an untrusted source. TLS is not affected because 414 record limits will reject an oversized certificate before it is parsed. 415 416 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 417 (CVE-2016-2182) 418 [Stephen Henson] 419 420 *) OOB read in TS_OBJ_print_bio() 421 422 The function TS_OBJ_print_bio() misuses OBJ_obj2txt(): the return value is 423 the total length the OID text representation would use and not the amount 424 of data written. This will result in OOB reads when large OIDs are 425 presented. 426 427 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 428 (CVE-2016-2180) 429 [Stephen Henson] 430 431 *) Pointer arithmetic undefined behaviour 432 433 Avoid some undefined pointer arithmetic 434 435 A common idiom in the codebase is to check limits in the following manner: 436 "p + len > limit" 437 438 Where "p" points to some malloc'd data of SIZE bytes and 439 limit == p + SIZE 440 441 "len" here could be from some externally supplied data (e.g. from a TLS 442 message). 443 444 The rules of C pointer arithmetic are such that "p + len" is only well 445 defined where len <= SIZE. Therefore the above idiom is actually 446 undefined behaviour. 447 448 For example this could cause problems if some malloc implementation 449 provides an address for "p" such that "p + len" actually overflows for 450 values of len that are too big and therefore p + len < limit. 451 452 This issue was reported to OpenSSL by Guido Vranken 453 (CVE-2016-2177) 454 [Matt Caswell] 455 456 *) Constant time flag not preserved in DSA signing 457 458 Operations in the DSA signing algorithm should run in constant time in 459 order to avoid side channel attacks. A flaw in the OpenSSL DSA 460 implementation means that a non-constant time codepath is followed for 461 certain operations. This has been demonstrated through a cache-timing 462 attack to be sufficient for an attacker to recover the private DSA key. 463 464 This issue was reported by C��sar Pereida (Aalto University), Billy Brumley 465 (Tampere University of Technology), and Yuval Yarom (The University of 466 Adelaide and NICTA). 467 (CVE-2016-2178) 468 [C��sar Pereida] 469 470 *) DTLS buffered message DoS 471 472 In a DTLS connection where handshake messages are delivered out-of-order 473 those messages that OpenSSL is not yet ready to process will be buffered 474 for later use. Under certain circumstances, a flaw in the logic means that 475 those messages do not get removed from the buffer even though the handshake 476 has been completed. An attacker could force up to approx. 15 messages to 477 remain in the buffer when they are no longer required. These messages will 478 be cleared when the DTLS connection is closed. The default maximum size for 479 a message is 100k. Therefore the attacker could force an additional 1500k 480 to be consumed per connection. By opening many simulataneous connections an 481 attacker could cause a DoS attack through memory exhaustion. 482 483 This issue was reported to OpenSSL by Quan Luo. 484 (CVE-2016-2179) 485 [Matt Caswell] 486 487 *) DTLS replay protection DoS 488 489 A flaw in the DTLS replay attack protection mechanism means that records 490 that arrive for future epochs update the replay protection "window" before 491 the MAC for the record has been validated. This could be exploited by an 492 attacker by sending a record for the next epoch (which does not have to 493 decrypt or have a valid MAC), with a very large sequence number. This means 494 that all subsequent legitimate packets are dropped causing a denial of 495 service for a specific DTLS connection. 496 497 This issue was reported to OpenSSL by the OCAP audit team. 498 (CVE-2016-2181) 499 [Matt Caswell] 500 501 *) Certificate message OOB reads 502 503 In OpenSSL 1.0.2 and earlier some missing message length checks can result 504 in OOB reads of up to 2 bytes beyond an allocated buffer. There is a 505 theoretical DoS risk but this has not been observed in practice on common 506 platforms. 507 508 The messages affected are client certificate, client certificate request 509 and server certificate. As a result the attack can only be performed 510 against a client or a server which enables client authentication. 511 512 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 513 (CVE-2016-6306) 514 [Stephen Henson] 515 516 Changes between 1.0.2g and 1.0.2h [3 May 2016] 517 518 *) Prevent padding oracle in AES-NI CBC MAC check 519 520 A MITM attacker can use a padding oracle attack to decrypt traffic 521 when the connection uses an AES CBC cipher and the server support 522 AES-NI. 523 524 This issue was introduced as part of the fix for Lucky 13 padding 525 attack (CVE-2013-0169). The padding check was rewritten to be in 526 constant time by making sure that always the same bytes are read and 527 compared against either the MAC or padding bytes. But it no longer 528 checked that there was enough data to have both the MAC and padding 529 bytes. 530 531 This issue was reported by Juraj Somorovsky using TLS-Attacker. 532 (CVE-2016-2107) 533 [Kurt Roeckx] 534 535 *) Fix EVP_EncodeUpdate overflow 536 537 An overflow can occur in the EVP_EncodeUpdate() function which is used for 538 Base64 encoding of binary data. If an attacker is able to supply very large 539 amounts of input data then a length check can overflow resulting in a heap 540 corruption. 541 542 Internally to OpenSSL the EVP_EncodeUpdate() function is primarly used by 543 the PEM_write_bio* family of functions. These are mainly used within the 544 OpenSSL command line applications, so any application which processes data 545 from an untrusted source and outputs it as a PEM file should be considered 546 vulnerable to this issue. User applications that call these APIs directly 547 with large amounts of untrusted data may also be vulnerable. 548 549 This issue was reported by Guido Vranken. 550 (CVE-2016-2105) 551 [Matt Caswell] 552 553 *) Fix EVP_EncryptUpdate overflow 554 555 An overflow can occur in the EVP_EncryptUpdate() function. If an attacker 556 is able to supply very large amounts of input data after a previous call to 557 EVP_EncryptUpdate() with a partial block then a length check can overflow 558 resulting in a heap corruption. Following an analysis of all OpenSSL 559 internal usage of the EVP_EncryptUpdate() function all usage is one of two 560 forms. The first form is where the EVP_EncryptUpdate() call is known to be 561 the first called function after an EVP_EncryptInit(), and therefore that 562 specific call must be safe. The second form is where the length passed to 563 EVP_EncryptUpdate() can be seen from the code to be some small value and 564 therefore there is no possibility of an overflow. Since all instances are 565 one of these two forms, it is believed that there can be no overflows in 566 internal code due to this problem. It should be noted that 567 EVP_DecryptUpdate() can call EVP_EncryptUpdate() in certain code paths. 568 Also EVP_CipherUpdate() is a synonym for EVP_EncryptUpdate(). All instances 569 of these calls have also been analysed too and it is believed there are no 570 instances in internal usage where an overflow could occur. 571 572 This issue was reported by Guido Vranken. 573 (CVE-2016-2106) 574 [Matt Caswell] 575 576 *) Prevent ASN.1 BIO excessive memory allocation 577 578 When ASN.1 data is read from a BIO using functions such as d2i_CMS_bio() 579 a short invalid encoding can casuse allocation of large amounts of memory 580 potentially consuming excessive resources or exhausting memory. 581 582 Any application parsing untrusted data through d2i BIO functions is 583 affected. The memory based functions such as d2i_X509() are *not* affected. 584 Since the memory based functions are used by the TLS library, TLS 585 applications are not affected. 586 587 This issue was reported by Brian Carpenter. 588 (CVE-2016-2109) 589 [Stephen Henson] 590 591 *) EBCDIC overread 592 593 ASN1 Strings that are over 1024 bytes can cause an overread in applications 594 using the X509_NAME_oneline() function on EBCDIC systems. This could result 595 in arbitrary stack data being returned in the buffer. 596 597 This issue was reported by Guido Vranken. 598 (CVE-2016-2176) 599 [Matt Caswell] 600 601 *) Modify behavior of ALPN to invoke callback after SNI/servername 602 callback, such that updates to the SSL_CTX affect ALPN. 603 [Todd Short] 604 605 *) Remove LOW from the DEFAULT cipher list. This removes singles DES from the 606 default. 607 [Kurt Roeckx] 608 609 *) Only remove the SSLv2 methods with the no-ssl2-method option. When the 610 methods are enabled and ssl2 is disabled the methods return NULL. 611 [Kurt Roeckx] 612 613 Changes between 1.0.2f and 1.0.2g [1 Mar 2016] 614 615 * Disable weak ciphers in SSLv3 and up in default builds of OpenSSL. 616 Builds that are not configured with "enable-weak-ssl-ciphers" will not 617 provide any "EXPORT" or "LOW" strength ciphers. 618 [Viktor Dukhovni] 619 620 * Disable SSLv2 default build, default negotiation and weak ciphers. SSLv2 621 is by default disabled at build-time. Builds that are not configured with 622 "enable-ssl2" will not support SSLv2. Even if "enable-ssl2" is used, 623 users who want to negotiate SSLv2 via the version-flexible SSLv23_method() 624 will need to explicitly call either of: 625 626 SSL_CTX_clear_options(ctx, SSL_OP_NO_SSLv2); 627 or 628 SSL_clear_options(ssl, SSL_OP_NO_SSLv2); 629 630 as appropriate. Even if either of those is used, or the application 631 explicitly uses the version-specific SSLv2_method() or its client and 632 server variants, SSLv2 ciphers vulnerable to exhaustive search key 633 recovery have been removed. Specifically, the SSLv2 40-bit EXPORT 634 ciphers, and SSLv2 56-bit DES are no longer available. 635 (CVE-2016-0800) 636 [Viktor Dukhovni] 637 638 *) Fix a double-free in DSA code 639 640 A double free bug was discovered when OpenSSL parses malformed DSA private 641 keys and could lead to a DoS attack or memory corruption for applications 642 that receive DSA private keys from untrusted sources. This scenario is 643 considered rare. 644 645 This issue was reported to OpenSSL by Adam Langley(Google/BoringSSL) using 646 libFuzzer. 647 (CVE-2016-0705) 648 [Stephen Henson] 649 650 *) Disable SRP fake user seed to address a server memory leak. 651 652 Add a new method SRP_VBASE_get1_by_user that handles the seed properly. 653 654 SRP_VBASE_get_by_user had inconsistent memory management behaviour. 655 In order to fix an unavoidable memory leak, SRP_VBASE_get_by_user 656 was changed to ignore the "fake user" SRP seed, even if the seed 657 is configured. 658 659 Users should use SRP_VBASE_get1_by_user instead. Note that in 660 SRP_VBASE_get1_by_user, caller must free the returned value. Note 661 also that even though configuring the SRP seed attempts to hide 662 invalid usernames by continuing the handshake with fake 663 credentials, this behaviour is not constant time and no strong 664 guarantees are made that the handshake is indistinguishable from 665 that of a valid user. 666 (CVE-2016-0798) 667 [Emilia K��sper] 668 669 *) Fix BN_hex2bn/BN_dec2bn NULL pointer deref/heap corruption 670 671 In the BN_hex2bn function the number of hex digits is calculated using an 672 int value |i|. Later |bn_expand| is called with a value of |i * 4|. For 673 large values of |i| this can result in |bn_expand| not allocating any 674 memory because |i * 4| is negative. This can leave the internal BIGNUM data 675 field as NULL leading to a subsequent NULL ptr deref. For very large values 676 of |i|, the calculation |i * 4| could be a positive value smaller than |i|. 677 In this case memory is allocated to the internal BIGNUM data field, but it 678 is insufficiently sized leading to heap corruption. A similar issue exists 679 in BN_dec2bn. This could have security consequences if BN_hex2bn/BN_dec2bn 680 is ever called by user applications with very large untrusted hex/dec data. 681 This is anticipated to be a rare occurrence. 682 683 All OpenSSL internal usage of these functions use data that is not expected 684 to be untrusted, e.g. config file data or application command line 685 arguments. If user developed applications generate config file data based 686 on untrusted data then it is possible that this could also lead to security 687 consequences. This is also anticipated to be rare. 688 689 This issue was reported to OpenSSL by Guido Vranken. 690 (CVE-2016-0797) 691 [Matt Caswell] 692 693 *) Fix memory issues in BIO_*printf functions 694 695 The internal |fmtstr| function used in processing a "%s" format string in 696 the BIO_*printf functions could overflow while calculating the length of a 697 string and cause an OOB read when printing very long strings. 698 699 Additionally the internal |doapr_outch| function can attempt to write to an 700 OOB memory location (at an offset from the NULL pointer) in the event of a 701 memory allocation failure. In 1.0.2 and below this could be caused where 702 the size of a buffer to be allocated is greater than INT_MAX. E.g. this 703 could be in processing a very long "%s" format string. Memory leaks can 704 also occur. 705 706 The first issue may mask the second issue dependent on compiler behaviour. 707 These problems could enable attacks where large amounts of untrusted data 708 is passed to the BIO_*printf functions. If applications use these functions 709 in this way then they could be vulnerable. OpenSSL itself uses these 710 functions when printing out human-readable dumps of ASN.1 data. Therefore 711 applications that print this data could be vulnerable if the data is from 712 untrusted sources. OpenSSL command line applications could also be 713 vulnerable where they print out ASN.1 data, or if untrusted data is passed 714 as command line arguments. 715 716 Libssl is not considered directly vulnerable. Additionally certificates etc 717 received via remote connections via libssl are also unlikely to be able to 718 trigger these issues because of message size limits enforced within libssl. 719 720 This issue was reported to OpenSSL Guido Vranken. 721 (CVE-2016-0799) 722 [Matt Caswell] 723 724 *) Side channel attack on modular exponentiation 725 726 A side-channel attack was found which makes use of cache-bank conflicts on 727 the Intel Sandy-Bridge microarchitecture which could lead to the recovery 728 of RSA keys. The ability to exploit this issue is limited as it relies on 729 an attacker who has control of code in a thread running on the same 730 hyper-threaded core as the victim thread which is performing decryptions. 731 732 This issue was reported to OpenSSL by Yuval Yarom, The University of 733 Adelaide and NICTA, Daniel Genkin, Technion and Tel Aviv University, and 734 Nadia Heninger, University of Pennsylvania with more information at 735 http://cachebleed.info. 736 (CVE-2016-0702) 737 [Andy Polyakov] 738 739 *) Change the req app to generate a 2048-bit RSA/DSA key by default, 740 if no keysize is specified with default_bits. This fixes an 741 omission in an earlier change that changed all RSA/DSA key generation 742 apps to use 2048 bits by default. 743 [Emilia K��sper] 744 745 Changes between 1.0.2e and 1.0.2f [28 Jan 2016] 746 747 *) DH small subgroups 748 749 Historically OpenSSL only ever generated DH parameters based on "safe" 750 primes. More recently (in version 1.0.2) support was provided for 751 generating X9.42 style parameter files such as those required for RFC 5114 752 support. The primes used in such files may not be "safe". Where an 753 application is using DH configured with parameters based on primes that are 754 not "safe" then an attacker could use this fact to find a peer's private 755 DH exponent. This attack requires that the attacker complete multiple 756 handshakes in which the peer uses the same private DH exponent. For example 757 this could be used to discover a TLS server's private DH exponent if it's 758 reusing the private DH exponent or it's using a static DH ciphersuite. 759 760 OpenSSL provides the option SSL_OP_SINGLE_DH_USE for ephemeral DH (DHE) in 761 TLS. It is not on by default. If the option is not set then the server 762 reuses the same private DH exponent for the life of the server process and 763 would be vulnerable to this attack. It is believed that many popular 764 applications do set this option and would therefore not be at risk. 765 766 The fix for this issue adds an additional check where a "q" parameter is 767 available (as is the case in X9.42 based parameters). This detects the 768 only known attack, and is the only possible defense for static DH 769 ciphersuites. This could have some performance impact. 770 771 Additionally the SSL_OP_SINGLE_DH_USE option has been switched on by 772 default and cannot be disabled. This could have some performance impact. 773 774 This issue was reported to OpenSSL by Antonio Sanso (Adobe). 775 (CVE-2016-0701) 776 [Matt Caswell] 777 778 *) SSLv2 doesn't block disabled ciphers 779 780 A malicious client can negotiate SSLv2 ciphers that have been disabled on 781 the server and complete SSLv2 handshakes even if all SSLv2 ciphers have 782 been disabled, provided that the SSLv2 protocol was not also disabled via 783 SSL_OP_NO_SSLv2. 784 785 This issue was reported to OpenSSL on 26th December 2015 by Nimrod Aviram 786 and Sebastian Schinzel. 787 (CVE-2015-3197) 788 [Viktor Dukhovni] 789 790 *) Reject DH handshakes with parameters shorter than 1024 bits. 791 [Kurt Roeckx] 792 793 Changes between 1.0.2d and 1.0.2e [3 Dec 2015] 794 795 *) BN_mod_exp may produce incorrect results on x86_64 796 797 There is a carry propagating bug in the x86_64 Montgomery squaring 798 procedure. No EC algorithms are affected. Analysis suggests that attacks 799 against RSA and DSA as a result of this defect would be very difficult to 800 perform and are not believed likely. Attacks against DH are considered just 801 feasible (although very difficult) because most of the work necessary to 802 deduce information about a private key may be performed offline. The amount 803 of resources required for such an attack would be very significant and 804 likely only accessible to a limited number of attackers. An attacker would 805 additionally need online access to an unpatched system using the target 806 private key in a scenario with persistent DH parameters and a private 807 key that is shared between multiple clients. For example this can occur by 808 default in OpenSSL DHE based SSL/TLS ciphersuites. 809 810 This issue was reported to OpenSSL by Hanno B��ck. 811 (CVE-2015-3193) 812 [Andy Polyakov] 813 814 *) Certificate verify crash with missing PSS parameter 815 816 The signature verification routines will crash with a NULL pointer 817 dereference if presented with an ASN.1 signature using the RSA PSS 818 algorithm and absent mask generation function parameter. Since these 819 routines are used to verify certificate signature algorithms this can be 820 used to crash any certificate verification operation and exploited in a 821 DoS attack. Any application which performs certificate verification is 822 vulnerable including OpenSSL clients and servers which enable client 823 authentication. 824 825 This issue was reported to OpenSSL by Lo��c Jonas Etienne (Qnective AG). 826 (CVE-2015-3194) 827 [Stephen Henson] 828 829 *) X509_ATTRIBUTE memory leak 830 831 When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak 832 memory. This structure is used by the PKCS#7 and CMS routines so any 833 application which reads PKCS#7 or CMS data from untrusted sources is 834 affected. SSL/TLS is not affected. 835 836 This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using 837 libFuzzer. 838 (CVE-2015-3195) 839 [Stephen Henson] 840 841 *) Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs. 842 This changes the decoding behaviour for some invalid messages, 843 though the change is mostly in the more lenient direction, and 844 legacy behaviour is preserved as much as possible. 845 [Emilia K��sper] 846 847 *) In DSA_generate_parameters_ex, if the provided seed is too short, 848 use a random seed, as already documented. 849 [Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>] 850 851 Changes between 1.0.2c and 1.0.2d [9 Jul 2015] 852 853 *) Alternate chains certificate forgery 854 855 During certificate verfification, OpenSSL will attempt to find an 856 alternative certificate chain if the first attempt to build such a chain 857 fails. An error in the implementation of this logic can mean that an 858 attacker could cause certain checks on untrusted certificates to be 859 bypassed, such as the CA flag, enabling them to use a valid leaf 860 certificate to act as a CA and "issue" an invalid certificate. 861 862 This issue was reported to OpenSSL by Adam Langley/David Benjamin 863 (Google/BoringSSL). 864 (CVE-2015-1793) 865 [Matt Caswell] 866 867 *) Race condition handling PSK identify hint 868 869 If PSK identity hints are received by a multi-threaded client then 870 the values are wrongly updated in the parent SSL_CTX structure. This can 871 result in a race condition potentially leading to a double free of the 872 identify hint data. 873 (CVE-2015-3196) 874 [Stephen Henson] 875 876 Changes between 1.0.2b and 1.0.2c [12 Jun 2015] 877 878 *) Fix HMAC ABI incompatibility. The previous version introduced an ABI 879 incompatibility in the handling of HMAC. The previous ABI has now been 880 restored. 881 882 Changes between 1.0.2a and 1.0.2b [11 Jun 2015] 883 884 *) Malformed ECParameters causes infinite loop 885 886 When processing an ECParameters structure OpenSSL enters an infinite loop 887 if the curve specified is over a specially malformed binary polynomial 888 field. 889 890 This can be used to perform denial of service against any 891 system which processes public keys, certificate requests or 892 certificates. This includes TLS clients and TLS servers with 893 client authentication enabled. 894 895 This issue was reported to OpenSSL by Joseph Barr-Pixton. 896 (CVE-2015-1788) 897 [Andy Polyakov] 898 899 *) Exploitable out-of-bounds read in X509_cmp_time 900 901 X509_cmp_time does not properly check the length of the ASN1_TIME 902 string and can read a few bytes out of bounds. In addition, 903 X509_cmp_time accepts an arbitrary number of fractional seconds in the 904 time string. 905 906 An attacker can use this to craft malformed certificates and CRLs of 907 various sizes and potentially cause a segmentation fault, resulting in 908 a DoS on applications that verify certificates or CRLs. TLS clients 909 that verify CRLs are affected. TLS clients and servers with client 910 authentication enabled may be affected if they use custom verification 911 callbacks. 912 913 This issue was reported to OpenSSL by Robert Swiecki (Google), and 914 independently by Hanno B��ck. 915 (CVE-2015-1789) 916 [Emilia K��sper] 917 918 *) PKCS7 crash with missing EnvelopedContent 919 920 The PKCS#7 parsing code does not handle missing inner EncryptedContent 921 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs 922 with missing content and trigger a NULL pointer dereference on parsing. 923 924 Applications that decrypt PKCS#7 data or otherwise parse PKCS#7 925 structures from untrusted sources are affected. OpenSSL clients and 926 servers are not affected. 927 928 This issue was reported to OpenSSL by Michal Zalewski (Google). 929 (CVE-2015-1790) 930 [Emilia K��sper] 931 932 *) CMS verify infinite loop with unknown hash function 933 934 When verifying a signedData message the CMS code can enter an infinite loop 935 if presented with an unknown hash function OID. This can be used to perform 936 denial of service against any system which verifies signedData messages using 937 the CMS code. 938 This issue was reported to OpenSSL by Johannes Bauer. 939 (CVE-2015-1792) 940 [Stephen Henson] 941 942 *) Race condition handling NewSessionTicket 943 944 If a NewSessionTicket is received by a multi-threaded client when attempting to 945 reuse a previous ticket then a race condition can occur potentially leading to 946 a double free of the ticket data. 947 (CVE-2015-1791) 948 [Matt Caswell] 949 950 *) Removed support for the two export grade static DH ciphersuites 951 EXP-DH-RSA-DES-CBC-SHA and EXP-DH-DSS-DES-CBC-SHA. These two ciphersuites 952 were newly added (along with a number of other static DH ciphersuites) to 953 1.0.2. However the two export ones have *never* worked since they were 954 introduced. It seems strange in any case to be adding new export 955 ciphersuites, and given "logjam" it also does not seem correct to fix them. 956 [Matt Caswell] 957 958 *) Only support 256-bit or stronger elliptic curves with the 959 'ecdh_auto' setting (server) or by default (client). Of supported 960 curves, prefer P-256 (both). 961 [Emilia Kasper] 962 963 *) Reject DH handshakes with parameters shorter than 768 bits. 964 [Kurt Roeckx and Emilia Kasper] 965 966 Changes between 1.0.2 and 1.0.2a [19 Mar 2015] 967 968 *) ClientHello sigalgs DoS fix 969 970 If a client connects to an OpenSSL 1.0.2 server and renegotiates with an 971 invalid signature algorithms extension a NULL pointer dereference will 972 occur. This can be exploited in a DoS attack against the server. 973 974 This issue was was reported to OpenSSL by David Ramos of Stanford 975 University. 976 (CVE-2015-0291) 977 [Stephen Henson and Matt Caswell] 978 979 *) Multiblock corrupted pointer fix 980 981 OpenSSL 1.0.2 introduced the "multiblock" performance improvement. This 982 feature only applies on 64 bit x86 architecture platforms that support AES 983 NI instructions. A defect in the implementation of "multiblock" can cause 984 OpenSSL's internal write buffer to become incorrectly set to NULL when 985 using non-blocking IO. Typically, when the user application is using a 986 socket BIO for writing, this will only result in a failed connection. 987 However if some other BIO is used then it is likely that a segmentation 988 fault will be triggered, thus enabling a potential DoS attack. 989 990 This issue was reported to OpenSSL by Daniel Danner and Rainer Mueller. 991 (CVE-2015-0290) 992 [Matt Caswell] 993 994 *) Segmentation fault in DTLSv1_listen fix 995 996 The DTLSv1_listen function is intended to be stateless and processes the 997 initial ClientHello from many peers. It is common for user code to loop 998 over the call to DTLSv1_listen until a valid ClientHello is received with 999 an associated cookie. A defect in the implementation of DTLSv1_listen means 1000 that state is preserved in the SSL object from one invocation to the next 1001 that can lead to a segmentation fault. Errors processing the initial 1002 ClientHello can trigger this scenario. An example of such an error could be 1003 that a DTLS1.0 only client is attempting to connect to a DTLS1.2 only 1004 server. 1005 1006 This issue was reported to OpenSSL by Per Allansson. 1007 (CVE-2015-0207) 1008 [Matt Caswell] 1009 1010 *) Segmentation fault in ASN1_TYPE_cmp fix 1011 1012 The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is 1013 made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check 1014 certificate signature algorithm consistency this can be used to crash any 1015 certificate verification operation and exploited in a DoS attack. Any 1016 application which performs certificate verification is vulnerable including 1017 OpenSSL clients and servers which enable client authentication. 1018 (CVE-2015-0286) 1019 [Stephen Henson] 1020 1021 *) Segmentation fault for invalid PSS parameters fix 1022 1023 The signature verification routines will crash with a NULL pointer 1024 dereference if presented with an ASN.1 signature using the RSA PSS 1025 algorithm and invalid parameters. Since these routines are used to verify 1026 certificate signature algorithms this can be used to crash any 1027 certificate verification operation and exploited in a DoS attack. Any 1028 application which performs certificate verification is vulnerable including 1029 OpenSSL clients and servers which enable client authentication. 1030 1031 This issue was was reported to OpenSSL by Brian Carpenter. 1032 (CVE-2015-0208) 1033 [Stephen Henson] 1034 1035 *) ASN.1 structure reuse memory corruption fix 1036 1037 Reusing a structure in ASN.1 parsing may allow an attacker to cause 1038 memory corruption via an invalid write. Such reuse is and has been 1039 strongly discouraged and is believed to be rare. 1040 1041 Applications that parse structures containing CHOICE or ANY DEFINED BY 1042 components may be affected. Certificate parsing (d2i_X509 and related 1043 functions) are however not affected. OpenSSL clients and servers are 1044 not affected. 1045 (CVE-2015-0287) 1046 [Stephen Henson] 1047 1048 *) PKCS7 NULL pointer dereferences fix 1049 1050 The PKCS#7 parsing code does not handle missing outer ContentInfo 1051 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with 1052 missing content and trigger a NULL pointer dereference on parsing. 1053 1054 Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or 1055 otherwise parse PKCS#7 structures from untrusted sources are 1056 affected. OpenSSL clients and servers are not affected. 1057 1058 This issue was reported to OpenSSL by Michal Zalewski (Google). 1059 (CVE-2015-0289) 1060 [Emilia K��sper] 1061 1062 *) DoS via reachable assert in SSLv2 servers fix 1063 1064 A malicious client can trigger an OPENSSL_assert (i.e., an abort) in 1065 servers that both support SSLv2 and enable export cipher suites by sending 1066 a specially crafted SSLv2 CLIENT-MASTER-KEY message. 1067 1068 This issue was discovered by Sean Burford (Google) and Emilia K��sper 1069 (OpenSSL development team). 1070 (CVE-2015-0293) 1071 [Emilia K��sper] 1072 1073 *) Empty CKE with client auth and DHE fix 1074 1075 If client auth is used then a server can seg fault in the event of a DHE 1076 ciphersuite being selected and a zero length ClientKeyExchange message 1077 being sent by the client. This could be exploited in a DoS attack. 1078 (CVE-2015-1787) 1079 [Matt Caswell] 1080 1081 *) Handshake with unseeded PRNG fix 1082 1083 Under certain conditions an OpenSSL 1.0.2 client can complete a handshake 1084 with an unseeded PRNG. The conditions are: 1085 - The client is on a platform where the PRNG has not been seeded 1086 automatically, and the user has not seeded manually 1087 - A protocol specific client method version has been used (i.e. not 1088 SSL_client_methodv23) 1089 - A ciphersuite is used that does not require additional random data from 1090 the PRNG beyond the initial ClientHello client random (e.g. PSK-RC4-SHA). 1091 1092 If the handshake succeeds then the client random that has been used will 1093 have been generated from a PRNG with insufficient entropy and therefore the 1094 output may be predictable. 1095 1096 For example using the following command with an unseeded openssl will 1097 succeed on an unpatched platform: 1098 1099 openssl s_client -psk 1a2b3c4d -tls1_2 -cipher PSK-RC4-SHA 1100 (CVE-2015-0285) 1101 [Matt Caswell] 1102 1103 *) Use After Free following d2i_ECPrivatekey error fix 1104 1105 A malformed EC private key file consumed via the d2i_ECPrivateKey function 1106 could cause a use after free condition. This, in turn, could cause a double 1107 free in several private key parsing functions (such as d2i_PrivateKey 1108 or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption 1109 for applications that receive EC private keys from untrusted 1110 sources. This scenario is considered rare. 1111 1112 This issue was discovered by the BoringSSL project and fixed in their 1113 commit 517073cd4b. 1114 (CVE-2015-0209) 1115 [Matt Caswell] 1116 1117 *) X509_to_X509_REQ NULL pointer deref fix 1118 1119 The function X509_to_X509_REQ will crash with a NULL pointer dereference if 1120 the certificate key is invalid. This function is rarely used in practice. 1121 1122 This issue was discovered by Brian Carpenter. 1123 (CVE-2015-0288) 1124 [Stephen Henson] 1125 1126 *) Removed the export ciphers from the DEFAULT ciphers 1127 [Kurt Roeckx] 1128 1129 Changes between 1.0.1l and 1.0.2 [22 Jan 2015] 1130 1131 *) Change RSA and DH/DSA key generation apps to generate 2048-bit 1132 keys by default. 1133 [Kurt Roeckx] 1134 1135 *) Facilitate "universal" ARM builds targeting range of ARM ISAs, e.g. 1136 ARMv5 through ARMv8, as opposite to "locking" it to single one. 1137 So far those who have to target multiple plaforms would compromise 1138 and argue that binary targeting say ARMv5 would still execute on 1139 ARMv8. "Universal" build resolves this compromise by providing 1140 near-optimal performance even on newer platforms. 1141 [Andy Polyakov] 1142 1143 *) Accelerated NIST P-256 elliptic curve implementation for x86_64 1144 (other platforms pending). 1145 [Shay Gueron & Vlad Krasnov (Intel Corp), Andy Polyakov] 1146 1147 *) Add support for the SignedCertificateTimestampList certificate and 1148 OCSP response extensions from RFC6962. 1149 [Rob Stradling] 1150 1151 *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.) 1152 for corner cases. (Certain input points at infinity could lead to 1153 bogus results, with non-infinity inputs mapped to infinity too.) 1154 [Bodo Moeller] 1155 1156 *) Initial support for PowerISA 2.0.7, first implemented in POWER8. 1157 This covers AES, SHA256/512 and GHASH. "Initial" means that most 1158 common cases are optimized and there still is room for further 1159 improvements. Vector Permutation AES for Altivec is also added. 1160 [Andy Polyakov] 1161 1162 *) Add support for little-endian ppc64 Linux target. 1163 [Marcelo Cerri (IBM)] 1164 1165 *) Initial support for AMRv8 ISA crypto extensions. This covers AES, 1166 SHA1, SHA256 and GHASH. "Initial" means that most common cases 1167 are optimized and there still is room for further improvements. 1168 Both 32- and 64-bit modes are supported. 1169 [Andy Polyakov, Ard Biesheuvel (Linaro)] 1170 1171 *) Improved ARMv7 NEON support. 1172 [Andy Polyakov] 1173 1174 *) Support for SPARC Architecture 2011 crypto extensions, first 1175 implemented in SPARC T4. This covers AES, DES, Camellia, SHA1, 1176 SHA256/512, MD5, GHASH and modular exponentiation. 1177 [Andy Polyakov, David Miller] 1178 1179 *) Accelerated modular exponentiation for Intel processors, a.k.a. 1180 RSAZ. 1181 [Shay Gueron & Vlad Krasnov (Intel Corp)] 1182 1183 *) Support for new and upcoming Intel processors, including AVX2, 1184 BMI and SHA ISA extensions. This includes additional "stitched" 1185 implementations, AESNI-SHA256 and GCM, and multi-buffer support 1186 for TLS encrypt. 1187 1188 This work was sponsored by Intel Corp. 1189 [Andy Polyakov] 1190 1191 *) Support for DTLS 1.2. This adds two sets of DTLS methods: DTLS_*_method() 1192 supports both DTLS 1.2 and 1.0 and should use whatever version the peer 1193 supports and DTLSv1_2_*_method() which supports DTLS 1.2 only. 1194 [Steve Henson] 1195 1196 *) Use algorithm specific chains in SSL_CTX_use_certificate_chain_file(): 1197 this fixes a limiation in previous versions of OpenSSL. 1198 [Steve Henson] 1199 1200 *) Extended RSA OAEP support via EVP_PKEY API. Options to specify digest, 1201 MGF1 digest and OAEP label. 1202 [Steve Henson] 1203 1204 *) Add EVP support for key wrapping algorithms, to avoid problems with 1205 existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in 1206 the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap 1207 algorithms and include tests cases. 1208 [Steve Henson] 1209 1210 *) Add functions to allocate and set the fields of an ECDSA_METHOD 1211 structure. 1212 [Douglas E. Engert, Steve Henson] 1213 1214 *) New functions OPENSSL_gmtime_diff and ASN1_TIME_diff to find the 1215 difference in days and seconds between two tm or ASN1_TIME structures. 1216 [Steve Henson] 1217 1218 *) Add -rev test option to s_server to just reverse order of characters 1219 received by client and send back to server. Also prints an abbreviated 1220 summary of the connection parameters. 1221 [Steve Henson] 1222 1223 *) New option -brief for s_client and s_server to print out a brief summary 1224 of connection parameters. 1225 [Steve Henson] 1226 1227 *) Add callbacks for arbitrary TLS extensions. 1228 [Trevor Perrin <trevp@trevp.net> and Ben Laurie] 1229 1230 *) New option -crl_download in several openssl utilities to download CRLs 1231 from CRLDP extension in certificates. 1232 [Steve Henson] 1233 1234 *) New options -CRL and -CRLform for s_client and s_server for CRLs. 1235 [Steve Henson] 1236 1237 *) New function X509_CRL_diff to generate a delta CRL from the difference 1238 of two full CRLs. Add support to "crl" utility. 1239 [Steve Henson] 1240 1241 *) New functions to set lookup_crls function and to retrieve 1242 X509_STORE from X509_STORE_CTX. 1243 [Steve Henson] 1244 1245 *) Print out deprecated issuer and subject unique ID fields in 1246 certificates. 1247 [Steve Henson] 1248 1249 *) Extend OCSP I/O functions so they can be used for simple general purpose 1250 HTTP as well as OCSP. New wrapper function which can be used to download 1251 CRLs using the OCSP API. 1252 [Steve Henson] 1253 1254 *) Delegate command line handling in s_client/s_server to SSL_CONF APIs. 1255 [Steve Henson] 1256 1257 *) SSL_CONF* functions. These provide a common framework for application 1258 configuration using configuration files or command lines. 1259 [Steve Henson] 1260 1261 *) SSL/TLS tracing code. This parses out SSL/TLS records using the 1262 message callback and prints the results. Needs compile time option 1263 "enable-ssl-trace". New options to s_client and s_server to enable 1264 tracing. 1265 [Steve Henson] 1266 1267 *) New ctrl and macro to retrieve supported points extensions. 1268 Print out extension in s_server and s_client. 1269 [Steve Henson] 1270 1271 *) New functions to retrieve certificate signature and signature 1272 OID NID. 1273 [Steve Henson] 1274 1275 *) Add functions to retrieve and manipulate the raw cipherlist sent by a 1276 client to OpenSSL. 1277 [Steve Henson] 1278 1279 *) New Suite B modes for TLS code. These use and enforce the requirements 1280 of RFC6460: restrict ciphersuites, only permit Suite B algorithms and 1281 only use Suite B curves. The Suite B modes can be set by using the 1282 strings "SUITEB128", "SUITEB192" or "SUITEB128ONLY" for the cipherstring. 1283 [Steve Henson] 1284 1285 *) New chain verification flags for Suite B levels of security. Check 1286 algorithms are acceptable when flags are set in X509_verify_cert. 1287 [Steve Henson] 1288 1289 *) Make tls1_check_chain return a set of flags indicating checks passed 1290 by a certificate chain. Add additional tests to handle client 1291 certificates: checks for matching certificate type and issuer name 1292 comparison. 1293 [Steve Henson] 1294 1295 *) If an attempt is made to use a signature algorithm not in the peer 1296 preference list abort the handshake. If client has no suitable 1297 signature algorithms in response to a certificate request do not 1298 use the certificate. 1299 [Steve Henson] 1300 1301 *) If server EC tmp key is not in client preference list abort handshake. 1302 [Steve Henson] 1303 1304 *) Add support for certificate stores in CERT structure. This makes it 1305 possible to have different stores per SSL structure or one store in 1306 the parent SSL_CTX. Include distint stores for certificate chain 1307 verification and chain building. New ctrl SSL_CTRL_BUILD_CERT_CHAIN 1308 to build and store a certificate chain in CERT structure: returing 1309 an error if the chain cannot be built: this will allow applications 1310 to test if a chain is correctly configured. 1311 1312 Note: if the CERT based stores are not set then the parent SSL_CTX 1313 store is used to retain compatibility with existing behaviour. 1314 1315 [Steve Henson] 1316 1317 *) New function ssl_set_client_disabled to set a ciphersuite disabled 1318 mask based on the current session, check mask when sending client 1319 hello and checking the requested ciphersuite. 1320 [Steve Henson] 1321 1322 *) New ctrls to retrieve and set certificate types in a certificate 1323 request message. Print out received values in s_client. If certificate 1324 types is not set with custom values set sensible values based on 1325 supported signature algorithms. 1326 [Steve Henson] 1327 1328 *) Support for distinct client and server supported signature algorithms. 1329 [Steve Henson] 1330 1331 *) Add certificate callback. If set this is called whenever a certificate 1332 is required by client or server. An application can decide which 1333 certificate chain to present based on arbitrary criteria: for example 1334 supported signature algorithms. Add very simple example to s_server. 1335 This fixes many of the problems and restrictions of the existing client 1336 certificate callback: for example you can now clear an existing 1337 certificate and specify the whole chain. 1338 [Steve Henson] 1339 1340 *) Add new "valid_flags" field to CERT_PKEY structure which determines what 1341 the certificate can be used for (if anything). Set valid_flags field 1342 in new tls1_check_chain function. Simplify ssl_set_cert_masks which used 1343 to have similar checks in it. 1344 1345 Add new "cert_flags" field to CERT structure and include a "strict mode". 1346 This enforces some TLS certificate requirements (such as only permitting 1347 certificate signature algorithms contained in the supported algorithms 1348 extension) which some implementations ignore: this option should be used 1349 with caution as it could cause interoperability issues. 1350 [Steve Henson] 1351 1352 *) Update and tidy signature algorithm extension processing. Work out 1353 shared signature algorithms based on preferences and peer algorithms 1354 and print them out in s_client and s_server. Abort handshake if no 1355 shared signature algorithms. 1356 [Steve Henson] 1357 1358 *) Add new functions to allow customised supported signature algorithms 1359 for SSL and SSL_CTX structures. Add options to s_client and s_server 1360 to support them. 1361 [Steve Henson] 1362 1363 *) New function SSL_certs_clear() to delete all references to certificates 1364 from an SSL structure. Before this once a certificate had been added 1365 it couldn't be removed. 1366 [Steve Henson] 1367 1368 *) Integrate hostname, email address and IP address checking with certificate 1369 verification. New verify options supporting checking in opensl utility. 1370 [Steve Henson] 1371 1372 *) Fixes and wildcard matching support to hostname and email checking 1373 functions. Add manual page. 1374 [Florian Weimer (Red Hat Product Security Team)] 1375 1376 *) New functions to check a hostname email or IP address against a 1377 certificate. Add options x509 utility to print results of checks against 1378 a certificate. 1379 [Steve Henson] 1380 1381 *) Fix OCSP checking. 1382 [Rob Stradling <rob.stradling@comodo.com> and Ben Laurie] 1383 1384 *) Initial experimental support for explicitly trusted non-root CAs. 1385 OpenSSL still tries to build a complete chain to a root but if an 1386 intermediate CA has a trust setting included that is used. The first 1387 setting is used: whether to trust (e.g., -addtrust option to the x509 1388 utility) or reject. 1389 [Steve Henson] 1390 1391 *) Add -trusted_first option which attempts to find certificates in the 1392 trusted store even if an untrusted chain is also supplied. 1393 [Steve Henson] 1394 1395 *) MIPS assembly pack updates: support for MIPS32r2 and SmartMIPS ASE, 1396 platform support for Linux and Android. 1397 [Andy Polyakov] 1398 1399 *) Support for linux-x32, ILP32 environment in x86_64 framework. 1400 [Andy Polyakov] 1401 1402 *) Experimental multi-implementation support for FIPS capable OpenSSL. 1403 When in FIPS mode the approved implementations are used as normal, 1404 when not in FIPS mode the internal unapproved versions are used instead. 1405 This means that the FIPS capable OpenSSL isn't forced to use the 1406 (often lower perfomance) FIPS implementations outside FIPS mode. 1407 [Steve Henson] 1408 1409 *) Transparently support X9.42 DH parameters when calling 1410 PEM_read_bio_DHparameters. This means existing applications can handle 1411 the new parameter format automatically. 1412 [Steve Henson] 1413 1414 *) Initial experimental support for X9.42 DH parameter format: mainly 1415 to support use of 'q' parameter for RFC5114 parameters. 1416 [Steve Henson] 1417 1418 *) Add DH parameters from RFC5114 including test data to dhtest. 1419 [Steve Henson] 1420 1421 *) Support for automatic EC temporary key parameter selection. If enabled 1422 the most preferred EC parameters are automatically used instead of 1423 hardcoded fixed parameters. Now a server just has to call: 1424 SSL_CTX_set_ecdh_auto(ctx, 1) and the server will automatically 1425 support ECDH and use the most appropriate parameters. 1426 [Steve Henson] 1427 1428 *) Enhance and tidy EC curve and point format TLS extension code. Use 1429 static structures instead of allocation if default values are used. 1430 New ctrls to set curves we wish to support and to retrieve shared curves. 1431 Print out shared curves in s_server. New options to s_server and s_client 1432 to set list of supported curves. 1433 [Steve Henson] 1434 1435 *) New ctrls to retrieve supported signature algorithms and 1436 supported curve values as an array of NIDs. Extend openssl utility 1437 to print out received values. 1438 [Steve Henson] 1439 1440 *) Add new APIs EC_curve_nist2nid and EC_curve_nid2nist which convert 1441 between NIDs and the more common NIST names such as "P-256". Enhance 1442 ecparam utility and ECC method to recognise the NIST names for curves. 1443 [Steve Henson] 1444 1445 *) Enhance SSL/TLS certificate chain handling to support different 1446 chains for each certificate instead of one chain in the parent SSL_CTX. 1447 [Steve Henson] 1448 1449 *) Support for fixed DH ciphersuite client authentication: where both 1450 server and client use DH certificates with common parameters. 1451 [Steve Henson] 1452 1453 *) Support for fixed DH ciphersuites: those requiring DH server 1454 certificates. 1455 [Steve Henson] 1456 1457 *) New function i2d_re_X509_tbs for re-encoding the TBS portion of 1458 the certificate. 1459 Note: Related 1.0.2-beta specific macros X509_get_cert_info, 1460 X509_CINF_set_modified, X509_CINF_get_issuer, X509_CINF_get_extensions and 1461 X509_CINF_get_signature were reverted post internal team review. 1462 1463 Changes between 1.0.1k and 1.0.1l [15 Jan 2015] 1464 1465 *) Build fixes for the Windows and OpenVMS platforms 1466 [Matt Caswell and Richard Levitte] 1467 1468 Changes between 1.0.1j and 1.0.1k [8 Jan 2015] 1469 1470 *) Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS 1471 message can cause a segmentation fault in OpenSSL due to a NULL pointer 1472 dereference. This could lead to a Denial Of Service attack. Thanks to 1473 Markus Stenberg of Cisco Systems, Inc. for reporting this issue. 1474 (CVE-2014-3571) 1475 [Steve Henson] 1476 1477 *) Fix DTLS memory leak in dtls1_buffer_record. A memory leak can occur in the 1478 dtls1_buffer_record function under certain conditions. In particular this 1479 could occur if an attacker sent repeated DTLS records with the same 1480 sequence number but for the next epoch. The memory leak could be exploited 1481 by an attacker in a Denial of Service attack through memory exhaustion. 1482 Thanks to Chris Mueller for reporting this issue. 1483 (CVE-2015-0206) 1484 [Matt Caswell] 1485 1486 *) Fix issue where no-ssl3 configuration sets method to NULL. When openssl is 1487 built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl 1488 method would be set to NULL which could later result in a NULL pointer 1489 dereference. Thanks to Frank Schmirler for reporting this issue. 1490 (CVE-2014-3569) 1491 [Kurt Roeckx] 1492 1493 *) Abort handshake if server key exchange message is omitted for ephemeral 1494 ECDH ciphersuites. 1495 1496 Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for 1497 reporting this issue. 1498 (CVE-2014-3572) 1499 [Steve Henson] 1500 1501 *) Remove non-export ephemeral RSA code on client and server. This code 1502 violated the TLS standard by allowing the use of temporary RSA keys in 1503 non-export ciphersuites and could be used by a server to effectively 1504 downgrade the RSA key length used to a value smaller than the server 1505 certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at 1506 INRIA or reporting this issue. 1507 (CVE-2015-0204) 1508 [Steve Henson] 1509 1510 *) Fixed issue where DH client certificates are accepted without verification. 1511 An OpenSSL server will accept a DH certificate for client authentication 1512 without the certificate verify message. This effectively allows a client to 1513 authenticate without the use of a private key. This only affects servers 1514 which trust a client certificate authority which issues certificates 1515 containing DH keys: these are extremely rare and hardly ever encountered. 1516 Thanks for Karthikeyan Bhargavan of the PROSECCO team at INRIA or reporting 1517 this issue. 1518 (CVE-2015-0205) 1519 [Steve Henson] 1520 1521 *) Ensure that the session ID context of an SSL is updated when its 1522 SSL_CTX is updated via SSL_set_SSL_CTX. 1523 1524 The session ID context is typically set from the parent SSL_CTX, 1525 and can vary with the CTX. 1526 [Adam Langley] 1527 1528 *) Fix various certificate fingerprint issues. 1529 1530 By using non-DER or invalid encodings outside the signed portion of a 1531 certificate the fingerprint can be changed without breaking the signature. 1532 Although no details of the signed portion of the certificate can be changed 1533 this can cause problems with some applications: e.g. those using the 1534 certificate fingerprint for blacklists. 1535 1536 1. Reject signatures with non zero unused bits. 1537 1538 If the BIT STRING containing the signature has non zero unused bits reject 1539 the signature. All current signature algorithms require zero unused bits. 1540 1541 2. Check certificate algorithm consistency. 1542 1543 Check the AlgorithmIdentifier inside TBS matches the one in the 1544 certificate signature. NB: this will result in signature failure 1545 errors for some broken certificates. 1546 1547 Thanks to Konrad Kraszewski from Google for reporting this issue. 1548 1549 3. Check DSA/ECDSA signatures use DER. 1550 1551 Reencode DSA/ECDSA signatures and compare with the original received 1552 signature. Return an error if there is a mismatch. 1553 1554 This will reject various cases including garbage after signature 1555 (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS 1556 program for discovering this case) and use of BER or invalid ASN.1 INTEGERs 1557 (negative or with leading zeroes). 1558 1559 Further analysis was conducted and fixes were developed by Stephen Henson 1560 of the OpenSSL core team. 1561 1562 (CVE-2014-8275) 1563 [Steve Henson] 1564 1565 *) Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect 1566 results on some platforms, including x86_64. This bug occurs at random 1567 with a very low probability, and is not known to be exploitable in any 1568 way, though its exact impact is difficult to determine. Thanks to Pieter 1569 Wuille (Blockstream) who reported this issue and also suggested an initial 1570 fix. Further analysis was conducted by the OpenSSL development team and 1571 Adam Langley of Google. The final fix was developed by Andy Polyakov of 1572 the OpenSSL core team. 1573 (CVE-2014-3570) 1574 [Andy Polyakov] 1575 1576 *) Do not resume sessions on the server if the negotiated protocol 1577 version does not match the session's version. Resuming with a different 1578 version, while not strictly forbidden by the RFC, is of questionable 1579 sanity and breaks all known clients. 1580 [David Benjamin, Emilia K��sper] 1581 1582 *) Tighten handling of the ChangeCipherSpec (CCS) message: reject 1583 early CCS messages during renegotiation. (Note that because 1584 renegotiation is encrypted, this early CCS was not exploitable.) 1585 [Emilia K��sper] 1586 1587 *) Tighten client-side session ticket handling during renegotiation: 1588 ensure that the client only accepts a session ticket if the server sends 1589 the extension anew in the ServerHello. Previously, a TLS client would 1590 reuse the old extension state and thus accept a session ticket if one was 1591 announced in the initial ServerHello. 1592 1593 Similarly, ensure that the client requires a session ticket if one 1594 was advertised in the ServerHello. Previously, a TLS client would 1595 ignore a missing NewSessionTicket message. 1596 [Emilia K��sper] 1597 1598 Changes between 1.0.1i and 1.0.1j [15 Oct 2014] 1599 1600 *) SRTP Memory Leak. 1601 1602 A flaw in the DTLS SRTP extension parsing code allows an attacker, who 1603 sends a carefully crafted handshake message, to cause OpenSSL to fail 1604 to free up to 64k of memory causing a memory leak. This could be 1605 exploited in a Denial Of Service attack. This issue affects OpenSSL 1606 1.0.1 server implementations for both SSL/TLS and DTLS regardless of 1607 whether SRTP is used or configured. Implementations of OpenSSL that 1608 have been compiled with OPENSSL_NO_SRTP defined are not affected. 1609 1610 The fix was developed by the OpenSSL team. 1611 (CVE-2014-3513) 1612 [OpenSSL team] 1613 1614 *) Session Ticket Memory Leak. 1615 1616 When an OpenSSL SSL/TLS/DTLS server receives a session ticket the 1617 integrity of that ticket is first verified. In the event of a session 1618 ticket integrity check failing, OpenSSL will fail to free memory 1619 causing a memory leak. By sending a large number of invalid session 1620 tickets an attacker could exploit this issue in a Denial Of Service 1621 attack. 1622 (CVE-2014-3567) 1623 [Steve Henson] 1624 1625 *) Build option no-ssl3 is incomplete. 1626 1627 When OpenSSL is configured with "no-ssl3" as a build option, servers 1628 could accept and complete a SSL 3.0 handshake, and clients could be 1629 configured to send them. 1630 (CVE-2014-3568) 1631 [Akamai and the OpenSSL team] 1632 1633 *) Add support for TLS_FALLBACK_SCSV. 1634 Client applications doing fallback retries should call 1635 SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV). 1636 (CVE-2014-3566) 1637 [Adam Langley, Bodo Moeller] 1638 1639 *) Add additional DigestInfo checks. 1640 1641 Reencode DigestInto in DER and check against the original when 1642 verifying RSA signature: this will reject any improperly encoded 1643 DigestInfo structures. 1644 1645 Note: this is a precautionary measure and no attacks are currently known. 1646 1647 [Steve Henson] 1648 1649 Changes between 1.0.1h and 1.0.1i [6 Aug 2014] 1650 1651 *) Fix SRP buffer overrun vulnerability. Invalid parameters passed to the 1652 SRP code can be overrun an internal buffer. Add sanity check that 1653 g, A, B < N to SRP code. 1654 1655 Thanks to Sean Devlin and Watson Ladd of Cryptography Services, NCC 1656 Group for discovering this issue. 1657 (CVE-2014-3512) 1658 [Steve Henson] 1659 1660 *) A flaw in the OpenSSL SSL/TLS server code causes the server to negotiate 1661 TLS 1.0 instead of higher protocol versions when the ClientHello message 1662 is badly fragmented. This allows a man-in-the-middle attacker to force a 1663 downgrade to TLS 1.0 even if both the server and the client support a 1664 higher protocol version, by modifying the client's TLS records. 1665 1666 Thanks to David Benjamin and Adam Langley (Google) for discovering and 1667 researching this issue. 1668 (CVE-2014-3511) 1669 [David Benjamin] 1670 1671 *) OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject 1672 to a denial of service attack. A malicious server can crash the client 1673 with a null pointer dereference (read) by specifying an anonymous (EC)DH 1674 ciphersuite and sending carefully crafted handshake messages. 1675 1676 Thanks to Felix Gr��bert (Google) for discovering and researching this 1677 issue. 1678 (CVE-2014-3510) 1679 [Emilia K��sper] 1680 1681 *) By sending carefully crafted DTLS packets an attacker could cause openssl 1682 to leak memory. This can be exploited through a Denial of Service attack. 1683 Thanks to Adam Langley for discovering and researching this issue. 1684 (CVE-2014-3507) 1685 [Adam Langley] 1686 1687 *) An attacker can force openssl to consume large amounts of memory whilst 1688 processing DTLS handshake messages. This can be exploited through a 1689 Denial of Service attack. 1690 Thanks to Adam Langley for discovering and researching this issue. 1691 (CVE-2014-3506) 1692 [Adam Langley] 1693 1694 *) An attacker can force an error condition which causes openssl to crash 1695 whilst processing DTLS packets due to memory being freed twice. This 1696 can be exploited through a Denial of Service attack. 1697 Thanks to Adam Langley and Wan-Teh Chang for discovering and researching 1698 this issue. 1699 (CVE-2014-3505) 1700 [Adam Langley] 1701 1702 *) If a multithreaded client connects to a malicious server using a resumed 1703 session and the server sends an ec point format extension it could write 1704 up to 255 bytes to freed memory. 1705 1706 Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this 1707 issue. 1708 (CVE-2014-3509) 1709 [Gabor Tyukasz] 1710 1711 *) A malicious server can crash an OpenSSL client with a null pointer 1712 dereference (read) by specifying an SRP ciphersuite even though it was not 1713 properly negotiated with the client. This can be exploited through a 1714 Denial of Service attack. 1715 1716 Thanks to Joonas Kuorilehto and Riku Hietam��ki (Codenomicon) for 1717 discovering and researching this issue. 1718 (CVE-2014-5139) 1719 [Steve Henson] 1720 1721 *) A flaw in OBJ_obj2txt may cause pretty printing functions such as 1722 X509_name_oneline, X509_name_print_ex et al. to leak some information 1723 from the stack. Applications may be affected if they echo pretty printing 1724 output to the attacker. 1725 1726 Thanks to Ivan Fratric (Google) for discovering this issue. 1727 (CVE-2014-3508) 1728 [Emilia K��sper, and Steve Henson] 1729 1730 *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.) 1731 for corner cases. (Certain input points at infinity could lead to 1732 bogus results, with non-infinity inputs mapped to infinity too.) 1733 [Bodo Moeller] 1734 1735 Changes between 1.0.1g and 1.0.1h [5 Jun 2014] 1736 1737 *) Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted 1738 handshake can force the use of weak keying material in OpenSSL 1739 SSL/TLS clients and servers. 1740 1741 Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and 1742 researching this issue. (CVE-2014-0224) 1743 [KIKUCHI Masashi, Steve Henson] 1744 1745 *) Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an 1746 OpenSSL DTLS client the code can be made to recurse eventually crashing 1747 in a DoS attack. 1748 1749 Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue. 1750 (CVE-2014-0221) 1751 [Imre Rad, Steve Henson] 1752 1753 *) Fix DTLS invalid fragment vulnerability. A buffer overrun attack can 1754 be triggered by sending invalid DTLS fragments to an OpenSSL DTLS 1755 client or server. This is potentially exploitable to run arbitrary 1756 code on a vulnerable client or server. 1757 1758 Thanks to J��ri Aedla for reporting this issue. (CVE-2014-0195) 1759 [J��ri Aedla, Steve Henson] 1760 1761 *) Fix bug in TLS code where clients enable anonymous ECDH ciphersuites 1762 are subject to a denial of service attack. 1763 1764 Thanks to Felix Gr��bert and Ivan Fratric at Google for discovering 1765 this issue. (CVE-2014-3470) 1766 [Felix Gr��bert, Ivan Fratric, Steve Henson] 1767 1768 *) Harmonize version and its documentation. -f flag is used to display 1769 compilation flags. 1770 [mancha <mancha1@zoho.com>] 1771 1772 *) Fix eckey_priv_encode so it immediately returns an error upon a failure 1773 in i2d_ECPrivateKey. Thanks to Ted Unangst for feedback on this issue. 1774 [mancha <mancha1@zoho.com>] 1775 1776 *) Fix some double frees. These are not thought to be exploitable. 1777 [mancha <mancha1@zoho.com>] 1778 1779 Changes between 1.0.1f and 1.0.1g [7 Apr 2014] 1780 1781 *) A missing bounds check in the handling of the TLS heartbeat extension 1782 can be used to reveal up to 64k of memory to a connected client or 1783 server. 1784 1785 Thanks for Neel Mehta of Google Security for discovering this bug and to 1786 Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for 1787 preparing the fix (CVE-2014-0160) 1788 [Adam Langley, Bodo Moeller] 1789 1790 *) Fix for the attack described in the paper "Recovering OpenSSL 1791 ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack" 1792 by Yuval Yarom and Naomi Benger. Details can be obtained from: 1793 http://eprint.iacr.org/2014/140 1794 1795 Thanks to Yuval Yarom and Naomi Benger for discovering this 1796 flaw and to Yuval Yarom for supplying a fix (CVE-2014-0076) 1797 [Yuval Yarom and Naomi Benger] 1798 1799 *) TLS pad extension: draft-agl-tls-padding-03 1800 1801 Workaround for the "TLS hang bug" (see FAQ and PR#2771): if the 1802 TLS client Hello record length value would otherwise be > 255 and 1803 less that 512 pad with a dummy extension containing zeroes so it 1804 is at least 512 bytes long. 1805 1806 [Adam Langley, Steve Henson] 1807 1808 Changes between 1.0.1e and 1.0.1f [6 Jan 2014] 1809 1810 *) Fix for TLS record tampering bug. A carefully crafted invalid 1811 handshake could crash OpenSSL with a NULL pointer exception. 1812 Thanks to Anton Johansson for reporting this issues. 1813 (CVE-2013-4353) 1814 1815 *) Keep original DTLS digest and encryption contexts in retransmission 1816 structures so we can use the previous session parameters if they need 1817 to be resent. (CVE-2013-6450) 1818 [Steve Henson] 1819 1820 *) Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which 1821 avoids preferring ECDHE-ECDSA ciphers when the client appears to be 1822 Safari on OS X. Safari on OS X 10.8..10.8.3 advertises support for 1823 several ECDHE-ECDSA ciphers, but fails to negotiate them. The bug 1824 is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing 1825 10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer. 1826 [Rob Stradling, Adam Langley] 1827 1828 Changes between 1.0.1d and 1.0.1e [11 Feb 2013] 1829 1830 *) Correct fix for CVE-2013-0169. The original didn't work on AES-NI 1831 supporting platforms or when small records were transferred. 1832 [Andy Polyakov, Steve Henson] 1833 1834 Changes between 1.0.1c and 1.0.1d [5 Feb 2013] 1835 1836 *) Make the decoding of SSLv3, TLS and DTLS CBC records constant time. 1837 1838 This addresses the flaw in CBC record processing discovered by 1839 Nadhem Alfardan and Kenny Paterson. Details of this attack can be found 1840 at: http://www.isg.rhul.ac.uk/tls/ 1841 1842 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 1843 Security Group at Royal Holloway, University of London 1844 (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and 1845 Emilia K��sper for the initial patch. 1846 (CVE-2013-0169) 1847 [Emilia K��sper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson] 1848 1849 *) Fix flaw in AESNI handling of TLS 1.2 and 1.1 records for CBC mode 1850 ciphersuites which can be exploited in a denial of service attack. 1851 Thanks go to and to Adam Langley <agl@chromium.org> for discovering 1852 and detecting this bug and to Wolfgang Ettlinger 1853 <wolfgang.ettlinger@gmail.com> for independently discovering this issue. 1854 (CVE-2012-2686) 1855 [Adam Langley] 1856 1857 *) Return an error when checking OCSP signatures when key is NULL. 1858 This fixes a DoS attack. (CVE-2013-0166) 1859 [Steve Henson] 1860 1861 *) Make openssl verify return errors. 1862 [Chris Palmer <palmer@google.com> and Ben Laurie] 1863 1864 *) Call OCSP Stapling callback after ciphersuite has been chosen, so 1865 the right response is stapled. Also change SSL_get_certificate() 1866 so it returns the certificate actually sent. 1867 See http://rt.openssl.org/Ticket/Display.html?id=2836. 1868 [Rob Stradling <rob.stradling@comodo.com>] 1869 1870 *) Fix possible deadlock when decoding public keys. 1871 [Steve Henson] 1872 1873 *) Don't use TLS 1.0 record version number in initial client hello 1874 if renegotiating. 1875 [Steve Henson] 1876 1877 Changes between 1.0.1b and 1.0.1c [10 May 2012] 1878 1879 *) Sanity check record length before skipping explicit IV in TLS 1880 1.2, 1.1 and DTLS to fix DoS attack. 1881 1882 Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic 1883 fuzzing as a service testing platform. 1884 (CVE-2012-2333) 1885 [Steve Henson] 1886 1887 *) Initialise tkeylen properly when encrypting CMS messages. 1888 Thanks to Solar Designer of Openwall for reporting this issue. 1889 [Steve Henson] 1890 1891 *) In FIPS mode don't try to use composite ciphers as they are not 1892 approved. 1893 [Steve Henson] 1894 1895 Changes between 1.0.1a and 1.0.1b [26 Apr 2012] 1896 1897 *) OpenSSL 1.0.0 sets SSL_OP_ALL to 0x80000FFFL and OpenSSL 1.0.1 and 1898 1.0.1a set SSL_OP_NO_TLSv1_1 to 0x00000400L which would unfortunately 1899 mean any application compiled against OpenSSL 1.0.0 headers setting 1900 SSL_OP_ALL would also set SSL_OP_NO_TLSv1_1, unintentionally disablng 1901 TLS 1.1 also. Fix this by changing the value of SSL_OP_NO_TLSv1_1 to 1902 0x10000000L Any application which was previously compiled against 1903 OpenSSL 1.0.1 or 1.0.1a headers and which cares about SSL_OP_NO_TLSv1_1 1904 will need to be recompiled as a result. Letting be results in 1905 inability to disable specifically TLS 1.1 and in client context, 1906 in unlike event, limit maximum offered version to TLS 1.0 [see below]. 1907 [Steve Henson] 1908 1909 *) In order to ensure interoperabilty SSL_OP_NO_protocolX does not 1910 disable just protocol X, but all protocols above X *if* there are 1911 protocols *below* X still enabled. In more practical terms it means 1912 that if application wants to disable TLS1.0 in favor of TLS1.1 and 1913 above, it's not sufficient to pass SSL_OP_NO_TLSv1, one has to pass 1914 SSL_OP_NO_TLSv1|SSL_OP_NO_SSLv3|SSL_OP_NO_SSLv2. This applies to 1915 client side. 1916 [Andy Polyakov] 1917 1918 Changes between 1.0.1 and 1.0.1a [19 Apr 2012] 1919 1920 *) Check for potentially exploitable overflows in asn1_d2i_read_bio 1921 BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer 1922 in CRYPTO_realloc_clean. 1923 1924 Thanks to Tavis Ormandy, Google Security Team, for discovering this 1925 issue and to Adam Langley <agl@chromium.org> for fixing it. 1926 (CVE-2012-2110) 1927 [Adam Langley (Google), Tavis Ormandy, Google Security Team] 1928 1929 *) Don't allow TLS 1.2 SHA-256 ciphersuites in TLS 1.0, 1.1 connections. 1930 [Adam Langley] 1931 1932 *) Workarounds for some broken servers that "hang" if a client hello 1933 record length exceeds 255 bytes. 1934 1935 1. Do not use record version number > TLS 1.0 in initial client 1936 hello: some (but not all) hanging servers will now work. 1937 2. If we set OPENSSL_MAX_TLS1_2_CIPHER_LENGTH this will truncate 1938 the number of ciphers sent in the client hello. This should be 1939 set to an even number, such as 50, for example by passing: 1940 -DOPENSSL_MAX_TLS1_2_CIPHER_LENGTH=50 to config or Configure. 1941 Most broken servers should now work. 1942 3. If all else fails setting OPENSSL_NO_TLS1_2_CLIENT will disable 1943 TLS 1.2 client support entirely. 1944 [Steve Henson] 1945 1946 *) Fix SEGV in Vector Permutation AES module observed in OpenSSH. 1947 [Andy Polyakov] 1948 1949 Changes between 1.0.0h and 1.0.1 [14 Mar 2012] 1950 1951 *) Add compatibility with old MDC2 signatures which use an ASN1 OCTET 1952 STRING form instead of a DigestInfo. 1953 [Steve Henson] 1954 1955 *) The format used for MDC2 RSA signatures is inconsistent between EVP 1956 and the RSA_sign/RSA_verify functions. This was made more apparent when 1957 OpenSSL used RSA_sign/RSA_verify for some RSA signatures in particular 1958 those which went through EVP_PKEY_METHOD in 1.0.0 and later. Detect 1959 the correct format in RSA_verify so both forms transparently work. 1960 [Steve Henson] 1961 1962 *) Some servers which support TLS 1.0 can choke if we initially indicate 1963 support for TLS 1.2 and later renegotiate using TLS 1.0 in the RSA 1964 encrypted premaster secret. As a workaround use the maximum pemitted 1965 client version in client hello, this should keep such servers happy 1966 and still work with previous versions of OpenSSL. 1967 [Steve Henson] 1968 1969 *) Add support for TLS/DTLS heartbeats. 1970 [Robin Seggelmann <seggelmann@fh-muenster.de>] 1971 1972 *) Add support for SCTP. 1973 [Robin Seggelmann <seggelmann@fh-muenster.de>] 1974 1975 *) Improved PRNG seeding for VOS. 1976 [Paul Green <Paul.Green@stratus.com>] 1977 1978 *) Extensive assembler packs updates, most notably: 1979 1980 - x86[_64]: AES-NI, PCLMULQDQ, RDRAND support; 1981 - x86[_64]: SSSE3 support (SHA1, vector-permutation AES); 1982 - x86_64: bit-sliced AES implementation; 1983 - ARM: NEON support, contemporary platforms optimizations; 1984 - s390x: z196 support; 1985 - *: GHASH and GF(2^m) multiplication implementations; 1986 1987 [Andy Polyakov] 1988 1989 *) Make TLS-SRP code conformant with RFC 5054 API cleanup 1990 (removal of unnecessary code) 1991 [Peter Sylvester <peter.sylvester@edelweb.fr>] 1992 1993 *) Add TLS key material exporter from RFC 5705. 1994 [Eric Rescorla] 1995 1996 *) Add DTLS-SRTP negotiation from RFC 5764. 1997 [Eric Rescorla] 1998 1999 *) Add Next Protocol Negotiation, 2000 http://tools.ietf.org/html/draft-agl-tls-nextprotoneg-00. Can be 2001 disabled with a no-npn flag to config or Configure. Code donated 2002 by Google. 2003 [Adam Langley <agl@google.com> and Ben Laurie] 2004 2005 *) Add optional 64-bit optimized implementations of elliptic curves NIST-P224, 2006 NIST-P256, NIST-P521, with constant-time single point multiplication on 2007 typical inputs. Compiler support for the nonstandard type __uint128_t is 2008 required to use this (present in gcc 4.4 and later, for 64-bit builds). 2009 Code made available under Apache License version 2.0. 2010 2011 Specify "enable-ec_nistp_64_gcc_128" on the Configure (or config) command 2012 line to include this in your build of OpenSSL, and run "make depend" (or 2013 "make update"). This enables the following EC_METHODs: 2014 2015 EC_GFp_nistp224_method() 2016 EC_GFp_nistp256_method() 2017 EC_GFp_nistp521_method() 2018 2019 EC_GROUP_new_by_curve_name() will automatically use these (while 2020 EC_GROUP_new_curve_GFp() currently prefers the more flexible 2021 implementations). 2022 [Emilia K��sper, Adam Langley, Bodo Moeller (Google)] 2023 2024 *) Use type ossl_ssize_t instad of ssize_t which isn't available on 2025 all platforms. Move ssize_t definition from e_os.h to the public 2026 header file e_os2.h as it now appears in public header file cms.h 2027 [Steve Henson] 2028 2029 *) New -sigopt option to the ca, req and x509 utilities. Additional 2030 signature parameters can be passed using this option and in 2031 particular PSS. 2032 [Steve Henson] 2033 2034 *) Add RSA PSS signing function. This will generate and set the 2035 appropriate AlgorithmIdentifiers for PSS based on those in the 2036 corresponding EVP_MD_CTX structure. No application support yet. 2037 [Steve Henson] 2038 2039 *) Support for companion algorithm specific ASN1 signing routines. 2040 New function ASN1_item_sign_ctx() signs a pre-initialised 2041 EVP_MD_CTX structure and sets AlgorithmIdentifiers based on 2042 the appropriate parameters. 2043 [Steve Henson] 2044 2045 *) Add new algorithm specific ASN1 verification initialisation function 2046 to EVP_PKEY_ASN1_METHOD: this is not in EVP_PKEY_METHOD since the ASN1 2047 handling will be the same no matter what EVP_PKEY_METHOD is used. 2048 Add a PSS handler to support verification of PSS signatures: checked 2049 against a number of sample certificates. 2050 [Steve Henson] 2051 2052 *) Add signature printing for PSS. Add PSS OIDs. 2053 [Steve Henson, Martin Kaiser <lists@kaiser.cx>] 2054 2055 *) Add algorithm specific signature printing. An individual ASN1 method 2056 can now print out signatures instead of the standard hex dump. 2057 2058 More complex signatures (e.g. PSS) can print out more meaningful 2059 information. Include DSA version that prints out the signature 2060 parameters r, s. 2061 [Steve Henson] 2062 2063 *) Password based recipient info support for CMS library: implementing 2064 RFC3211. 2065 [Steve Henson] 2066 2067 *) Split password based encryption into PBES2 and PBKDF2 functions. This 2068 neatly separates the code into cipher and PBE sections and is required 2069 for some algorithms that split PBES2 into separate pieces (such as 2070 password based CMS). 2071 [Steve Henson] 2072 2073 *) Session-handling fixes: 2074 - Fix handling of connections that are resuming with a session ID, 2075 but also support Session Tickets. 2076 - Fix a bug that suppressed issuing of a new ticket if the client 2077 presented a ticket with an expired session. 2078 - Try to set the ticket lifetime hint to something reasonable. 2079 - Make tickets shorter by excluding irrelevant information. 2080 - On the client side, don't ignore renewed tickets. 2081 [Adam Langley, Bodo Moeller (Google)] 2082 2083 *) Fix PSK session representation. 2084 [Bodo Moeller] 2085 2086 *) Add RC4-MD5 and AESNI-SHA1 "stitched" implementations. 2087 2088 This work was sponsored by Intel. 2089 [Andy Polyakov] 2090 2091 *) Add GCM support to TLS library. Some custom code is needed to split 2092 the IV between the fixed (from PRF) and explicit (from TLS record) 2093 portions. This adds all GCM ciphersuites supported by RFC5288 and 2094 RFC5289. Generalise some AES* cipherstrings to inlclude GCM and 2095 add a special AESGCM string for GCM only. 2096 [Steve Henson] 2097 2098 *) Expand range of ctrls for AES GCM. Permit setting invocation 2099 field on decrypt and retrieval of invocation field only on encrypt. 2100 [Steve Henson] 2101 2102 *) Add HMAC ECC ciphersuites from RFC5289. Include SHA384 PRF support. 2103 As required by RFC5289 these ciphersuites cannot be used if for 2104 versions of TLS earlier than 1.2. 2105 [Steve Henson] 2106 2107 *) For FIPS capable OpenSSL interpret a NULL default public key method 2108 as unset and return the appopriate default but do *not* set the default. 2109 This means we can return the appopriate method in applications that 2110 swicth between FIPS and non-FIPS modes. 2111 [Steve Henson] 2112 2113 *) Redirect HMAC and CMAC operations to FIPS module in FIPS mode. If an 2114 ENGINE is used then we cannot handle that in the FIPS module so we 2115 keep original code iff non-FIPS operations are allowed. 2116 [Steve Henson] 2117 2118 *) Add -attime option to openssl utilities. 2119 [Peter Eckersley <pde@eff.org>, Ben Laurie and Steve Henson] 2120 2121 *) Redirect DSA and DH operations to FIPS module in FIPS mode. 2122 [Steve Henson] 2123 2124 *) Redirect ECDSA and ECDH operations to FIPS module in FIPS mode. Also use 2125 FIPS EC methods unconditionally for now. 2126 [Steve Henson] 2127 2128 *) New build option no-ec2m to disable characteristic 2 code. 2129 [Steve Henson] 2130 2131 *) Backport libcrypto audit of return value checking from 1.1.0-dev; not 2132 all cases can be covered as some introduce binary incompatibilities. 2133 [Steve Henson] 2134 2135 *) Redirect RSA operations to FIPS module including keygen, 2136 encrypt, decrypt, sign and verify. Block use of non FIPS RSA methods. 2137 [Steve Henson] 2138 2139 *) Add similar low level API blocking to ciphers. 2140 [Steve Henson] 2141 2142 *) Low level digest APIs are not approved in FIPS mode: any attempt 2143 to use these will cause a fatal error. Applications that *really* want 2144 to use them can use the private_* version instead. 2145 [Steve Henson] 2146 2147 *) Redirect cipher operations to FIPS module for FIPS builds. 2148 [Steve Henson] 2149 2150 *) Redirect digest operations to FIPS module for FIPS builds. 2151 [Steve Henson] 2152 2153 *) Update build system to add "fips" flag which will link in fipscanister.o 2154 for static and shared library builds embedding a signature if needed. 2155 [Steve Henson] 2156 2157 *) Output TLS supported curves in preference order instead of numerical 2158 order. This is currently hardcoded for the highest order curves first. 2159 This should be configurable so applications can judge speed vs strength. 2160 [Steve Henson] 2161 2162 *) Add TLS v1.2 server support for client authentication. 2163 [Steve Henson] 2164 2165 *) Add support for FIPS mode in ssl library: disable SSLv3, non-FIPS ciphers 2166 and enable MD5. 2167 [Steve Henson] 2168 2169 *) Functions FIPS_mode_set() and FIPS_mode() which call the underlying 2170 FIPS modules versions. 2171 [Steve Henson] 2172 2173 *) Add TLS v1.2 client side support for client authentication. Keep cache 2174 of handshake records longer as we don't know the hash algorithm to use 2175 until after the certificate request message is received. 2176 [Steve Henson] 2177 2178 *) Initial TLS v1.2 client support. Add a default signature algorithms 2179 extension including all the algorithms we support. Parse new signature 2180 format in client key exchange. Relax some ECC signing restrictions for 2181 TLS v1.2 as indicated in RFC5246. 2182 [Steve Henson] 2183 2184 *) Add server support for TLS v1.2 signature algorithms extension. Switch 2185 to new signature format when needed using client digest preference. 2186 All server ciphersuites should now work correctly in TLS v1.2. No client 2187 support yet and no support for client certificates. 2188 [Steve Henson] 2189 2190 *) Initial TLS v1.2 support. Add new SHA256 digest to ssl code, switch 2191 to SHA256 for PRF when using TLS v1.2 and later. Add new SHA256 based 2192 ciphersuites. At present only RSA key exchange ciphersuites work with 2193 TLS v1.2. Add new option for TLS v1.2 replacing the old and obsolete 2194 SSL_OP_PKCS1_CHECK flags with SSL_OP_NO_TLSv1_2. New TLSv1.2 methods 2195 and version checking. 2196 [Steve Henson] 2197 2198 *) New option OPENSSL_NO_SSL_INTERN. If an application can be compiled 2199 with this defined it will not be affected by any changes to ssl internal 2200 structures. Add several utility functions to allow openssl application 2201 to work with OPENSSL_NO_SSL_INTERN defined. 2202 [Steve Henson] 2203 2204 *) A long standing patch to add support for SRP from EdelWeb (Peter 2205 Sylvester and Christophe Renou) was integrated. 2206 [Christophe Renou <christophe.renou@edelweb.fr>, Peter Sylvester 2207 <peter.sylvester@edelweb.fr>, Tom Wu <tjw@cs.stanford.edu>, and 2208 Ben Laurie] 2209 2210 *) Add functions to copy EVP_PKEY_METHOD and retrieve flags and id. 2211 [Steve Henson] 2212 2213 *) Permit abbreviated handshakes when renegotiating using the function 2214 SSL_renegotiate_abbreviated(). 2215 [Robin Seggelmann <seggelmann@fh-muenster.de>] 2216 2217 *) Add call to ENGINE_register_all_complete() to 2218 ENGINE_load_builtin_engines(), so some implementations get used 2219 automatically instead of needing explicit application support. 2220 [Steve Henson] 2221 2222 *) Add support for TLS key exporter as described in RFC5705. 2223 [Robin Seggelmann <seggelmann@fh-muenster.de>, Steve Henson] 2224 2225 *) Initial TLSv1.1 support. Since TLSv1.1 is very similar to TLS v1.0 only 2226 a few changes are required: 2227 2228 Add SSL_OP_NO_TLSv1_1 flag. 2229 Add TLSv1_1 methods. 2230 Update version checking logic to handle version 1.1. 2231 Add explicit IV handling (ported from DTLS code). 2232 Add command line options to s_client/s_server. 2233 [Steve Henson] 2234 2235 Changes between 1.0.0g and 1.0.0h [12 Mar 2012] 2236 2237 *) Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness 2238 in CMS and PKCS7 code. When RSA decryption fails use a random key for 2239 content decryption and always return the same error. Note: this attack 2240 needs on average 2^20 messages so it only affects automated senders. The 2241 old behaviour can be reenabled in the CMS code by setting the 2242 CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where 2243 an MMA defence is not necessary. 2244 Thanks to Ivan Nestlerode <inestlerode@us.ibm.com> for discovering 2245 this issue. (CVE-2012-0884) 2246 [Steve Henson] 2247 2248 *) Fix CVE-2011-4619: make sure we really are receiving a 2249 client hello before rejecting multiple SGC restarts. Thanks to 2250 Ivan Nestlerode <inestlerode@us.ibm.com> for discovering this bug. 2251 [Steve Henson] 2252 2253 Changes between 1.0.0f and 1.0.0g [18 Jan 2012] 2254 2255 *) Fix for DTLS DoS issue introduced by fix for CVE-2011-4109. 2256 Thanks to Antonio Martin, Enterprise Secure Access Research and 2257 Development, Cisco Systems, Inc. for discovering this bug and 2258 preparing a fix. (CVE-2012-0050) 2259 [Antonio Martin] 2260 2261 Changes between 1.0.0e and 1.0.0f [4 Jan 2012] 2262 2263 *) Nadhem Alfardan and Kenny Paterson have discovered an extension 2264 of the Vaudenay padding oracle attack on CBC mode encryption 2265 which enables an efficient plaintext recovery attack against 2266 the OpenSSL implementation of DTLS. Their attack exploits timing 2267 differences arising during decryption processing. A research 2268 paper describing this attack can be found at: 2269 http://www.isg.rhul.ac.uk/~kp/dtls.pdf 2270 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 2271 Security Group at Royal Holloway, University of London 2272 (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann 2273 <seggelmann@fh-muenster.de> and Michael Tuexen <tuexen@fh-muenster.de> 2274 for preparing the fix. (CVE-2011-4108) 2275 [Robin Seggelmann, Michael Tuexen] 2276 2277 *) Clear bytes used for block padding of SSL 3.0 records. 2278 (CVE-2011-4576) 2279 [Adam Langley (Google)] 2280 2281 *) Only allow one SGC handshake restart for SSL/TLS. Thanks to George 2282 Kadianakis <desnacked@gmail.com> for discovering this issue and 2283 Adam Langley for preparing the fix. (CVE-2011-4619) 2284 [Adam Langley (Google)] 2285 2286 *) Check parameters are not NULL in GOST ENGINE. (CVE-2012-0027) 2287 [Andrey Kulikov <amdeich@gmail.com>] 2288 2289 *) Prevent malformed RFC3779 data triggering an assertion failure. 2290 Thanks to Andrew Chi, BBN Technologies, for discovering the flaw 2291 and Rob Austein <sra@hactrn.net> for fixing it. (CVE-2011-4577) 2292 [Rob Austein <sra@hactrn.net>] 2293 2294 *) Improved PRNG seeding for VOS. 2295 [Paul Green <Paul.Green@stratus.com>] 2296 2297 *) Fix ssl_ciph.c set-up race. 2298 [Adam Langley (Google)] 2299 2300 *) Fix spurious failures in ecdsatest.c. 2301 [Emilia K��sper (Google)] 2302 2303 *) Fix the BIO_f_buffer() implementation (which was mixing different 2304 interpretations of the '..._len' fields). 2305 [Adam Langley (Google)] 2306 2307 *) Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than 2308 BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent 2309 threads won't reuse the same blinding coefficients. 2310 2311 This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING 2312 lock to call BN_BLINDING_invert_ex, and avoids one use of 2313 BN_BLINDING_update for each BN_BLINDING structure (previously, 2314 the last update always remained unused). 2315 [Emilia K��sper (Google)] 2316 2317 *) In ssl3_clear, preserve s3->init_extra along with s3->rbuf. 2318 [Bob Buckholz (Google)] 2319 2320 Changes between 1.0.0d and 1.0.0e [6 Sep 2011] 2321 2322 *) Fix bug where CRLs with nextUpdate in the past are sometimes accepted 2323 by initialising X509_STORE_CTX properly. (CVE-2011-3207) 2324 [Kaspar Brand <ossl@velox.ch>] 2325 2326 *) Fix SSL memory handling for (EC)DH ciphersuites, in particular 2327 for multi-threaded use of ECDH. (CVE-2011-3210) 2328 [Adam Langley (Google)] 2329 2330 *) Fix x509_name_ex_d2i memory leak on bad inputs. 2331 [Bodo Moeller] 2332 2333 *) Remove hard coded ecdsaWithSHA1 signature tests in ssl code and check 2334 signature public key algorithm by using OID xref utilities instead. 2335 Before this you could only use some ECC ciphersuites with SHA1 only. 2336 [Steve Henson] 2337 2338 *) Add protection against ECDSA timing attacks as mentioned in the paper 2339 by Billy Bob Brumley and Nicola Tuveri, see: 2340 2341 http://eprint.iacr.org/2011/232.pdf 2342 2343 [Billy Bob Brumley and Nicola Tuveri] 2344 2345 Changes between 1.0.0c and 1.0.0d [8 Feb 2011] 2346 2347 *) Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014 2348 [Neel Mehta, Adam Langley, Bodo Moeller (Google)] 2349 2350 *) Fix bug in string printing code: if *any* escaping is enabled we must 2351 escape the escape character (backslash) or the resulting string is 2352 ambiguous. 2353 [Steve Henson] 2354 2355 Changes between 1.0.0b and 1.0.0c [2 Dec 2010] 2356 2357 *) Disable code workaround for ancient and obsolete Netscape browsers 2358 and servers: an attacker can use it in a ciphersuite downgrade attack. 2359 Thanks to Martin Rex for discovering this bug. CVE-2010-4180 2360 [Steve Henson] 2361 2362 *) Fixed J-PAKE implementation error, originally discovered by 2363 Sebastien Martini, further info and confirmation from Stefan 2364 Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252 2365 [Ben Laurie] 2366 2367 Changes between 1.0.0a and 1.0.0b [16 Nov 2010] 2368 2369 *) Fix extension code to avoid race conditions which can result in a buffer 2370 overrun vulnerability: resumed sessions must not be modified as they can 2371 be shared by multiple threads. CVE-2010-3864 2372 [Steve Henson] 2373 2374 *) Fix WIN32 build system to correctly link an ENGINE directory into 2375 a DLL. 2376 [Steve Henson] 2377 2378 Changes between 1.0.0 and 1.0.0a [01 Jun 2010] 2379 2380 *) Check return value of int_rsa_verify in pkey_rsa_verifyrecover 2381 (CVE-2010-1633) 2382 [Steve Henson, Peter-Michael Hager <hager@dortmund.net>] 2383 2384 Changes between 0.9.8n and 1.0.0 [29 Mar 2010] 2385 2386 *) Add "missing" function EVP_CIPHER_CTX_copy(). This copies a cipher 2387 context. The operation can be customised via the ctrl mechanism in 2388 case ENGINEs want to include additional functionality. 2389 [Steve Henson] 2390 2391 *) Tolerate yet another broken PKCS#8 key format: private key value negative. 2392 [Steve Henson] 2393 2394 *) Add new -subject_hash_old and -issuer_hash_old options to x509 utility to 2395 output hashes compatible with older versions of OpenSSL. 2396 [Willy Weisz <weisz@vcpc.univie.ac.at>] 2397 2398 *) Fix compression algorithm handling: if resuming a session use the 2399 compression algorithm of the resumed session instead of determining 2400 it from client hello again. Don't allow server to change algorithm. 2401 [Steve Henson] 2402 2403 *) Add load_crls() function to apps tidying load_certs() too. Add option 2404 to verify utility to allow additional CRLs to be included. 2405 [Steve Henson] 2406 2407 *) Update OCSP request code to permit adding custom headers to the request: 2408 some responders need this. 2409 [Steve Henson] 2410 2411 *) The function EVP_PKEY_sign() returns <=0 on error: check return code 2412 correctly. 2413 [Julia Lawall <julia@diku.dk>] 2414 2415 *) Update verify callback code in apps/s_cb.c and apps/verify.c, it 2416 needlessly dereferenced structures, used obsolete functions and 2417 didn't handle all updated verify codes correctly. 2418 [Steve Henson] 2419 2420 *) Disable MD2 in the default configuration. 2421 [Steve Henson] 2422 2423 *) In BIO_pop() and BIO_push() use the ctrl argument (which was NULL) to 2424 indicate the initial BIO being pushed or popped. This makes it possible 2425 to determine whether the BIO is the one explicitly called or as a result 2426 of the ctrl being passed down the chain. Fix BIO_pop() and SSL BIOs so 2427 it handles reference counts correctly and doesn't zero out the I/O bio 2428 when it is not being explicitly popped. WARNING: applications which 2429 included workarounds for the old buggy behaviour will need to be modified 2430 or they could free up already freed BIOs. 2431 [Steve Henson] 2432 2433 *) Extend the uni2asc/asc2uni => OPENSSL_uni2asc/OPENSSL_asc2uni 2434 renaming to all platforms (within the 0.9.8 branch, this was 2435 done conditionally on Netware platforms to avoid a name clash). 2436 [Guenter <lists@gknw.net>] 2437 2438 *) Add ECDHE and PSK support to DTLS. 2439 [Michael Tuexen <tuexen@fh-muenster.de>] 2440 2441 *) Add CHECKED_STACK_OF macro to safestack.h, otherwise safestack can't 2442 be used on C++. 2443 [Steve Henson] 2444 2445 *) Add "missing" function EVP_MD_flags() (without this the only way to 2446 retrieve a digest flags is by accessing the structure directly. Update 2447 EVP_MD_do_all*() and EVP_CIPHER_do_all*() to include the name a digest 2448 or cipher is registered as in the "from" argument. Print out all 2449 registered digests in the dgst usage message instead of manually 2450 attempting to work them out. 2451 [Steve Henson] 2452 2453 *) If no SSLv2 ciphers are used don't use an SSLv2 compatible client hello: 2454 this allows the use of compression and extensions. Change default cipher 2455 string to remove SSLv2 ciphersuites. This effectively avoids ancient SSLv2 2456 by default unless an application cipher string requests it. 2457 [Steve Henson] 2458 2459 *) Alter match criteria in PKCS12_parse(). It used to try to use local 2460 key ids to find matching certificates and keys but some PKCS#12 files 2461 don't follow the (somewhat unwritten) rules and this strategy fails. 2462 Now just gather all certificates together and the first private key 2463 then look for the first certificate that matches the key. 2464 [Steve Henson] 2465 2466 *) Support use of registered digest and cipher names for dgst and cipher 2467 commands instead of having to add each one as a special case. So now 2468 you can do: 2469 2470 openssl sha256 foo 2471 2472 as well as: 2473 2474 openssl dgst -sha256 foo 2475 2476 and this works for ENGINE based algorithms too. 2477 2478 [Steve Henson] 2479 2480 *) Update Gost ENGINE to support parameter files. 2481 [Victor B. Wagner <vitus@cryptocom.ru>] 2482 2483 *) Support GeneralizedTime in ca utility. 2484 [Oliver Martin <oliver@volatilevoid.net>, Steve Henson] 2485 2486 *) Enhance the hash format used for certificate directory links. The new 2487 form uses the canonical encoding (meaning equivalent names will work 2488 even if they aren't identical) and uses SHA1 instead of MD5. This form 2489 is incompatible with the older format and as a result c_rehash should 2490 be used to rebuild symbolic links. 2491 [Steve Henson] 2492 2493 *) Make PKCS#8 the default write format for private keys, replacing the 2494 traditional format. This form is standardised, more secure and doesn't 2495 include an implicit MD5 dependency. 2496 [Steve Henson] 2497 2498 *) Add a $gcc_devteam_warn option to Configure. The idea is that any code 2499 committed to OpenSSL should pass this lot as a minimum. 2500 [Steve Henson] 2501 2502 *) Add session ticket override functionality for use by EAP-FAST. 2503 [Jouni Malinen <j@w1.fi>] 2504 2505 *) Modify HMAC functions to return a value. Since these can be implemented 2506 in an ENGINE errors can occur. 2507 [Steve Henson] 2508 2509 *) Type-checked OBJ_bsearch_ex. 2510 [Ben Laurie] 2511 2512 *) Type-checked OBJ_bsearch. Also some constification necessitated 2513 by type-checking. Still to come: TXT_DB, bsearch(?), 2514 OBJ_bsearch_ex, qsort, CRYPTO_EX_DATA, ASN1_VALUE, ASN1_STRING, 2515 CONF_VALUE. 2516 [Ben Laurie] 2517 2518 *) New function OPENSSL_gmtime_adj() to add a specific number of days and 2519 seconds to a tm structure directly, instead of going through OS 2520 specific date routines. This avoids any issues with OS routines such 2521 as the year 2038 bug. New *_adj() functions for ASN1 time structures 2522 and X509_time_adj_ex() to cover the extended range. The existing 2523 X509_time_adj() is still usable and will no longer have any date issues. 2524 [Steve Henson] 2525 2526 *) Delta CRL support. New use deltas option which will attempt to locate 2527 and search any appropriate delta CRLs available. 2528 2529 This work was sponsored by Google. 2530 [Steve Henson] 2531 2532 *) Support for CRLs partitioned by reason code. Reorganise CRL processing 2533 code and add additional score elements. Validate alternate CRL paths 2534 as part of the CRL checking and indicate a new error "CRL path validation 2535 error" in this case. Applications wanting additional details can use 2536 the verify callback and check the new "parent" field. If this is not 2537 NULL CRL path validation is taking place. Existing applications wont 2538 see this because it requires extended CRL support which is off by 2539 default. 2540 2541 This work was sponsored by Google. 2542 [Steve Henson] 2543 2544 *) Support for freshest CRL extension. 2545 2546 This work was sponsored by Google. 2547 [Steve Henson] 2548 2549 *) Initial indirect CRL support. Currently only supported in the CRLs 2550 passed directly and not via lookup. Process certificate issuer 2551 CRL entry extension and lookup CRL entries by bother issuer name 2552 and serial number. Check and process CRL issuer entry in IDP extension. 2553 2554 This work was sponsored by Google. 2555 [Steve Henson] 2556 2557 *) Add support for distinct certificate and CRL paths. The CRL issuer 2558 certificate is validated separately in this case. Only enabled if 2559 an extended CRL support flag is set: this flag will enable additional 2560 CRL functionality in future. 2561 2562 This work was sponsored by Google. 2563 [Steve Henson] 2564 2565 *) Add support for policy mappings extension. 2566 2567 This work was sponsored by Google. 2568 [Steve Henson] 2569 2570 *) Fixes to pathlength constraint, self issued certificate handling, 2571 policy processing to align with RFC3280 and PKITS tests. 2572 2573 This work was sponsored by Google. 2574 [Steve Henson] 2575 2576 *) Support for name constraints certificate extension. DN, email, DNS 2577 and URI types are currently supported. 2578 2579 This work was sponsored by Google. 2580 [Steve Henson] 2581 2582 *) To cater for systems that provide a pointer-based thread ID rather 2583 than numeric, deprecate the current numeric thread ID mechanism and 2584 replace it with a structure and associated callback type. This 2585 mechanism allows a numeric "hash" to be extracted from a thread ID in 2586 either case, and on platforms where pointers are larger than 'long', 2587 mixing is done to help ensure the numeric 'hash' is usable even if it 2588 can't be guaranteed unique. The default mechanism is to use "&errno" 2589 as a pointer-based thread ID to distinguish between threads. 2590 2591 Applications that want to provide their own thread IDs should now use 2592 CRYPTO_THREADID_set_callback() to register a callback that will call 2593 either CRYPTO_THREADID_set_numeric() or CRYPTO_THREADID_set_pointer(). 2594 2595 Note that ERR_remove_state() is now deprecated, because it is tied 2596 to the assumption that thread IDs are numeric. ERR_remove_state(0) 2597 to free the current thread's error state should be replaced by 2598 ERR_remove_thread_state(NULL). 2599 2600 (This new approach replaces the functions CRYPTO_set_idptr_callback(), 2601 CRYPTO_get_idptr_callback(), and CRYPTO_thread_idptr() that existed in 2602 OpenSSL 0.9.9-dev between June 2006 and August 2008. Also, if an 2603 application was previously providing a numeric thread callback that 2604 was inappropriate for distinguishing threads, then uniqueness might 2605 have been obtained with &errno that happened immediately in the 2606 intermediate development versions of OpenSSL; this is no longer the 2607 case, the numeric thread callback will now override the automatic use 2608 of &errno.) 2609 [Geoff Thorpe, with help from Bodo Moeller] 2610 2611 *) Initial support for different CRL issuing certificates. This covers a 2612 simple case where the self issued certificates in the chain exist and 2613 the real CRL issuer is higher in the existing chain. 2614 2615 This work was sponsored by Google. 2616 [Steve Henson] 2617 2618 *) Removed effectively defunct crypto/store from the build. 2619 [Ben Laurie] 2620 2621 *) Revamp of STACK to provide stronger type-checking. Still to come: 2622 TXT_DB, bsearch(?), OBJ_bsearch, qsort, CRYPTO_EX_DATA, ASN1_VALUE, 2623 ASN1_STRING, CONF_VALUE. 2624 [Ben Laurie] 2625 2626 *) Add a new SSL_MODE_RELEASE_BUFFERS mode flag to release unused buffer 2627 RAM on SSL connections. This option can save about 34k per idle SSL. 2628 [Nick Mathewson] 2629 2630 *) Revamp of LHASH to provide stronger type-checking. Still to come: 2631 STACK, TXT_DB, bsearch, qsort. 2632 [Ben Laurie] 2633 2634 *) Initial support for Cryptographic Message Syntax (aka CMS) based 2635 on RFC3850, RFC3851 and RFC3852. New cms directory and cms utility, 2636 support for data, signedData, compressedData, digestedData and 2637 encryptedData, envelopedData types included. Scripts to check against 2638 RFC4134 examples draft and interop and consistency checks of many 2639 content types and variants. 2640 [Steve Henson] 2641 2642 *) Add options to enc utility to support use of zlib compression BIO. 2643 [Steve Henson] 2644 2645 *) Extend mk1mf to support importing of options and assembly language 2646 files from Configure script, currently only included in VC-WIN32. 2647 The assembly language rules can now optionally generate the source 2648 files from the associated perl scripts. 2649 [Steve Henson] 2650 2651 *) Implement remaining functionality needed to support GOST ciphersuites. 2652 Interop testing has been performed using CryptoPro implementations. 2653 [Victor B. Wagner <vitus@cryptocom.ru>] 2654 2655 *) s390x assembler pack. 2656 [Andy Polyakov] 2657 2658 *) ARMv4 assembler pack. ARMv4 refers to v4 and later ISA, not CPU 2659 "family." 2660 [Andy Polyakov] 2661 2662 *) Implement Opaque PRF Input TLS extension as specified in 2663 draft-rescorla-tls-opaque-prf-input-00.txt. Since this is not an 2664 official specification yet and no extension type assignment by 2665 IANA exists, this extension (for now) will have to be explicitly 2666 enabled when building OpenSSL by providing the extension number 2667 to use. For example, specify an option 2668 2669 -DTLSEXT_TYPE_opaque_prf_input=0x9527 2670 2671 to the "config" or "Configure" script to enable the extension, 2672 assuming extension number 0x9527 (which is a completely arbitrary 2673 and unofficial assignment based on the MD5 hash of the Internet 2674 Draft). Note that by doing so, you potentially lose 2675 interoperability with other TLS implementations since these might 2676 be using the same extension number for other purposes. 2677 2678 SSL_set_tlsext_opaque_prf_input(ssl, src, len) is used to set the 2679 opaque PRF input value to use in the handshake. This will create 2680 an interal copy of the length-'len' string at 'src', and will 2681 return non-zero for success. 2682 2683 To get more control and flexibility, provide a callback function 2684 by using 2685 2686 SSL_CTX_set_tlsext_opaque_prf_input_callback(ctx, cb) 2687 SSL_CTX_set_tlsext_opaque_prf_input_callback_arg(ctx, arg) 2688 2689 where 2690 2691 int (*cb)(SSL *, void *peerinput, size_t len, void *arg); 2692 void *arg; 2693 2694 Callback function 'cb' will be called in handshakes, and is 2695 expected to use SSL_set_tlsext_opaque_prf_input() as appropriate. 2696 Argument 'arg' is for application purposes (the value as given to 2697 SSL_CTX_set_tlsext_opaque_prf_input_callback_arg() will directly 2698 be provided to the callback function). The callback function 2699 has to return non-zero to report success: usually 1 to use opaque 2700 PRF input just if possible, or 2 to enforce use of the opaque PRF 2701 input. In the latter case, the library will abort the handshake 2702 if opaque PRF input is not successfully negotiated. 2703 2704 Arguments 'peerinput' and 'len' given to the callback function 2705 will always be NULL and 0 in the case of a client. A server will 2706 see the client's opaque PRF input through these variables if 2707 available (NULL and 0 otherwise). Note that if the server 2708 provides an opaque PRF input, the length must be the same as the 2709 length of the client's opaque PRF input. 2710 2711 Note that the callback function will only be called when creating 2712 a new session (session resumption can resume whatever was 2713 previously negotiated), and will not be called in SSL 2.0 2714 handshakes; thus, SSL_CTX_set_options(ctx, SSL_OP_NO_SSLv2) or 2715 SSL_set_options(ssl, SSL_OP_NO_SSLv2) is especially recommended 2716 for applications that need to enforce opaque PRF input. 2717 2718 [Bodo Moeller] 2719 2720 *) Update ssl code to support digests other than SHA1+MD5 for handshake 2721 MAC. 2722 2723 [Victor B. Wagner <vitus@cryptocom.ru>] 2724 2725 *) Add RFC4507 support to OpenSSL. This includes the corrections in 2726 RFC4507bis. The encrypted ticket format is an encrypted encoded 2727 SSL_SESSION structure, that way new session features are automatically 2728 supported. 2729 2730 If a client application caches session in an SSL_SESSION structure 2731 support is transparent because tickets are now stored in the encoded 2732 SSL_SESSION. 2733 2734 The SSL_CTX structure automatically generates keys for ticket 2735 protection in servers so again support should be possible 2736 with no application modification. 2737 2738 If a client or server wishes to disable RFC4507 support then the option 2739 SSL_OP_NO_TICKET can be set. 2740 2741 Add a TLS extension debugging callback to allow the contents of any client 2742 or server extensions to be examined. 2743 2744 This work was sponsored by Google. 2745 [Steve Henson] 2746 2747 *) Final changes to avoid use of pointer pointer casts in OpenSSL. 2748 OpenSSL should now compile cleanly on gcc 4.2 2749 [Peter Hartley <pdh@utter.chaos.org.uk>, Steve Henson] 2750 2751 *) Update SSL library to use new EVP_PKEY MAC API. Include generic MAC 2752 support including streaming MAC support: this is required for GOST 2753 ciphersuite support. 2754 [Victor B. Wagner <vitus@cryptocom.ru>, Steve Henson] 2755 2756 *) Add option -stream to use PKCS#7 streaming in smime utility. New 2757 function i2d_PKCS7_bio_stream() and PEM_write_PKCS7_bio_stream() 2758 to output in BER and PEM format. 2759 [Steve Henson] 2760 2761 *) Experimental support for use of HMAC via EVP_PKEY interface. This 2762 allows HMAC to be handled via the EVP_DigestSign*() interface. The 2763 EVP_PKEY "key" in this case is the HMAC key, potentially allowing 2764 ENGINE support for HMAC keys which are unextractable. New -mac and 2765 -macopt options to dgst utility. 2766 [Steve Henson] 2767 2768 *) New option -sigopt to dgst utility. Update dgst to use 2769 EVP_Digest{Sign,Verify}*. These two changes make it possible to use 2770 alternative signing paramaters such as X9.31 or PSS in the dgst 2771 utility. 2772 [Steve Henson] 2773 2774 *) Change ssl_cipher_apply_rule(), the internal function that does 2775 the work each time a ciphersuite string requests enabling 2776 ("foo+bar"), moving ("+foo+bar"), disabling ("-foo+bar", or 2777 removing ("!foo+bar") a class of ciphersuites: Now it maintains 2778 the order of disabled ciphersuites such that those ciphersuites 2779 that most recently went from enabled to disabled not only stay 2780 in order with respect to each other, but also have higher priority 2781 than other disabled ciphersuites the next time ciphersuites are 2782 enabled again. 2783 2784 This means that you can now say, e.g., "PSK:-PSK:HIGH" to enable 2785 the same ciphersuites as with "HIGH" alone, but in a specific 2786 order where the PSK ciphersuites come first (since they are the 2787 most recently disabled ciphersuites when "HIGH" is parsed). 2788 2789 Also, change ssl_create_cipher_list() (using this new 2790 funcionality) such that between otherwise identical 2791 cihpersuites, ephemeral ECDH is preferred over ephemeral DH in 2792 the default order. 2793 [Bodo Moeller] 2794 2795 *) Change ssl_create_cipher_list() so that it automatically 2796 arranges the ciphersuites in reasonable order before starting 2797 to process the rule string. Thus, the definition for "DEFAULT" 2798 (SSL_DEFAULT_CIPHER_LIST) now is just "ALL:!aNULL:!eNULL", but 2799 remains equivalent to "AES:ALL:!aNULL:!eNULL:+aECDH:+kRSA:+RC4:@STRENGTH". 2800 This makes it much easier to arrive at a reasonable default order 2801 in applications for which anonymous ciphers are OK (meaning 2802 that you can't actually use DEFAULT). 2803 [Bodo Moeller; suggested by Victor Duchovni] 2804 2805 *) Split the SSL/TLS algorithm mask (as used for ciphersuite string 2806 processing) into multiple integers instead of setting 2807 "SSL_MKEY_MASK" bits, "SSL_AUTH_MASK" bits, "SSL_ENC_MASK", 2808 "SSL_MAC_MASK", and "SSL_SSL_MASK" bits all in a single integer. 2809 (These masks as well as the individual bit definitions are hidden 2810 away into the non-exported interface ssl/ssl_locl.h, so this 2811 change to the definition of the SSL_CIPHER structure shouldn't 2812 affect applications.) This give us more bits for each of these 2813 categories, so there is no longer a need to coagulate AES128 and 2814 AES256 into a single algorithm bit, and to coagulate Camellia128 2815 and Camellia256 into a single algorithm bit, which has led to all 2816 kinds of kludges. 2817 2818 Thus, among other things, the kludge introduced in 0.9.7m and 2819 0.9.8e for masking out AES256 independently of AES128 or masking 2820 out Camellia256 independently of AES256 is not needed here in 0.9.9. 2821 2822 With the change, we also introduce new ciphersuite aliases that 2823 so far were missing: "AES128", "AES256", "CAMELLIA128", and 2824 "CAMELLIA256". 2825 [Bodo Moeller] 2826 2827 *) Add support for dsa-with-SHA224 and dsa-with-SHA256. 2828 Use the leftmost N bytes of the signature input if the input is 2829 larger than the prime q (with N being the size in bytes of q). 2830 [Nils Larsch] 2831 2832 *) Very *very* experimental PKCS#7 streaming encoder support. Nothing uses 2833 it yet and it is largely untested. 2834 [Steve Henson] 2835 2836 *) Add support for the ecdsa-with-SHA224/256/384/512 signature types. 2837 [Nils Larsch] 2838 2839 *) Initial incomplete changes to avoid need for function casts in OpenSSL 2840 some compilers (gcc 4.2 and later) reject their use. Safestack is 2841 reimplemented. Update ASN1 to avoid use of legacy functions. 2842 [Steve Henson] 2843 2844 *) Win32/64 targets are linked with Winsock2. 2845 [Andy Polyakov] 2846 2847 *) Add an X509_CRL_METHOD structure to allow CRL processing to be redirected 2848 to external functions. This can be used to increase CRL handling 2849 efficiency especially when CRLs are very large by (for example) storing 2850 the CRL revoked certificates in a database. 2851 [Steve Henson] 2852 2853 *) Overhaul of by_dir code. Add support for dynamic loading of CRLs so 2854 new CRLs added to a directory can be used. New command line option 2855 -verify_return_error to s_client and s_server. This causes real errors 2856 to be returned by the verify callback instead of carrying on no matter 2857 what. This reflects the way a "real world" verify callback would behave. 2858 [Steve Henson] 2859 2860 *) GOST engine, supporting several GOST algorithms and public key formats. 2861 Kindly donated by Cryptocom. 2862 [Cryptocom] 2863 2864 *) Partial support for Issuing Distribution Point CRL extension. CRLs 2865 partitioned by DP are handled but no indirect CRL or reason partitioning 2866 (yet). Complete overhaul of CRL handling: now the most suitable CRL is 2867 selected via a scoring technique which handles IDP and AKID in CRLs. 2868 [Steve Henson] 2869 2870 *) New X509_STORE_CTX callbacks lookup_crls() and lookup_certs() which 2871 will ultimately be used for all verify operations: this will remove the 2872 X509_STORE dependency on certificate verification and allow alternative 2873 lookup methods. X509_STORE based implementations of these two callbacks. 2874 [Steve Henson] 2875 2876 *) Allow multiple CRLs to exist in an X509_STORE with matching issuer names. 2877 Modify get_crl() to find a valid (unexpired) CRL if possible. 2878 [Steve Henson] 2879 2880 *) New function X509_CRL_match() to check if two CRLs are identical. Normally 2881 this would be called X509_CRL_cmp() but that name is already used by 2882 a function that just compares CRL issuer names. Cache several CRL 2883 extensions in X509_CRL structure and cache CRLDP in X509. 2884 [Steve Henson] 2885 2886 *) Store a "canonical" representation of X509_NAME structure (ASN1 Name) 2887 this maps equivalent X509_NAME structures into a consistent structure. 2888 Name comparison can then be performed rapidly using memcmp(). 2889 [Steve Henson] 2890 2891 *) Non-blocking OCSP request processing. Add -timeout option to ocsp 2892 utility. 2893 [Steve Henson] 2894 2895 *) Allow digests to supply their own micalg string for S/MIME type using 2896 the ctrl EVP_MD_CTRL_MICALG. 2897 [Steve Henson] 2898 2899 *) During PKCS7 signing pass the PKCS7 SignerInfo structure to the 2900 EVP_PKEY_METHOD before and after signing via the EVP_PKEY_CTRL_PKCS7_SIGN 2901 ctrl. It can then customise the structure before and/or after signing 2902 if necessary. 2903 [Steve Henson] 2904 2905 *) New function OBJ_add_sigid() to allow application defined signature OIDs 2906 to be added to OpenSSLs internal tables. New function OBJ_sigid_free() 2907 to free up any added signature OIDs. 2908 [Steve Henson] 2909 2910 *) New functions EVP_CIPHER_do_all(), EVP_CIPHER_do_all_sorted(), 2911 EVP_MD_do_all() and EVP_MD_do_all_sorted() to enumerate internal 2912 digest and cipher tables. New options added to openssl utility: 2913 list-message-digest-algorithms and list-cipher-algorithms. 2914 [Steve Henson] 2915 2916 *) Change the array representation of binary polynomials: the list 2917 of degrees of non-zero coefficients is now terminated with -1. 2918 Previously it was terminated with 0, which was also part of the 2919 value; thus, the array representation was not applicable to 2920 polynomials where t^0 has coefficient zero. This change makes 2921 the array representation useful in a more general context. 2922 [Douglas Stebila] 2923 2924 *) Various modifications and fixes to SSL/TLS cipher string 2925 handling. For ECC, the code now distinguishes between fixed ECDH 2926 with RSA certificates on the one hand and with ECDSA certificates 2927 on the other hand, since these are separate ciphersuites. The 2928 unused code for Fortezza ciphersuites has been removed. 2929 2930 For consistency with EDH, ephemeral ECDH is now called "EECDH" 2931 (not "ECDHE"). For consistency with the code for DH 2932 certificates, use of ECDH certificates is now considered ECDH 2933 authentication, not RSA or ECDSA authentication (the latter is 2934 merely the CA's signing algorithm and not actively used in the 2935 protocol). 2936 2937 The temporary ciphersuite alias "ECCdraft" is no longer 2938 available, and ECC ciphersuites are no longer excluded from "ALL" 2939 and "DEFAULT". The following aliases now exist for RFC 4492 2940 ciphersuites, most of these by analogy with the DH case: 2941 2942 kECDHr - ECDH cert, signed with RSA 2943 kECDHe - ECDH cert, signed with ECDSA 2944 kECDH - ECDH cert (signed with either RSA or ECDSA) 2945 kEECDH - ephemeral ECDH 2946 ECDH - ECDH cert or ephemeral ECDH 2947 2948 aECDH - ECDH cert 2949 aECDSA - ECDSA cert 2950 ECDSA - ECDSA cert 2951 2952 AECDH - anonymous ECDH 2953 EECDH - non-anonymous ephemeral ECDH (equivalent to "kEECDH:-AECDH") 2954 2955 [Bodo Moeller] 2956 2957 *) Add additional S/MIME capabilities for AES and GOST ciphers if supported. 2958 Use correct micalg parameters depending on digest(s) in signed message. 2959 [Steve Henson] 2960 2961 *) Add engine support for EVP_PKEY_ASN1_METHOD. Add functions to process 2962 an ENGINE asn1 method. Support ENGINE lookups in the ASN1 code. 2963 [Steve Henson] 2964 2965 *) Initial engine support for EVP_PKEY_METHOD. New functions to permit 2966 an engine to register a method. Add ENGINE lookups for methods and 2967 functional reference processing. 2968 [Steve Henson] 2969 2970 *) New functions EVP_Digest{Sign,Verify)*. These are enchance versions of 2971 EVP_{Sign,Verify}* which allow an application to customise the signature 2972 process. 2973 [Steve Henson] 2974 2975 *) New -resign option to smime utility. This adds one or more signers 2976 to an existing PKCS#7 signedData structure. Also -md option to use an 2977 alternative message digest algorithm for signing. 2978 [Steve Henson] 2979 2980 *) Tidy up PKCS#7 routines and add new functions to make it easier to 2981 create PKCS7 structures containing multiple signers. Update smime 2982 application to support multiple signers. 2983 [Steve Henson] 2984 2985 *) New -macalg option to pkcs12 utility to allow setting of an alternative 2986 digest MAC. 2987 [Steve Henson] 2988 2989 *) Initial support for PKCS#5 v2.0 PRFs other than default SHA1 HMAC. 2990 Reorganize PBE internals to lookup from a static table using NIDs, 2991 add support for HMAC PBE OID translation. Add a EVP_CIPHER ctrl: 2992 EVP_CTRL_PBE_PRF_NID this allows a cipher to specify an alternative 2993 PRF which will be automatically used with PBES2. 2994 [Steve Henson] 2995 2996 *) Replace the algorithm specific calls to generate keys in "req" with the 2997 new API. 2998 [Steve Henson] 2999 3000 *) Update PKCS#7 enveloped data routines to use new API. This is now 3001 supported by any public key method supporting the encrypt operation. A 3002 ctrl is added to allow the public key algorithm to examine or modify 3003 the PKCS#7 RecipientInfo structure if it needs to: for RSA this is 3004 a no op. 3005 [Steve Henson] 3006 3007 *) Add a ctrl to asn1 method to allow a public key algorithm to express 3008 a default digest type to use. In most cases this will be SHA1 but some 3009 algorithms (such as GOST) need to specify an alternative digest. The 3010 return value indicates how strong the prefernce is 1 means optional and 3011 2 is mandatory (that is it is the only supported type). Modify 3012 ASN1_item_sign() to accept a NULL digest argument to indicate it should 3013 use the default md. Update openssl utilities to use the default digest 3014 type for signing if it is not explicitly indicated. 3015 [Steve Henson] 3016 3017 *) Use OID cross reference table in ASN1_sign() and ASN1_verify(). New 3018 EVP_MD flag EVP_MD_FLAG_PKEY_METHOD_SIGNATURE. This uses the relevant 3019 signing method from the key type. This effectively removes the link 3020 between digests and public key types. 3021 [Steve Henson] 3022 3023 *) Add an OID cross reference table and utility functions. Its purpose is to 3024 translate between signature OIDs such as SHA1WithrsaEncryption and SHA1, 3025 rsaEncryption. This will allow some of the algorithm specific hackery 3026 needed to use the correct OID to be removed. 3027 [Steve Henson] 3028 3029 *) Remove algorithm specific dependencies when setting PKCS7_SIGNER_INFO 3030 structures for PKCS7_sign(). They are now set up by the relevant public 3031 key ASN1 method. 3032 [Steve Henson] 3033 3034 *) Add provisional EC pkey method with support for ECDSA and ECDH. 3035 [Steve Henson] 3036 3037 *) Add support for key derivation (agreement) in the API, DH method and 3038 pkeyutl. 3039 [Steve Henson] 3040 3041 *) Add DSA pkey method and DH pkey methods, extend DH ASN1 method to support 3042 public and private key formats. As a side effect these add additional 3043 command line functionality not previously available: DSA signatures can be 3044 generated and verified using pkeyutl and DH key support and generation in 3045 pkey, genpkey. 3046 [Steve Henson] 3047 3048 *) BeOS support. 3049 [Oliver Tappe <zooey@hirschkaefer.de>] 3050 3051 *) New make target "install_html_docs" installs HTML renditions of the 3052 manual pages. 3053 [Oliver Tappe <zooey@hirschkaefer.de>] 3054 3055 *) New utility "genpkey" this is analagous to "genrsa" etc except it can 3056 generate keys for any algorithm. Extend and update EVP_PKEY_METHOD to 3057 support key and parameter generation and add initial key generation 3058 functionality for RSA. 3059 [Steve Henson] 3060 3061 *) Add functions for main EVP_PKEY_method operations. The undocumented 3062 functions EVP_PKEY_{encrypt,decrypt} have been renamed to 3063 EVP_PKEY_{encrypt,decrypt}_old. 3064 [Steve Henson] 3065 3066 *) Initial definitions for EVP_PKEY_METHOD. This will be a high level public 3067 key API, doesn't do much yet. 3068 [Steve Henson] 3069 3070 *) New function EVP_PKEY_asn1_get0_info() to retrieve information about 3071 public key algorithms. New option to openssl utility: 3072 "list-public-key-algorithms" to print out info. 3073 [Steve Henson] 3074 3075 *) Implement the Supported Elliptic Curves Extension for 3076 ECC ciphersuites from draft-ietf-tls-ecc-12.txt. 3077 [Douglas Stebila] 3078 3079 *) Don't free up OIDs in OBJ_cleanup() if they are in use by EVP_MD or 3080 EVP_CIPHER structures to avoid later problems in EVP_cleanup(). 3081 [Steve Henson] 3082 3083 *) New utilities pkey and pkeyparam. These are similar to algorithm specific 3084 utilities such as rsa, dsa, dsaparam etc except they process any key 3085 type. 3086 [Steve Henson] 3087 3088 *) Transfer public key printing routines to EVP_PKEY_ASN1_METHOD. New 3089 functions EVP_PKEY_print_public(), EVP_PKEY_print_private(), 3090 EVP_PKEY_print_param() to print public key data from an EVP_PKEY 3091 structure. 3092 [Steve Henson] 3093 3094 *) Initial support for pluggable public key ASN1. 3095 De-spaghettify the public key ASN1 handling. Move public and private 3096 key ASN1 handling to a new EVP_PKEY_ASN1_METHOD structure. Relocate 3097 algorithm specific handling to a single module within the relevant 3098 algorithm directory. Add functions to allow (near) opaque processing 3099 of public and private key structures. 3100 [Steve Henson] 3101 3102 *) Implement the Supported Point Formats Extension for 3103 ECC ciphersuites from draft-ietf-tls-ecc-12.txt. 3104 [Douglas Stebila] 3105 3106 *) Add initial support for RFC 4279 PSK TLS ciphersuites. Add members 3107 for the psk identity [hint] and the psk callback functions to the 3108 SSL_SESSION, SSL and SSL_CTX structure. 3109 3110 New ciphersuites: 3111 PSK-RC4-SHA, PSK-3DES-EDE-CBC-SHA, PSK-AES128-CBC-SHA, 3112 PSK-AES256-CBC-SHA 3113 3114 New functions: 3115 SSL_CTX_use_psk_identity_hint 3116 SSL_get_psk_identity_hint 3117 SSL_get_psk_identity 3118 SSL_use_psk_identity_hint 3119 3120 [Mika Kousa and Pasi Eronen of Nokia Corporation] 3121 3122 *) Add RFC 3161 compliant time stamp request creation, response generation 3123 and response verification functionality. 3124 [Zolt��n Gl��zik <zglozik@opentsa.org>, The OpenTSA Project] 3125 3126 *) Add initial support for TLS extensions, specifically for the server_name 3127 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now 3128 have new members for a host name. The SSL data structure has an 3129 additional member SSL_CTX *initial_ctx so that new sessions can be 3130 stored in that context to allow for session resumption, even after the 3131 SSL has been switched to a new SSL_CTX in reaction to a client's 3132 server_name extension. 3133 3134 New functions (subject to change): 3135 3136 SSL_get_servername() 3137 SSL_get_servername_type() 3138 SSL_set_SSL_CTX() 3139 3140 New CTRL codes and macros (subject to change): 3141 3142 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB 3143 - SSL_CTX_set_tlsext_servername_callback() 3144 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG 3145 - SSL_CTX_set_tlsext_servername_arg() 3146 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name() 3147 3148 openssl s_client has a new '-servername ...' option. 3149 3150 openssl s_server has new options '-servername_host ...', '-cert2 ...', 3151 '-key2 ...', '-servername_fatal' (subject to change). This allows 3152 testing the HostName extension for a specific single host name ('-cert' 3153 and '-key' remain fallbacks for handshakes without HostName 3154 negotiation). If the unrecogninzed_name alert has to be sent, this by 3155 default is a warning; it becomes fatal with the '-servername_fatal' 3156 option. 3157 3158 [Peter Sylvester, Remy Allais, Christophe Renou] 3159 3160 *) Whirlpool hash implementation is added. 3161 [Andy Polyakov] 3162 3163 *) BIGNUM code on 64-bit SPARCv9 targets is switched from bn(64,64) to 3164 bn(64,32). Because of instruction set limitations it doesn't have 3165 any negative impact on performance. This was done mostly in order 3166 to make it possible to share assembler modules, such as bn_mul_mont 3167 implementations, between 32- and 64-bit builds without hassle. 3168 [Andy Polyakov] 3169 3170 *) Move code previously exiled into file crypto/ec/ec2_smpt.c 3171 to ec2_smpl.c, and no longer require the OPENSSL_EC_BIN_PT_COMP 3172 macro. 3173 [Bodo Moeller] 3174 3175 *) New candidate for BIGNUM assembler implementation, bn_mul_mont, 3176 dedicated Montgomery multiplication procedure, is introduced. 3177 BN_MONT_CTX is modified to allow bn_mul_mont to reach for higher 3178 "64-bit" performance on certain 32-bit targets. 3179 [Andy Polyakov] 3180 3181 *) New option SSL_OP_NO_COMP to disable use of compression selectively 3182 in SSL structures. New SSL ctrl to set maximum send fragment size. 3183 Save memory by seeting the I/O buffer sizes dynamically instead of 3184 using the maximum available value. 3185 [Steve Henson] 3186 3187 *) New option -V for 'openssl ciphers'. This prints the ciphersuite code 3188 in addition to the text details. 3189 [Bodo Moeller] 3190 3191 *) Very, very preliminary EXPERIMENTAL support for printing of general 3192 ASN1 structures. This currently produces rather ugly output and doesn't 3193 handle several customised structures at all. 3194 [Steve Henson] 3195 3196 *) Integrated support for PVK file format and some related formats such 3197 as MS PUBLICKEYBLOB and PRIVATEKEYBLOB. Command line switches to support 3198 these in the 'rsa' and 'dsa' utilities. 3199 [Steve Henson] 3200 3201 *) Support for PKCS#1 RSAPublicKey format on rsa utility command line. 3202 [Steve Henson] 3203 3204 *) Remove the ancient ASN1_METHOD code. This was only ever used in one 3205 place for the (very old) "NETSCAPE" format certificates which are now 3206 handled using new ASN1 code equivalents. 3207 [Steve Henson] 3208 3209 *) Let the TLSv1_method() etc. functions return a 'const' SSL_METHOD 3210 pointer and make the SSL_METHOD parameter in SSL_CTX_new, 3211 SSL_CTX_set_ssl_version and SSL_set_ssl_method 'const'. 3212 [Nils Larsch] 3213 3214 *) Modify CRL distribution points extension code to print out previously 3215 unsupported fields. Enhance extension setting code to allow setting of 3216 all fields. 3217 [Steve Henson] 3218 3219 *) Add print and set support for Issuing Distribution Point CRL extension. 3220 [Steve Henson] 3221 3222 *) Change 'Configure' script to enable Camellia by default. 3223 [NTT] 3224 3225 Changes between 0.9.8m and 0.9.8n [24 Mar 2010] 3226 3227 *) When rejecting SSL/TLS records due to an incorrect version number, never 3228 update s->server with a new major version number. As of 3229 - OpenSSL 0.9.8m if 'short' is a 16-bit type, 3230 - OpenSSL 0.9.8f if 'short' is longer than 16 bits, 3231 the previous behavior could result in a read attempt at NULL when 3232 receiving specific incorrect SSL/TLS records once record payload 3233 protection is active. (CVE-2010-0740) 3234 [Bodo Moeller, Adam Langley <agl@chromium.org>] 3235 3236 *) Fix for CVE-2010-0433 where some kerberos enabled versions of OpenSSL 3237 could be crashed if the relevant tables were not present (e.g. chrooted). 3238 [Tomas Hoger <thoger@redhat.com>] 3239 3240 Changes between 0.9.8l and 0.9.8m [25 Feb 2010] 3241 3242 *) Always check bn_wexpend() return values for failure. (CVE-2009-3245) 3243 [Martin Olsson, Neel Mehta] 3244 3245 *) Fix X509_STORE locking: Every 'objs' access requires a lock (to 3246 accommodate for stack sorting, always a write lock!). 3247 [Bodo Moeller] 3248 3249 *) On some versions of WIN32 Heap32Next is very slow. This can cause 3250 excessive delays in the RAND_poll(): over a minute. As a workaround 3251 include a time check in the inner Heap32Next loop too. 3252 [Steve Henson] 3253 3254 *) The code that handled flushing of data in SSL/TLS originally used the 3255 BIO_CTRL_INFO ctrl to see if any data was pending first. This caused 3256 the problem outlined in PR#1949. The fix suggested there however can 3257 trigger problems with buggy BIO_CTRL_WPENDING (e.g. some versions 3258 of Apache). So instead simplify the code to flush unconditionally. 3259 This should be fine since flushing with no data to flush is a no op. 3260 [Steve Henson] 3261 3262 *) Handle TLS versions 2.0 and later properly and correctly use the 3263 highest version of TLS/SSL supported. Although TLS >= 2.0 is some way 3264 off ancient servers have a habit of sticking around for a while... 3265 [Steve Henson] 3266 3267 *) Modify compression code so it frees up structures without using the 3268 ex_data callbacks. This works around a problem where some applications 3269 call CRYPTO_cleanup_all_ex_data() before application exit (e.g. when 3270 restarting) then use compression (e.g. SSL with compression) later. 3271 This results in significant per-connection memory leaks and 3272 has caused some security issues including CVE-2008-1678 and 3273 CVE-2009-4355. 3274 [Steve Henson] 3275 3276 *) Constify crypto/cast (i.e., <openssl/cast.h>): a CAST_KEY doesn't 3277 change when encrypting or decrypting. 3278 [Bodo Moeller] 3279 3280 *) Add option SSL_OP_LEGACY_SERVER_CONNECT which will allow clients to 3281 connect and renegotiate with servers which do not support RI. 3282 Until RI is more widely deployed this option is enabled by default. 3283 [Steve Henson] 3284 3285 *) Add "missing" ssl ctrls to clear options and mode. 3286 [Steve Henson] 3287 3288 *) If client attempts to renegotiate and doesn't support RI respond with 3289 a no_renegotiation alert as required by RFC5746. Some renegotiating 3290 TLS clients will continue a connection gracefully when they receive 3291 the alert. Unfortunately OpenSSL mishandled this alert and would hang 3292 waiting for a server hello which it will never receive. Now we treat a 3293 received no_renegotiation alert as a fatal error. This is because 3294 applications requesting a renegotiation might well expect it to succeed 3295 and would have no code in place to handle the server denying it so the 3296 only safe thing to do is to terminate the connection. 3297 [Steve Henson] 3298 3299 *) Add ctrl macro SSL_get_secure_renegotiation_support() which returns 1 if 3300 peer supports secure renegotiation and 0 otherwise. Print out peer 3301 renegotiation support in s_client/s_server. 3302 [Steve Henson] 3303 3304 *) Replace the highly broken and deprecated SPKAC certification method with 3305 the updated NID creation version. This should correctly handle UTF8. 3306 [Steve Henson] 3307 3308 *) Implement RFC5746. Re-enable renegotiation but require the extension 3309 as needed. Unfortunately, SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION 3310 turns out to be a bad idea. It has been replaced by 3311 SSL_OP_ALLOW_UNSAFE_LEGACY_RENEGOTIATION which can be set with 3312 SSL_CTX_set_options(). This is really not recommended unless you 3313 know what you are doing. 3314 [Eric Rescorla <ekr@networkresonance.com>, Ben Laurie, Steve Henson] 3315 3316 *) Fixes to stateless session resumption handling. Use initial_ctx when 3317 issuing and attempting to decrypt tickets in case it has changed during 3318 servername handling. Use a non-zero length session ID when attempting 3319 stateless session resumption: this makes it possible to determine if 3320 a resumption has occurred immediately after receiving server hello 3321 (several places in OpenSSL subtly assume this) instead of later in 3322 the handshake. 3323 [Steve Henson] 3324 3325 *) The functions ENGINE_ctrl(), OPENSSL_isservice(), 3326 CMS_get1_RecipientRequest() and RAND_bytes() can return <=0 on error 3327 fixes for a few places where the return code is not checked 3328 correctly. 3329 [Julia Lawall <julia@diku.dk>] 3330 3331 *) Add --strict-warnings option to Configure script to include devteam 3332 warnings in other configurations. 3333 [Steve Henson] 3334 3335 *) Add support for --libdir option and LIBDIR variable in makefiles. This 3336 makes it possible to install openssl libraries in locations which 3337 have names other than "lib", for example "/usr/lib64" which some 3338 systems need. 3339 [Steve Henson, based on patch from Jeremy Utley] 3340 3341 *) Don't allow the use of leading 0x80 in OIDs. This is a violation of 3342 X690 8.9.12 and can produce some misleading textual output of OIDs. 3343 [Steve Henson, reported by Dan Kaminsky] 3344 3345 *) Delete MD2 from algorithm tables. This follows the recommendation in 3346 several standards that it is not used in new applications due to 3347 several cryptographic weaknesses. For binary compatibility reasons 3348 the MD2 API is still compiled in by default. 3349 [Steve Henson] 3350 3351 *) Add compression id to {d2i,i2d}_SSL_SESSION so it is correctly saved 3352 and restored. 3353 [Steve Henson] 3354 3355 *) Rename uni2asc and asc2uni functions to OPENSSL_uni2asc and 3356 OPENSSL_asc2uni conditionally on Netware platforms to avoid a name 3357 clash. 3358 [Guenter <lists@gknw.net>] 3359 3360 *) Fix the server certificate chain building code to use X509_verify_cert(), 3361 it used to have an ad-hoc builder which was unable to cope with anything 3362 other than a simple chain. 3363 [David Woodhouse <dwmw2@infradead.org>, Steve Henson] 3364 3365 *) Don't check self signed certificate signatures in X509_verify_cert() 3366 by default (a flag can override this): it just wastes time without 3367 adding any security. As a useful side effect self signed root CAs 3368 with non-FIPS digests are now usable in FIPS mode. 3369 [Steve Henson] 3370 3371 *) In dtls1_process_out_of_seq_message() the check if the current message 3372 is already buffered was missing. For every new message was memory 3373 allocated, allowing an attacker to perform an denial of service attack 3374 with sending out of seq handshake messages until there is no memory 3375 left. Additionally every future messege was buffered, even if the 3376 sequence number made no sense and would be part of another handshake. 3377 So only messages with sequence numbers less than 10 in advance will be 3378 buffered. (CVE-2009-1378) 3379 [Robin Seggelmann, discovered by Daniel Mentz] 3380 3381 *) Records are buffered if they arrive with a future epoch to be 3382 processed after finishing the corresponding handshake. There is 3383 currently no limitation to this buffer allowing an attacker to perform 3384 a DOS attack with sending records with future epochs until there is no 3385 memory left. This patch adds the pqueue_size() function to detemine 3386 the size of a buffer and limits the record buffer to 100 entries. 3387 (CVE-2009-1377) 3388 [Robin Seggelmann, discovered by Daniel Mentz] 3389 3390 *) Keep a copy of frag->msg_header.frag_len so it can be used after the 3391 parent structure is freed. (CVE-2009-1379) 3392 [Daniel Mentz] 3393 3394 *) Handle non-blocking I/O properly in SSL_shutdown() call. 3395 [Darryl Miles <darryl-mailinglists@netbauds.net>] 3396 3397 *) Add 2.5.4.* OIDs 3398 [Ilya O. <vrghost@gmail.com>] 3399 3400 Changes between 0.9.8k and 0.9.8l [5 Nov 2009] 3401 3402 *) Disable renegotiation completely - this fixes a severe security 3403 problem (CVE-2009-3555) at the cost of breaking all 3404 renegotiation. Renegotiation can be re-enabled by setting 3405 SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION in s3->flags at 3406 run-time. This is really not recommended unless you know what 3407 you're doing. 3408 [Ben Laurie] 3409 3410 Changes between 0.9.8j and 0.9.8k [25 Mar 2009] 3411 3412 *) Don't set val to NULL when freeing up structures, it is freed up by 3413 underlying code. If sizeof(void *) > sizeof(long) this can result in 3414 zeroing past the valid field. (CVE-2009-0789) 3415 [Paolo Ganci <Paolo.Ganci@AdNovum.CH>] 3416 3417 *) Fix bug where return value of CMS_SignerInfo_verify_content() was not 3418 checked correctly. This would allow some invalid signed attributes to 3419 appear to verify correctly. (CVE-2009-0591) 3420 [Ivan Nestlerode <inestlerode@us.ibm.com>] 3421 3422 *) Reject UniversalString and BMPString types with invalid lengths. This 3423 prevents a crash in ASN1_STRING_print_ex() which assumes the strings have 3424 a legal length. (CVE-2009-0590) 3425 [Steve Henson] 3426 3427 *) Set S/MIME signing as the default purpose rather than setting it 3428 unconditionally. This allows applications to override it at the store 3429 level. 3430 [Steve Henson] 3431 3432 *) Permit restricted recursion of ASN1 strings. This is needed in practice 3433 to handle some structures. 3434 [Steve Henson] 3435 3436 *) Improve efficiency of mem_gets: don't search whole buffer each time 3437 for a '\n' 3438 [Jeremy Shapiro <jnshapir@us.ibm.com>] 3439 3440 *) New -hex option for openssl rand. 3441 [Matthieu Herrb] 3442 3443 *) Print out UTF8String and NumericString when parsing ASN1. 3444 [Steve Henson] 3445 3446 *) Support NumericString type for name components. 3447 [Steve Henson] 3448 3449 *) Allow CC in the environment to override the automatically chosen 3450 compiler. Note that nothing is done to ensure flags work with the 3451 chosen compiler. 3452 [Ben Laurie] 3453 3454 Changes between 0.9.8i and 0.9.8j [07 Jan 2009] 3455 3456 *) Properly check EVP_VerifyFinal() and similar return values 3457 (CVE-2008-5077). 3458 [Ben Laurie, Bodo Moeller, Google Security Team] 3459 3460 *) Enable TLS extensions by default. 3461 [Ben Laurie] 3462 3463 *) Allow the CHIL engine to be loaded, whether the application is 3464 multithreaded or not. (This does not release the developer from the 3465 obligation to set up the dynamic locking callbacks.) 3466 [Sander Temme <sander@temme.net>] 3467 3468 *) Use correct exit code if there is an error in dgst command. 3469 [Steve Henson; problem pointed out by Roland Dirlewanger] 3470 3471 *) Tweak Configure so that you need to say "experimental-jpake" to enable 3472 JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications. 3473 [Bodo Moeller] 3474 3475 *) Add experimental JPAKE support, including demo authentication in 3476 s_client and s_server. 3477 [Ben Laurie] 3478 3479 *) Set the comparison function in v3_addr_canonize(). 3480 [Rob Austein <sra@hactrn.net>] 3481 3482 *) Add support for XMPP STARTTLS in s_client. 3483 [Philip Paeps <philip@freebsd.org>] 3484 3485 *) Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior 3486 to ensure that even with this option, only ciphersuites in the 3487 server's preference list will be accepted. (Note that the option 3488 applies only when resuming a session, so the earlier behavior was 3489 just about the algorithm choice for symmetric cryptography.) 3490 [Bodo Moeller] 3491 3492 Changes between 0.9.8h and 0.9.8i [15 Sep 2008] 3493 3494 *) Fix NULL pointer dereference if a DTLS server received 3495 ChangeCipherSpec as first record (CVE-2009-1386). 3496 [PR #1679] 3497 3498 *) Fix a state transitition in s3_srvr.c and d1_srvr.c 3499 (was using SSL3_ST_CW_CLNT_HELLO_B, should be ..._ST_SW_SRVR_...). 3500 [Nagendra Modadugu] 3501 3502 *) The fix in 0.9.8c that supposedly got rid of unsafe 3503 double-checked locking was incomplete for RSA blinding, 3504 addressing just one layer of what turns out to have been 3505 doubly unsafe triple-checked locking. 3506 3507 So now fix this for real by retiring the MONT_HELPER macro 3508 in crypto/rsa/rsa_eay.c. 3509 3510 [Bodo Moeller; problem pointed out by Marius Schilder] 3511 3512 *) Various precautionary measures: 3513 3514 - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h). 3515 3516 - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c). 3517 (NB: This would require knowledge of the secret session ticket key 3518 to exploit, in which case you'd be SOL either way.) 3519 3520 - Change bn_nist.c so that it will properly handle input BIGNUMs 3521 outside the expected range. 3522 3523 - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG 3524 builds. 3525 3526 [Neel Mehta, Bodo Moeller] 3527 3528 *) Allow engines to be "soft loaded" - i.e. optionally don't die if 3529 the load fails. Useful for distros. 3530 [Ben Laurie and the FreeBSD team] 3531 3532 *) Add support for Local Machine Keyset attribute in PKCS#12 files. 3533 [Steve Henson] 3534 3535 *) Fix BN_GF2m_mod_arr() top-bit cleanup code. 3536 [Huang Ying] 3537 3538 *) Expand ENGINE to support engine supplied SSL client certificate functions. 3539 3540 This work was sponsored by Logica. 3541 [Steve Henson] 3542 3543 *) Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows 3544 keystores. Support for SSL/TLS client authentication too. 3545 Not compiled unless enable-capieng specified to Configure. 3546 3547 This work was sponsored by Logica. 3548 [Steve Henson] 3549 3550 *) Fix bug in X509_ATTRIBUTE creation: dont set attribute using 3551 ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain 3552 attribute creation routines such as certifcate requests and PKCS#12 3553 files. 3554 [Steve Henson] 3555 3556 Changes between 0.9.8g and 0.9.8h [28 May 2008] 3557 3558 *) Fix flaw if 'Server Key exchange message' is omitted from a TLS 3559 handshake which could lead to a cilent crash as found using the 3560 Codenomicon TLS test suite (CVE-2008-1672) 3561 [Steve Henson, Mark Cox] 3562 3563 *) Fix double free in TLS server name extensions which could lead to 3564 a remote crash found by Codenomicon TLS test suite (CVE-2008-0891) 3565 [Joe Orton] 3566 3567 *) Clear error queue in SSL_CTX_use_certificate_chain_file() 3568 3569 Clear the error queue to ensure that error entries left from 3570 older function calls do not interfere with the correct operation. 3571 [Lutz Jaenicke, Erik de Castro Lopo] 3572 3573 *) Remove root CA certificates of commercial CAs: 3574 3575 The OpenSSL project does not recommend any specific CA and does not 3576 have any policy with respect to including or excluding any CA. 3577 Therefore it does not make any sense to ship an arbitrary selection 3578 of root CA certificates with the OpenSSL software. 3579 [Lutz Jaenicke] 3580 3581 *) RSA OAEP patches to fix two separate invalid memory reads. 3582 The first one involves inputs when 'lzero' is greater than 3583 'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes 3584 before the beginning of from). The second one involves inputs where 3585 the 'db' section contains nothing but zeroes (there is a one-byte 3586 invalid read after the end of 'db'). 3587 [Ivan Nestlerode <inestlerode@us.ibm.com>] 3588 3589 *) Partial backport from 0.9.9-dev: 3590 3591 Introduce bn_mul_mont (dedicated Montgomery multiplication 3592 procedure) as a candidate for BIGNUM assembler implementation. 3593 While 0.9.9-dev uses assembler for various architectures, only 3594 x86_64 is available by default here in the 0.9.8 branch, and 3595 32-bit x86 is available through a compile-time setting. 3596 3597 To try the 32-bit x86 assembler implementation, use Configure 3598 option "enable-montasm" (which exists only for this backport). 3599 3600 As "enable-montasm" for 32-bit x86 disclaims code stability 3601 anyway, in this constellation we activate additional code 3602 backported from 0.9.9-dev for further performance improvements, 3603 namely BN_from_montgomery_word. (To enable this otherwise, 3604 e.g. x86_64, try "-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD".) 3605 3606 [Andy Polyakov (backport partially by Bodo Moeller)] 3607 3608 *) Add TLS session ticket callback. This allows an application to set 3609 TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed 3610 values. This is useful for key rollover for example where several key 3611 sets may exist with different names. 3612 [Steve Henson] 3613 3614 *) Reverse ENGINE-internal logic for caching default ENGINE handles. 3615 This was broken until now in 0.9.8 releases, such that the only way 3616 a registered ENGINE could be used (assuming it initialises 3617 successfully on the host) was to explicitly set it as the default 3618 for the relevant algorithms. This is in contradiction with 0.9.7 3619 behaviour and the documentation. With this fix, when an ENGINE is 3620 registered into a given algorithm's table of implementations, the 3621 'uptodate' flag is reset so that auto-discovery will be used next 3622 time a new context for that algorithm attempts to select an 3623 implementation. 3624 [Ian Lister (tweaked by Geoff Thorpe)] 3625 3626 *) Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9 3627 implemention in the following ways: 3628 3629 Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be 3630 hard coded. 3631 3632 Lack of BER streaming support means one pass streaming processing is 3633 only supported if data is detached: setting the streaming flag is 3634 ignored for embedded content. 3635 3636 CMS support is disabled by default and must be explicitly enabled 3637 with the enable-cms configuration option. 3638 [Steve Henson] 3639 3640 *) Update the GMP engine glue to do direct copies between BIGNUM and 3641 mpz_t when openssl and GMP use the same limb size. Otherwise the 3642 existing "conversion via a text string export" trick is still used. 3643 [Paul Sheer <paulsheer@gmail.com>] 3644 3645 *) Zlib compression BIO. This is a filter BIO which compressed and 3646 uncompresses any data passed through it. 3647 [Steve Henson] 3648 3649 *) Add AES_wrap_key() and AES_unwrap_key() functions to implement 3650 RFC3394 compatible AES key wrapping. 3651 [Steve Henson] 3652 3653 *) Add utility functions to handle ASN1 structures. ASN1_STRING_set0(): 3654 sets string data without copying. X509_ALGOR_set0() and 3655 X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier) 3656 data. Attribute function X509at_get0_data_by_OBJ(): retrieves data 3657 from an X509_ATTRIBUTE structure optionally checking it occurs only 3658 once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied 3659 data. 3660 [Steve Henson] 3661 3662 *) Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set() 3663 to get the expected BN_FLG_CONSTTIME behavior. 3664 [Bodo Moeller (Google)] 3665 3666 *) Netware support: 3667 3668 - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets 3669 - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT) 3670 - added some more tests to do_tests.pl 3671 - fixed RunningProcess usage so that it works with newer LIBC NDKs too 3672 - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency 3673 - added new Configure targets netware-clib-bsdsock, netware-clib-gcc, 3674 netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc 3675 - various changes to netware.pl to enable gcc-cross builds on Win32 3676 platform 3677 - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD) 3678 - various changes to fix missing prototype warnings 3679 - fixed x86nasm.pl to create correct asm files for NASM COFF output 3680 - added AES, WHIRLPOOL and CPUID assembler code to build files 3681 - added missing AES assembler make rules to mk1mf.pl 3682 - fixed order of includes in apps/ocsp.c so that e_os.h settings apply 3683 [Guenter Knauf <eflash@gmx.net>] 3684 3685 *) Implement certificate status request TLS extension defined in RFC3546. 3686 A client can set the appropriate parameters and receive the encoded 3687 OCSP response via a callback. A server can query the supplied parameters 3688 and set the encoded OCSP response in the callback. Add simplified examples 3689 to s_client and s_server. 3690 [Steve Henson] 3691 3692 Changes between 0.9.8f and 0.9.8g [19 Oct 2007] 3693 3694 *) Fix various bugs: 3695 + Binary incompatibility of ssl_ctx_st structure 3696 + DTLS interoperation with non-compliant servers 3697 + Don't call get_session_cb() without proposed session 3698 + Fix ia64 assembler code 3699 [Andy Polyakov, Steve Henson] 3700 3701 Changes between 0.9.8e and 0.9.8f [11 Oct 2007] 3702 3703 *) DTLS Handshake overhaul. There were longstanding issues with 3704 OpenSSL DTLS implementation, which were making it impossible for 3705 RFC 4347 compliant client to communicate with OpenSSL server. 3706 Unfortunately just fixing these incompatibilities would "cut off" 3707 pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e 3708 server keeps tolerating non RFC compliant syntax. The opposite is 3709 not true, 0.9.8f client can not communicate with earlier server. 3710 This update even addresses CVE-2007-4995. 3711 [Andy Polyakov] 3712 3713 *) Changes to avoid need for function casts in OpenSSL: some compilers 3714 (gcc 4.2 and later) reject their use. 3715 [Kurt Roeckx <kurt@roeckx.be>, Peter Hartley <pdh@utter.chaos.org.uk>, 3716 Steve Henson] 3717 3718 *) Add RFC4507 support to OpenSSL. This includes the corrections in 3719 RFC4507bis. The encrypted ticket format is an encrypted encoded 3720 SSL_SESSION structure, that way new session features are automatically 3721 supported. 3722 3723 If a client application caches session in an SSL_SESSION structure 3724 support is transparent because tickets are now stored in the encoded 3725 SSL_SESSION. 3726 3727 The SSL_CTX structure automatically generates keys for ticket 3728 protection in servers so again support should be possible 3729 with no application modification. 3730 3731 If a client or server wishes to disable RFC4507 support then the option 3732 SSL_OP_NO_TICKET can be set. 3733 3734 Add a TLS extension debugging callback to allow the contents of any client 3735 or server extensions to be examined. 3736 3737 This work was sponsored by Google. 3738 [Steve Henson] 3739 3740 *) Add initial support for TLS extensions, specifically for the server_name 3741 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now 3742 have new members for a host name. The SSL data structure has an 3743 additional member SSL_CTX *initial_ctx so that new sessions can be 3744 stored in that context to allow for session resumption, even after the 3745 SSL has been switched to a new SSL_CTX in reaction to a client's 3746 server_name extension. 3747 3748 New functions (subject to change): 3749 3750 SSL_get_servername() 3751 SSL_get_servername_type() 3752 SSL_set_SSL_CTX() 3753 3754 New CTRL codes and macros (subject to change): 3755 3756 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB 3757 - SSL_CTX_set_tlsext_servername_callback() 3758 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG 3759 - SSL_CTX_set_tlsext_servername_arg() 3760 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name() 3761 3762 openssl s_client has a new '-servername ...' option. 3763 3764 openssl s_server has new options '-servername_host ...', '-cert2 ...', 3765 '-key2 ...', '-servername_fatal' (subject to change). This allows 3766 testing the HostName extension for a specific single host name ('-cert' 3767 and '-key' remain fallbacks for handshakes without HostName 3768 negotiation). If the unrecogninzed_name alert has to be sent, this by 3769 default is a warning; it becomes fatal with the '-servername_fatal' 3770 option. 3771 3772 [Peter Sylvester, Remy Allais, Christophe Renou, Steve Henson] 3773 3774 *) Add AES and SSE2 assembly language support to VC++ build. 3775 [Steve Henson] 3776 3777 *) Mitigate attack on final subtraction in Montgomery reduction. 3778 [Andy Polyakov] 3779 3780 *) Fix crypto/ec/ec_mult.c to work properly with scalars of value 0 3781 (which previously caused an internal error). 3782 [Bodo Moeller] 3783 3784 *) Squeeze another 10% out of IGE mode when in != out. 3785 [Ben Laurie] 3786 3787 *) AES IGE mode speedup. 3788 [Dean Gaudet (Google)] 3789 3790 *) Add the Korean symmetric 128-bit cipher SEED (see 3791 http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp) and 3792 add SEED ciphersuites from RFC 4162: 3793 3794 TLS_RSA_WITH_SEED_CBC_SHA = "SEED-SHA" 3795 TLS_DHE_DSS_WITH_SEED_CBC_SHA = "DHE-DSS-SEED-SHA" 3796 TLS_DHE_RSA_WITH_SEED_CBC_SHA = "DHE-RSA-SEED-SHA" 3797 TLS_DH_anon_WITH_SEED_CBC_SHA = "ADH-SEED-SHA" 3798 3799 To minimize changes between patchlevels in the OpenSSL 0.9.8 3800 series, SEED remains excluded from compilation unless OpenSSL 3801 is configured with 'enable-seed'. 3802 [KISA, Bodo Moeller] 3803 3804 *) Mitigate branch prediction attacks, which can be practical if a 3805 single processor is shared, allowing a spy process to extract 3806 information. For detailed background information, see 3807 http://eprint.iacr.org/2007/039 (O. Aciicmez, S. Gueron, 3808 J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL 3809 and Necessary Software Countermeasures"). The core of the change 3810 are new versions BN_div_no_branch() and 3811 BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(), 3812 respectively, which are slower, but avoid the security-relevant 3813 conditional branches. These are automatically called by BN_div() 3814 and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one 3815 of the input BIGNUMs. Also, BN_is_bit_set() has been changed to 3816 remove a conditional branch. 3817 3818 BN_FLG_CONSTTIME is the new name for the previous 3819 BN_FLG_EXP_CONSTTIME flag, since it now affects more than just 3820 modular exponentiation. (Since OpenSSL 0.9.7h, setting this flag 3821 in the exponent causes BN_mod_exp_mont() to use the alternative 3822 implementation in BN_mod_exp_mont_consttime().) The old name 3823 remains as a deprecated alias. 3824 3825 Similary, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general 3826 RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses 3827 constant-time implementations for more than just exponentiation. 3828 Here too the old name is kept as a deprecated alias. 3829 3830 BN_BLINDING_new() will now use BN_dup() for the modulus so that 3831 the BN_BLINDING structure gets an independent copy of the 3832 modulus. This means that the previous "BIGNUM *m" argument to 3833 BN_BLINDING_new() and to BN_BLINDING_create_param() now 3834 essentially becomes "const BIGNUM *m", although we can't actually 3835 change this in the header file before 0.9.9. It allows 3836 RSA_setup_blinding() to use BN_with_flags() on the modulus to 3837 enable BN_FLG_CONSTTIME. 3838 3839 [Matthew D Wood (Intel Corp)] 3840 3841 *) In the SSL/TLS server implementation, be strict about session ID 3842 context matching (which matters if an application uses a single 3843 external cache for different purposes). Previously, 3844 out-of-context reuse was forbidden only if SSL_VERIFY_PEER was 3845 set. This did ensure strict client verification, but meant that, 3846 with applications using a single external cache for quite 3847 different requirements, clients could circumvent ciphersuite 3848 restrictions for a given session ID context by starting a session 3849 in a different context. 3850 [Bodo Moeller] 3851 3852 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 3853 a ciphersuite string such as "DEFAULT:RSA" cannot enable 3854 authentication-only ciphersuites. 3855 [Bodo Moeller] 3856 3857 *) Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was 3858 not complete and could lead to a possible single byte overflow 3859 (CVE-2007-5135) [Ben Laurie] 3860 3861 Changes between 0.9.8d and 0.9.8e [23 Feb 2007] 3862 3863 *) Since AES128 and AES256 (and similarly Camellia128 and 3864 Camellia256) share a single mask bit in the logic of 3865 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 3866 kludge to work properly if AES128 is available and AES256 isn't 3867 (or if Camellia128 is available and Camellia256 isn't). 3868 [Victor Duchovni] 3869 3870 *) Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c 3871 (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters): 3872 When a point or a seed is encoded in a BIT STRING, we need to 3873 prevent the removal of trailing zero bits to get the proper DER 3874 encoding. (By default, crypto/asn1/a_bitstr.c assumes the case 3875 of a NamedBitList, for which trailing 0 bits need to be removed.) 3876 [Bodo Moeller] 3877 3878 *) Have SSL/TLS server implementation tolerate "mismatched" record 3879 protocol version while receiving ClientHello even if the 3880 ClientHello is fragmented. (The server can't insist on the 3881 particular protocol version it has chosen before the ServerHello 3882 message has informed the client about his choice.) 3883 [Bodo Moeller] 3884 3885 *) Add RFC 3779 support. 3886 [Rob Austein for ARIN, Ben Laurie] 3887 3888 *) Load error codes if they are not already present instead of using a 3889 static variable. This allows them to be cleanly unloaded and reloaded. 3890 Improve header file function name parsing. 3891 [Steve Henson] 3892 3893 *) extend SMTP and IMAP protocol emulation in s_client to use EHLO 3894 or CAPABILITY handshake as required by RFCs. 3895 [Goetz Babin-Ebell] 3896 3897 Changes between 0.9.8c and 0.9.8d [28 Sep 2006] 3898 3899 *) Introduce limits to prevent malicious keys being able to 3900 cause a denial of service. (CVE-2006-2940) 3901 [Steve Henson, Bodo Moeller] 3902 3903 *) Fix ASN.1 parsing of certain invalid structures that can result 3904 in a denial of service. (CVE-2006-2937) [Steve Henson] 3905 3906 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 3907 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 3908 3909 *) Fix SSL client code which could crash if connecting to a 3910 malicious SSLv2 server. (CVE-2006-4343) 3911 [Tavis Ormandy and Will Drewry, Google Security Team] 3912 3913 *) Since 0.9.8b, ciphersuite strings naming explicit ciphersuites 3914 match only those. Before that, "AES256-SHA" would be interpreted 3915 as a pattern and match "AES128-SHA" too (since AES128-SHA got 3916 the same strength classification in 0.9.7h) as we currently only 3917 have a single AES bit in the ciphersuite description bitmap. 3918 That change, however, also applied to ciphersuite strings such as 3919 "RC4-MD5" that intentionally matched multiple ciphersuites -- 3920 namely, SSL 2.0 ciphersuites in addition to the more common ones 3921 from SSL 3.0/TLS 1.0. 3922 3923 So we change the selection algorithm again: Naming an explicit 3924 ciphersuite selects this one ciphersuite, and any other similar 3925 ciphersuite (same bitmap) from *other* protocol versions. 3926 Thus, "RC4-MD5" again will properly select both the SSL 2.0 3927 ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite. 3928 3929 Since SSL 2.0 does not have any ciphersuites for which the 3930 128/256 bit distinction would be relevant, this works for now. 3931 The proper fix will be to use different bits for AES128 and 3932 AES256, which would have avoided the problems from the beginning; 3933 however, bits are scarce, so we can only do this in a new release 3934 (not just a patchlevel) when we can change the SSL_CIPHER 3935 definition to split the single 'unsigned long mask' bitmap into 3936 multiple values to extend the available space. 3937 3938 [Bodo Moeller] 3939 3940 Changes between 0.9.8b and 0.9.8c [05 Sep 2006] 3941 3942 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 3943 (CVE-2006-4339) [Ben Laurie and Google Security Team] 3944 3945 *) Add AES IGE and biIGE modes. 3946 [Ben Laurie] 3947 3948 *) Change the Unix randomness entropy gathering to use poll() when 3949 possible instead of select(), since the latter has some 3950 undesirable limitations. 3951 [Darryl Miles via Richard Levitte and Bodo Moeller] 3952 3953 *) Disable "ECCdraft" ciphersuites more thoroughly. Now special 3954 treatment in ssl/ssl_ciph.s makes sure that these ciphersuites 3955 cannot be implicitly activated as part of, e.g., the "AES" alias. 3956 However, please upgrade to OpenSSL 0.9.9[-dev] for 3957 non-experimental use of the ECC ciphersuites to get TLS extension 3958 support, which is required for curve and point format negotiation 3959 to avoid potential handshake problems. 3960 [Bodo Moeller] 3961 3962 *) Disable rogue ciphersuites: 3963 3964 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 3965 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 3966 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 3967 3968 The latter two were purportedly from 3969 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 3970 appear there. 3971 3972 Also deactivate the remaining ciphersuites from 3973 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 3974 unofficial, and the ID has long expired. 3975 [Bodo Moeller] 3976 3977 *) Fix RSA blinding Heisenbug (problems sometimes occured on 3978 dual-core machines) and other potential thread-safety issues. 3979 [Bodo Moeller] 3980 3981 *) Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key 3982 versions), which is now available for royalty-free use 3983 (see http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html). 3984 Also, add Camellia TLS ciphersuites from RFC 4132. 3985 3986 To minimize changes between patchlevels in the OpenSSL 0.9.8 3987 series, Camellia remains excluded from compilation unless OpenSSL 3988 is configured with 'enable-camellia'. 3989 [NTT] 3990 3991 *) Disable the padding bug check when compression is in use. The padding 3992 bug check assumes the first packet is of even length, this is not 3993 necessarily true if compresssion is enabled and can result in false 3994 positives causing handshake failure. The actual bug test is ancient 3995 code so it is hoped that implementations will either have fixed it by 3996 now or any which still have the bug do not support compression. 3997 [Steve Henson] 3998 3999 Changes between 0.9.8a and 0.9.8b [04 May 2006] 4000 4001 *) When applying a cipher rule check to see if string match is an explicit 4002 cipher suite and only match that one cipher suite if it is. 4003 [Steve Henson] 4004 4005 *) Link in manifests for VC++ if needed. 4006 [Austin Ziegler <halostatue@gmail.com>] 4007 4008 *) Update support for ECC-based TLS ciphersuites according to 4009 draft-ietf-tls-ecc-12.txt with proposed changes (but without 4010 TLS extensions, which are supported starting with the 0.9.9 4011 branch, not in the OpenSSL 0.9.8 branch). 4012 [Douglas Stebila] 4013 4014 *) New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support 4015 opaque EVP_CIPHER_CTX handling. 4016 [Steve Henson] 4017 4018 *) Fixes and enhancements to zlib compression code. We now only use 4019 "zlib1.dll" and use the default __cdecl calling convention on Win32 4020 to conform with the standards mentioned here: 4021 http://www.zlib.net/DLL_FAQ.txt 4022 Static zlib linking now works on Windows and the new --with-zlib-include 4023 --with-zlib-lib options to Configure can be used to supply the location 4024 of the headers and library. Gracefully handle case where zlib library 4025 can't be loaded. 4026 [Steve Henson] 4027 4028 *) Several fixes and enhancements to the OID generation code. The old code 4029 sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't 4030 handle numbers larger than ULONG_MAX, truncated printing and had a 4031 non standard OBJ_obj2txt() behaviour. 4032 [Steve Henson] 4033 4034 *) Add support for building of engines under engine/ as shared libraries 4035 under VC++ build system. 4036 [Steve Henson] 4037 4038 *) Corrected the numerous bugs in the Win32 path splitter in DSO. 4039 Hopefully, we will not see any false combination of paths any more. 4040 [Richard Levitte] 4041 4042 Changes between 0.9.8 and 0.9.8a [11 Oct 2005] 4043 4044 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 4045 (part of SSL_OP_ALL). This option used to disable the 4046 countermeasure against man-in-the-middle protocol-version 4047 rollback in the SSL 2.0 server implementation, which is a bad 4048 idea. (CVE-2005-2969) 4049 4050 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 4051 for Information Security, National Institute of Advanced Industrial 4052 Science and Technology [AIST], Japan)] 4053 4054 *) Add two function to clear and return the verify parameter flags. 4055 [Steve Henson] 4056 4057 *) Keep cipherlists sorted in the source instead of sorting them at 4058 runtime, thus removing the need for a lock. 4059 [Nils Larsch] 4060 4061 *) Avoid some small subgroup attacks in Diffie-Hellman. 4062 [Nick Mathewson and Ben Laurie] 4063 4064 *) Add functions for well-known primes. 4065 [Nick Mathewson] 4066 4067 *) Extended Windows CE support. 4068 [Satoshi Nakamura and Andy Polyakov] 4069 4070 *) Initialize SSL_METHOD structures at compile time instead of during 4071 runtime, thus removing the need for a lock. 4072 [Steve Henson] 4073 4074 *) Make PKCS7_decrypt() work even if no certificate is supplied by 4075 attempting to decrypt each encrypted key in turn. Add support to 4076 smime utility. 4077 [Steve Henson] 4078 4079 Changes between 0.9.7h and 0.9.8 [05 Jul 2005] 4080 4081 [NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after 4082 OpenSSL 0.9.8.] 4083 4084 *) Add libcrypto.pc and libssl.pc for those who feel they need them. 4085 [Richard Levitte] 4086 4087 *) Change CA.sh and CA.pl so they don't bundle the CSR and the private 4088 key into the same file any more. 4089 [Richard Levitte] 4090 4091 *) Add initial support for Win64, both IA64 and AMD64/x64 flavors. 4092 [Andy Polyakov] 4093 4094 *) Add -utf8 command line and config file option to 'ca'. 4095 [Stefan <stf@udoma.org] 4096 4097 *) Removed the macro des_crypt(), as it seems to conflict with some 4098 libraries. Use DES_crypt(). 4099 [Richard Levitte] 4100 4101 *) Correct naming of the 'chil' and '4758cca' ENGINEs. This 4102 involves renaming the source and generated shared-libs for 4103 both. The engines will accept the corrected or legacy ids 4104 ('ncipher' and '4758_cca' respectively) when binding. NB, 4105 this only applies when building 'shared'. 4106 [Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe] 4107 4108 *) Add attribute functions to EVP_PKEY structure. Modify 4109 PKCS12_create() to recognize a CSP name attribute and 4110 use it. Make -CSP option work again in pkcs12 utility. 4111 [Steve Henson] 4112 4113 *) Add new functionality to the bn blinding code: 4114 - automatic re-creation of the BN_BLINDING parameters after 4115 a fixed number of uses (currently 32) 4116 - add new function for parameter creation 4117 - introduce flags to control the update behaviour of the 4118 BN_BLINDING parameters 4119 - hide BN_BLINDING structure 4120 Add a second BN_BLINDING slot to the RSA structure to improve 4121 performance when a single RSA object is shared among several 4122 threads. 4123 [Nils Larsch] 4124 4125 *) Add support for DTLS. 4126 [Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie] 4127 4128 *) Add support for DER encoded private keys (SSL_FILETYPE_ASN1) 4129 to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file() 4130 [Walter Goulet] 4131 4132 *) Remove buggy and incompletet DH cert support from 4133 ssl/ssl_rsa.c and ssl/s3_both.c 4134 [Nils Larsch] 4135 4136 *) Use SHA-1 instead of MD5 as the default digest algorithm for 4137 the apps/openssl applications. 4138 [Nils Larsch] 4139 4140 *) Compile clean with "-Wall -Wmissing-prototypes 4141 -Wstrict-prototypes -Wmissing-declarations -Werror". Currently 4142 DEBUG_SAFESTACK must also be set. 4143 [Ben Laurie] 4144 4145 *) Change ./Configure so that certain algorithms can be disabled by default. 4146 The new counterpiece to "no-xxx" is "enable-xxx". 4147 4148 The patented RC5 and MDC2 algorithms will now be disabled unless 4149 "enable-rc5" and "enable-mdc2", respectively, are specified. 4150 4151 (IDEA remains enabled despite being patented. This is because IDEA 4152 is frequently required for interoperability, and there is no license 4153 fee for non-commercial use. As before, "no-idea" can be used to 4154 avoid this algorithm.) 4155 4156 [Bodo Moeller] 4157 4158 *) Add processing of proxy certificates (see RFC 3820). This work was 4159 sponsored by KTH (The Royal Institute of Technology in Stockholm) and 4160 EGEE (Enabling Grids for E-science in Europe). 4161 [Richard Levitte] 4162 4163 *) RC4 performance overhaul on modern architectures/implementations, such 4164 as Intel P4, IA-64 and AMD64. 4165 [Andy Polyakov] 4166 4167 *) New utility extract-section.pl. This can be used specify an alternative 4168 section number in a pod file instead of having to treat each file as 4169 a separate case in Makefile. This can be done by adding two lines to the 4170 pod file: 4171 4172 =for comment openssl_section:XXX 4173 4174 The blank line is mandatory. 4175 4176 [Steve Henson] 4177 4178 *) New arguments -certform, -keyform and -pass for s_client and s_server 4179 to allow alternative format key and certificate files and passphrase 4180 sources. 4181 [Steve Henson] 4182 4183 *) New structure X509_VERIFY_PARAM which combines current verify parameters, 4184 update associated structures and add various utility functions. 4185 4186 Add new policy related verify parameters, include policy checking in 4187 standard verify code. Enhance 'smime' application with extra parameters 4188 to support policy checking and print out. 4189 [Steve Henson] 4190 4191 *) Add a new engine to support VIA PadLock ACE extensions in the VIA C3 4192 Nehemiah processors. These extensions support AES encryption in hardware 4193 as well as RNG (though RNG support is currently disabled). 4194 [Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov] 4195 4196 *) Deprecate BN_[get|set]_params() functions (they were ignored internally). 4197 [Geoff Thorpe] 4198 4199 *) New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented. 4200 [Andy Polyakov and a number of other people] 4201 4202 *) Improved PowerPC platform support. Most notably BIGNUM assembler 4203 implementation contributed by IBM. 4204 [Suresh Chari, Peter Waltenberg, Andy Polyakov] 4205 4206 *) The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public 4207 exponent rather than 'unsigned long'. There is a corresponding change to 4208 the new 'rsa_keygen' element of the RSA_METHOD structure. 4209 [Jelte Jansen, Geoff Thorpe] 4210 4211 *) Functionality for creating the initial serial number file is now 4212 moved from CA.pl to the 'ca' utility with a new option -create_serial. 4213 4214 (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial 4215 number file to 1, which is bound to cause problems. To avoid 4216 the problems while respecting compatibility between different 0.9.7 4217 patchlevels, 0.9.7e employed 'openssl x509 -next_serial' in 4218 CA.pl for serial number initialization. With the new release 0.9.8, 4219 we can fix the problem directly in the 'ca' utility.) 4220 [Steve Henson] 4221 4222 *) Reduced header interdepencies by declaring more opaque objects in 4223 ossl_typ.h. As a consequence, including some headers (eg. engine.h) will 4224 give fewer recursive includes, which could break lazy source code - so 4225 this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always, 4226 developers should define this symbol when building and using openssl to 4227 ensure they track the recommended behaviour, interfaces, [etc], but 4228 backwards-compatible behaviour prevails when this isn't defined. 4229 [Geoff Thorpe] 4230 4231 *) New function X509_POLICY_NODE_print() which prints out policy nodes. 4232 [Steve Henson] 4233 4234 *) Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality. 4235 This will generate a random key of the appropriate length based on the 4236 cipher context. The EVP_CIPHER can provide its own random key generation 4237 routine to support keys of a specific form. This is used in the des and 4238 3des routines to generate a key of the correct parity. Update S/MIME 4239 code to use new functions and hence generate correct parity DES keys. 4240 Add EVP_CHECK_DES_KEY #define to return an error if the key is not 4241 valid (weak or incorrect parity). 4242 [Steve Henson] 4243 4244 *) Add a local set of CRLs that can be used by X509_verify_cert() as well 4245 as looking them up. This is useful when the verified structure may contain 4246 CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs 4247 present unless the new PKCS7_NO_CRL flag is asserted. 4248 [Steve Henson] 4249 4250 *) Extend ASN1 oid configuration module. It now additionally accepts the 4251 syntax: 4252 4253 shortName = some long name, 1.2.3.4 4254 [Steve Henson] 4255 4256 *) Reimplemented the BN_CTX implementation. There is now no more static 4257 limitation on the number of variables it can handle nor the depth of the 4258 "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack 4259 information can now expand as required, and rather than having a single 4260 static array of bignums, BN_CTX now uses a linked-list of such arrays 4261 allowing it to expand on demand whilst maintaining the usefulness of 4262 BN_CTX's "bundling". 4263 [Geoff Thorpe] 4264 4265 *) Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD 4266 to allow all RSA operations to function using a single BN_CTX. 4267 [Geoff Thorpe] 4268 4269 *) Preliminary support for certificate policy evaluation and checking. This 4270 is initially intended to pass the tests outlined in "Conformance Testing 4271 of Relying Party Client Certificate Path Processing Logic" v1.07. 4272 [Steve Henson] 4273 4274 *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and 4275 remained unused and not that useful. A variety of other little bignum 4276 tweaks and fixes have also been made continuing on from the audit (see 4277 below). 4278 [Geoff Thorpe] 4279 4280 *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with 4281 associated ASN1, EVP and SSL functions and old ASN1 macros. 4282 [Richard Levitte] 4283 4284 *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results, 4285 and this should never fail. So the return value from the use of 4286 BN_set_word() (which can fail due to needless expansion) is now deprecated; 4287 if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro. 4288 [Geoff Thorpe] 4289 4290 *) BN_CTX_get() should return zero-valued bignums, providing the same 4291 initialised value as BN_new(). 4292 [Geoff Thorpe, suggested by Ulf M��ller] 4293 4294 *) Support for inhibitAnyPolicy certificate extension. 4295 [Steve Henson] 4296 4297 *) An audit of the BIGNUM code is underway, for which debugging code is 4298 enabled when BN_DEBUG is defined. This makes stricter enforcements on what 4299 is considered valid when processing BIGNUMs, and causes execution to 4300 assert() when a problem is discovered. If BN_DEBUG_RAND is defined, 4301 further steps are taken to deliberately pollute unused data in BIGNUM 4302 structures to try and expose faulty code further on. For now, openssl will 4303 (in its default mode of operation) continue to tolerate the inconsistent 4304 forms that it has tolerated in the past, but authors and packagers should 4305 consider trying openssl and their own applications when compiled with 4306 these debugging symbols defined. It will help highlight potential bugs in 4307 their own code, and will improve the test coverage for OpenSSL itself. At 4308 some point, these tighter rules will become openssl's default to improve 4309 maintainability, though the assert()s and other overheads will remain only 4310 in debugging configurations. See bn.h for more details. 4311 [Geoff Thorpe, Nils Larsch, Ulf M��ller] 4312 4313 *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure 4314 that can only be obtained through BN_CTX_new() (which implicitly 4315 initialises it). The presence of this function only made it possible 4316 to overwrite an existing structure (and cause memory leaks). 4317 [Geoff Thorpe] 4318 4319 *) Because of the callback-based approach for implementing LHASH as a 4320 template type, lh_insert() adds opaque objects to hash-tables and 4321 lh_doall() or lh_doall_arg() are typically used with a destructor callback 4322 to clean up those corresponding objects before destroying the hash table 4323 (and losing the object pointers). So some over-zealous constifications in 4324 LHASH have been relaxed so that lh_insert() does not take (nor store) the 4325 objects as "const" and the lh_doall[_arg] callback wrappers are not 4326 prototyped to have "const" restrictions on the object pointers they are 4327 given (and so aren't required to cast them away any more). 4328 [Geoff Thorpe] 4329 4330 *) The tmdiff.h API was so ugly and minimal that our own timing utility 4331 (speed) prefers to use its own implementation. The two implementations 4332 haven't been consolidated as yet (volunteers?) but the tmdiff API has had 4333 its object type properly exposed (MS_TM) instead of casting to/from "char 4334 *". This may still change yet if someone realises MS_TM and "ms_time_***" 4335 aren't necessarily the greatest nomenclatures - but this is what was used 4336 internally to the implementation so I've used that for now. 4337 [Geoff Thorpe] 4338 4339 *) Ensure that deprecated functions do not get compiled when 4340 OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of 4341 the self-tests were still using deprecated key-generation functions so 4342 these have been updated also. 4343 [Geoff Thorpe] 4344 4345 *) Reorganise PKCS#7 code to separate the digest location functionality 4346 into PKCS7_find_digest(), digest addtion into PKCS7_bio_add_digest(). 4347 New function PKCS7_set_digest() to set the digest type for PKCS#7 4348 digestedData type. Add additional code to correctly generate the 4349 digestedData type and add support for this type in PKCS7 initialization 4350 functions. 4351 [Steve Henson] 4352 4353 *) New function PKCS7_set0_type_other() this initializes a PKCS7 4354 structure of type "other". 4355 [Steve Henson] 4356 4357 *) Fix prime generation loop in crypto/bn/bn_prime.pl by making 4358 sure the loop does correctly stop and breaking ("division by zero") 4359 modulus operations are not performed. The (pre-generated) prime 4360 table crypto/bn/bn_prime.h was already correct, but it could not be 4361 re-generated on some platforms because of the "division by zero" 4362 situation in the script. 4363 [Ralf S. Engelschall] 4364 4365 *) Update support for ECC-based TLS ciphersuites according to 4366 draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with 4367 SHA-1 now is only used for "small" curves (where the 4368 representation of a field element takes up to 24 bytes); for 4369 larger curves, the field element resulting from ECDH is directly 4370 used as premaster secret. 4371 [Douglas Stebila (Sun Microsystems Laboratories)] 4372 4373 *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2 4374 curve secp160r1 to the tests. 4375 [Douglas Stebila (Sun Microsystems Laboratories)] 4376 4377 *) Add the possibility to load symbols globally with DSO. 4378 [G��tz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte] 4379 4380 *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better 4381 control of the error stack. 4382 [Richard Levitte] 4383 4384 *) Add support for STORE in ENGINE. 4385 [Richard Levitte] 4386 4387 *) Add the STORE type. The intention is to provide a common interface 4388 to certificate and key stores, be they simple file-based stores, or 4389 HSM-type store, or LDAP stores, or... 4390 NOTE: The code is currently UNTESTED and isn't really used anywhere. 4391 [Richard Levitte] 4392 4393 *) Add a generic structure called OPENSSL_ITEM. This can be used to 4394 pass a list of arguments to any function as well as provide a way 4395 for a function to pass data back to the caller. 4396 [Richard Levitte] 4397 4398 *) Add the functions BUF_strndup() and BUF_memdup(). BUF_strndup() 4399 works like BUF_strdup() but can be used to duplicate a portion of 4400 a string. The copy gets NUL-terminated. BUF_memdup() duplicates 4401 a memory area. 4402 [Richard Levitte] 4403 4404 *) Add the function sk_find_ex() which works like sk_find(), but will 4405 return an index to an element even if an exact match couldn't be 4406 found. The index is guaranteed to point at the element where the 4407 searched-for key would be inserted to preserve sorting order. 4408 [Richard Levitte] 4409 4410 *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but 4411 takes an extra flags argument for optional functionality. Currently, 4412 the following flags are defined: 4413 4414 OBJ_BSEARCH_VALUE_ON_NOMATCH 4415 This one gets OBJ_bsearch_ex() to return a pointer to the first 4416 element where the comparing function returns a negative or zero 4417 number. 4418 4419 OBJ_BSEARCH_FIRST_VALUE_ON_MATCH 4420 This one gets OBJ_bsearch_ex() to return a pointer to the first 4421 element where the comparing function returns zero. This is useful 4422 if there are more than one element where the comparing function 4423 returns zero. 4424 [Richard Levitte] 4425 4426 *) Make it possible to create self-signed certificates with 'openssl ca' 4427 in such a way that the self-signed certificate becomes part of the 4428 CA database and uses the same mechanisms for serial number generation 4429 as all other certificate signing. The new flag '-selfsign' enables 4430 this functionality. Adapt CA.sh and CA.pl.in. 4431 [Richard Levitte] 4432 4433 *) Add functionality to check the public key of a certificate request 4434 against a given private. This is useful to check that a certificate 4435 request can be signed by that key (self-signing). 4436 [Richard Levitte] 4437 4438 *) Make it possible to have multiple active certificates with the same 4439 subject in the CA index file. This is done only if the keyword 4440 'unique_subject' is set to 'no' in the main CA section (default 4441 if 'CA_default') of the configuration file. The value is saved 4442 with the database itself in a separate index attribute file, 4443 named like the index file with '.attr' appended to the name. 4444 [Richard Levitte] 4445 4446 *) Generate muti valued AVAs using '+' notation in config files for 4447 req and dirName. 4448 [Steve Henson] 4449 4450 *) Support for nameConstraints certificate extension. 4451 [Steve Henson] 4452 4453 *) Support for policyConstraints certificate extension. 4454 [Steve Henson] 4455 4456 *) Support for policyMappings certificate extension. 4457 [Steve Henson] 4458 4459 *) Make sure the default DSA_METHOD implementation only uses its 4460 dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL, 4461 and change its own handlers to be NULL so as to remove unnecessary 4462 indirection. This lets alternative implementations fallback to the 4463 default implementation more easily. 4464 [Geoff Thorpe] 4465 4466 *) Support for directoryName in GeneralName related extensions 4467 in config files. 4468 [Steve Henson] 4469 4470 *) Make it possible to link applications using Makefile.shared. 4471 Make that possible even when linking against static libraries! 4472 [Richard Levitte] 4473 4474 *) Support for single pass processing for S/MIME signing. This now 4475 means that S/MIME signing can be done from a pipe, in addition 4476 cleartext signing (multipart/signed type) is effectively streaming 4477 and the signed data does not need to be all held in memory. 4478 4479 This is done with a new flag PKCS7_STREAM. When this flag is set 4480 PKCS7_sign() only initializes the PKCS7 structure and the actual signing 4481 is done after the data is output (and digests calculated) in 4482 SMIME_write_PKCS7(). 4483 [Steve Henson] 4484 4485 *) Add full support for -rpath/-R, both in shared libraries and 4486 applications, at least on the platforms where it's known how 4487 to do it. 4488 [Richard Levitte] 4489 4490 *) In crypto/ec/ec_mult.c, implement fast point multiplication with 4491 precomputation, based on wNAF splitting: EC_GROUP_precompute_mult() 4492 will now compute a table of multiples of the generator that 4493 makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul() 4494 faster (notably in the case of a single point multiplication, 4495 scalar * generator). 4496 [Nils Larsch, Bodo Moeller] 4497 4498 *) IPv6 support for certificate extensions. The various extensions 4499 which use the IP:a.b.c.d can now take IPv6 addresses using the 4500 formats of RFC1884 2.2 . IPv6 addresses are now also displayed 4501 correctly. 4502 [Steve Henson] 4503 4504 *) Added an ENGINE that implements RSA by performing private key 4505 exponentiations with the GMP library. The conversions to and from 4506 GMP's mpz_t format aren't optimised nor are any montgomery forms 4507 cached, and on x86 it appears OpenSSL's own performance has caught up. 4508 However there are likely to be other architectures where GMP could 4509 provide a boost. This ENGINE is not built in by default, but it can be 4510 specified at Configure time and should be accompanied by the necessary 4511 linker additions, eg; 4512 ./config -DOPENSSL_USE_GMP -lgmp 4513 [Geoff Thorpe] 4514 4515 *) "openssl engine" will not display ENGINE/DSO load failure errors when 4516 testing availability of engines with "-t" - the old behaviour is 4517 produced by increasing the feature's verbosity with "-tt". 4518 [Geoff Thorpe] 4519 4520 *) ECDSA routines: under certain error conditions uninitialized BN objects 4521 could be freed. Solution: make sure initialization is performed early 4522 enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de> 4523 via PR#459) 4524 [Lutz Jaenicke] 4525 4526 *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD 4527 and DH_METHOD (eg. by ENGINE implementations) to override the normal 4528 software implementations. For DSA and DH, parameter generation can 4529 also be overriden by providing the appropriate method callbacks. 4530 [Geoff Thorpe] 4531 4532 *) Change the "progress" mechanism used in key-generation and 4533 primality testing to functions that take a new BN_GENCB pointer in 4534 place of callback/argument pairs. The new API functions have "_ex" 4535 postfixes and the older functions are reimplemented as wrappers for 4536 the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide 4537 declarations of the old functions to help (graceful) attempts to 4538 migrate to the new functions. Also, the new key-generation API 4539 functions operate on a caller-supplied key-structure and return 4540 success/failure rather than returning a key or NULL - this is to 4541 help make "keygen" another member function of RSA_METHOD etc. 4542 4543 Example for using the new callback interface: 4544 4545 int (*my_callback)(int a, int b, BN_GENCB *cb) = ...; 4546 void *my_arg = ...; 4547 BN_GENCB my_cb; 4548 4549 BN_GENCB_set(&my_cb, my_callback, my_arg); 4550 4551 return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb); 4552 /* For the meaning of a, b in calls to my_callback(), see the 4553 * documentation of the function that calls the callback. 4554 * cb will point to my_cb; my_arg can be retrieved as cb->arg. 4555 * my_callback should return 1 if it wants BN_is_prime_ex() 4556 * to continue, or 0 to stop. 4557 */ 4558 4559 [Geoff Thorpe] 4560 4561 *) Change the ZLIB compression method to be stateful, and make it 4562 available to TLS with the number defined in 4563 draft-ietf-tls-compression-04.txt. 4564 [Richard Levitte] 4565 4566 *) Add the ASN.1 structures and functions for CertificatePair, which 4567 is defined as follows (according to X.509_4thEditionDraftV6.pdf): 4568 4569 CertificatePair ::= SEQUENCE { 4570 forward [0] Certificate OPTIONAL, 4571 reverse [1] Certificate OPTIONAL, 4572 -- at least one of the pair shall be present -- } 4573 4574 Also implement the PEM functions to read and write certificate 4575 pairs, and defined the PEM tag as "CERTIFICATE PAIR". 4576 4577 This needed to be defined, mostly for the sake of the LDAP 4578 attribute crossCertificatePair, but may prove useful elsewhere as 4579 well. 4580 [Richard Levitte] 4581 4582 *) Make it possible to inhibit symlinking of shared libraries in 4583 Makefile.shared, for Cygwin's sake. 4584 [Richard Levitte] 4585 4586 *) Extend the BIGNUM API by creating a function 4587 void BN_set_negative(BIGNUM *a, int neg); 4588 and a macro that behave like 4589 int BN_is_negative(const BIGNUM *a); 4590 4591 to avoid the need to access 'a->neg' directly in applications. 4592 [Nils Larsch] 4593 4594 *) Implement fast modular reduction for pseudo-Mersenne primes 4595 used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c). 4596 EC_GROUP_new_curve_GFp() will now automatically use this 4597 if applicable. 4598 [Nils Larsch <nla@trustcenter.de>] 4599 4600 *) Add new lock type (CRYPTO_LOCK_BN). 4601 [Bodo Moeller] 4602 4603 *) Change the ENGINE framework to automatically load engines 4604 dynamically from specific directories unless they could be 4605 found to already be built in or loaded. Move all the 4606 current engines except for the cryptodev one to a new 4607 directory engines/. 4608 The engines in engines/ are built as shared libraries if 4609 the "shared" options was given to ./Configure or ./config. 4610 Otherwise, they are inserted in libcrypto.a. 4611 /usr/local/ssl/engines is the default directory for dynamic 4612 engines, but that can be overriden at configure time through 4613 the usual use of --prefix and/or --openssldir, and at run 4614 time with the environment variable OPENSSL_ENGINES. 4615 [Geoff Thorpe and Richard Levitte] 4616 4617 *) Add Makefile.shared, a helper makefile to build shared 4618 libraries. Addapt Makefile.org. 4619 [Richard Levitte] 4620 4621 *) Add version info to Win32 DLLs. 4622 [Peter 'Luna' Runestig" <peter@runestig.com>] 4623 4624 *) Add new 'medium level' PKCS#12 API. Certificates and keys 4625 can be added using this API to created arbitrary PKCS#12 4626 files while avoiding the low level API. 4627 4628 New options to PKCS12_create(), key or cert can be NULL and 4629 will then be omitted from the output file. The encryption 4630 algorithm NIDs can be set to -1 for no encryption, the mac 4631 iteration count can be set to 0 to omit the mac. 4632 4633 Enhance pkcs12 utility by making the -nokeys and -nocerts 4634 options work when creating a PKCS#12 file. New option -nomac 4635 to omit the mac, NONE can be set for an encryption algorithm. 4636 New code is modified to use the enhanced PKCS12_create() 4637 instead of the low level API. 4638 [Steve Henson] 4639 4640 *) Extend ASN1 encoder to support indefinite length constructed 4641 encoding. This can output sequences tags and octet strings in 4642 this form. Modify pk7_asn1.c to support indefinite length 4643 encoding. This is experimental and needs additional code to 4644 be useful, such as an ASN1 bio and some enhanced streaming 4645 PKCS#7 code. 4646 4647 Extend template encode functionality so that tagging is passed 4648 down to the template encoder. 4649 [Steve Henson] 4650 4651 *) Let 'openssl req' fail if an argument to '-newkey' is not 4652 recognized instead of using RSA as a default. 4653 [Bodo Moeller] 4654 4655 *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt. 4656 As these are not official, they are not included in "ALL"; 4657 the "ECCdraft" ciphersuite group alias can be used to select them. 4658 [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)] 4659 4660 *) Add ECDH engine support. 4661 [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)] 4662 4663 *) Add ECDH in new directory crypto/ecdh/. 4664 [Douglas Stebila (Sun Microsystems Laboratories)] 4665 4666 *) Let BN_rand_range() abort with an error after 100 iterations 4667 without success (which indicates a broken PRNG). 4668 [Bodo Moeller] 4669 4670 *) Change BN_mod_sqrt() so that it verifies that the input value 4671 is really the square of the return value. (Previously, 4672 BN_mod_sqrt would show GIGO behaviour.) 4673 [Bodo Moeller] 4674 4675 *) Add named elliptic curves over binary fields from X9.62, SECG, 4676 and WAP/WTLS; add OIDs that were still missing. 4677 4678 [Sheueling Chang Shantz and Douglas Stebila 4679 (Sun Microsystems Laboratories)] 4680 4681 *) Extend the EC library for elliptic curves over binary fields 4682 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/). 4683 New EC_METHOD: 4684 4685 EC_GF2m_simple_method 4686 4687 New API functions: 4688 4689 EC_GROUP_new_curve_GF2m 4690 EC_GROUP_set_curve_GF2m 4691 EC_GROUP_get_curve_GF2m 4692 EC_POINT_set_affine_coordinates_GF2m 4693 EC_POINT_get_affine_coordinates_GF2m 4694 EC_POINT_set_compressed_coordinates_GF2m 4695 4696 Point compression for binary fields is disabled by default for 4697 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to 4698 enable it). 4699 4700 As binary polynomials are represented as BIGNUMs, various members 4701 of the EC_GROUP and EC_POINT data structures can be shared 4702 between the implementations for prime fields and binary fields; 4703 the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m) 4704 are essentially identical to their ..._GFp counterparts. 4705 (For simplicity, the '..._GFp' prefix has been dropped from 4706 various internal method names.) 4707 4708 An internal 'field_div' method (similar to 'field_mul' and 4709 'field_sqr') has been added; this is used only for binary fields. 4710 4711 [Sheueling Chang Shantz and Douglas Stebila 4712 (Sun Microsystems Laboratories)] 4713 4714 *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult() 4715 through methods ('mul', 'precompute_mult'). 4716 4717 The generic implementations (now internally called 'ec_wNAF_mul' 4718 and 'ec_wNAF_precomputed_mult') remain the default if these 4719 methods are undefined. 4720 4721 [Sheueling Chang Shantz and Douglas Stebila 4722 (Sun Microsystems Laboratories)] 4723 4724 *) New function EC_GROUP_get_degree, which is defined through 4725 EC_METHOD. For curves over prime fields, this returns the bit 4726 length of the modulus. 4727 4728 [Sheueling Chang Shantz and Douglas Stebila 4729 (Sun Microsystems Laboratories)] 4730 4731 *) New functions EC_GROUP_dup, EC_POINT_dup. 4732 (These simply call ..._new and ..._copy). 4733 4734 [Sheueling Chang Shantz and Douglas Stebila 4735 (Sun Microsystems Laboratories)] 4736 4737 *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c. 4738 Polynomials are represented as BIGNUMs (where the sign bit is not 4739 used) in the following functions [macros]: 4740 4741 BN_GF2m_add 4742 BN_GF2m_sub [= BN_GF2m_add] 4743 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr] 4744 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr] 4745 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr] 4746 BN_GF2m_mod_inv 4747 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr] 4748 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr] 4749 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr] 4750 BN_GF2m_cmp [= BN_ucmp] 4751 4752 (Note that only the 'mod' functions are actually for fields GF(2^m). 4753 BN_GF2m_add() is misnomer, but this is for the sake of consistency.) 4754 4755 For some functions, an the irreducible polynomial defining a 4756 field can be given as an 'unsigned int[]' with strictly 4757 decreasing elements giving the indices of those bits that are set; 4758 i.e., p[] represents the polynomial 4759 f(t) = t^p[0] + t^p[1] + ... + t^p[k] 4760 where 4761 p[0] > p[1] > ... > p[k] = 0. 4762 This applies to the following functions: 4763 4764 BN_GF2m_mod_arr 4765 BN_GF2m_mod_mul_arr 4766 BN_GF2m_mod_sqr_arr 4767 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv] 4768 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div] 4769 BN_GF2m_mod_exp_arr 4770 BN_GF2m_mod_sqrt_arr 4771 BN_GF2m_mod_solve_quad_arr 4772 BN_GF2m_poly2arr 4773 BN_GF2m_arr2poly 4774 4775 Conversion can be performed by the following functions: 4776 4777 BN_GF2m_poly2arr 4778 BN_GF2m_arr2poly 4779 4780 bntest.c has additional tests for binary polynomial arithmetic. 4781 4782 Two implementations for BN_GF2m_mod_div() are available. 4783 The default algorithm simply uses BN_GF2m_mod_inv() and 4784 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only 4785 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the 4786 copyright notice in crypto/bn/bn_gf2m.c before enabling it). 4787 4788 [Sheueling Chang Shantz and Douglas Stebila 4789 (Sun Microsystems Laboratories)] 4790 4791 *) Add new error code 'ERR_R_DISABLED' that can be used when some 4792 functionality is disabled at compile-time. 4793 [Douglas Stebila <douglas.stebila@sun.com>] 4794 4795 *) Change default behaviour of 'openssl asn1parse' so that more 4796 information is visible when viewing, e.g., a certificate: 4797 4798 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump' 4799 mode the content of non-printable OCTET STRINGs is output in a 4800 style similar to INTEGERs, but with '[HEX DUMP]' prepended to 4801 avoid the appearance of a printable string. 4802 [Nils Larsch <nla@trustcenter.de>] 4803 4804 *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access 4805 functions 4806 EC_GROUP_set_asn1_flag() 4807 EC_GROUP_get_asn1_flag() 4808 EC_GROUP_set_point_conversion_form() 4809 EC_GROUP_get_point_conversion_form() 4810 These control ASN1 encoding details: 4811 - Curves (i.e., groups) are encoded explicitly unless asn1_flag 4812 has been set to OPENSSL_EC_NAMED_CURVE. 4813 - Points are encoded in uncompressed form by default; options for 4814 asn1_for are as for point2oct, namely 4815 POINT_CONVERSION_COMPRESSED 4816 POINT_CONVERSION_UNCOMPRESSED 4817 POINT_CONVERSION_HYBRID 4818 4819 Also add 'seed' and 'seed_len' members to EC_GROUP with access 4820 functions 4821 EC_GROUP_set_seed() 4822 EC_GROUP_get0_seed() 4823 EC_GROUP_get_seed_len() 4824 This is used only for ASN1 purposes (so far). 4825 [Nils Larsch <nla@trustcenter.de>] 4826 4827 *) Add 'field_type' member to EC_METHOD, which holds the NID 4828 of the appropriate field type OID. The new function 4829 EC_METHOD_get_field_type() returns this value. 4830 [Nils Larsch <nla@trustcenter.de>] 4831 4832 *) Add functions 4833 EC_POINT_point2bn() 4834 EC_POINT_bn2point() 4835 EC_POINT_point2hex() 4836 EC_POINT_hex2point() 4837 providing useful interfaces to EC_POINT_point2oct() and 4838 EC_POINT_oct2point(). 4839 [Nils Larsch <nla@trustcenter.de>] 4840 4841 *) Change internals of the EC library so that the functions 4842 EC_GROUP_set_generator() 4843 EC_GROUP_get_generator() 4844 EC_GROUP_get_order() 4845 EC_GROUP_get_cofactor() 4846 are implemented directly in crypto/ec/ec_lib.c and not dispatched 4847 to methods, which would lead to unnecessary code duplication when 4848 adding different types of curves. 4849 [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller] 4850 4851 *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM 4852 arithmetic, and such that modified wNAFs are generated 4853 (which avoid length expansion in many cases). 4854 [Bodo Moeller] 4855 4856 *) Add a function EC_GROUP_check_discriminant() (defined via 4857 EC_METHOD) that verifies that the curve discriminant is non-zero. 4858 4859 Add a function EC_GROUP_check() that makes some sanity tests 4860 on a EC_GROUP, its generator and order. This includes 4861 EC_GROUP_check_discriminant(). 4862 [Nils Larsch <nla@trustcenter.de>] 4863 4864 *) Add ECDSA in new directory crypto/ecdsa/. 4865 4866 Add applications 'openssl ecparam' and 'openssl ecdsa' 4867 (these are based on 'openssl dsaparam' and 'openssl dsa'). 4868 4869 ECDSA support is also included in various other files across the 4870 library. Most notably, 4871 - 'openssl req' now has a '-newkey ecdsa:file' option; 4872 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA; 4873 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and 4874 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make 4875 them suitable for ECDSA where domain parameters must be 4876 extracted before the specific public key; 4877 - ECDSA engine support has been added. 4878 [Nils Larsch <nla@trustcenter.de>] 4879 4880 *) Include some named elliptic curves, and add OIDs from X9.62, 4881 SECG, and WAP/WTLS. Each curve can be obtained from the new 4882 function 4883 EC_GROUP_new_by_curve_name(), 4884 and the list of available named curves can be obtained with 4885 EC_get_builtin_curves(). 4886 Also add a 'curve_name' member to EC_GROUP objects, which can be 4887 accessed via 4888 EC_GROUP_set_curve_name() 4889 EC_GROUP_get_curve_name() 4890 [Nils Larsch <larsch@trustcenter.de, Bodo Moeller] 4891 4892 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 4893 was actually never needed) and in BN_mul(). The removal in BN_mul() 4894 required a small change in bn_mul_part_recursive() and the addition 4895 of the functions bn_cmp_part_words(), bn_sub_part_words() and 4896 bn_add_part_words(), which do the same thing as bn_cmp_words(), 4897 bn_sub_words() and bn_add_words() except they take arrays with 4898 differing sizes. 4899 [Richard Levitte] 4900 4901 Changes between 0.9.7l and 0.9.7m [23 Feb 2007] 4902 4903 *) Cleanse PEM buffers before freeing them since they may contain 4904 sensitive data. 4905 [Benjamin Bennett <ben@psc.edu>] 4906 4907 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 4908 a ciphersuite string such as "DEFAULT:RSA" cannot enable 4909 authentication-only ciphersuites. 4910 [Bodo Moeller] 4911 4912 *) Since AES128 and AES256 share a single mask bit in the logic of 4913 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 4914 kludge to work properly if AES128 is available and AES256 isn't. 4915 [Victor Duchovni] 4916 4917 *) Expand security boundary to match 1.1.1 module. 4918 [Steve Henson] 4919 4920 *) Remove redundant features: hash file source, editing of test vectors 4921 modify fipsld to use external fips_premain.c signature. 4922 [Steve Henson] 4923 4924 *) New perl script mkfipsscr.pl to create shell scripts or batch files to 4925 run algorithm test programs. 4926 [Steve Henson] 4927 4928 *) Make algorithm test programs more tolerant of whitespace. 4929 [Steve Henson] 4930 4931 *) Have SSL/TLS server implementation tolerate "mismatched" record 4932 protocol version while receiving ClientHello even if the 4933 ClientHello is fragmented. (The server can't insist on the 4934 particular protocol version it has chosen before the ServerHello 4935 message has informed the client about his choice.) 4936 [Bodo Moeller] 4937 4938 *) Load error codes if they are not already present instead of using a 4939 static variable. This allows them to be cleanly unloaded and reloaded. 4940 [Steve Henson] 4941 4942 Changes between 0.9.7k and 0.9.7l [28 Sep 2006] 4943 4944 *) Introduce limits to prevent malicious keys being able to 4945 cause a denial of service. (CVE-2006-2940) 4946 [Steve Henson, Bodo Moeller] 4947 4948 *) Fix ASN.1 parsing of certain invalid structures that can result 4949 in a denial of service. (CVE-2006-2937) [Steve Henson] 4950 4951 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 4952 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 4953 4954 *) Fix SSL client code which could crash if connecting to a 4955 malicious SSLv2 server. (CVE-2006-4343) 4956 [Tavis Ormandy and Will Drewry, Google Security Team] 4957 4958 *) Change ciphersuite string processing so that an explicit 4959 ciphersuite selects this one ciphersuite (so that "AES256-SHA" 4960 will no longer include "AES128-SHA"), and any other similar 4961 ciphersuite (same bitmap) from *other* protocol versions (so that 4962 "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the 4963 SSL 3.0/TLS 1.0 ciphersuite). This is a backport combining 4964 changes from 0.9.8b and 0.9.8d. 4965 [Bodo Moeller] 4966 4967 Changes between 0.9.7j and 0.9.7k [05 Sep 2006] 4968 4969 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 4970 (CVE-2006-4339) [Ben Laurie and Google Security Team] 4971 4972 *) Change the Unix randomness entropy gathering to use poll() when 4973 possible instead of select(), since the latter has some 4974 undesirable limitations. 4975 [Darryl Miles via Richard Levitte and Bodo Moeller] 4976 4977 *) Disable rogue ciphersuites: 4978 4979 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 4980 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 4981 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 4982 4983 The latter two were purportedly from 4984 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 4985 appear there. 4986 4987 Also deactive the remaining ciphersuites from 4988 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 4989 unofficial, and the ID has long expired. 4990 [Bodo Moeller] 4991 4992 *) Fix RSA blinding Heisenbug (problems sometimes occured on 4993 dual-core machines) and other potential thread-safety issues. 4994 [Bodo Moeller] 4995 4996 Changes between 0.9.7i and 0.9.7j [04 May 2006] 4997 4998 *) Adapt fipsld and the build system to link against the validated FIPS 4999 module in FIPS mode. 5000 [Steve Henson] 5001 5002 *) Fixes for VC++ 2005 build under Windows. 5003 [Steve Henson] 5004 5005 *) Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make 5006 from a Windows bash shell such as MSYS. It is autodetected from the 5007 "config" script when run from a VC++ environment. Modify standard VC++ 5008 build to use fipscanister.o from the GNU make build. 5009 [Steve Henson] 5010 5011 Changes between 0.9.7h and 0.9.7i [14 Oct 2005] 5012 5013 *) Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS. 5014 The value now differs depending on if you build for FIPS or not. 5015 BEWARE! A program linked with a shared FIPSed libcrypto can't be 5016 safely run with a non-FIPSed libcrypto, as it may crash because of 5017 the difference induced by this change. 5018 [Andy Polyakov] 5019 5020 Changes between 0.9.7g and 0.9.7h [11 Oct 2005] 5021 5022 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 5023 (part of SSL_OP_ALL). This option used to disable the 5024 countermeasure against man-in-the-middle protocol-version 5025 rollback in the SSL 2.0 server implementation, which is a bad 5026 idea. (CVE-2005-2969) 5027 5028 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 5029 for Information Security, National Institute of Advanced Industrial 5030 Science and Technology [AIST], Japan)] 5031 5032 *) Minimal support for X9.31 signatures and PSS padding modes. This is 5033 mainly for FIPS compliance and not fully integrated at this stage. 5034 [Steve Henson] 5035 5036 *) For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform 5037 the exponentiation using a fixed-length exponent. (Otherwise, 5038 the information leaked through timing could expose the secret key 5039 after many signatures; cf. Bleichenbacher's attack on DSA with 5040 biased k.) 5041 [Bodo Moeller] 5042 5043 *) Make a new fixed-window mod_exp implementation the default for 5044 RSA, DSA, and DH private-key operations so that the sequence of 5045 squares and multiplies and the memory access pattern are 5046 independent of the particular secret key. This will mitigate 5047 cache-timing and potential related attacks. 5048 5049 BN_mod_exp_mont_consttime() is the new exponentiation implementation, 5050 and this is automatically used by BN_mod_exp_mont() if the new flag 5051 BN_FLG_EXP_CONSTTIME is set for the exponent. RSA, DSA, and DH 5052 will use this BN flag for private exponents unless the flag 5053 RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or 5054 DH_FLAG_NO_EXP_CONSTTIME, respectively, is set. 5055 5056 [Matthew D Wood (Intel Corp), with some changes by Bodo Moeller] 5057 5058 *) Change the client implementation for SSLv23_method() and 5059 SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0 5060 Client Hello message format if the SSL_OP_NO_SSLv2 option is set. 5061 (Previously, the SSL 2.0 backwards compatible Client Hello 5062 message format would be used even with SSL_OP_NO_SSLv2.) 5063 [Bodo Moeller] 5064 5065 *) Add support for smime-type MIME parameter in S/MIME messages which some 5066 clients need. 5067 [Steve Henson] 5068 5069 *) New function BN_MONT_CTX_set_locked() to set montgomery parameters in 5070 a threadsafe manner. Modify rsa code to use new function and add calls 5071 to dsa and dh code (which had race conditions before). 5072 [Steve Henson] 5073 5074 *) Include the fixed error library code in the C error file definitions 5075 instead of fixing them up at runtime. This keeps the error code 5076 structures constant. 5077 [Steve Henson] 5078 5079 Changes between 0.9.7f and 0.9.7g [11 Apr 2005] 5080 5081 [NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after 5082 OpenSSL 0.9.8.] 5083 5084 *) Fixes for newer kerberos headers. NB: the casts are needed because 5085 the 'length' field is signed on one version and unsigned on another 5086 with no (?) obvious way to tell the difference, without these VC++ 5087 complains. Also the "definition" of FAR (blank) is no longer included 5088 nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up 5089 some needed definitions. 5090 [Steve Henson] 5091 5092 *) Undo Cygwin change. 5093 [Ulf M��ller] 5094 5095 *) Added support for proxy certificates according to RFC 3820. 5096 Because they may be a security thread to unaware applications, 5097 they must be explicitely allowed in run-time. See 5098 docs/HOWTO/proxy_certificates.txt for further information. 5099 [Richard Levitte] 5100 5101 Changes between 0.9.7e and 0.9.7f [22 Mar 2005] 5102 5103 *) Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating 5104 server and client random values. Previously 5105 (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in 5106 less random data when sizeof(time_t) > 4 (some 64 bit platforms). 5107 5108 This change has negligible security impact because: 5109 5110 1. Server and client random values still have 24 bytes of pseudo random 5111 data. 5112 5113 2. Server and client random values are sent in the clear in the initial 5114 handshake. 5115 5116 3. The master secret is derived using the premaster secret (48 bytes in 5117 size for static RSA ciphersuites) as well as client server and random 5118 values. 5119 5120 The OpenSSL team would like to thank the UK NISCC for bringing this issue 5121 to our attention. 5122 5123 [Stephen Henson, reported by UK NISCC] 5124 5125 *) Use Windows randomness collection on Cygwin. 5126 [Ulf M��ller] 5127 5128 *) Fix hang in EGD/PRNGD query when communication socket is closed 5129 prematurely by EGD/PRNGD. 5130 [Darren Tucker <dtucker@zip.com.au> via Lutz J��nicke, resolves #1014] 5131 5132 *) Prompt for pass phrases when appropriate for PKCS12 input format. 5133 [Steve Henson] 5134 5135 *) Back-port of selected performance improvements from development 5136 branch, as well as improved support for PowerPC platforms. 5137 [Andy Polyakov] 5138 5139 *) Add lots of checks for memory allocation failure, error codes to indicate 5140 failure and freeing up memory if a failure occurs. 5141 [Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson] 5142 5143 *) Add new -passin argument to dgst. 5144 [Steve Henson] 5145 5146 *) Perform some character comparisons of different types in X509_NAME_cmp: 5147 this is needed for some certificates that reencode DNs into UTF8Strings 5148 (in violation of RFC3280) and can't or wont issue name rollover 5149 certificates. 5150 [Steve Henson] 5151 5152 *) Make an explicit check during certificate validation to see that 5153 the CA setting in each certificate on the chain is correct. As a 5154 side effect always do the following basic checks on extensions, 5155 not just when there's an associated purpose to the check: 5156 5157 - if there is an unhandled critical extension (unless the user 5158 has chosen to ignore this fault) 5159 - if the path length has been exceeded (if one is set at all) 5160 - that certain extensions fit the associated purpose (if one has 5161 been given) 5162 [Richard Levitte] 5163 5164 Changes between 0.9.7d and 0.9.7e [25 Oct 2004] 5165 5166 *) Avoid a race condition when CRLs are checked in a multi threaded 5167 environment. This would happen due to the reordering of the revoked 5168 entries during signature checking and serial number lookup. Now the 5169 encoding is cached and the serial number sort performed under a lock. 5170 Add new STACK function sk_is_sorted(). 5171 [Steve Henson] 5172 5173 *) Add Delta CRL to the extension code. 5174 [Steve Henson] 5175 5176 *) Various fixes to s3_pkt.c so alerts are sent properly. 5177 [David Holmes <d.holmes@f5.com>] 5178 5179 *) Reduce the chances of duplicate issuer name and serial numbers (in 5180 violation of RFC3280) using the OpenSSL certificate creation utilities. 5181 This is done by creating a random 64 bit value for the initial serial 5182 number when a serial number file is created or when a self signed 5183 certificate is created using 'openssl req -x509'. The initial serial 5184 number file is created using 'openssl x509 -next_serial' in CA.pl 5185 rather than being initialized to 1. 5186 [Steve Henson] 5187 5188 Changes between 0.9.7c and 0.9.7d [17 Mar 2004] 5189 5190 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 5191 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 5192 [Joe Orton, Steve Henson] 5193 5194 *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites 5195 (CVE-2004-0112) 5196 [Joe Orton, Steve Henson] 5197 5198 *) Make it possible to have multiple active certificates with the same 5199 subject in the CA index file. This is done only if the keyword 5200 'unique_subject' is set to 'no' in the main CA section (default 5201 if 'CA_default') of the configuration file. The value is saved 5202 with the database itself in a separate index attribute file, 5203 named like the index file with '.attr' appended to the name. 5204 [Richard Levitte] 5205 5206 *) X509 verify fixes. Disable broken certificate workarounds when 5207 X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if 5208 keyUsage extension present. Don't accept CRLs with unhandled critical 5209 extensions: since verify currently doesn't process CRL extensions this 5210 rejects a CRL with *any* critical extensions. Add new verify error codes 5211 for these cases. 5212 [Steve Henson] 5213 5214 *) When creating an OCSP nonce use an OCTET STRING inside the extnValue. 5215 A clarification of RFC2560 will require the use of OCTET STRINGs and 5216 some implementations cannot handle the current raw format. Since OpenSSL 5217 copies and compares OCSP nonces as opaque blobs without any attempt at 5218 parsing them this should not create any compatibility issues. 5219 [Steve Henson] 5220 5221 *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when 5222 calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without 5223 this HMAC (and other) operations are several times slower than OpenSSL 5224 < 0.9.7. 5225 [Steve Henson] 5226 5227 *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex(). 5228 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 5229 5230 *) Use the correct content when signing type "other". 5231 [Steve Henson] 5232 5233 Changes between 0.9.7b and 0.9.7c [30 Sep 2003] 5234 5235 *) Fix various bugs revealed by running the NISCC test suite: 5236 5237 Stop out of bounds reads in the ASN1 code when presented with 5238 invalid tags (CVE-2003-0543 and CVE-2003-0544). 5239 5240 Free up ASN1_TYPE correctly if ANY type is invalid (CVE-2003-0545). 5241 5242 If verify callback ignores invalid public key errors don't try to check 5243 certificate signature with the NULL public key. 5244 5245 [Steve Henson] 5246 5247 *) New -ignore_err option in ocsp application to stop the server 5248 exiting on the first error in a request. 5249 [Steve Henson] 5250 5251 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 5252 if the server requested one: as stated in TLS 1.0 and SSL 3.0 5253 specifications. 5254 [Steve Henson] 5255 5256 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 5257 extra data after the compression methods not only for TLS 1.0 5258 but also for SSL 3.0 (as required by the specification). 5259 [Bodo Moeller; problem pointed out by Matthias Loepfe] 5260 5261 *) Change X509_certificate_type() to mark the key as exported/exportable 5262 when it's 512 *bits* long, not 512 bytes. 5263 [Richard Levitte] 5264 5265 *) Change AES_cbc_encrypt() so it outputs exact multiple of 5266 blocks during encryption. 5267 [Richard Levitte] 5268 5269 *) Various fixes to base64 BIO and non blocking I/O. On write 5270 flushes were not handled properly if the BIO retried. On read 5271 data was not being buffered properly and had various logic bugs. 5272 This also affects blocking I/O when the data being decoded is a 5273 certain size. 5274 [Steve Henson] 5275 5276 *) Various S/MIME bugfixes and compatibility changes: 5277 output correct application/pkcs7 MIME type if 5278 PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures. 5279 Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening 5280 of files as .eml work). Correctly handle very long lines in MIME 5281 parser. 5282 [Steve Henson] 5283 5284 Changes between 0.9.7a and 0.9.7b [10 Apr 2003] 5285 5286 *) Countermeasure against the Klima-Pokorny-Rosa extension of 5287 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 5288 a protocol version number mismatch like a decryption error 5289 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 5290 [Bodo Moeller] 5291 5292 *) Turn on RSA blinding by default in the default implementation 5293 to avoid a timing attack. Applications that don't want it can call 5294 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 5295 They would be ill-advised to do so in most cases. 5296 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 5297 5298 *) Change RSA blinding code so that it works when the PRNG is not 5299 seeded (in this case, the secret RSA exponent is abused as 5300 an unpredictable seed -- if it is not unpredictable, there 5301 is no point in blinding anyway). Make RSA blinding thread-safe 5302 by remembering the creator's thread ID in rsa->blinding and 5303 having all other threads use local one-time blinding factors 5304 (this requires more computation than sharing rsa->blinding, but 5305 avoids excessive locking; and if an RSA object is not shared 5306 between threads, blinding will still be very fast). 5307 [Bodo Moeller] 5308 5309 *) Fixed a typo bug that would cause ENGINE_set_default() to set an 5310 ENGINE as defaults for all supported algorithms irrespective of 5311 the 'flags' parameter. 'flags' is now honoured, so applications 5312 should make sure they are passing it correctly. 5313 [Geoff Thorpe] 5314 5315 *) Target "mingw" now allows native Windows code to be generated in 5316 the Cygwin environment as well as with the MinGW compiler. 5317 [Ulf Moeller] 5318 5319 Changes between 0.9.7 and 0.9.7a [19 Feb 2003] 5320 5321 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 5322 via timing by performing a MAC computation even if incorrrect 5323 block cipher padding has been found. This is a countermeasure 5324 against active attacks where the attacker has to distinguish 5325 between bad padding and a MAC verification error. (CVE-2003-0078) 5326 5327 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 5328 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 5329 Martin Vuagnoux (EPFL, Ilion)] 5330 5331 *) Make the no-err option work as intended. The intention with no-err 5332 is not to have the whole error stack handling routines removed from 5333 libcrypto, it's only intended to remove all the function name and 5334 reason texts, thereby removing some of the footprint that may not 5335 be interesting if those errors aren't displayed anyway. 5336 5337 NOTE: it's still possible for any application or module to have it's 5338 own set of error texts inserted. The routines are there, just not 5339 used by default when no-err is given. 5340 [Richard Levitte] 5341 5342 *) Add support for FreeBSD on IA64. 5343 [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454] 5344 5345 *) Adjust DES_cbc_cksum() so it returns the same value as the MIT 5346 Kerberos function mit_des_cbc_cksum(). Before this change, 5347 the value returned by DES_cbc_cksum() was like the one from 5348 mit_des_cbc_cksum(), except the bytes were swapped. 5349 [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte] 5350 5351 *) Allow an application to disable the automatic SSL chain building. 5352 Before this a rather primitive chain build was always performed in 5353 ssl3_output_cert_chain(): an application had no way to send the 5354 correct chain if the automatic operation produced an incorrect result. 5355 5356 Now the chain builder is disabled if either: 5357 5358 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert(). 5359 5360 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set. 5361 5362 The reasoning behind this is that an application would not want the 5363 auto chain building to take place if extra chain certificates are 5364 present and it might also want a means of sending no additional 5365 certificates (for example the chain has two certificates and the 5366 root is omitted). 5367 [Steve Henson] 5368 5369 *) Add the possibility to build without the ENGINE framework. 5370 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 5371 5372 *) Under Win32 gmtime() can return NULL: check return value in 5373 OPENSSL_gmtime(). Add error code for case where gmtime() fails. 5374 [Steve Henson] 5375 5376 *) DSA routines: under certain error conditions uninitialized BN objects 5377 could be freed. Solution: make sure initialization is performed early 5378 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>, 5379 Nils Larsch <nla@trustcenter.de> via PR#459) 5380 [Lutz Jaenicke] 5381 5382 *) Another fix for SSLv2 session ID handling: the session ID was incorrectly 5383 checked on reconnect on the client side, therefore session resumption 5384 could still fail with a "ssl session id is different" error. This 5385 behaviour is masked when SSL_OP_ALL is used due to 5386 SSL_OP_MICROSOFT_SESS_ID_BUG being set. 5387 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 5388 followup to PR #377. 5389 [Lutz Jaenicke] 5390 5391 *) IA-32 assembler support enhancements: unified ELF targets, support 5392 for SCO/Caldera platforms, fix for Cygwin shared build. 5393 [Andy Polyakov] 5394 5395 *) Add support for FreeBSD on sparc64. As a consequence, support for 5396 FreeBSD on non-x86 processors is separate from x86 processors on 5397 the config script, much like the NetBSD support. 5398 [Richard Levitte & Kris Kennaway <kris@obsecurity.org>] 5399 5400 Changes between 0.9.6h and 0.9.7 [31 Dec 2002] 5401 5402 [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after 5403 OpenSSL 0.9.7.] 5404 5405 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED 5406 code (06) was taken as the first octet of the session ID and the last 5407 octet was ignored consequently. As a result SSLv2 client side session 5408 caching could not have worked due to the session ID mismatch between 5409 client and server. 5410 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 5411 PR #377. 5412 [Lutz Jaenicke] 5413 5414 *) Change the declaration of needed Kerberos libraries to use EX_LIBS 5415 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is 5416 removed entirely. 5417 [Richard Levitte] 5418 5419 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it 5420 seems that in spite of existing for more than a year, many application 5421 author have done nothing to provide the necessary callbacks, which 5422 means that this particular engine will not work properly anywhere. 5423 This is a very unfortunate situation which forces us, in the name 5424 of usability, to give the hw_ncipher.c a static lock, which is part 5425 of libcrypto. 5426 NOTE: This is for the 0.9.7 series ONLY. This hack will never 5427 appear in 0.9.8 or later. We EXPECT application authors to have 5428 dealt properly with this when 0.9.8 is released (unless we actually 5429 make such changes in the libcrypto locking code that changes will 5430 have to be made anyway). 5431 [Richard Levitte] 5432 5433 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content 5434 octets have been read, EOF or an error occurs. Without this change 5435 some truncated ASN1 structures will not produce an error. 5436 [Steve Henson] 5437 5438 *) Disable Heimdal support, since it hasn't been fully implemented. 5439 Still give the possibility to force the use of Heimdal, but with 5440 warnings and a request that patches get sent to openssl-dev. 5441 [Richard Levitte] 5442 5443 *) Add the VC-CE target, introduce the WINCE sysname, and add 5444 INSTALL.WCE and appropriate conditionals to make it build. 5445 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 5446 5447 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and 5448 cygssl-x.y.z.dll, where x, y and z are the major, minor and 5449 edit numbers of the version. 5450 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 5451 5452 *) Introduce safe string copy and catenation functions 5453 (BUF_strlcpy() and BUF_strlcat()). 5454 [Ben Laurie (CHATS) and Richard Levitte] 5455 5456 *) Avoid using fixed-size buffers for one-line DNs. 5457 [Ben Laurie (CHATS)] 5458 5459 *) Add BUF_MEM_grow_clean() to avoid information leakage when 5460 resizing buffers containing secrets, and use where appropriate. 5461 [Ben Laurie (CHATS)] 5462 5463 *) Avoid using fixed size buffers for configuration file location. 5464 [Ben Laurie (CHATS)] 5465 5466 *) Avoid filename truncation for various CA files. 5467 [Ben Laurie (CHATS)] 5468 5469 *) Use sizeof in preference to magic numbers. 5470 [Ben Laurie (CHATS)] 5471 5472 *) Avoid filename truncation in cert requests. 5473 [Ben Laurie (CHATS)] 5474 5475 *) Add assertions to check for (supposedly impossible) buffer 5476 overflows. 5477 [Ben Laurie (CHATS)] 5478 5479 *) Don't cache truncated DNS entries in the local cache (this could 5480 potentially lead to a spoofing attack). 5481 [Ben Laurie (CHATS)] 5482 5483 *) Fix various buffers to be large enough for hex/decimal 5484 representations in a platform independent manner. 5485 [Ben Laurie (CHATS)] 5486 5487 *) Add CRYPTO_realloc_clean() to avoid information leakage when 5488 resizing buffers containing secrets, and use where appropriate. 5489 [Ben Laurie (CHATS)] 5490 5491 *) Add BIO_indent() to avoid much slightly worrying code to do 5492 indents. 5493 [Ben Laurie (CHATS)] 5494 5495 *) Convert sprintf()/BIO_puts() to BIO_printf(). 5496 [Ben Laurie (CHATS)] 5497 5498 *) buffer_gets() could terminate with the buffer only half 5499 full. Fixed. 5500 [Ben Laurie (CHATS)] 5501 5502 *) Add assertions to prevent user-supplied crypto functions from 5503 overflowing internal buffers by having large block sizes, etc. 5504 [Ben Laurie (CHATS)] 5505 5506 *) New OPENSSL_assert() macro (similar to assert(), but enabled 5507 unconditionally). 5508 [Ben Laurie (CHATS)] 5509 5510 *) Eliminate unused copy of key in RC4. 5511 [Ben Laurie (CHATS)] 5512 5513 *) Eliminate unused and incorrectly sized buffers for IV in pem.h. 5514 [Ben Laurie (CHATS)] 5515 5516 *) Fix off-by-one error in EGD path. 5517 [Ben Laurie (CHATS)] 5518 5519 *) If RANDFILE path is too long, ignore instead of truncating. 5520 [Ben Laurie (CHATS)] 5521 5522 *) Eliminate unused and incorrectly sized X.509 structure 5523 CBCParameter. 5524 [Ben Laurie (CHATS)] 5525 5526 *) Eliminate unused and dangerous function knumber(). 5527 [Ben Laurie (CHATS)] 5528 5529 *) Eliminate unused and dangerous structure, KSSL_ERR. 5530 [Ben Laurie (CHATS)] 5531 5532 *) Protect against overlong session ID context length in an encoded 5533 session object. Since these are local, this does not appear to be 5534 exploitable. 5535 [Ben Laurie (CHATS)] 5536 5537 *) Change from security patch (see 0.9.6e below) that did not affect 5538 the 0.9.6 release series: 5539 5540 Remote buffer overflow in SSL3 protocol - an attacker could 5541 supply an oversized master key in Kerberos-enabled versions. 5542 (CVE-2002-0657) 5543 [Ben Laurie (CHATS)] 5544 5545 *) Change the SSL kerb5 codes to match RFC 2712. 5546 [Richard Levitte] 5547 5548 *) Make -nameopt work fully for req and add -reqopt switch. 5549 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson] 5550 5551 *) The "block size" for block ciphers in CFB and OFB mode should be 1. 5552 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>] 5553 5554 *) Make sure tests can be performed even if the corresponding algorithms 5555 have been removed entirely. This was also the last step to make 5556 OpenSSL compilable with DJGPP under all reasonable conditions. 5557 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>] 5558 5559 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT 5560 to allow version independent disabling of normally unselected ciphers, 5561 which may be activated as a side-effect of selecting a single cipher. 5562 5563 (E.g., cipher list string "RSA" enables ciphersuites that are left 5564 out of "ALL" because they do not provide symmetric encryption. 5565 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.) 5566 [Lutz Jaenicke, Bodo Moeller] 5567 5568 *) Add appropriate support for separate platform-dependent build 5569 directories. The recommended way to make a platform-dependent 5570 build directory is the following (tested on Linux), maybe with 5571 some local tweaks: 5572 5573 # Place yourself outside of the OpenSSL source tree. In 5574 # this example, the environment variable OPENSSL_SOURCE 5575 # is assumed to contain the absolute OpenSSL source directory. 5576 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`" 5577 cd objtree/"`uname -s`-`uname -r`-`uname -m`" 5578 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do 5579 mkdir -p `dirname $F` 5580 ln -s $OPENSSL_SOURCE/$F $F 5581 done 5582 5583 To be absolutely sure not to disturb the source tree, a "make clean" 5584 is a good thing. If it isn't successfull, don't worry about it, 5585 it probably means the source directory is very clean. 5586 [Richard Levitte] 5587 5588 *) Make sure any ENGINE control commands make local copies of string 5589 pointers passed to them whenever necessary. Otherwise it is possible 5590 the caller may have overwritten (or deallocated) the original string 5591 data when a later ENGINE operation tries to use the stored values. 5592 [G��tz Babin-Ebell <babinebell@trustcenter.de>] 5593 5594 *) Improve diagnostics in file reading and command-line digests. 5595 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>] 5596 5597 *) Add AES modes CFB and OFB to the object database. Correct an 5598 error in AES-CFB decryption. 5599 [Richard Levitte] 5600 5601 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this 5602 allows existing EVP_CIPHER_CTX structures to be reused after 5603 calling EVP_*Final(). This behaviour is used by encryption 5604 BIOs and some applications. This has the side effect that 5605 applications must explicitly clean up cipher contexts with 5606 EVP_CIPHER_CTX_cleanup() or they will leak memory. 5607 [Steve Henson] 5608 5609 *) Check the values of dna and dnb in bn_mul_recursive before calling 5610 bn_mul_comba (a non zero value means the a or b arrays do not contain 5611 n2 elements) and fallback to bn_mul_normal if either is not zero. 5612 [Steve Henson] 5613 5614 *) Fix escaping of non-ASCII characters when using the -subj option 5615 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>) 5616 [Lutz Jaenicke] 5617 5618 *) Make object definitions compliant to LDAP (RFC2256): SN is the short 5619 form for "surname", serialNumber has no short form. 5620 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798; 5621 therefore remove "mail" short name for "internet 7". 5622 The OID for unique identifiers in X509 certificates is 5623 x500UniqueIdentifier, not uniqueIdentifier. 5624 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>) 5625 [Lutz Jaenicke] 5626 5627 *) Add an "init" command to the ENGINE config module and auto initialize 5628 ENGINEs. Without any "init" command the ENGINE will be initialized 5629 after all ctrl commands have been executed on it. If init=1 the 5630 ENGINE is initailized at that point (ctrls before that point are run 5631 on the uninitialized ENGINE and after on the initialized one). If 5632 init=0 then the ENGINE will not be iniatialized at all. 5633 [Steve Henson] 5634 5635 *) Fix the 'app_verify_callback' interface so that the user-defined 5636 argument is actually passed to the callback: In the 5637 SSL_CTX_set_cert_verify_callback() prototype, the callback 5638 declaration has been changed from 5639 int (*cb)() 5640 into 5641 int (*cb)(X509_STORE_CTX *,void *); 5642 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call 5643 i=s->ctx->app_verify_callback(&ctx) 5644 has been changed into 5645 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg). 5646 5647 To update applications using SSL_CTX_set_cert_verify_callback(), 5648 a dummy argument can be added to their callback functions. 5649 [D. K. Smetters <smetters@parc.xerox.com>] 5650 5651 *) Added the '4758cca' ENGINE to support IBM 4758 cards. 5652 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe] 5653 5654 *) Add and OPENSSL_LOAD_CONF define which will cause 5655 OpenSSL_add_all_algorithms() to load the openssl.cnf config file. 5656 This allows older applications to transparently support certain 5657 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading. 5658 Two new functions OPENSSL_add_all_algorithms_noconf() which will never 5659 load the config file and OPENSSL_add_all_algorithms_conf() which will 5660 always load it have also been added. 5661 [Steve Henson] 5662 5663 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES. 5664 Adjust NIDs and EVP layer. 5665 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 5666 5667 *) Config modules support in openssl utility. 5668 5669 Most commands now load modules from the config file, 5670 though in a few (such as version) this isn't done 5671 because it couldn't be used for anything. 5672 5673 In the case of ca and req the config file used is 5674 the same as the utility itself: that is the -config 5675 command line option can be used to specify an 5676 alternative file. 5677 [Steve Henson] 5678 5679 *) Move default behaviour from OPENSSL_config(). If appname is NULL 5680 use "openssl_conf" if filename is NULL use default openssl config file. 5681 [Steve Henson] 5682 5683 *) Add an argument to OPENSSL_config() to allow the use of an alternative 5684 config section name. Add a new flag to tolerate a missing config file 5685 and move code to CONF_modules_load_file(). 5686 [Steve Henson] 5687 5688 *) Support for crypto accelerator cards from Accelerated Encryption 5689 Processing, www.aep.ie. (Use engine 'aep') 5690 The support was copied from 0.9.6c [engine] and adapted/corrected 5691 to work with the new engine framework. 5692 [AEP Inc. and Richard Levitte] 5693 5694 *) Support for SureWare crypto accelerator cards from Baltimore 5695 Technologies. (Use engine 'sureware') 5696 The support was copied from 0.9.6c [engine] and adapted 5697 to work with the new engine framework. 5698 [Richard Levitte] 5699 5700 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually 5701 make the newer ENGINE framework commands for the CHIL engine work. 5702 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte] 5703 5704 *) Make it possible to produce shared libraries on ReliantUNIX. 5705 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte] 5706 5707 *) Add the configuration target debug-linux-ppro. 5708 Make 'openssl rsa' use the general key loading routines 5709 implemented in apps.c, and make those routines able to 5710 handle the key format FORMAT_NETSCAPE and the variant 5711 FORMAT_IISSGC. 5712 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 5713 5714 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 5715 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 5716 5717 *) Add -keyform to rsautl, and document -engine. 5718 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>] 5719 5720 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new 5721 BIO_R_NO_SUCH_FILE error code rather than the generic 5722 ERR_R_SYS_LIB error code if fopen() fails with ENOENT. 5723 [Ben Laurie] 5724 5725 *) Add new functions 5726 ERR_peek_last_error 5727 ERR_peek_last_error_line 5728 ERR_peek_last_error_line_data. 5729 These are similar to 5730 ERR_peek_error 5731 ERR_peek_error_line 5732 ERR_peek_error_line_data, 5733 but report on the latest error recorded rather than the first one 5734 still in the error queue. 5735 [Ben Laurie, Bodo Moeller] 5736 5737 *) default_algorithms option in ENGINE config module. This allows things 5738 like: 5739 default_algorithms = ALL 5740 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS 5741 [Steve Henson] 5742 5743 *) Prelminary ENGINE config module. 5744 [Steve Henson] 5745 5746 *) New experimental application configuration code. 5747 [Steve Henson] 5748 5749 *) Change the AES code to follow the same name structure as all other 5750 symmetric ciphers, and behave the same way. Move everything to 5751 the directory crypto/aes, thereby obsoleting crypto/rijndael. 5752 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 5753 5754 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c. 5755 [Ben Laurie and Theo de Raadt] 5756 5757 *) Add option to output public keys in req command. 5758 [Massimiliano Pala madwolf@openca.org] 5759 5760 *) Use wNAFs in EC_POINTs_mul() for improved efficiency 5761 (up to about 10% better than before for P-192 and P-224). 5762 [Bodo Moeller] 5763 5764 *) New functions/macros 5765 5766 SSL_CTX_set_msg_callback(ctx, cb) 5767 SSL_CTX_set_msg_callback_arg(ctx, arg) 5768 SSL_set_msg_callback(ssl, cb) 5769 SSL_set_msg_callback_arg(ssl, arg) 5770 5771 to request calling a callback function 5772 5773 void cb(int write_p, int version, int content_type, 5774 const void *buf, size_t len, SSL *ssl, void *arg) 5775 5776 whenever a protocol message has been completely received 5777 (write_p == 0) or sent (write_p == 1). Here 'version' is the 5778 protocol version according to which the SSL library interprets 5779 the current protocol message (SSL2_VERSION, SSL3_VERSION, or 5780 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or 5781 the content type as defined in the SSL 3.0/TLS 1.0 protocol 5782 specification (change_cipher_spec(20), alert(21), handshake(22)). 5783 'buf' and 'len' point to the actual message, 'ssl' to the 5784 SSL object, and 'arg' is the application-defined value set by 5785 SSL[_CTX]_set_msg_callback_arg(). 5786 5787 'openssl s_client' and 'openssl s_server' have new '-msg' options 5788 to enable a callback that displays all protocol messages. 5789 [Bodo Moeller] 5790 5791 *) Change the shared library support so shared libraries are built as 5792 soon as the corresponding static library is finished, and thereby get 5793 openssl and the test programs linked against the shared library. 5794 This still only happens when the keyword "shard" has been given to 5795 the configuration scripts. 5796 5797 NOTE: shared library support is still an experimental thing, and 5798 backward binary compatibility is still not guaranteed. 5799 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte] 5800 5801 *) Add support for Subject Information Access extension. 5802 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 5803 5804 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero 5805 additional bytes when new memory had to be allocated, not just 5806 when reusing an existing buffer. 5807 [Bodo Moeller] 5808 5809 *) New command line and configuration option 'utf8' for the req command. 5810 This allows field values to be specified as UTF8 strings. 5811 [Steve Henson] 5812 5813 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel 5814 runs for the former and machine-readable output for the latter. 5815 [Ben Laurie] 5816 5817 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion 5818 of the e-mail address in the DN (i.e., it will go into a certificate 5819 extension only). The new configuration file option 'email_in_dn = no' 5820 has the same effect. 5821 [Massimiliano Pala madwolf@openca.org] 5822 5823 *) Change all functions with names starting with des_ to be starting 5824 with DES_ instead. Add wrappers that are compatible with libdes, 5825 but are named _ossl_old_des_*. Finally, add macros that map the 5826 des_* symbols to the corresponding _ossl_old_des_* if libdes 5827 compatibility is desired. If OpenSSL 0.9.6c compatibility is 5828 desired, the des_* symbols will be mapped to DES_*, with one 5829 exception. 5830 5831 Since we provide two compatibility mappings, the user needs to 5832 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes 5833 compatibility is desired. The default (i.e., when that macro 5834 isn't defined) is OpenSSL 0.9.6c compatibility. 5835 5836 There are also macros that enable and disable the support of old 5837 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT 5838 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those 5839 are defined, the default will apply: to support the old des routines. 5840 5841 In either case, one must include openssl/des.h to get the correct 5842 definitions. Do not try to just include openssl/des_old.h, that 5843 won't work. 5844 5845 NOTE: This is a major break of an old API into a new one. Software 5846 authors are encouraged to switch to the DES_ style functions. Some 5847 time in the future, des_old.h and the libdes compatibility functions 5848 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the 5849 default), and then completely removed. 5850 [Richard Levitte] 5851 5852 *) Test for certificates which contain unsupported critical extensions. 5853 If such a certificate is found during a verify operation it is 5854 rejected by default: this behaviour can be overridden by either 5855 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or 5856 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function 5857 X509_supported_extension() has also been added which returns 1 if a 5858 particular extension is supported. 5859 [Steve Henson] 5860 5861 *) Modify the behaviour of EVP cipher functions in similar way to digests 5862 to retain compatibility with existing code. 5863 [Steve Henson] 5864 5865 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain 5866 compatibility with existing code. In particular the 'ctx' parameter does 5867 not have to be to be initialized before the call to EVP_DigestInit() and 5868 it is tidied up after a call to EVP_DigestFinal(). New function 5869 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function 5870 EVP_MD_CTX_copy() changed to not require the destination to be 5871 initialized valid and new function EVP_MD_CTX_copy_ex() added which 5872 requires the destination to be valid. 5873 5874 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(), 5875 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex(). 5876 [Steve Henson] 5877 5878 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it 5879 so that complete 'Handshake' protocol structures are kept in memory 5880 instead of overwriting 'msg_type' and 'length' with 'body' data. 5881 [Bodo Moeller] 5882 5883 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32. 5884 [Massimo Santin via Richard Levitte] 5885 5886 *) Major restructuring to the underlying ENGINE code. This includes 5887 reduction of linker bloat, separation of pure "ENGINE" manipulation 5888 (initialisation, etc) from functionality dealing with implementations 5889 of specific crypto iterfaces. This change also introduces integrated 5890 support for symmetric ciphers and digest implementations - so ENGINEs 5891 can now accelerate these by providing EVP_CIPHER and EVP_MD 5892 implementations of their own. This is detailed in crypto/engine/README 5893 as it couldn't be adequately described here. However, there are a few 5894 API changes worth noting - some RSA, DSA, DH, and RAND functions that 5895 were changed in the original introduction of ENGINE code have now 5896 reverted back - the hooking from this code to ENGINE is now a good 5897 deal more passive and at run-time, operations deal directly with 5898 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than 5899 dereferencing through an ENGINE pointer any more. Also, the ENGINE 5900 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed - 5901 they were not being used by the framework as there is no concept of a 5902 BIGNUM_METHOD and they could not be generalised to the new 5903 'ENGINE_TABLE' mechanism that underlies the new code. Similarly, 5904 ENGINE_cpy() has been removed as it cannot be consistently defined in 5905 the new code. 5906 [Geoff Thorpe] 5907 5908 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds. 5909 [Steve Henson] 5910 5911 *) Change mkdef.pl to sort symbols that get the same entry number, 5912 and make sure the automatically generated functions ERR_load_* 5913 become part of libeay.num as well. 5914 [Richard Levitte] 5915 5916 *) New function SSL_renegotiate_pending(). This returns true once 5917 renegotiation has been requested (either SSL_renegotiate() call 5918 or HelloRequest/ClientHello receveived from the peer) and becomes 5919 false once a handshake has been completed. 5920 (For servers, SSL_renegotiate() followed by SSL_do_handshake() 5921 sends a HelloRequest, but does not ensure that a handshake takes 5922 place. SSL_renegotiate_pending() is useful for checking if the 5923 client has followed the request.) 5924 [Bodo Moeller] 5925 5926 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION. 5927 By default, clients may request session resumption even during 5928 renegotiation (if session ID contexts permit); with this option, 5929 session resumption is possible only in the first handshake. 5930 5931 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes 5932 more bits available for options that should not be part of 5933 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION). 5934 [Bodo Moeller] 5935 5936 *) Add some demos for certificate and certificate request creation. 5937 [Steve Henson] 5938 5939 *) Make maximum certificate chain size accepted from the peer application 5940 settable (SSL*_get/set_max_cert_list()), as proposed by 5941 "Douglas E. Engert" <deengert@anl.gov>. 5942 [Lutz Jaenicke] 5943 5944 *) Add support for shared libraries for Unixware-7 5945 (Boyd Lynn Gerber <gerberb@zenez.com>). 5946 [Lutz Jaenicke] 5947 5948 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to 5949 be done prior to destruction. Use this to unload error strings from 5950 ENGINEs that load their own error strings. NB: This adds two new API 5951 functions to "get" and "set" this destroy handler in an ENGINE. 5952 [Geoff Thorpe] 5953 5954 *) Alter all existing ENGINE implementations (except "openssl" and 5955 "openbsd") to dynamically instantiate their own error strings. This 5956 makes them more flexible to be built both as statically-linked ENGINEs 5957 and self-contained shared-libraries loadable via the "dynamic" ENGINE. 5958 Also, add stub code to each that makes building them as self-contained 5959 shared-libraries easier (see README.ENGINE). 5960 [Geoff Thorpe] 5961 5962 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE 5963 implementations into applications that are completely implemented in 5964 self-contained shared-libraries. The "dynamic" ENGINE exposes control 5965 commands that can be used to configure what shared-library to load and 5966 to control aspects of the way it is handled. Also, made an update to 5967 the README.ENGINE file that brings its information up-to-date and 5968 provides some information and instructions on the "dynamic" ENGINE 5969 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc). 5970 [Geoff Thorpe] 5971 5972 *) Make it possible to unload ranges of ERR strings with a new 5973 "ERR_unload_strings" function. 5974 [Geoff Thorpe] 5975 5976 *) Add a copy() function to EVP_MD. 5977 [Ben Laurie] 5978 5979 *) Make EVP_MD routines take a context pointer instead of just the 5980 md_data void pointer. 5981 [Ben Laurie] 5982 5983 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates 5984 that the digest can only process a single chunk of data 5985 (typically because it is provided by a piece of 5986 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application 5987 is only going to provide a single chunk of data, and hence the 5988 framework needn't accumulate the data for oneshot drivers. 5989 [Ben Laurie] 5990 5991 *) As with "ERR", make it possible to replace the underlying "ex_data" 5992 functions. This change also alters the storage and management of global 5993 ex_data state - it's now all inside ex_data.c and all "class" code (eg. 5994 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class 5995 index counters. The API functions that use this state have been changed 5996 to take a "class_index" rather than pointers to the class's local STACK 5997 and counter, and there is now an API function to dynamically create new 5998 classes. This centralisation allows us to (a) plug a lot of the 5999 thread-safety problems that existed, and (b) makes it possible to clean 6000 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b) 6001 such data would previously have always leaked in application code and 6002 workarounds were in place to make the memory debugging turn a blind eye 6003 to it. Application code that doesn't use this new function will still 6004 leak as before, but their memory debugging output will announce it now 6005 rather than letting it slide. 6006 6007 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change 6008 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now 6009 has a return value to indicate success or failure. 6010 [Geoff Thorpe] 6011 6012 *) Make it possible to replace the underlying "ERR" functions such that the 6013 global state (2 LHASH tables and 2 locks) is only used by the "default" 6014 implementation. This change also adds two functions to "get" and "set" 6015 the implementation prior to it being automatically set the first time 6016 any other ERR function takes place. Ie. an application can call "get", 6017 pass the return value to a module it has just loaded, and that module 6018 can call its own "set" function using that value. This means the 6019 module's "ERR" operations will use (and modify) the error state in the 6020 application and not in its own statically linked copy of OpenSSL code. 6021 [Geoff Thorpe] 6022 6023 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment 6024 reference counts. This performs normal REF_PRINT/REF_CHECK macros on 6025 the operation, and provides a more encapsulated way for external code 6026 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code 6027 to use these functions rather than manually incrementing the counts. 6028 6029 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()". 6030 [Geoff Thorpe] 6031 6032 *) Add EVP test program. 6033 [Ben Laurie] 6034 6035 *) Add symmetric cipher support to ENGINE. Expect the API to change! 6036 [Ben Laurie] 6037 6038 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name() 6039 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(), 6040 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate(). 6041 These allow a CRL to be built without having to access X509_CRL fields 6042 directly. Modify 'ca' application to use new functions. 6043 [Steve Henson] 6044 6045 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended 6046 bug workarounds. Rollback attack detection is a security feature. 6047 The problem will only arise on OpenSSL servers when TLSv1 is not 6048 available (sslv3_server_method() or SSL_OP_NO_TLSv1). 6049 Software authors not wanting to support TLSv1 will have special reasons 6050 for their choice and can explicitly enable this option. 6051 [Bodo Moeller, Lutz Jaenicke] 6052 6053 *) Rationalise EVP so it can be extended: don't include a union of 6054 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX 6055 (similar to those existing for EVP_CIPHER_CTX). 6056 Usage example: 6057 6058 EVP_MD_CTX md; 6059 6060 EVP_MD_CTX_init(&md); /* new function call */ 6061 EVP_DigestInit(&md, EVP_sha1()); 6062 EVP_DigestUpdate(&md, in, len); 6063 EVP_DigestFinal(&md, out, NULL); 6064 EVP_MD_CTX_cleanup(&md); /* new function call */ 6065 6066 [Ben Laurie] 6067 6068 *) Make DES key schedule conform to the usual scheme, as well as 6069 correcting its structure. This means that calls to DES functions 6070 now have to pass a pointer to a des_key_schedule instead of a 6071 plain des_key_schedule (which was actually always a pointer 6072 anyway): E.g., 6073 6074 des_key_schedule ks; 6075 6076 des_set_key_checked(..., &ks); 6077 des_ncbc_encrypt(..., &ks, ...); 6078 6079 (Note that a later change renames 'des_...' into 'DES_...'.) 6080 [Ben Laurie] 6081 6082 *) Initial reduction of linker bloat: the use of some functions, such as 6083 PEM causes large amounts of unused functions to be linked in due to 6084 poor organisation. For example pem_all.c contains every PEM function 6085 which has a knock on effect of linking in large amounts of (unused) 6086 ASN1 code. Grouping together similar functions and splitting unrelated 6087 functions prevents this. 6088 [Steve Henson] 6089 6090 *) Cleanup of EVP macros. 6091 [Ben Laurie] 6092 6093 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the 6094 correct _ecb suffix. 6095 [Ben Laurie] 6096 6097 *) Add initial OCSP responder support to ocsp application. The 6098 revocation information is handled using the text based index 6099 use by the ca application. The responder can either handle 6100 requests generated internally, supplied in files (for example 6101 via a CGI script) or using an internal minimal server. 6102 [Steve Henson] 6103 6104 *) Add configuration choices to get zlib compression for TLS. 6105 [Richard Levitte] 6106 6107 *) Changes to Kerberos SSL for RFC 2712 compliance: 6108 1. Implemented real KerberosWrapper, instead of just using 6109 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>] 6110 2. Implemented optional authenticator field of KerberosWrapper. 6111 6112 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req, 6113 and authenticator structs; see crypto/krb5/. 6114 6115 Generalized Kerberos calls to support multiple Kerberos libraries. 6116 [Vern Staats <staatsvr@asc.hpc.mil>, 6117 Jeffrey Altman <jaltman@columbia.edu> 6118 via Richard Levitte] 6119 6120 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it 6121 already does with RSA. testdsa.h now has 'priv_key/pub_key' 6122 values for each of the key sizes rather than having just 6123 parameters (and 'speed' generating keys each time). 6124 [Geoff Thorpe] 6125 6126 *) Speed up EVP routines. 6127 Before: 6128encrypt 6129type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes 6130des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k 6131des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k 6132des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k 6133decrypt 6134des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k 6135des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k 6136des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k 6137 After: 6138encrypt 6139des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k 6140decrypt 6141des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k 6142 [Ben Laurie] 6143 6144 *) Added the OS2-EMX target. 6145 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte] 6146 6147 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions 6148 to support NCONF routines in extension code. New function CONF_set_nconf() 6149 to allow functions which take an NCONF to also handle the old LHASH 6150 structure: this means that the old CONF compatible routines can be 6151 retained (in particular wrt extensions) without having to duplicate the 6152 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack. 6153 [Steve Henson] 6154 6155 *) Enhance the general user interface with mechanisms for inner control 6156 and with possibilities to have yes/no kind of prompts. 6157 [Richard Levitte] 6158 6159 *) Change all calls to low level digest routines in the library and 6160 applications to use EVP. Add missing calls to HMAC_cleanup() and 6161 don't assume HMAC_CTX can be copied using memcpy(). 6162 [Verdon Walker <VWalker@novell.com>, Steve Henson] 6163 6164 *) Add the possibility to control engines through control names but with 6165 arbitrary arguments instead of just a string. 6166 Change the key loaders to take a UI_METHOD instead of a callback 6167 function pointer. NOTE: this breaks binary compatibility with earlier 6168 versions of OpenSSL [engine]. 6169 Adapt the nCipher code for these new conditions and add a card insertion 6170 callback. 6171 [Richard Levitte] 6172 6173 *) Enhance the general user interface with mechanisms to better support 6174 dialog box interfaces, application-defined prompts, the possibility 6175 to use defaults (for example default passwords from somewhere else) 6176 and interrupts/cancellations. 6177 [Richard Levitte] 6178 6179 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name 6180 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility. 6181 [Steve Henson] 6182 6183 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also 6184 tidy up some unnecessarily weird code in 'sk_new()'). 6185 [Geoff, reported by Diego Tartara <dtartara@novamens.com>] 6186 6187 *) Change the key loading routines for ENGINEs to use the same kind 6188 callback (pem_password_cb) as all other routines that need this 6189 kind of callback. 6190 [Richard Levitte] 6191 6192 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with 6193 256 bit (=32 byte) keys. Of course seeding with more entropy bytes 6194 than this minimum value is recommended. 6195 [Lutz Jaenicke] 6196 6197 *) New random seeder for OpenVMS, using the system process statistics 6198 that are easily reachable. 6199 [Richard Levitte] 6200 6201 *) Windows apparently can't transparently handle global 6202 variables defined in DLLs. Initialisations such as: 6203 6204 const ASN1_ITEM *it = &ASN1_INTEGER_it; 6205 6206 wont compile. This is used by the any applications that need to 6207 declare their own ASN1 modules. This was fixed by adding the option 6208 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly 6209 needed for static libraries under Win32. 6210 [Steve Henson] 6211 6212 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle 6213 setting of purpose and trust fields. New X509_STORE trust and 6214 purpose functions and tidy up setting in other SSL functions. 6215 [Steve Henson] 6216 6217 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE 6218 structure. These are inherited by X509_STORE_CTX when it is 6219 initialised. This allows various defaults to be set in the 6220 X509_STORE structure (such as flags for CRL checking and custom 6221 purpose or trust settings) for functions which only use X509_STORE_CTX 6222 internally such as S/MIME. 6223 6224 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and 6225 trust settings if they are not set in X509_STORE. This allows X509_STORE 6226 purposes and trust (in S/MIME for example) to override any set by default. 6227 6228 Add command line options for CRL checking to smime, s_client and s_server 6229 applications. 6230 [Steve Henson] 6231 6232 *) Initial CRL based revocation checking. If the CRL checking flag(s) 6233 are set then the CRL is looked up in the X509_STORE structure and 6234 its validity and signature checked, then if the certificate is found 6235 in the CRL the verify fails with a revoked error. 6236 6237 Various new CRL related callbacks added to X509_STORE_CTX structure. 6238 6239 Command line options added to 'verify' application to support this. 6240 6241 This needs some additional work, such as being able to handle multiple 6242 CRLs with different times, extension based lookup (rather than just 6243 by subject name) and ultimately more complete V2 CRL extension 6244 handling. 6245 [Steve Henson] 6246 6247 *) Add a general user interface API (crypto/ui/). This is designed 6248 to replace things like des_read_password and friends (backward 6249 compatibility functions using this new API are provided). 6250 The purpose is to remove prompting functions from the DES code 6251 section as well as provide for prompting through dialog boxes in 6252 a window system and the like. 6253 [Richard Levitte] 6254 6255 *) Add "ex_data" support to ENGINE so implementations can add state at a 6256 per-structure level rather than having to store it globally. 6257 [Geoff] 6258 6259 *) Make it possible for ENGINE structures to be copied when retrieved by 6260 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY. 6261 This causes the "original" ENGINE structure to act like a template, 6262 analogous to the RSA vs. RSA_METHOD type of separation. Because of this 6263 operational state can be localised to each ENGINE structure, despite the 6264 fact they all share the same "methods". New ENGINE structures returned in 6265 this case have no functional references and the return value is the single 6266 structural reference. This matches the single structural reference returned 6267 by ENGINE_by_id() normally, when it is incremented on the pre-existing 6268 ENGINE structure. 6269 [Geoff] 6270 6271 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this 6272 needs to match any other type at all we need to manually clear the 6273 tag cache. 6274 [Steve Henson] 6275 6276 *) Changes to the "openssl engine" utility to include; 6277 - verbosity levels ('-v', '-vv', and '-vvv') that provide information 6278 about an ENGINE's available control commands. 6279 - executing control commands from command line arguments using the 6280 '-pre' and '-post' switches. '-post' is only used if '-t' is 6281 specified and the ENGINE is successfully initialised. The syntax for 6282 the individual commands are colon-separated, for example; 6283 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so 6284 [Geoff] 6285 6286 *) New dynamic control command support for ENGINEs. ENGINEs can now 6287 declare their own commands (numbers), names (strings), descriptions, 6288 and input types for run-time discovery by calling applications. A 6289 subset of these commands are implicitly classed as "executable" 6290 depending on their input type, and only these can be invoked through 6291 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this 6292 can be based on user input, config files, etc). The distinction is 6293 that "executable" commands cannot return anything other than a boolean 6294 result and can only support numeric or string input, whereas some 6295 discoverable commands may only be for direct use through 6296 ENGINE_ctrl(), eg. supporting the exchange of binary data, function 6297 pointers, or other custom uses. The "executable" commands are to 6298 support parameterisations of ENGINE behaviour that can be 6299 unambiguously defined by ENGINEs and used consistently across any 6300 OpenSSL-based application. Commands have been added to all the 6301 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow 6302 control over shared-library paths without source code alterations. 6303 [Geoff] 6304 6305 *) Changed all ENGINE implementations to dynamically allocate their 6306 ENGINEs rather than declaring them statically. Apart from this being 6307 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction, 6308 this also allows the implementations to compile without using the 6309 internal engine_int.h header. 6310 [Geoff] 6311 6312 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a 6313 'const' value. Any code that should be able to modify a RAND_METHOD 6314 should already have non-const pointers to it (ie. they should only 6315 modify their own ones). 6316 [Geoff] 6317 6318 *) Made a variety of little tweaks to the ENGINE code. 6319 - "atalla" and "ubsec" string definitions were moved from header files 6320 to C code. "nuron" string definitions were placed in variables 6321 rather than hard-coded - allowing parameterisation of these values 6322 later on via ctrl() commands. 6323 - Removed unused "#if 0"'d code. 6324 - Fixed engine list iteration code so it uses ENGINE_free() to release 6325 structural references. 6326 - Constified the RAND_METHOD element of ENGINE structures. 6327 - Constified various get/set functions as appropriate and added 6328 missing functions (including a catch-all ENGINE_cpy that duplicates 6329 all ENGINE values onto a new ENGINE except reference counts/state). 6330 - Removed NULL parameter checks in get/set functions. Setting a method 6331 or function to NULL is a way of cancelling out a previously set 6332 value. Passing a NULL ENGINE parameter is just plain stupid anyway 6333 and doesn't justify the extra error symbols and code. 6334 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for 6335 flags from engine_int.h to engine.h. 6336 - Changed prototypes for ENGINE handler functions (init(), finish(), 6337 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter. 6338 [Geoff] 6339 6340 *) Implement binary inversion algorithm for BN_mod_inverse in addition 6341 to the algorithm using long division. The binary algorithm can be 6342 used only if the modulus is odd. On 32-bit systems, it is faster 6343 only for relatively small moduli (roughly 20-30% for 128-bit moduli, 6344 roughly 5-15% for 256-bit moduli), so we use it only for moduli 6345 up to 450 bits. In 64-bit environments, the binary algorithm 6346 appears to be advantageous for much longer moduli; here we use it 6347 for moduli up to 2048 bits. 6348 [Bodo Moeller] 6349 6350 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code 6351 could not support the combine flag in choice fields. 6352 [Steve Henson] 6353 6354 *) Add a 'copy_extensions' option to the 'ca' utility. This copies 6355 extensions from a certificate request to the certificate. 6356 [Steve Henson] 6357 6358 *) Allow multiple 'certopt' and 'nameopt' options to be separated 6359 by commas. Add 'namopt' and 'certopt' options to the 'ca' config 6360 file: this allows the display of the certificate about to be 6361 signed to be customised, to allow certain fields to be included 6362 or excluded and extension details. The old system didn't display 6363 multicharacter strings properly, omitted fields not in the policy 6364 and couldn't display additional details such as extensions. 6365 [Steve Henson] 6366 6367 *) Function EC_POINTs_mul for multiple scalar multiplication 6368 of an arbitrary number of elliptic curve points 6369 \sum scalars[i]*points[i], 6370 optionally including the generator defined for the EC_GROUP: 6371 scalar*generator + \sum scalars[i]*points[i]. 6372 6373 EC_POINT_mul is a simple wrapper function for the typical case 6374 that the point list has just one item (besides the optional 6375 generator). 6376 [Bodo Moeller] 6377 6378 *) First EC_METHODs for curves over GF(p): 6379 6380 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr 6381 operations and provides various method functions that can also 6382 operate with faster implementations of modular arithmetic. 6383 6384 EC_GFp_mont_method() reuses most functions that are part of 6385 EC_GFp_simple_method, but uses Montgomery arithmetic. 6386 6387 [Bodo Moeller; point addition and point doubling 6388 implementation directly derived from source code provided by 6389 Lenka Fibikova <fibikova@exp-math.uni-essen.de>] 6390 6391 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h, 6392 crypto/ec/ec_lib.c): 6393 6394 Curves are EC_GROUP objects (with an optional group generator) 6395 based on EC_METHODs that are built into the library. 6396 6397 Points are EC_POINT objects based on EC_GROUP objects. 6398 6399 Most of the framework would be able to handle curves over arbitrary 6400 finite fields, but as there are no obvious types for fields other 6401 than GF(p), some functions are limited to that for now. 6402 [Bodo Moeller] 6403 6404 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires 6405 that the file contains a complete HTTP response. 6406 [Richard Levitte] 6407 6408 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl 6409 change the def and num file printf format specifier from "%-40sXXX" 6410 to "%-39s XXX". The latter will always guarantee a space after the 6411 field while the former will cause them to run together if the field 6412 is 40 of more characters long. 6413 [Steve Henson] 6414 6415 *) Constify the cipher and digest 'method' functions and structures 6416 and modify related functions to take constant EVP_MD and EVP_CIPHER 6417 pointers. 6418 [Steve Henson] 6419 6420 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them 6421 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32. 6422 [Bodo Moeller] 6423 6424 *) Modify EVP_Digest*() routines so they now return values. Although the 6425 internal software routines can never fail additional hardware versions 6426 might. 6427 [Steve Henson] 6428 6429 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts: 6430 6431 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7 6432 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32. 6433 6434 ASN1 error codes 6435 ERR_R_NESTED_ASN1_ERROR 6436 ... 6437 ERR_R_MISSING_ASN1_EOS 6438 were 4 .. 9, conflicting with 6439 ERR_LIB_RSA (= ERR_R_RSA_LIB) 6440 ... 6441 ERR_LIB_PEM (= ERR_R_PEM_LIB). 6442 They are now 58 .. 63 (i.e., just below ERR_R_FATAL). 6443 6444 Add new error code 'ERR_R_INTERNAL_ERROR'. 6445 [Bodo Moeller] 6446 6447 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock 6448 suffices. 6449 [Bodo Moeller] 6450 6451 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This 6452 sets the subject name for a new request or supersedes the 6453 subject name in a given request. Formats that can be parsed are 6454 'CN=Some Name, OU=myOU, C=IT' 6455 and 6456 'CN=Some Name/OU=myOU/C=IT'. 6457 6458 Add options '-batch' and '-verbose' to 'openssl req'. 6459 [Massimiliano Pala <madwolf@hackmasters.net>] 6460 6461 *) Introduce the possibility to access global variables through 6462 functions on platform were that's the best way to handle exporting 6463 global variables in shared libraries. To enable this functionality, 6464 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro 6465 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter 6466 is normally done by Configure or something similar). 6467 6468 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL 6469 in the source file (foo.c) like this: 6470 6471 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1; 6472 OPENSSL_IMPLEMENT_GLOBAL(double,bar); 6473 6474 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL 6475 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this: 6476 6477 OPENSSL_DECLARE_GLOBAL(int,foo); 6478 #define foo OPENSSL_GLOBAL_REF(foo) 6479 OPENSSL_DECLARE_GLOBAL(double,bar); 6480 #define bar OPENSSL_GLOBAL_REF(bar) 6481 6482 The #defines are very important, and therefore so is including the 6483 header file everywhere where the defined globals are used. 6484 6485 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition 6486 of ASN.1 items, but that structure is a bit different. 6487 6488 The largest change is in util/mkdef.pl which has been enhanced with 6489 better and easier to understand logic to choose which symbols should 6490 go into the Windows .def files as well as a number of fixes and code 6491 cleanup (among others, algorithm keywords are now sorted 6492 lexicographically to avoid constant rewrites). 6493 [Richard Levitte] 6494 6495 *) In BN_div() keep a copy of the sign of 'num' before writing the 6496 result to 'rm' because if rm==num the value will be overwritten 6497 and produce the wrong result if 'num' is negative: this caused 6498 problems with BN_mod() and BN_nnmod(). 6499 [Steve Henson] 6500 6501 *) Function OCSP_request_verify(). This checks the signature on an 6502 OCSP request and verifies the signer certificate. The signer 6503 certificate is just checked for a generic purpose and OCSP request 6504 trust settings. 6505 [Steve Henson] 6506 6507 *) Add OCSP_check_validity() function to check the validity of OCSP 6508 responses. OCSP responses are prepared in real time and may only 6509 be a few seconds old. Simply checking that the current time lies 6510 between thisUpdate and nextUpdate max reject otherwise valid responses 6511 caused by either OCSP responder or client clock inaccuracy. Instead 6512 we allow thisUpdate and nextUpdate to fall within a certain period of 6513 the current time. The age of the response can also optionally be 6514 checked. Two new options -validity_period and -status_age added to 6515 ocsp utility. 6516 [Steve Henson] 6517 6518 *) If signature or public key algorithm is unrecognized print out its 6519 OID rather that just UNKNOWN. 6520 [Steve Henson] 6521 6522 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and 6523 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate 6524 ID to be generated from the issuer certificate alone which can then be 6525 passed to OCSP_id_issuer_cmp(). 6526 [Steve Henson] 6527 6528 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new 6529 ASN1 modules to export functions returning ASN1_ITEM pointers 6530 instead of the ASN1_ITEM structures themselves. This adds several 6531 new macros which allow the underlying ASN1 function/structure to 6532 be accessed transparently. As a result code should not use ASN1_ITEM 6533 references directly (such as &X509_it) but instead use the relevant 6534 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow 6535 use of the new ASN1 code on platforms where exporting structures 6536 is problematical (for example in shared libraries) but exporting 6537 functions returning pointers to structures is not. 6538 [Steve Henson] 6539 6540 *) Add support for overriding the generation of SSL/TLS session IDs. 6541 These callbacks can be registered either in an SSL_CTX or per SSL. 6542 The purpose of this is to allow applications to control, if they wish, 6543 the arbitrary values chosen for use as session IDs, particularly as it 6544 can be useful for session caching in multiple-server environments. A 6545 command-line switch for testing this (and any client code that wishes 6546 to use such a feature) has been added to "s_server". 6547 [Geoff Thorpe, Lutz Jaenicke] 6548 6549 *) Modify mkdef.pl to recognise and parse preprocessor conditionals 6550 of the form '#if defined(...) || defined(...) || ...' and 6551 '#if !defined(...) && !defined(...) && ...'. This also avoids 6552 the growing number of special cases it was previously handling. 6553 [Richard Levitte] 6554 6555 *) Make all configuration macros available for application by making 6556 sure they are available in opensslconf.h, by giving them names starting 6557 with "OPENSSL_" to avoid conflicts with other packages and by making 6558 sure e_os2.h will cover all platform-specific cases together with 6559 opensslconf.h. 6560 Additionally, it is now possible to define configuration/platform- 6561 specific names (called "system identities"). In the C code, these 6562 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another 6563 macro with the name beginning with "OPENSSL_SYS_", which is determined 6564 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on 6565 what is available. 6566 [Richard Levitte] 6567 6568 *) New option -set_serial to 'req' and 'x509' this allows the serial 6569 number to use to be specified on the command line. Previously self 6570 signed certificates were hard coded with serial number 0 and the 6571 CA options of 'x509' had to use a serial number in a file which was 6572 auto incremented. 6573 [Steve Henson] 6574 6575 *) New options to 'ca' utility to support V2 CRL entry extensions. 6576 Currently CRL reason, invalidity date and hold instruction are 6577 supported. Add new CRL extensions to V3 code and some new objects. 6578 [Steve Henson] 6579 6580 *) New function EVP_CIPHER_CTX_set_padding() this is used to 6581 disable standard block padding (aka PKCS#5 padding) in the EVP 6582 API, which was previously mandatory. This means that the data is 6583 not padded in any way and so the total length much be a multiple 6584 of the block size, otherwise an error occurs. 6585 [Steve Henson] 6586 6587 *) Initial (incomplete) OCSP SSL support. 6588 [Steve Henson] 6589 6590 *) New function OCSP_parse_url(). This splits up a URL into its host, 6591 port and path components: primarily to parse OCSP URLs. New -url 6592 option to ocsp utility. 6593 [Steve Henson] 6594 6595 *) New nonce behavior. The return value of OCSP_check_nonce() now 6596 reflects the various checks performed. Applications can decide 6597 whether to tolerate certain situations such as an absent nonce 6598 in a response when one was present in a request: the ocsp application 6599 just prints out a warning. New function OCSP_add1_basic_nonce() 6600 this is to allow responders to include a nonce in a response even if 6601 the request is nonce-less. 6602 [Steve Henson] 6603 6604 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are 6605 skipped when using openssl x509 multiple times on a single input file, 6606 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs". 6607 [Bodo Moeller] 6608 6609 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string() 6610 set string type: to handle setting ASN1_TIME structures. Fix ca 6611 utility to correctly initialize revocation date of CRLs. 6612 [Steve Henson] 6613 6614 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override 6615 the clients preferred ciphersuites and rather use its own preferences. 6616 Should help to work around M$ SGC (Server Gated Cryptography) bug in 6617 Internet Explorer by ensuring unchanged hash method during stepup. 6618 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.) 6619 [Lutz Jaenicke] 6620 6621 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael 6622 to aes and add a new 'exist' option to print out symbols that don't 6623 appear to exist. 6624 [Steve Henson] 6625 6626 *) Additional options to ocsp utility to allow flags to be set and 6627 additional certificates supplied. 6628 [Steve Henson] 6629 6630 *) Add the option -VAfile to 'openssl ocsp', so the user can give the 6631 OCSP client a number of certificate to only verify the response 6632 signature against. 6633 [Richard Levitte] 6634 6635 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to 6636 handle the new API. Currently only ECB, CBC modes supported. Add new 6637 AES OIDs. 6638 6639 Add TLS AES ciphersuites as described in RFC3268, "Advanced 6640 Encryption Standard (AES) Ciphersuites for Transport Layer 6641 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were 6642 not enabled by default and were not part of the "ALL" ciphersuite 6643 alias because they were not yet official; they could be 6644 explicitly requested by specifying the "AESdraft" ciphersuite 6645 group alias. In the final release of OpenSSL 0.9.7, the group 6646 alias is called "AES" and is part of "ALL".) 6647 [Ben Laurie, Steve Henson, Bodo Moeller] 6648 6649 *) New function OCSP_copy_nonce() to copy nonce value (if present) from 6650 request to response. 6651 [Steve Henson] 6652 6653 *) Functions for OCSP responders. OCSP_request_onereq_count(), 6654 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info() 6655 extract information from a certificate request. OCSP_response_create() 6656 creates a response and optionally adds a basic response structure. 6657 OCSP_basic_add1_status() adds a complete single response to a basic 6658 response and returns the OCSP_SINGLERESP structure just added (to allow 6659 extensions to be included for example). OCSP_basic_add1_cert() adds a 6660 certificate to a basic response and OCSP_basic_sign() signs a basic 6661 response with various flags. New helper functions ASN1_TIME_check() 6662 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime() 6663 (converts ASN1_TIME to GeneralizedTime). 6664 [Steve Henson] 6665 6666 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}() 6667 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key 6668 structure from a certificate. X509_pubkey_digest() digests the public_key 6669 contents: this is used in various key identifiers. 6670 [Steve Henson] 6671 6672 *) Make sk_sort() tolerate a NULL argument. 6673 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>] 6674 6675 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates 6676 passed by the function are trusted implicitly. If any of them signed the 6677 response then it is assumed to be valid and is not verified. 6678 [Steve Henson] 6679 6680 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT 6681 to data. This was previously part of the PKCS7 ASN1 code. This 6682 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures. 6683 [Steve Henson, reported by Kenneth R. Robinette 6684 <support@securenetterm.com>] 6685 6686 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1 6687 routines: without these tracing memory leaks is very painful. 6688 Fix leaks in PKCS12 and PKCS7 routines. 6689 [Steve Henson] 6690 6691 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new(). 6692 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which 6693 effectively meant GeneralizedTime would never be used. Now it 6694 is initialised to -1 but X509_time_adj() now has to check the value 6695 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or 6696 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime. 6697 [Steve Henson, reported by Kenneth R. Robinette 6698 <support@securenetterm.com>] 6699 6700 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously 6701 result in a zero length in the ASN1_INTEGER structure which was 6702 not consistent with the structure when d2i_ASN1_INTEGER() was used 6703 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER() 6704 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER() 6705 where it did not print out a minus for negative ASN1_INTEGER. 6706 [Steve Henson] 6707 6708 *) Add summary printout to ocsp utility. The various functions which 6709 convert status values to strings have been renamed to: 6710 OCSP_response_status_str(), OCSP_cert_status_str() and 6711 OCSP_crl_reason_str() and are no longer static. New options 6712 to verify nonce values and to disable verification. OCSP response 6713 printout format cleaned up. 6714 [Steve Henson] 6715 6716 *) Add additional OCSP certificate checks. These are those specified 6717 in RFC2560. This consists of two separate checks: the CA of the 6718 certificate being checked must either be the OCSP signer certificate 6719 or the issuer of the OCSP signer certificate. In the latter case the 6720 OCSP signer certificate must contain the OCSP signing extended key 6721 usage. This check is performed by attempting to match the OCSP 6722 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash 6723 in the OCSP_CERTID structures of the response. 6724 [Steve Henson] 6725 6726 *) Initial OCSP certificate verification added to OCSP_basic_verify() 6727 and related routines. This uses the standard OpenSSL certificate 6728 verify routines to perform initial checks (just CA validity) and 6729 to obtain the certificate chain. Then additional checks will be 6730 performed on the chain. Currently the root CA is checked to see 6731 if it is explicitly trusted for OCSP signing. This is used to set 6732 a root CA as a global signing root: that is any certificate that 6733 chains to that CA is an acceptable OCSP signing certificate. 6734 [Steve Henson] 6735 6736 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3 6737 extensions from a separate configuration file. 6738 As when reading extensions from the main configuration file, 6739 the '-extensions ...' option may be used for specifying the 6740 section to use. 6741 [Massimiliano Pala <madwolf@comune.modena.it>] 6742 6743 *) New OCSP utility. Allows OCSP requests to be generated or 6744 read. The request can be sent to a responder and the output 6745 parsed, outputed or printed in text form. Not complete yet: 6746 still needs to check the OCSP response validity. 6747 [Steve Henson] 6748 6749 *) New subcommands for 'openssl ca': 6750 'openssl ca -status <serial>' prints the status of the cert with 6751 the given serial number (according to the index file). 6752 'openssl ca -updatedb' updates the expiry status of certificates 6753 in the index file. 6754 [Massimiliano Pala <madwolf@comune.modena.it>] 6755 6756 *) New '-newreq-nodes' command option to CA.pl. This is like 6757 '-newreq', but calls 'openssl req' with the '-nodes' option 6758 so that the resulting key is not encrypted. 6759 [Damien Miller <djm@mindrot.org>] 6760 6761 *) New configuration for the GNU Hurd. 6762 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte] 6763 6764 *) Initial code to implement OCSP basic response verify. This 6765 is currently incomplete. Currently just finds the signer's 6766 certificate and verifies the signature on the response. 6767 [Steve Henson] 6768 6769 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in 6770 value of OPENSSLDIR. This is available via the new '-d' option 6771 to 'openssl version', and is also included in 'openssl version -a'. 6772 [Bodo Moeller] 6773 6774 *) Allowing defining memory allocation callbacks that will be given 6775 file name and line number information in additional arguments 6776 (a const char* and an int). The basic functionality remains, as 6777 well as the original possibility to just replace malloc(), 6778 realloc() and free() by functions that do not know about these 6779 additional arguments. To register and find out the current 6780 settings for extended allocation functions, the following 6781 functions are provided: 6782 6783 CRYPTO_set_mem_ex_functions 6784 CRYPTO_set_locked_mem_ex_functions 6785 CRYPTO_get_mem_ex_functions 6786 CRYPTO_get_locked_mem_ex_functions 6787 6788 These work the same way as CRYPTO_set_mem_functions and friends. 6789 CRYPTO_get_[locked_]mem_functions now writes 0 where such an 6790 extended allocation function is enabled. 6791 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where 6792 a conventional allocation function is enabled. 6793 [Richard Levitte, Bodo Moeller] 6794 6795 *) Finish off removing the remaining LHASH function pointer casts. 6796 There should no longer be any prototype-casting required when using 6797 the LHASH abstraction, and any casts that remain are "bugs". See 6798 the callback types and macros at the head of lhash.h for details 6799 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example). 6800 [Geoff Thorpe] 6801 6802 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant. 6803 If /dev/[u]random devices are not available or do not return enough 6804 entropy, EGD style sockets (served by EGD or PRNGD) will automatically 6805 be queried. 6806 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and 6807 /etc/entropy will be queried once each in this sequence, quering stops 6808 when enough entropy was collected without querying more sockets. 6809 [Lutz Jaenicke] 6810 6811 *) Change the Unix RAND_poll() variant to be able to poll several 6812 random devices, as specified by DEVRANDOM, until a sufficient amount 6813 of data has been collected. We spend at most 10 ms on each file 6814 (select timeout) and read in non-blocking mode. DEVRANDOM now 6815 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom" 6816 (previously it was just the string "/dev/urandom"), so on typical 6817 platforms the 10 ms delay will never occur. 6818 Also separate out the Unix variant to its own file, rand_unix.c. 6819 For VMS, there's a currently-empty rand_vms.c. 6820 [Richard Levitte] 6821 6822 *) Move OCSP client related routines to ocsp_cl.c. These 6823 provide utility functions which an application needing 6824 to issue a request to an OCSP responder and analyse the 6825 response will typically need: as opposed to those which an 6826 OCSP responder itself would need which will be added later. 6827 6828 OCSP_request_sign() signs an OCSP request with an API similar 6829 to PKCS7_sign(). OCSP_response_status() returns status of OCSP 6830 response. OCSP_response_get1_basic() extracts basic response 6831 from response. OCSP_resp_find_status(): finds and extracts status 6832 information from an OCSP_CERTID structure (which will be created 6833 when the request structure is built). These are built from lower 6834 level functions which work on OCSP_SINGLERESP structures but 6835 wont normally be used unless the application wishes to examine 6836 extensions in the OCSP response for example. 6837 6838 Replace nonce routines with a pair of functions. 6839 OCSP_request_add1_nonce() adds a nonce value and optionally 6840 generates a random value. OCSP_check_nonce() checks the 6841 validity of the nonce in an OCSP response. 6842 [Steve Henson] 6843 6844 *) Change function OCSP_request_add() to OCSP_request_add0_id(). 6845 This doesn't copy the supplied OCSP_CERTID and avoids the 6846 need to free up the newly created id. Change return type 6847 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure. 6848 This can then be used to add extensions to the request. 6849 Deleted OCSP_request_new(), since most of its functionality 6850 is now in OCSP_REQUEST_new() (and the case insensitive name 6851 clash) apart from the ability to set the request name which 6852 will be added elsewhere. 6853 [Steve Henson] 6854 6855 *) Update OCSP API. Remove obsolete extensions argument from 6856 various functions. Extensions are now handled using the new 6857 OCSP extension code. New simple OCSP HTTP function which 6858 can be used to send requests and parse the response. 6859 [Steve Henson] 6860 6861 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new 6862 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN 6863 uses the special reorder version of SET OF to sort the attributes 6864 and reorder them to match the encoded order. This resolves a long 6865 standing problem: a verify on a PKCS7 structure just after signing 6866 it used to fail because the attribute order did not match the 6867 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes: 6868 it uses the received order. This is necessary to tolerate some broken 6869 software that does not order SET OF. This is handled by encoding 6870 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class) 6871 to produce the required SET OF. 6872 [Steve Henson] 6873 6874 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and 6875 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header 6876 files to get correct declarations of the ASN.1 item variables. 6877 [Richard Levitte] 6878 6879 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many 6880 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs: 6881 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was 6882 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i(). 6883 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant 6884 ASN1_ITEM and no wrapper functions. 6885 [Steve Henson] 6886 6887 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These 6888 replace the old function pointer based I/O routines. Change most of 6889 the *_d2i_bio() and *_d2i_fp() functions to use these. 6890 [Steve Henson] 6891 6892 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor 6893 lines, recognice more "algorithms" that can be deselected, and make 6894 it complain about algorithm deselection that isn't recognised. 6895 [Richard Levitte] 6896 6897 *) New ASN1 functions to handle dup, sign, verify, digest, pack and 6898 unpack operations in terms of ASN1_ITEM. Modify existing wrappers 6899 to use new functions. Add NO_ASN1_OLD which can be set to remove 6900 some old style ASN1 functions: this can be used to determine if old 6901 code will still work when these eventually go away. 6902 [Steve Henson] 6903 6904 *) New extension functions for OCSP structures, these follow the 6905 same conventions as certificates and CRLs. 6906 [Steve Henson] 6907 6908 *) New function X509V3_add1_i2d(). This automatically encodes and 6909 adds an extension. Its behaviour can be customised with various 6910 flags to append, replace or delete. Various wrappers added for 6911 certifcates and CRLs. 6912 [Steve Henson] 6913 6914 *) Fix to avoid calling the underlying ASN1 print routine when 6915 an extension cannot be parsed. Correct a typo in the 6916 OCSP_SERVICELOC extension. Tidy up print OCSP format. 6917 [Steve Henson] 6918 6919 *) Make mkdef.pl parse some of the ASN1 macros and add apropriate 6920 entries for variables. 6921 [Steve Henson] 6922 6923 *) Add functionality to apps/openssl.c for detecting locking 6924 problems: As the program is single-threaded, all we have 6925 to do is register a locking callback using an array for 6926 storing which locks are currently held by the program. 6927 [Bodo Moeller] 6928 6929 *) Use a lock around the call to CRYPTO_get_ex_new_index() in 6930 SSL_get_ex_data_X509_STORE_idx(), which is used in 6931 ssl_verify_cert_chain() and thus can be called at any time 6932 during TLS/SSL handshakes so that thread-safety is essential. 6933 Unfortunately, the ex_data design is not at all suited 6934 for multi-threaded use, so it probably should be abolished. 6935 [Bodo Moeller] 6936 6937 *) Added Broadcom "ubsec" ENGINE to OpenSSL. 6938 [Broadcom, tweaked and integrated by Geoff Thorpe] 6939 6940 *) Move common extension printing code to new function 6941 X509V3_print_extensions(). Reorganise OCSP print routines and 6942 implement some needed OCSP ASN1 functions. Add OCSP extensions. 6943 [Steve Henson] 6944 6945 *) New function X509_signature_print() to remove duplication in some 6946 print routines. 6947 [Steve Henson] 6948 6949 *) Add a special meaning when SET OF and SEQUENCE OF flags are both 6950 set (this was treated exactly the same as SET OF previously). This 6951 is used to reorder the STACK representing the structure to match the 6952 encoding. This will be used to get round a problem where a PKCS7 6953 structure which was signed could not be verified because the STACK 6954 order did not reflect the encoded order. 6955 [Steve Henson] 6956 6957 *) Reimplement the OCSP ASN1 module using the new code. 6958 [Steve Henson] 6959 6960 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure 6961 for its ASN1 operations. The old style function pointers still exist 6962 for now but they will eventually go away. 6963 [Steve Henson] 6964 6965 *) Merge in replacement ASN1 code from the ASN1 branch. This almost 6966 completely replaces the old ASN1 functionality with a table driven 6967 encoder and decoder which interprets an ASN1_ITEM structure describing 6968 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is 6969 largely maintained. Almost all of the old asn1_mac.h macro based ASN1 6970 has also been converted to the new form. 6971 [Steve Henson] 6972 6973 *) Change BN_mod_exp_recp so that negative moduli are tolerated 6974 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set 6975 so that BN_mod_exp_mont and BN_mod_exp_mont_word work 6976 for negative moduli. 6977 [Bodo Moeller] 6978 6979 *) Fix BN_uadd and BN_usub: Always return non-negative results instead 6980 of not touching the result's sign bit. 6981 [Bodo Moeller] 6982 6983 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be 6984 set. 6985 [Bodo Moeller] 6986 6987 *) Changed the LHASH code to use prototypes for callbacks, and created 6988 macros to declare and implement thin (optionally static) functions 6989 that provide type-safety and avoid function pointer casting for the 6990 type-specific callbacks. 6991 [Geoff Thorpe] 6992 6993 *) Added Kerberos Cipher Suites to be used with TLS, as written in 6994 RFC 2712. 6995 [Veers Staats <staatsvr@asc.hpc.mil>, 6996 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte] 6997 6998 *) Reformat the FAQ so the different questions and answers can be divided 6999 in sections depending on the subject. 7000 [Richard Levitte] 7001 7002 *) Have the zlib compression code load ZLIB.DLL dynamically under 7003 Windows. 7004 [Richard Levitte] 7005 7006 *) New function BN_mod_sqrt for computing square roots modulo a prime 7007 (using the probabilistic Tonelli-Shanks algorithm unless 7008 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can 7009 be handled deterministically). 7010 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 7011 7012 *) Make BN_mod_inverse faster by explicitly handling small quotients 7013 in the Euclid loop. (Speed gain about 20% for small moduli [256 or 7014 512 bits], about 30% for larger ones [1024 or 2048 bits].) 7015 [Bodo Moeller] 7016 7017 *) New function BN_kronecker. 7018 [Bodo Moeller] 7019 7020 *) Fix BN_gcd so that it works on negative inputs; the result is 7021 positive unless both parameters are zero. 7022 Previously something reasonably close to an infinite loop was 7023 possible because numbers could be growing instead of shrinking 7024 in the implementation of Euclid's algorithm. 7025 [Bodo Moeller] 7026 7027 *) Fix BN_is_word() and BN_is_one() macros to take into account the 7028 sign of the number in question. 7029 7030 Fix BN_is_word(a,w) to work correctly for w == 0. 7031 7032 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w) 7033 because its test if the absolute value of 'a' equals 'w'. 7034 Note that BN_abs_is_word does *not* handle w == 0 reliably; 7035 it exists mostly for use in the implementations of BN_is_zero(), 7036 BN_is_one(), and BN_is_word(). 7037 [Bodo Moeller] 7038 7039 *) New function BN_swap. 7040 [Bodo Moeller] 7041 7042 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that 7043 the exponentiation functions are more likely to produce reasonable 7044 results on negative inputs. 7045 [Bodo Moeller] 7046 7047 *) Change BN_mod_mul so that the result is always non-negative. 7048 Previously, it could be negative if one of the factors was negative; 7049 I don't think anyone really wanted that behaviour. 7050 [Bodo Moeller] 7051 7052 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c 7053 (except for exponentiation, which stays in crypto/bn/bn_exp.c, 7054 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c) 7055 and add new functions: 7056 7057 BN_nnmod 7058 BN_mod_sqr 7059 BN_mod_add 7060 BN_mod_add_quick 7061 BN_mod_sub 7062 BN_mod_sub_quick 7063 BN_mod_lshift1 7064 BN_mod_lshift1_quick 7065 BN_mod_lshift 7066 BN_mod_lshift_quick 7067 7068 These functions always generate non-negative results. 7069 7070 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r 7071 such that |m| < r < 0, BN_nnmod will output rem + |m| instead). 7072 7073 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as 7074 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b] 7075 be reduced modulo m. 7076 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 7077 7078#if 0 7079 The following entry accidentily appeared in the CHANGES file 7080 distributed with OpenSSL 0.9.7. The modifications described in 7081 it do *not* apply to OpenSSL 0.9.7. 7082 7083 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 7084 was actually never needed) and in BN_mul(). The removal in BN_mul() 7085 required a small change in bn_mul_part_recursive() and the addition 7086 of the functions bn_cmp_part_words(), bn_sub_part_words() and 7087 bn_add_part_words(), which do the same thing as bn_cmp_words(), 7088 bn_sub_words() and bn_add_words() except they take arrays with 7089 differing sizes. 7090 [Richard Levitte] 7091#endif 7092 7093 *) In 'openssl passwd', verify passwords read from the terminal 7094 unless the '-salt' option is used (which usually means that 7095 verification would just waste user's time since the resulting 7096 hash is going to be compared with some given password hash) 7097 or the new '-noverify' option is used. 7098 7099 This is an incompatible change, but it does not affect 7100 non-interactive use of 'openssl passwd' (passwords on the command 7101 line, '-stdin' option, '-in ...' option) and thus should not 7102 cause any problems. 7103 [Bodo Moeller] 7104 7105 *) Remove all references to RSAref, since there's no more need for it. 7106 [Richard Levitte] 7107 7108 *) Make DSO load along a path given through an environment variable 7109 (SHLIB_PATH) with shl_load(). 7110 [Richard Levitte] 7111 7112 *) Constify the ENGINE code as a result of BIGNUM constification. 7113 Also constify the RSA code and most things related to it. In a 7114 few places, most notable in the depth of the ASN.1 code, ugly 7115 casts back to non-const were required (to be solved at a later 7116 time) 7117 [Richard Levitte] 7118 7119 *) Make it so the openssl application has all engines loaded by default. 7120 [Richard Levitte] 7121 7122 *) Constify the BIGNUM routines a little more. 7123 [Richard Levitte] 7124 7125 *) Add the following functions: 7126 7127 ENGINE_load_cswift() 7128 ENGINE_load_chil() 7129 ENGINE_load_atalla() 7130 ENGINE_load_nuron() 7131 ENGINE_load_builtin_engines() 7132 7133 That way, an application can itself choose if external engines that 7134 are built-in in OpenSSL shall ever be used or not. The benefit is 7135 that applications won't have to be linked with libdl or other dso 7136 libraries unless it's really needed. 7137 7138 Changed 'openssl engine' to load all engines on demand. 7139 Changed the engine header files to avoid the duplication of some 7140 declarations (they differed!). 7141 [Richard Levitte] 7142 7143 *) 'openssl engine' can now list capabilities. 7144 [Richard Levitte] 7145 7146 *) Better error reporting in 'openssl engine'. 7147 [Richard Levitte] 7148 7149 *) Never call load_dh_param(NULL) in s_server. 7150 [Bodo Moeller] 7151 7152 *) Add engine application. It can currently list engines by name and 7153 identity, and test if they are actually available. 7154 [Richard Levitte] 7155 7156 *) Improve RPM specification file by forcing symbolic linking and making 7157 sure the installed documentation is also owned by root.root. 7158 [Damien Miller <djm@mindrot.org>] 7159 7160 *) Give the OpenSSL applications more possibilities to make use of 7161 keys (public as well as private) handled by engines. 7162 [Richard Levitte] 7163 7164 *) Add OCSP code that comes from CertCo. 7165 [Richard Levitte] 7166 7167 *) Add VMS support for the Rijndael code. 7168 [Richard Levitte] 7169 7170 *) Added untested support for Nuron crypto accelerator. 7171 [Ben Laurie] 7172 7173 *) Add support for external cryptographic devices. This code was 7174 previously distributed separately as the "engine" branch. 7175 [Geoff Thorpe, Richard Levitte] 7176 7177 *) Rework the filename-translation in the DSO code. It is now possible to 7178 have far greater control over how a "name" is turned into a filename 7179 depending on the operating environment and any oddities about the 7180 different shared library filenames on each system. 7181 [Geoff Thorpe] 7182 7183 *) Support threads on FreeBSD-elf in Configure. 7184 [Richard Levitte] 7185 7186 *) Fix for SHA1 assembly problem with MASM: it produces 7187 warnings about corrupt line number information when assembling 7188 with debugging information. This is caused by the overlapping 7189 of two sections. 7190 [Bernd Matthes <mainbug@celocom.de>, Steve Henson] 7191 7192 *) NCONF changes. 7193 NCONF_get_number() has no error checking at all. As a replacement, 7194 NCONF_get_number_e() is defined (_e for "error checking") and is 7195 promoted strongly. The old NCONF_get_number is kept around for 7196 binary backward compatibility. 7197 Make it possible for methods to load from something other than a BIO, 7198 by providing a function pointer that is given a name instead of a BIO. 7199 For example, this could be used to load configuration data from an 7200 LDAP server. 7201 [Richard Levitte] 7202 7203 *) Fix for non blocking accept BIOs. Added new I/O special reason 7204 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs 7205 with non blocking I/O was not possible because no retry code was 7206 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover 7207 this case. 7208 [Steve Henson] 7209 7210 *) Added the beginnings of Rijndael support. 7211 [Ben Laurie] 7212 7213 *) Fix for bug in DirectoryString mask setting. Add support for 7214 X509_NAME_print_ex() in 'req' and X509_print_ex() function 7215 to allow certificate printing to more controllable, additional 7216 'certopt' option to 'x509' to allow new printing options to be 7217 set. 7218 [Steve Henson] 7219 7220 *) Clean old EAY MD5 hack from e_os.h. 7221 [Richard Levitte] 7222 7223 Changes between 0.9.6l and 0.9.6m [17 Mar 2004] 7224 7225 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 7226 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 7227 [Joe Orton, Steve Henson] 7228 7229 Changes between 0.9.6k and 0.9.6l [04 Nov 2003] 7230 7231 *) Fix additional bug revealed by the NISCC test suite: 7232 7233 Stop bug triggering large recursion when presented with 7234 certain ASN.1 tags (CVE-2003-0851) 7235 [Steve Henson] 7236 7237 Changes between 0.9.6j and 0.9.6k [30 Sep 2003] 7238 7239 *) Fix various bugs revealed by running the NISCC test suite: 7240 7241 Stop out of bounds reads in the ASN1 code when presented with 7242 invalid tags (CVE-2003-0543 and CVE-2003-0544). 7243 7244 If verify callback ignores invalid public key errors don't try to check 7245 certificate signature with the NULL public key. 7246 7247 [Steve Henson] 7248 7249 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 7250 if the server requested one: as stated in TLS 1.0 and SSL 3.0 7251 specifications. 7252 [Steve Henson] 7253 7254 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 7255 extra data after the compression methods not only for TLS 1.0 7256 but also for SSL 3.0 (as required by the specification). 7257 [Bodo Moeller; problem pointed out by Matthias Loepfe] 7258 7259 *) Change X509_certificate_type() to mark the key as exported/exportable 7260 when it's 512 *bits* long, not 512 bytes. 7261 [Richard Levitte] 7262 7263 Changes between 0.9.6i and 0.9.6j [10 Apr 2003] 7264 7265 *) Countermeasure against the Klima-Pokorny-Rosa extension of 7266 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 7267 a protocol version number mismatch like a decryption error 7268 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 7269 [Bodo Moeller] 7270 7271 *) Turn on RSA blinding by default in the default implementation 7272 to avoid a timing attack. Applications that don't want it can call 7273 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 7274 They would be ill-advised to do so in most cases. 7275 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 7276 7277 *) Change RSA blinding code so that it works when the PRNG is not 7278 seeded (in this case, the secret RSA exponent is abused as 7279 an unpredictable seed -- if it is not unpredictable, there 7280 is no point in blinding anyway). Make RSA blinding thread-safe 7281 by remembering the creator's thread ID in rsa->blinding and 7282 having all other threads use local one-time blinding factors 7283 (this requires more computation than sharing rsa->blinding, but 7284 avoids excessive locking; and if an RSA object is not shared 7285 between threads, blinding will still be very fast). 7286 [Bodo Moeller] 7287 7288 Changes between 0.9.6h and 0.9.6i [19 Feb 2003] 7289 7290 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 7291 via timing by performing a MAC computation even if incorrrect 7292 block cipher padding has been found. This is a countermeasure 7293 against active attacks where the attacker has to distinguish 7294 between bad padding and a MAC verification error. (CVE-2003-0078) 7295 7296 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 7297 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 7298 Martin Vuagnoux (EPFL, Ilion)] 7299 7300 Changes between 0.9.6g and 0.9.6h [5 Dec 2002] 7301 7302 *) New function OPENSSL_cleanse(), which is used to cleanse a section of 7303 memory from it's contents. This is done with a counter that will 7304 place alternating values in each byte. This can be used to solve 7305 two issues: 1) the removal of calls to memset() by highly optimizing 7306 compilers, and 2) cleansing with other values than 0, since those can 7307 be read through on certain media, for example a swap space on disk. 7308 [Geoff Thorpe] 7309 7310 *) Bugfix: client side session caching did not work with external caching, 7311 because the session->cipher setting was not restored when reloading 7312 from the external cache. This problem was masked, when 7313 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set. 7314 (Found by Steve Haslam <steve@araqnid.ddts.net>.) 7315 [Lutz Jaenicke] 7316 7317 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total 7318 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33. 7319 [Zeev Lieber <zeev-l@yahoo.com>] 7320 7321 *) Undo an undocumented change introduced in 0.9.6e which caused 7322 repeated calls to OpenSSL_add_all_ciphers() and 7323 OpenSSL_add_all_digests() to be ignored, even after calling 7324 EVP_cleanup(). 7325 [Richard Levitte] 7326 7327 *) Change the default configuration reader to deal with last line not 7328 being properly terminated. 7329 [Richard Levitte] 7330 7331 *) Change X509_NAME_cmp() so it applies the special rules on handling 7332 DN values that are of type PrintableString, as well as RDNs of type 7333 emailAddress where the value has the type ia5String. 7334 [stefank@valicert.com via Richard Levitte] 7335 7336 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half 7337 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently 7338 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be 7339 the bitwise-OR of the two for use by the majority of applications 7340 wanting this behaviour, and update the docs. The documented 7341 behaviour and actual behaviour were inconsistent and had been 7342 changing anyway, so this is more a bug-fix than a behavioural 7343 change. 7344 [Geoff Thorpe, diagnosed by Nadav Har'El] 7345 7346 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c 7347 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes). 7348 [Bodo Moeller] 7349 7350 *) Fix initialization code race conditions in 7351 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(), 7352 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(), 7353 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(), 7354 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(), 7355 ssl2_get_cipher_by_char(), 7356 ssl3_get_cipher_by_char(). 7357 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller] 7358 7359 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after 7360 the cached sessions are flushed, as the remove_cb() might use ex_data 7361 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com> 7362 (see [openssl.org #212]). 7363 [Geoff Thorpe, Lutz Jaenicke] 7364 7365 *) Fix typo in OBJ_txt2obj which incorrectly passed the content 7366 length, instead of the encoding length to d2i_ASN1_OBJECT. 7367 [Steve Henson] 7368 7369 Changes between 0.9.6f and 0.9.6g [9 Aug 2002] 7370 7371 *) [In 0.9.6g-engine release:] 7372 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall'). 7373 [Lynn Gazis <lgazis@rainbow.com>] 7374 7375 Changes between 0.9.6e and 0.9.6f [8 Aug 2002] 7376 7377 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX 7378 and get fix the header length calculation. 7379 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>, 7380 Alon Kantor <alonk@checkpoint.com> (and others), 7381 Steve Henson] 7382 7383 *) Use proper error handling instead of 'assertions' in buffer 7384 overflow checks added in 0.9.6e. This prevents DoS (the 7385 assertions could call abort()). 7386 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller] 7387 7388 Changes between 0.9.6d and 0.9.6e [30 Jul 2002] 7389 7390 *) Add various sanity checks to asn1_get_length() to reject 7391 the ASN1 length bytes if they exceed sizeof(long), will appear 7392 negative or the content length exceeds the length of the 7393 supplied buffer. 7394 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 7395 7396 *) Fix cipher selection routines: ciphers without encryption had no flags 7397 for the cipher strength set and where therefore not handled correctly 7398 by the selection routines (PR #130). 7399 [Lutz Jaenicke] 7400 7401 *) Fix EVP_dsa_sha macro. 7402 [Nils Larsch] 7403 7404 *) New option 7405 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS 7406 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure 7407 that was added in OpenSSL 0.9.6d. 7408 7409 As the countermeasure turned out to be incompatible with some 7410 broken SSL implementations, the new option is part of SSL_OP_ALL. 7411 SSL_OP_ALL is usually employed when compatibility with weird SSL 7412 implementations is desired (e.g. '-bugs' option to 's_client' and 7413 's_server'), so the new option is automatically set in many 7414 applications. 7415 [Bodo Moeller] 7416 7417 *) Changes in security patch: 7418 7419 Changes marked "(CHATS)" were sponsored by the Defense Advanced 7420 Research Projects Agency (DARPA) and Air Force Research Laboratory, 7421 Air Force Materiel Command, USAF, under agreement number 7422 F30602-01-2-0537. 7423 7424 *) Add various sanity checks to asn1_get_length() to reject 7425 the ASN1 length bytes if they exceed sizeof(long), will appear 7426 negative or the content length exceeds the length of the 7427 supplied buffer. (CVE-2002-0659) 7428 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 7429 7430 *) Assertions for various potential buffer overflows, not known to 7431 happen in practice. 7432 [Ben Laurie (CHATS)] 7433 7434 *) Various temporary buffers to hold ASCII versions of integers were 7435 too small for 64 bit platforms. (CVE-2002-0655) 7436 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)> 7437 7438 *) Remote buffer overflow in SSL3 protocol - an attacker could 7439 supply an oversized session ID to a client. (CVE-2002-0656) 7440 [Ben Laurie (CHATS)] 7441 7442 *) Remote buffer overflow in SSL2 protocol - an attacker could 7443 supply an oversized client master key. (CVE-2002-0656) 7444 [Ben Laurie (CHATS)] 7445 7446 Changes between 0.9.6c and 0.9.6d [9 May 2002] 7447 7448 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not 7449 encoded as NULL) with id-dsa-with-sha1. 7450 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller] 7451 7452 *) Check various X509_...() return values in apps/req.c. 7453 [Nils Larsch <nla@trustcenter.de>] 7454 7455 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines: 7456 an end-of-file condition would erronously be flagged, when the CRLF 7457 was just at the end of a processed block. The bug was discovered when 7458 processing data through a buffering memory BIO handing the data to a 7459 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov 7460 <ptsekov@syntrex.com> and Nedelcho Stanev. 7461 [Lutz Jaenicke] 7462 7463 *) Implement a countermeasure against a vulnerability recently found 7464 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment 7465 before application data chunks to avoid the use of known IVs 7466 with data potentially chosen by the attacker. 7467 [Bodo Moeller] 7468 7469 *) Fix length checks in ssl3_get_client_hello(). 7470 [Bodo Moeller] 7471 7472 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently 7473 to prevent ssl3_read_internal() from incorrectly assuming that 7474 ssl3_read_bytes() found application data while handshake 7475 processing was enabled when in fact s->s3->in_read_app_data was 7476 merely automatically cleared during the initial handshake. 7477 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>] 7478 7479 *) Fix object definitions for Private and Enterprise: they were not 7480 recognized in their shortname (=lowercase) representation. Extend 7481 obj_dat.pl to issue an error when using undefined keywords instead 7482 of silently ignoring the problem (Svenning Sorensen 7483 <sss@sss.dnsalias.net>). 7484 [Lutz Jaenicke] 7485 7486 *) Fix DH_generate_parameters() so that it works for 'non-standard' 7487 generators, i.e. generators other than 2 and 5. (Previously, the 7488 code did not properly initialise the 'add' and 'rem' values to 7489 BN_generate_prime().) 7490 7491 In the new general case, we do not insist that 'generator' is 7492 actually a primitive root: This requirement is rather pointless; 7493 a generator of the order-q subgroup is just as good, if not 7494 better. 7495 [Bodo Moeller] 7496 7497 *) Map new X509 verification errors to alerts. Discovered and submitted by 7498 Tom Wu <tom@arcot.com>. 7499 [Lutz Jaenicke] 7500 7501 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from 7502 returning non-zero before the data has been completely received 7503 when using non-blocking I/O. 7504 [Bodo Moeller; problem pointed out by John Hughes] 7505 7506 *) Some of the ciphers missed the strength entry (SSL_LOW etc). 7507 [Ben Laurie, Lutz Jaenicke] 7508 7509 *) Fix bug in SSL_clear(): bad sessions were not removed (found by 7510 Yoram Zahavi <YoramZ@gilian.com>). 7511 [Lutz Jaenicke] 7512 7513 *) Add information about CygWin 1.3 and on, and preserve proper 7514 configuration for the versions before that. 7515 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 7516 7517 *) Make removal from session cache (SSL_CTX_remove_session()) more robust: 7518 check whether we deal with a copy of a session and do not delete from 7519 the cache in this case. Problem reported by "Izhar Shoshani Levi" 7520 <izhar@checkpoint.com>. 7521 [Lutz Jaenicke] 7522 7523 *) Do not store session data into the internal session cache, if it 7524 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP 7525 flag is set). Proposed by Aslam <aslam@funk.com>. 7526 [Lutz Jaenicke] 7527 7528 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested 7529 value is 0. 7530 [Richard Levitte] 7531 7532 *) [In 0.9.6d-engine release:] 7533 Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 7534 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 7535 7536 *) Add the configuration target linux-s390x. 7537 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte] 7538 7539 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of 7540 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag 7541 variable as an indication that a ClientHello message has been 7542 received. As the flag value will be lost between multiple 7543 invocations of ssl3_accept when using non-blocking I/O, the 7544 function may not be aware that a handshake has actually taken 7545 place, thus preventing a new session from being added to the 7546 session cache. 7547 7548 To avoid this problem, we now set s->new_session to 2 instead of 7549 using a local variable. 7550 [Lutz Jaenicke, Bodo Moeller] 7551 7552 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c) 7553 if the SSL_R_LENGTH_MISMATCH error is detected. 7554 [Geoff Thorpe, Bodo Moeller] 7555 7556 *) New 'shared_ldflag' column in Configure platform table. 7557 [Richard Levitte] 7558 7559 *) Fix EVP_CIPHER_mode macro. 7560 ["Dan S. Camper" <dan@bti.net>] 7561 7562 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown 7563 type, we must throw them away by setting rr->length to 0. 7564 [D P Chang <dpc@qualys.com>] 7565 7566 Changes between 0.9.6b and 0.9.6c [21 dec 2001] 7567 7568 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl 7569 <Dominikus.Scherkl@biodata.com>. (The previous implementation 7570 worked incorrectly for those cases where range = 10..._2 and 7571 3*range is two bits longer than range.) 7572 [Bodo Moeller] 7573 7574 *) Only add signing time to PKCS7 structures if it is not already 7575 present. 7576 [Steve Henson] 7577 7578 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce", 7579 OBJ_ld_ce should be OBJ_id_ce. 7580 Also some ip-pda OIDs in crypto/objects/objects.txt were 7581 incorrect (cf. RFC 3039). 7582 [Matt Cooper, Frederic Giudicelli, Bodo Moeller] 7583 7584 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid() 7585 returns early because it has nothing to do. 7586 [Andy Schneider <andy.schneider@bjss.co.uk>] 7587 7588 *) [In 0.9.6c-engine release:] 7589 Fix mutex callback return values in crypto/engine/hw_ncipher.c. 7590 [Andy Schneider <andy.schneider@bjss.co.uk>] 7591 7592 *) [In 0.9.6c-engine release:] 7593 Add support for Cryptographic Appliance's keyserver technology. 7594 (Use engine 'keyclient') 7595 [Cryptographic Appliances and Geoff Thorpe] 7596 7597 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89' 7598 is called via tools/c89.sh because arguments have to be 7599 rearranged (all '-L' options must appear before the first object 7600 modules). 7601 [Richard Shapiro <rshapiro@abinitio.com>] 7602 7603 *) [In 0.9.6c-engine release:] 7604 Add support for Broadcom crypto accelerator cards, backported 7605 from 0.9.7. 7606 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox] 7607 7608 *) [In 0.9.6c-engine release:] 7609 Add support for SureWare crypto accelerator cards from 7610 Baltimore Technologies. (Use engine 'sureware') 7611 [Baltimore Technologies and Mark Cox] 7612 7613 *) [In 0.9.6c-engine release:] 7614 Add support for crypto accelerator cards from Accelerated 7615 Encryption Processing, www.aep.ie. (Use engine 'aep') 7616 [AEP Inc. and Mark Cox] 7617 7618 *) Add a configuration entry for gcc on UnixWare. 7619 [Gary Benson <gbenson@redhat.com>] 7620 7621 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake 7622 messages are stored in a single piece (fixed-length part and 7623 variable-length part combined) and fix various bugs found on the way. 7624 [Bodo Moeller] 7625 7626 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname() 7627 instead. BIO_gethostbyname() does not know what timeouts are 7628 appropriate, so entries would stay in cache even when they have 7629 become invalid. 7630 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com> 7631 7632 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when 7633 faced with a pathologically small ClientHello fragment that does 7634 not contain client_version: Instead of aborting with an error, 7635 simply choose the highest available protocol version (i.e., 7636 TLS 1.0 unless it is disabled). In practice, ClientHello 7637 messages are never sent like this, but this change gives us 7638 strictly correct behaviour at least for TLS. 7639 [Bodo Moeller] 7640 7641 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear() 7642 never resets s->method to s->ctx->method when called from within 7643 one of the SSL handshake functions. 7644 [Bodo Moeller; problem pointed out by Niko Baric] 7645 7646 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert 7647 (sent using the client's version number) if client_version is 7648 smaller than the protocol version in use. Also change 7649 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if 7650 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then 7651 the client will at least see that alert. 7652 [Bodo Moeller] 7653 7654 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation 7655 correctly. 7656 [Bodo Moeller] 7657 7658 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a 7659 client receives HelloRequest while in a handshake. 7660 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>] 7661 7662 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C 7663 should end in 'break', not 'goto end' which circuments various 7664 cleanups done in state SSL_ST_OK. But session related stuff 7665 must be disabled for SSL_ST_OK in the case that we just sent a 7666 HelloRequest. 7667 7668 Also avoid some overhead by not calling ssl_init_wbio_buffer() 7669 before just sending a HelloRequest. 7670 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>] 7671 7672 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't 7673 reveal whether illegal block cipher padding was found or a MAC 7674 verification error occured. (Neither SSLerr() codes nor alerts 7675 are directly visible to potential attackers, but the information 7676 may leak via logfiles.) 7677 7678 Similar changes are not required for the SSL 2.0 implementation 7679 because the number of padding bytes is sent in clear for SSL 2.0, 7680 and the extra bytes are just ignored. However ssl/s2_pkt.c 7681 failed to verify that the purported number of padding bytes is in 7682 the legal range. 7683 [Bodo Moeller] 7684 7685 *) Add OpenUNIX-8 support including shared libraries 7686 (Boyd Lynn Gerber <gerberb@zenez.com>). 7687 [Lutz Jaenicke] 7688 7689 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid 7690 'wristwatch attack' using huge encoding parameters (cf. 7691 James H. Manger's CRYPTO 2001 paper). Note that the 7692 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use 7693 encoding parameters and hence was not vulnerable. 7694 [Bodo Moeller] 7695 7696 *) BN_sqr() bug fix. 7697 [Ulf M��ller, reported by Jim Ellis <jim.ellis@cavium.com>] 7698 7699 *) Rabin-Miller test analyses assume uniformly distributed witnesses, 7700 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand() 7701 followed by modular reduction. 7702 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>] 7703 7704 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range() 7705 equivalent based on BN_pseudo_rand() instead of BN_rand(). 7706 [Bodo Moeller] 7707 7708 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB). 7709 This function was broken, as the check for a new client hello message 7710 to handle SGC did not allow these large messages. 7711 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.) 7712 [Lutz Jaenicke] 7713 7714 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long](). 7715 [Lutz Jaenicke] 7716 7717 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl() 7718 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>). 7719 [Lutz Jaenicke] 7720 7721 *) Rework the configuration and shared library support for Tru64 Unix. 7722 The configuration part makes use of modern compiler features and 7723 still retains old compiler behavior for those that run older versions 7724 of the OS. The shared library support part includes a variant that 7725 uses the RPATH feature, and is available through the special 7726 configuration target "alpha-cc-rpath", which will never be selected 7727 automatically. 7728 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte] 7729 7730 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message() 7731 with the same message size as in ssl3_get_certificate_request(). 7732 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest 7733 messages might inadvertently be reject as too long. 7734 [Petr Lampa <lampa@fee.vutbr.cz>] 7735 7736 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX). 7737 [Andy Polyakov] 7738 7739 *) Modified SSL library such that the verify_callback that has been set 7740 specificly for an SSL object with SSL_set_verify() is actually being 7741 used. Before the change, a verify_callback set with this function was 7742 ignored and the verify_callback() set in the SSL_CTX at the time of 7743 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced 7744 to allow the necessary settings. 7745 [Lutz Jaenicke] 7746 7747 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c 7748 explicitly to NULL, as at least on Solaris 8 this seems not always to be 7749 done automatically (in contradiction to the requirements of the C 7750 standard). This made problems when used from OpenSSH. 7751 [Lutz Jaenicke] 7752 7753 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored 7754 dh->length and always used 7755 7756 BN_rand_range(priv_key, dh->p). 7757 7758 BN_rand_range() is not necessary for Diffie-Hellman, and this 7759 specific range makes Diffie-Hellman unnecessarily inefficient if 7760 dh->length (recommended exponent length) is much smaller than the 7761 length of dh->p. We could use BN_rand_range() if the order of 7762 the subgroup was stored in the DH structure, but we only have 7763 dh->length. 7764 7765 So switch back to 7766 7767 BN_rand(priv_key, l, ...) 7768 7769 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1 7770 otherwise. 7771 [Bodo Moeller] 7772 7773 *) In 7774 7775 RSA_eay_public_encrypt 7776 RSA_eay_private_decrypt 7777 RSA_eay_private_encrypt (signing) 7778 RSA_eay_public_decrypt (signature verification) 7779 7780 (default implementations for RSA_public_encrypt, 7781 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt), 7782 always reject numbers >= n. 7783 [Bodo Moeller] 7784 7785 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2 7786 to synchronize access to 'locking_thread'. This is necessary on 7787 systems where access to 'locking_thread' (an 'unsigned long' 7788 variable) is not atomic. 7789 [Bodo Moeller] 7790 7791 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID 7792 *before* setting the 'crypto_lock_rand' flag. The previous code had 7793 a race condition if 0 is a valid thread ID. 7794 [Travis Vitek <vitek@roguewave.com>] 7795 7796 *) Add support for shared libraries under Irix. 7797 [Albert Chin-A-Young <china@thewrittenword.com>] 7798 7799 *) Add configuration option to build on Linux on both big-endian and 7800 little-endian MIPS. 7801 [Ralf Baechle <ralf@uni-koblenz.de>] 7802 7803 *) Add the possibility to create shared libraries on HP-UX. 7804 [Richard Levitte] 7805 7806 Changes between 0.9.6a and 0.9.6b [9 Jul 2001] 7807 7808 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c) 7809 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by 7810 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>: 7811 PRNG state recovery was possible based on the output of 7812 one PRNG request appropriately sized to gain knowledge on 7813 'md' followed by enough consecutive 1-byte PRNG requests 7814 to traverse all of 'state'. 7815 7816 1. When updating 'md_local' (the current thread's copy of 'md') 7817 during PRNG output generation, hash all of the previous 7818 'md_local' value, not just the half used for PRNG output. 7819 7820 2. Make the number of bytes from 'state' included into the hash 7821 independent from the number of PRNG bytes requested. 7822 7823 The first measure alone would be sufficient to avoid 7824 Markku-Juhani's attack. (Actually it had never occurred 7825 to me that the half of 'md_local' used for chaining was the 7826 half from which PRNG output bytes were taken -- I had always 7827 assumed that the secret half would be used.) The second 7828 measure makes sure that additional data from 'state' is never 7829 mixed into 'md_local' in small portions; this heuristically 7830 further strengthens the PRNG. 7831 [Bodo Moeller] 7832 7833 *) Fix crypto/bn/asm/mips3.s. 7834 [Andy Polyakov] 7835 7836 *) When only the key is given to "enc", the IV is undefined. Print out 7837 an error message in this case. 7838 [Lutz Jaenicke] 7839 7840 *) Handle special case when X509_NAME is empty in X509 printing routines. 7841 [Steve Henson] 7842 7843 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are 7844 positive and less than q. 7845 [Bodo Moeller] 7846 7847 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is 7848 used: it isn't thread safe and the add_lock_callback should handle 7849 that itself. 7850 [Paul Rose <Paul.Rose@bridge.com>] 7851 7852 *) Verify that incoming data obeys the block size in 7853 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c). 7854 [Bodo Moeller] 7855 7856 *) Fix OAEP check. 7857 [Ulf M��ller, Bodo M��ller] 7858 7859 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5 7860 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5 7861 when fixing the server behaviour for backwards-compatible 'client 7862 hello' messages. (Note that the attack is impractical against 7863 SSL 3.0 and TLS 1.0 anyway because length and version checking 7864 means that the probability of guessing a valid ciphertext is 7865 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98 7866 paper.) 7867 7868 Before 0.9.5, the countermeasure (hide the error by generating a 7869 random 'decryption result') did not work properly because 7870 ERR_clear_error() was missing, meaning that SSL_get_error() would 7871 detect the supposedly ignored error. 7872 7873 Both problems are now fixed. 7874 [Bodo Moeller] 7875 7876 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096 7877 (previously it was 1024). 7878 [Bodo Moeller] 7879 7880 *) Fix for compatibility mode trust settings: ignore trust settings 7881 unless some valid trust or reject settings are present. 7882 [Steve Henson] 7883 7884 *) Fix for blowfish EVP: its a variable length cipher. 7885 [Steve Henson] 7886 7887 *) Fix various bugs related to DSA S/MIME verification. Handle missing 7888 parameters in DSA public key structures and return an error in the 7889 DSA routines if parameters are absent. 7890 [Steve Henson] 7891 7892 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd" 7893 in the current directory if neither $RANDFILE nor $HOME was set. 7894 RAND_file_name() in 0.9.6a returned NULL in this case. This has 7895 caused some confusion to Windows users who haven't defined $HOME. 7896 Thus RAND_file_name() is changed again: e_os.h can define a 7897 DEFAULT_HOME, which will be used if $HOME is not set. 7898 For Windows, we use "C:"; on other platforms, we still require 7899 environment variables. 7900 7901 *) Move 'if (!initialized) RAND_poll()' into regions protected by 7902 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids 7903 having multiple threads call RAND_poll() concurrently. 7904 [Bodo Moeller] 7905 7906 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a 7907 combination of a flag and a thread ID variable. 7908 Otherwise while one thread is in ssleay_rand_bytes (which sets the 7909 flag), *other* threads can enter ssleay_add_bytes without obeying 7910 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock 7911 that they do not hold after the first thread unsets add_do_not_lock). 7912 [Bodo Moeller] 7913 7914 *) Change bctest again: '-x' expressions are not available in all 7915 versions of 'test'. 7916 [Bodo Moeller] 7917 7918 Changes between 0.9.6 and 0.9.6a [5 Apr 2001] 7919 7920 *) Fix a couple of memory leaks in PKCS7_dataDecode() 7921 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>] 7922 7923 *) Change Configure and Makefiles to provide EXE_EXT, which will contain 7924 the default extension for executables, if any. Also, make the perl 7925 scripts that use symlink() to test if it really exists and use "cp" 7926 if it doesn't. All this made OpenSSL compilable and installable in 7927 CygWin. 7928 [Richard Levitte] 7929 7930 *) Fix for asn1_GetSequence() for indefinite length constructed data. 7931 If SEQUENCE is length is indefinite just set c->slen to the total 7932 amount of data available. 7933 [Steve Henson, reported by shige@FreeBSD.org] 7934 [This change does not apply to 0.9.7.] 7935 7936 *) Change bctest to avoid here-documents inside command substitution 7937 (workaround for FreeBSD /bin/sh bug). 7938 For compatibility with Ultrix, avoid shell functions (introduced 7939 in the bctest version that searches along $PATH). 7940 [Bodo Moeller] 7941 7942 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes 7943 with des_encrypt() defined on some operating systems, like Solaris 7944 and UnixWare. 7945 [Richard Levitte] 7946 7947 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton: 7948 On the Importance of Eliminating Errors in Cryptographic 7949 Computations, J. Cryptology 14 (2001) 2, 101-119, 7950 http://theory.stanford.edu/~dabo/papers/faults.ps.gz). 7951 [Ulf Moeller] 7952 7953 *) MIPS assembler BIGNUM division bug fix. 7954 [Andy Polyakov] 7955 7956 *) Disabled incorrect Alpha assembler code. 7957 [Richard Levitte] 7958 7959 *) Fix PKCS#7 decode routines so they correctly update the length 7960 after reading an EOC for the EXPLICIT tag. 7961 [Steve Henson] 7962 [This change does not apply to 0.9.7.] 7963 7964 *) Fix bug in PKCS#12 key generation routines. This was triggered 7965 if a 3DES key was generated with a 0 initial byte. Include 7966 PKCS12_BROKEN_KEYGEN compilation option to retain the old 7967 (but broken) behaviour. 7968 [Steve Henson] 7969 7970 *) Enhance bctest to search for a working bc along $PATH and print 7971 it when found. 7972 [Tim Rice <tim@multitalents.net> via Richard Levitte] 7973 7974 *) Fix memory leaks in err.c: free err_data string if necessary; 7975 don't write to the wrong index in ERR_set_error_data. 7976 [Bodo Moeller] 7977 7978 *) Implement ssl23_peek (analogous to ssl23_read), which previously 7979 did not exist. 7980 [Bodo Moeller] 7981 7982 *) Replace rdtsc with _emit statements for VC++ version 5. 7983 [Jeremy Cooper <jeremy@baymoo.org>] 7984 7985 *) Make it possible to reuse SSLv2 sessions. 7986 [Richard Levitte] 7987 7988 *) In copy_email() check for >= 0 as a return value for 7989 X509_NAME_get_index_by_NID() since 0 is a valid index. 7990 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>] 7991 7992 *) Avoid coredump with unsupported or invalid public keys by checking if 7993 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when 7994 PKCS7_verify() fails with non detached data. 7995 [Steve Henson] 7996 7997 *) Don't use getenv in library functions when run as setuid/setgid. 7998 New function OPENSSL_issetugid(). 7999 [Ulf Moeller] 8000 8001 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c) 8002 due to incorrect handling of multi-threading: 8003 8004 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl(). 8005 8006 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on(). 8007 8008 3. Count how many times MemCheck_off() has been called so that 8009 nested use can be treated correctly. This also avoids 8010 inband-signalling in the previous code (which relied on the 8011 assumption that thread ID 0 is impossible). 8012 [Bodo Moeller] 8013 8014 *) Add "-rand" option also to s_client and s_server. 8015 [Lutz Jaenicke] 8016 8017 *) Fix CPU detection on Irix 6.x. 8018 [Kurt Hockenbury <khockenb@stevens-tech.edu> and 8019 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 8020 8021 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME 8022 was empty. 8023 [Steve Henson] 8024 [This change does not apply to 0.9.7.] 8025 8026 *) Use the cached encoding of an X509_NAME structure rather than 8027 copying it. This is apparently the reason for the libsafe "errors" 8028 but the code is actually correct. 8029 [Steve Henson] 8030 8031 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent 8032 Bleichenbacher's DSA attack. 8033 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits 8034 to be set and top=0 forces the highest bit to be set; top=-1 is new 8035 and leaves the highest bit random. 8036 [Ulf Moeller, Bodo Moeller] 8037 8038 *) In the NCONF_...-based implementations for CONF_... queries 8039 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using 8040 a temporary CONF structure with the data component set to NULL 8041 (which gives segmentation faults in lh_retrieve). 8042 Instead, use NULL for the CONF pointer in CONF_get_string and 8043 CONF_get_number (which may use environment variables) and directly 8044 return NULL from CONF_get_section. 8045 [Bodo Moeller] 8046 8047 *) Fix potential buffer overrun for EBCDIC. 8048 [Ulf Moeller] 8049 8050 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign 8051 keyUsage if basicConstraints absent for a CA. 8052 [Steve Henson] 8053 8054 *) Make SMIME_write_PKCS7() write mail header values with a format that 8055 is more generally accepted (no spaces before the semicolon), since 8056 some programs can't parse those values properly otherwise. Also make 8057 sure BIO's that break lines after each write do not create invalid 8058 headers. 8059 [Richard Levitte] 8060 8061 *) Make the CRL encoding routines work with empty SEQUENCE OF. The 8062 macros previously used would not encode an empty SEQUENCE OF 8063 and break the signature. 8064 [Steve Henson] 8065 [This change does not apply to 0.9.7.] 8066 8067 *) Zero the premaster secret after deriving the master secret in 8068 DH ciphersuites. 8069 [Steve Henson] 8070 8071 *) Add some EVP_add_digest_alias registrations (as found in 8072 OpenSSL_add_all_digests()) to SSL_library_init() 8073 aka OpenSSL_add_ssl_algorithms(). This provides improved 8074 compatibility with peers using X.509 certificates 8075 with unconventional AlgorithmIdentifier OIDs. 8076 [Bodo Moeller] 8077 8078 *) Fix for Irix with NO_ASM. 8079 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 8080 8081 *) ./config script fixes. 8082 [Ulf Moeller, Richard Levitte] 8083 8084 *) Fix 'openssl passwd -1'. 8085 [Bodo Moeller] 8086 8087 *) Change PKCS12_key_gen_asc() so it can cope with non null 8088 terminated strings whose length is passed in the passlen 8089 parameter, for example from PEM callbacks. This was done 8090 by adding an extra length parameter to asc2uni(). 8091 [Steve Henson, reported by <oddissey@samsung.co.kr>] 8092 8093 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn 8094 call failed, free the DSA structure. 8095 [Bodo Moeller] 8096 8097 *) Fix to uni2asc() to cope with zero length Unicode strings. 8098 These are present in some PKCS#12 files. 8099 [Steve Henson] 8100 8101 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c). 8102 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits 8103 when writing a 32767 byte record. 8104 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>] 8105 8106 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c), 8107 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}. 8108 8109 (RSA objects have a reference count access to which is protected 8110 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c], 8111 so they are meant to be shared between threads.) 8112 [Bodo Moeller, Geoff Thorpe; original patch submitted by 8113 "Reddie, Steven" <Steven.Reddie@ca.com>] 8114 8115 *) Fix a deadlock in CRYPTO_mem_leaks(). 8116 [Bodo Moeller] 8117 8118 *) Use better test patterns in bntest. 8119 [Ulf M��ller] 8120 8121 *) rand_win.c fix for Borland C. 8122 [Ulf M��ller] 8123 8124 *) BN_rshift bugfix for n == 0. 8125 [Bodo Moeller] 8126 8127 *) Add a 'bctest' script that checks for some known 'bc' bugs 8128 so that 'make test' does not abort just because 'bc' is broken. 8129 [Bodo Moeller] 8130 8131 *) Store verify_result within SSL_SESSION also for client side to 8132 avoid potential security hole. (Re-used sessions on the client side 8133 always resulted in verify_result==X509_V_OK, not using the original 8134 result of the server certificate verification.) 8135 [Lutz Jaenicke] 8136 8137 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type 8138 SSL3_RT_APPLICATION_DATA, return 0. 8139 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true. 8140 [Bodo Moeller] 8141 8142 *) Fix SSL_peek: 8143 Both ssl2_peek and ssl3_peek, which were totally broken in earlier 8144 releases, have been re-implemented by renaming the previous 8145 implementations of ssl2_read and ssl3_read to ssl2_read_internal 8146 and ssl3_read_internal, respectively, and adding 'peek' parameters 8147 to them. The new ssl[23]_{read,peek} functions are calls to 8148 ssl[23]_read_internal with the 'peek' flag set appropriately. 8149 A 'peek' parameter has also been added to ssl3_read_bytes, which 8150 does the actual work for ssl3_read_internal. 8151 [Bodo Moeller] 8152 8153 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling 8154 the method-specific "init()" handler. Also clean up ex_data after 8155 calling the method-specific "finish()" handler. Previously, this was 8156 happening the other way round. 8157 [Geoff Thorpe] 8158 8159 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16. 8160 The previous value, 12, was not always sufficient for BN_mod_exp(). 8161 [Bodo Moeller] 8162 8163 *) Make sure that shared libraries get the internal name engine with 8164 the full version number and not just 0. This should mark the 8165 shared libraries as not backward compatible. Of course, this should 8166 be changed again when we can guarantee backward binary compatibility. 8167 [Richard Levitte] 8168 8169 *) Fix typo in get_cert_by_subject() in by_dir.c 8170 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>] 8171 8172 *) Rework the system to generate shared libraries: 8173 8174 - Make note of the expected extension for the shared libraries and 8175 if there is a need for symbolic links from for example libcrypto.so.0 8176 to libcrypto.so.0.9.7. There is extended info in Configure for 8177 that. 8178 8179 - Make as few rebuilds of the shared libraries as possible. 8180 8181 - Still avoid linking the OpenSSL programs with the shared libraries. 8182 8183 - When installing, install the shared libraries separately from the 8184 static ones. 8185 [Richard Levitte] 8186 8187 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument. 8188 8189 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new 8190 and not in SSL_clear because the latter is also used by the 8191 accept/connect functions; previously, the settings made by 8192 SSL_set_read_ahead would be lost during the handshake. 8193 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>] 8194 8195 *) Correct util/mkdef.pl to be selective about disabled algorithms. 8196 Previously, it would create entries for disableed algorithms no 8197 matter what. 8198 [Richard Levitte] 8199 8200 *) Added several new manual pages for SSL_* function. 8201 [Lutz Jaenicke] 8202 8203 Changes between 0.9.5a and 0.9.6 [24 Sep 2000] 8204 8205 *) In ssl23_get_client_hello, generate an error message when faced 8206 with an initial SSL 3.0/TLS record that is too small to contain the 8207 first two bytes of the ClientHello message, i.e. client_version. 8208 (Note that this is a pathologic case that probably has never happened 8209 in real life.) The previous approach was to use the version number 8210 from the record header as a substitute; but our protocol choice 8211 should not depend on that one because it is not authenticated 8212 by the Finished messages. 8213 [Bodo Moeller] 8214 8215 *) More robust randomness gathering functions for Windows. 8216 [Jeffrey Altman <jaltman@columbia.edu>] 8217 8218 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is 8219 not set then we don't setup the error code for issuer check errors 8220 to avoid possibly overwriting other errors which the callback does 8221 handle. If an application does set the flag then we assume it knows 8222 what it is doing and can handle the new informational codes 8223 appropriately. 8224 [Steve Henson] 8225 8226 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for 8227 a general "ANY" type, as such it should be able to decode anything 8228 including tagged types. However it didn't check the class so it would 8229 wrongly interpret tagged types in the same way as their universal 8230 counterpart and unknown types were just rejected. Changed so that the 8231 tagged and unknown types are handled in the same way as a SEQUENCE: 8232 that is the encoding is stored intact. There is also a new type 8233 "V_ASN1_OTHER" which is used when the class is not universal, in this 8234 case we have no idea what the actual type is so we just lump them all 8235 together. 8236 [Steve Henson] 8237 8238 *) On VMS, stdout may very well lead to a file that is written to 8239 in a record-oriented fashion. That means that every write() will 8240 write a separate record, which will be read separately by the 8241 programs trying to read from it. This can be very confusing. 8242 8243 The solution is to put a BIO filter in the way that will buffer 8244 text until a linefeed is reached, and then write everything a 8245 line at a time, so every record written will be an actual line, 8246 not chunks of lines and not (usually doesn't happen, but I've 8247 seen it once) several lines in one record. BIO_f_linebuffer() is 8248 the answer. 8249 8250 Currently, it's a VMS-only method, because that's where it has 8251 been tested well enough. 8252 [Richard Levitte] 8253 8254 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery, 8255 it can return incorrect results. 8256 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a, 8257 but it was in 0.9.6-beta[12].) 8258 [Bodo Moeller] 8259 8260 *) Disable the check for content being present when verifying detached 8261 signatures in pk7_smime.c. Some versions of Netscape (wrongly) 8262 include zero length content when signing messages. 8263 [Steve Henson] 8264 8265 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR 8266 BIO_ctrl (for BIO pairs). 8267 [Bodo M��ller] 8268 8269 *) Add DSO method for VMS. 8270 [Richard Levitte] 8271 8272 *) Bug fix: Montgomery multiplication could produce results with the 8273 wrong sign. 8274 [Ulf M��ller] 8275 8276 *) Add RPM specification openssl.spec and modify it to build three 8277 packages. The default package contains applications, application 8278 documentation and run-time libraries. The devel package contains 8279 include files, static libraries and function documentation. The 8280 doc package contains the contents of the doc directory. The original 8281 openssl.spec was provided by Damien Miller <djm@mindrot.org>. 8282 [Richard Levitte] 8283 8284 *) Add a large number of documentation files for many SSL routines. 8285 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 8286 8287 *) Add a configuration entry for Sony News 4. 8288 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>] 8289 8290 *) Don't set the two most significant bits to one when generating a 8291 random number < q in the DSA library. 8292 [Ulf M��ller] 8293 8294 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default 8295 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if 8296 the underlying transport is blocking) if a handshake took place. 8297 (The default behaviour is needed by applications such as s_client 8298 and s_server that use select() to determine when to use SSL_read; 8299 but for applications that know in advance when to expect data, it 8300 just makes things more complicated.) 8301 [Bodo Moeller] 8302 8303 *) Add RAND_egd_bytes(), which gives control over the number of bytes read 8304 from EGD. 8305 [Ben Laurie] 8306 8307 *) Add a few more EBCDIC conditionals that make `req' and `x509' 8308 work better on such systems. 8309 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 8310 8311 *) Add two demo programs for PKCS12_parse() and PKCS12_create(). 8312 Update PKCS12_parse() so it copies the friendlyName and the 8313 keyid to the certificates aux info. 8314 [Steve Henson] 8315 8316 *) Fix bug in PKCS7_verify() which caused an infinite loop 8317 if there was more than one signature. 8318 [Sven Uszpelkat <su@celocom.de>] 8319 8320 *) Major change in util/mkdef.pl to include extra information 8321 about each symbol, as well as presentig variables as well 8322 as functions. This change means that there's n more need 8323 to rebuild the .num files when some algorithms are excluded. 8324 [Richard Levitte] 8325 8326 *) Allow the verify time to be set by an application, 8327 rather than always using the current time. 8328 [Steve Henson] 8329 8330 *) Phase 2 verify code reorganisation. The certificate 8331 verify code now looks up an issuer certificate by a 8332 number of criteria: subject name, authority key id 8333 and key usage. It also verifies self signed certificates 8334 by the same criteria. The main comparison function is 8335 X509_check_issued() which performs these checks. 8336 8337 Lot of changes were necessary in order to support this 8338 without completely rewriting the lookup code. 8339 8340 Authority and subject key identifier are now cached. 8341 8342 The LHASH 'certs' is X509_STORE has now been replaced 8343 by a STACK_OF(X509_OBJECT). This is mainly because an 8344 LHASH can't store or retrieve multiple objects with 8345 the same hash value. 8346 8347 As a result various functions (which were all internal 8348 use only) have changed to handle the new X509_STORE 8349 structure. This will break anything that messed round 8350 with X509_STORE internally. 8351 8352 The functions X509_STORE_add_cert() now checks for an 8353 exact match, rather than just subject name. 8354 8355 The X509_STORE API doesn't directly support the retrieval 8356 of multiple certificates matching a given criteria, however 8357 this can be worked round by performing a lookup first 8358 (which will fill the cache with candidate certificates) 8359 and then examining the cache for matches. This is probably 8360 the best we can do without throwing out X509_LOOKUP 8361 entirely (maybe later...). 8362 8363 The X509_VERIFY_CTX structure has been enhanced considerably. 8364 8365 All certificate lookup operations now go via a get_issuer() 8366 callback. Although this currently uses an X509_STORE it 8367 can be replaced by custom lookups. This is a simple way 8368 to bypass the X509_STORE hackery necessary to make this 8369 work and makes it possible to use more efficient techniques 8370 in future. A very simple version which uses a simple 8371 STACK for its trusted certificate store is also provided 8372 using X509_STORE_CTX_trusted_stack(). 8373 8374 The verify_cb() and verify() callbacks now have equivalents 8375 in the X509_STORE_CTX structure. 8376 8377 X509_STORE_CTX also has a 'flags' field which can be used 8378 to customise the verify behaviour. 8379 [Steve Henson] 8380 8381 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which 8382 excludes S/MIME capabilities. 8383 [Steve Henson] 8384 8385 *) When a certificate request is read in keep a copy of the 8386 original encoding of the signed data and use it when outputing 8387 again. Signatures then use the original encoding rather than 8388 a decoded, encoded version which may cause problems if the 8389 request is improperly encoded. 8390 [Steve Henson] 8391 8392 *) For consistency with other BIO_puts implementations, call 8393 buffer_write(b, ...) directly in buffer_puts instead of calling 8394 BIO_write(b, ...). 8395 8396 In BIO_puts, increment b->num_write as in BIO_write. 8397 [Peter.Sylvester@EdelWeb.fr] 8398 8399 *) Fix BN_mul_word for the case where the word is 0. (We have to use 8400 BN_zero, we may not return a BIGNUM with an array consisting of 8401 words set to zero.) 8402 [Bodo Moeller] 8403 8404 *) Avoid calling abort() from within the library when problems are 8405 detected, except if preprocessor symbols have been defined 8406 (such as REF_CHECK, BN_DEBUG etc.). 8407 [Bodo Moeller] 8408 8409 *) New openssl application 'rsautl'. This utility can be 8410 used for low level RSA operations. DER public key 8411 BIO/fp routines also added. 8412 [Steve Henson] 8413 8414 *) New Configure entry and patches for compiling on QNX 4. 8415 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>] 8416 8417 *) A demo state-machine implementation was sponsored by 8418 Nuron (http://www.nuron.com/) and is now available in 8419 demos/state_machine. 8420 [Ben Laurie] 8421 8422 *) New options added to the 'dgst' utility for signature 8423 generation and verification. 8424 [Steve Henson] 8425 8426 *) Unrecognized PKCS#7 content types are now handled via a 8427 catch all ASN1_TYPE structure. This allows unsupported 8428 types to be stored as a "blob" and an application can 8429 encode and decode it manually. 8430 [Steve Henson] 8431 8432 *) Fix various signed/unsigned issues to make a_strex.c 8433 compile under VC++. 8434 [Oscar Jacobsson <oscar.jacobsson@celocom.com>] 8435 8436 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct 8437 length if passed a buffer. ASN1_INTEGER_to_BN failed 8438 if passed a NULL BN and its argument was negative. 8439 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>] 8440 8441 *) Modification to PKCS#7 encoding routines to output definite 8442 length encoding. Since currently the whole structures are in 8443 memory there's not real point in using indefinite length 8444 constructed encoding. However if OpenSSL is compiled with 8445 the flag PKCS7_INDEFINITE_ENCODING the old form is used. 8446 [Steve Henson] 8447 8448 *) Added BIO_vprintf() and BIO_vsnprintf(). 8449 [Richard Levitte] 8450 8451 *) Added more prefixes to parse for in the the strings written 8452 through a logging bio, to cover all the levels that are available 8453 through syslog. The prefixes are now: 8454 8455 PANIC, EMERG, EMR => LOG_EMERG 8456 ALERT, ALR => LOG_ALERT 8457 CRIT, CRI => LOG_CRIT 8458 ERROR, ERR => LOG_ERR 8459 WARNING, WARN, WAR => LOG_WARNING 8460 NOTICE, NOTE, NOT => LOG_NOTICE 8461 INFO, INF => LOG_INFO 8462 DEBUG, DBG => LOG_DEBUG 8463 8464 and as before, if none of those prefixes are present at the 8465 beginning of the string, LOG_ERR is chosen. 8466 8467 On Win32, the LOG_* levels are mapped according to this: 8468 8469 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE 8470 LOG_WARNING => EVENTLOG_WARNING_TYPE 8471 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE 8472 8473 [Richard Levitte] 8474 8475 *) Made it possible to reconfigure with just the configuration 8476 argument "reconf" or "reconfigure". The command line arguments 8477 are stored in Makefile.ssl in the variable CONFIGURE_ARGS, 8478 and are retrieved from there when reconfiguring. 8479 [Richard Levitte] 8480 8481 *) MD4 implemented. 8482 [Assar Westerlund <assar@sics.se>, Richard Levitte] 8483 8484 *) Add the arguments -CAfile and -CApath to the pkcs12 utility. 8485 [Richard Levitte] 8486 8487 *) The obj_dat.pl script was messing up the sorting of object 8488 names. The reason was that it compared the quoted version 8489 of strings as a result "OCSP" > "OCSP Signing" because 8490 " > SPACE. Changed script to store unquoted versions of 8491 names and add quotes on output. It was also omitting some 8492 names from the lookup table if they were given a default 8493 value (that is if SN is missing it is given the same 8494 value as LN and vice versa), these are now added on the 8495 grounds that if an object has a name we should be able to 8496 look it up. Finally added warning output when duplicate 8497 short or long names are found. 8498 [Steve Henson] 8499 8500 *) Changes needed for Tandem NSK. 8501 [Scott Uroff <scott@xypro.com>] 8502 8503 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in 8504 RSA_padding_check_SSLv23(), special padding was never detected 8505 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol 8506 version rollback attacks was not effective. 8507 8508 In s23_clnt.c, don't use special rollback-attack detection padding 8509 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the 8510 client; similarly, in s23_srvr.c, don't do the rollback check if 8511 SSL 2.0 is the only protocol enabled in the server. 8512 [Bodo Moeller] 8513 8514 *) Make it possible to get hexdumps of unprintable data with 'openssl 8515 asn1parse'. By implication, the functions ASN1_parse_dump() and 8516 BIO_dump_indent() are added. 8517 [Richard Levitte] 8518 8519 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex() 8520 these print out strings and name structures based on various 8521 flags including RFC2253 support and proper handling of 8522 multibyte characters. Added options to the 'x509' utility 8523 to allow the various flags to be set. 8524 [Steve Henson] 8525 8526 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME. 8527 Also change the functions X509_cmp_current_time() and 8528 X509_gmtime_adj() work with an ASN1_TIME structure, 8529 this will enable certificates using GeneralizedTime in validity 8530 dates to be checked. 8531 [Steve Henson] 8532 8533 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid 8534 negative public key encodings) on by default, 8535 NO_NEG_PUBKEY_BUG can be set to disable it. 8536 [Steve Henson] 8537 8538 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT 8539 content octets. An i2c_ASN1_OBJECT is unnecessary because 8540 the encoding can be trivially obtained from the structure. 8541 [Steve Henson] 8542 8543 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock), 8544 not read locks (CRYPTO_r_[un]lock). 8545 [Bodo Moeller] 8546 8547 *) A first attempt at creating official support for shared 8548 libraries through configuration. I've kept it so the 8549 default is static libraries only, and the OpenSSL programs 8550 are always statically linked for now, but there are 8551 preparations for dynamic linking in place. 8552 This has been tested on Linux and Tru64. 8553 [Richard Levitte] 8554 8555 *) Randomness polling function for Win9x, as described in: 8556 Peter Gutmann, Software Generation of Practically Strong 8557 Random Numbers. 8558 [Ulf M��ller] 8559 8560 *) Fix so PRNG is seeded in req if using an already existing 8561 DSA key. 8562 [Steve Henson] 8563 8564 *) New options to smime application. -inform and -outform 8565 allow alternative formats for the S/MIME message including 8566 PEM and DER. The -content option allows the content to be 8567 specified separately. This should allow things like Netscape 8568 form signing output easier to verify. 8569 [Steve Henson] 8570 8571 *) Fix the ASN1 encoding of tags using the 'long form'. 8572 [Steve Henson] 8573 8574 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT 8575 STRING types. These convert content octets to and from the 8576 underlying type. The actual tag and length octets are 8577 already assumed to have been read in and checked. These 8578 are needed because all other string types have virtually 8579 identical handling apart from the tag. By having versions 8580 of the ASN1 functions that just operate on content octets 8581 IMPLICIT tagging can be handled properly. It also allows 8582 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED 8583 and ASN1_INTEGER are identical apart from the tag. 8584 [Steve Henson] 8585 8586 *) Change the handling of OID objects as follows: 8587 8588 - New object identifiers are inserted in objects.txt, following 8589 the syntax given in objects.README. 8590 - objects.pl is used to process obj_mac.num and create a new 8591 obj_mac.h. 8592 - obj_dat.pl is used to create a new obj_dat.h, using the data in 8593 obj_mac.h. 8594 8595 This is currently kind of a hack, and the perl code in objects.pl 8596 isn't very elegant, but it works as I intended. The simplest way 8597 to check that it worked correctly is to look in obj_dat.h and 8598 check the array nid_objs and make sure the objects haven't moved 8599 around (this is important!). Additions are OK, as well as 8600 consistent name changes. 8601 [Richard Levitte] 8602 8603 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1'). 8604 [Bodo Moeller] 8605 8606 *) Addition of the command line parameter '-rand file' to 'openssl req'. 8607 The given file adds to whatever has already been seeded into the 8608 random pool through the RANDFILE configuration file option or 8609 environment variable, or the default random state file. 8610 [Richard Levitte] 8611 8612 *) mkstack.pl now sorts each macro group into lexical order. 8613 Previously the output order depended on the order the files 8614 appeared in the directory, resulting in needless rewriting 8615 of safestack.h . 8616 [Steve Henson] 8617 8618 *) Patches to make OpenSSL compile under Win32 again. Mostly 8619 work arounds for the VC++ problem that it treats func() as 8620 func(void). Also stripped out the parts of mkdef.pl that 8621 added extra typesafe functions: these no longer exist. 8622 [Steve Henson] 8623 8624 *) Reorganisation of the stack code. The macros are now all 8625 collected in safestack.h . Each macro is defined in terms of 8626 a "stack macro" of the form SKM_<name>(type, a, b). The 8627 DEBUG_SAFESTACK is now handled in terms of function casts, 8628 this has the advantage of retaining type safety without the 8629 use of additional functions. If DEBUG_SAFESTACK is not defined 8630 then the non typesafe macros are used instead. Also modified the 8631 mkstack.pl script to handle the new form. Needs testing to see 8632 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK 8633 the default if no major problems. Similar behaviour for ASN1_SET_OF 8634 and PKCS12_STACK_OF. 8635 [Steve Henson] 8636 8637 *) When some versions of IIS use the 'NET' form of private key the 8638 key derivation algorithm is different. Normally MD5(password) is 8639 used as a 128 bit RC4 key. In the modified case 8640 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some 8641 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same 8642 as the old Netscape_RSA functions except they have an additional 8643 'sgckey' parameter which uses the modified algorithm. Also added 8644 an -sgckey command line option to the rsa utility. Thanks to 8645 Adrian Peck <bertie@ncipher.com> for posting details of the modified 8646 algorithm to openssl-dev. 8647 [Steve Henson] 8648 8649 *) The evp_local.h macros were using 'c.##kname' which resulted in 8650 invalid expansion on some systems (SCO 5.0.5 for example). 8651 Corrected to 'c.kname'. 8652 [Phillip Porch <root@theporch.com>] 8653 8654 *) New X509_get1_email() and X509_REQ_get1_email() functions that return 8655 a STACK of email addresses from a certificate or request, these look 8656 in the subject name and the subject alternative name extensions and 8657 omit any duplicate addresses. 8658 [Steve Henson] 8659 8660 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows. 8661 This makes DSA verification about 2 % faster. 8662 [Bodo Moeller] 8663 8664 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5 8665 (meaning that now 2^5 values will be precomputed, which is only 4 KB 8666 plus overhead for 1024 bit moduli). 8667 This makes exponentiations about 0.5 % faster for 1024 bit 8668 exponents (as measured by "openssl speed rsa2048"). 8669 [Bodo Moeller] 8670 8671 *) Rename memory handling macros to avoid conflicts with other 8672 software: 8673 Malloc => OPENSSL_malloc 8674 Malloc_locked => OPENSSL_malloc_locked 8675 Realloc => OPENSSL_realloc 8676 Free => OPENSSL_free 8677 [Richard Levitte] 8678 8679 *) New function BN_mod_exp_mont_word for small bases (roughly 15% 8680 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange). 8681 [Bodo Moeller] 8682 8683 *) CygWin32 support. 8684 [John Jarvie <jjarvie@newsguy.com>] 8685 8686 *) The type-safe stack code has been rejigged. It is now only compiled 8687 in when OpenSSL is configured with the DEBUG_SAFESTACK option and 8688 by default all type-specific stack functions are "#define"d back to 8689 standard stack functions. This results in more streamlined output 8690 but retains the type-safety checking possibilities of the original 8691 approach. 8692 [Geoff Thorpe] 8693 8694 *) The STACK code has been cleaned up, and certain type declarations 8695 that didn't make a lot of sense have been brought in line. This has 8696 also involved a cleanup of sorts in safestack.h to more correctly 8697 map type-safe stack functions onto their plain stack counterparts. 8698 This work has also resulted in a variety of "const"ifications of 8699 lots of the code, especially "_cmp" operations which should normally 8700 be prototyped with "const" parameters anyway. 8701 [Geoff Thorpe] 8702 8703 *) When generating bytes for the first time in md_rand.c, 'stir the pool' 8704 by seeding with STATE_SIZE dummy bytes (with zero entropy count). 8705 (The PRNG state consists of two parts, the large pool 'state' and 'md', 8706 where all of 'md' is used each time the PRNG is used, but 'state' 8707 is used only indexed by a cyclic counter. As entropy may not be 8708 well distributed from the beginning, 'md' is important as a 8709 chaining variable. However, the output function chains only half 8710 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains 8711 all of 'md', and seeding with STATE_SIZE dummy bytes will result 8712 in all of 'state' being rewritten, with the new values depending 8713 on virtually all of 'md'. This overcomes the 80 bit limitation.) 8714 [Bodo Moeller] 8715 8716 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when 8717 the handshake is continued after ssl_verify_cert_chain(); 8718 otherwise, if SSL_VERIFY_NONE is set, remaining error codes 8719 can lead to 'unexplainable' connection aborts later. 8720 [Bodo Moeller; problem tracked down by Lutz Jaenicke] 8721 8722 *) Major EVP API cipher revision. 8723 Add hooks for extra EVP features. This allows various cipher 8724 parameters to be set in the EVP interface. Support added for variable 8725 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and 8726 setting of RC2 and RC5 parameters. 8727 8728 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length 8729 ciphers. 8730 8731 Remove lots of duplicated code from the EVP library. For example *every* 8732 cipher init() function handles the 'iv' in the same way according to the 8733 cipher mode. They also all do nothing if the 'key' parameter is NULL and 8734 for CFB and OFB modes they zero ctx->num. 8735 8736 New functionality allows removal of S/MIME code RC2 hack. 8737 8738 Most of the routines have the same form and so can be declared in terms 8739 of macros. 8740 8741 By shifting this to the top level EVP_CipherInit() it can be removed from 8742 all individual ciphers. If the cipher wants to handle IVs or keys 8743 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT 8744 flags. 8745 8746 Change lots of functions like EVP_EncryptUpdate() to now return a 8747 value: although software versions of the algorithms cannot fail 8748 any installed hardware versions can. 8749 [Steve Henson] 8750 8751 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if 8752 this option is set, tolerate broken clients that send the negotiated 8753 protocol version number instead of the requested protocol version 8754 number. 8755 [Bodo Moeller] 8756 8757 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag; 8758 i.e. non-zero for export ciphersuites, zero otherwise. 8759 Previous versions had this flag inverted, inconsistent with 8760 rsa_tmp_cb (..._TMP_RSA_CB). 8761 [Bodo Moeller; problem reported by Amit Chopra] 8762 8763 *) Add missing DSA library text string. Work around for some IIS 8764 key files with invalid SEQUENCE encoding. 8765 [Steve Henson] 8766 8767 *) Add a document (doc/standards.txt) that list all kinds of standards 8768 and so on that are implemented in OpenSSL. 8769 [Richard Levitte] 8770 8771 *) Enhance c_rehash script. Old version would mishandle certificates 8772 with the same subject name hash and wouldn't handle CRLs at all. 8773 Added -fingerprint option to crl utility, to support new c_rehash 8774 features. 8775 [Steve Henson] 8776 8777 *) Eliminate non-ANSI declarations in crypto.h and stack.h. 8778 [Ulf M��ller] 8779 8780 *) Fix for SSL server purpose checking. Server checking was 8781 rejecting certificates which had extended key usage present 8782 but no ssl client purpose. 8783 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>] 8784 8785 *) Make PKCS#12 code work with no password. The PKCS#12 spec 8786 is a little unclear about how a blank password is handled. 8787 Since the password in encoded as a BMPString with terminating 8788 double NULL a zero length password would end up as just the 8789 double NULL. However no password at all is different and is 8790 handled differently in the PKCS#12 key generation code. NS 8791 treats a blank password as zero length. MSIE treats it as no 8792 password on export: but it will try both on import. We now do 8793 the same: PKCS12_parse() tries zero length and no password if 8794 the password is set to "" or NULL (NULL is now a valid password: 8795 it wasn't before) as does the pkcs12 application. 8796 [Steve Henson] 8797 8798 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use 8799 perror when PEM_read_bio_X509_REQ fails, the error message must 8800 be obtained from the error queue. 8801 [Bodo Moeller] 8802 8803 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing 8804 it in ERR_remove_state if appropriate, and change ERR_get_state 8805 accordingly to avoid race conditions (this is necessary because 8806 thread_hash is no longer constant once set). 8807 [Bodo Moeller] 8808 8809 *) Bugfix for linux-elf makefile.one. 8810 [Ulf M��ller] 8811 8812 *) RSA_get_default_method() will now cause a default 8813 RSA_METHOD to be chosen if one doesn't exist already. 8814 Previously this was only set during a call to RSA_new() 8815 or RSA_new_method(NULL) meaning it was possible for 8816 RSA_get_default_method() to return NULL. 8817 [Geoff Thorpe] 8818 8819 *) Added native name translation to the existing DSO code 8820 that will convert (if the flag to do so is set) filenames 8821 that are sufficiently small and have no path information 8822 into a canonical native form. Eg. "blah" converted to 8823 "libblah.so" or "blah.dll" etc. 8824 [Geoff Thorpe] 8825 8826 *) New function ERR_error_string_n(e, buf, len) which is like 8827 ERR_error_string(e, buf), but writes at most 'len' bytes 8828 including the 0 terminator. For ERR_error_string_n, 'buf' 8829 may not be NULL. 8830 [Damien Miller <djm@mindrot.org>, Bodo Moeller] 8831 8832 *) CONF library reworked to become more general. A new CONF 8833 configuration file reader "class" is implemented as well as a 8834 new functions (NCONF_*, for "New CONF") to handle it. The now 8835 old CONF_* functions are still there, but are reimplemented to 8836 work in terms of the new functions. Also, a set of functions 8837 to handle the internal storage of the configuration data is 8838 provided to make it easier to write new configuration file 8839 reader "classes" (I can definitely see something reading a 8840 configuration file in XML format, for example), called _CONF_*, 8841 or "the configuration storage API"... 8842 8843 The new configuration file reading functions are: 8844 8845 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio, 8846 NCONF_get_section, NCONF_get_string, NCONF_get_numbre 8847 8848 NCONF_default, NCONF_WIN32 8849 8850 NCONF_dump_fp, NCONF_dump_bio 8851 8852 NCONF_default and NCONF_WIN32 are method (or "class") choosers, 8853 NCONF_new creates a new CONF object. This works in the same way 8854 as other interfaces in OpenSSL, like the BIO interface. 8855 NCONF_dump_* dump the internal storage of the configuration file, 8856 which is useful for debugging. All other functions take the same 8857 arguments as the old CONF_* functions wth the exception of the 8858 first that must be a `CONF *' instead of a `LHASH *'. 8859 8860 To make it easer to use the new classes with the old CONF_* functions, 8861 the function CONF_set_default_method is provided. 8862 [Richard Levitte] 8863 8864 *) Add '-tls1' option to 'openssl ciphers', which was already 8865 mentioned in the documentation but had not been implemented. 8866 (This option is not yet really useful because even the additional 8867 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.) 8868 [Bodo Moeller] 8869 8870 *) Initial DSO code added into libcrypto for letting OpenSSL (and 8871 OpenSSL-based applications) load shared libraries and bind to 8872 them in a portable way. 8873 [Geoff Thorpe, with contributions from Richard Levitte] 8874 8875 Changes between 0.9.5 and 0.9.5a [1 Apr 2000] 8876 8877 *) Make sure _lrotl and _lrotr are only used with MSVC. 8878 8879 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status 8880 (the default implementation of RAND_status). 8881 8882 *) Rename openssl x509 option '-crlext', which was added in 0.9.5, 8883 to '-clrext' (= clear extensions), as intended and documented. 8884 [Bodo Moeller; inconsistency pointed out by Michael Attili 8885 <attili@amaxo.com>] 8886 8887 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length 8888 was larger than the MD block size. 8889 [Steve Henson, pointed out by Yost William <YostW@tce.com>] 8890 8891 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument 8892 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set() 8893 using the passed key: if the passed key was a private key the result 8894 of X509_print(), for example, would be to print out all the private key 8895 components. 8896 [Steve Henson] 8897 8898 *) des_quad_cksum() byte order bug fix. 8899 [Ulf M��ller, using the problem description in krb4-0.9.7, where 8900 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>] 8901 8902 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly 8903 discouraged. 8904 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>] 8905 8906 *) For easily testing in shell scripts whether some command 8907 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX' 8908 returns with exit code 0 iff no command of the given name is available. 8909 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases, 8910 the output goes to stdout and nothing is printed to stderr. 8911 Additional arguments are always ignored. 8912 8913 Since for each cipher there is a command of the same name, 8914 the 'no-cipher' compilation switches can be tested this way. 8915 8916 ('openssl no-XXX' is not able to detect pseudo-commands such 8917 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.) 8918 [Bodo Moeller] 8919 8920 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration. 8921 [Bodo Moeller] 8922 8923 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE 8924 is set; it will be thrown away anyway because each handshake creates 8925 its own key. 8926 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition 8927 to parameters -- in previous versions (since OpenSSL 0.9.3) the 8928 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining 8929 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro. 8930 [Bodo Moeller] 8931 8932 *) New s_client option -ign_eof: EOF at stdin is ignored, and 8933 'Q' and 'R' lose their special meanings (quit/renegotiate). 8934 This is part of what -quiet does; unlike -quiet, -ign_eof 8935 does not suppress any output. 8936 [Richard Levitte] 8937 8938 *) Add compatibility options to the purpose and trust code. The 8939 purpose X509_PURPOSE_ANY is "any purpose" which automatically 8940 accepts a certificate or CA, this was the previous behaviour, 8941 with all the associated security issues. 8942 8943 X509_TRUST_COMPAT is the old trust behaviour: only and 8944 automatically trust self signed roots in certificate store. A 8945 new trust setting X509_TRUST_DEFAULT is used to specify that 8946 a purpose has no associated trust setting and it should instead 8947 use the value in the default purpose. 8948 [Steve Henson] 8949 8950 *) Fix the PKCS#8 DSA private key code so it decodes keys again 8951 and fix a memory leak. 8952 [Steve Henson] 8953 8954 *) In util/mkerr.pl (which implements 'make errors'), preserve 8955 reason strings from the previous version of the .c file, as 8956 the default to have only downcase letters (and digits) in 8957 automatically generated reasons codes is not always appropriate. 8958 [Bodo Moeller] 8959 8960 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table 8961 using strerror. Previously, ERR_reason_error_string() returned 8962 library names as reason strings for SYSerr; but SYSerr is a special 8963 case where small numbers are errno values, not library numbers. 8964 [Bodo Moeller] 8965 8966 *) Add '-dsaparam' option to 'openssl dhparam' application. This 8967 converts DSA parameters into DH parameters. (When creating parameters, 8968 DSA_generate_parameters is used.) 8969 [Bodo Moeller] 8970 8971 *) Include 'length' (recommended exponent length) in C code generated 8972 by 'openssl dhparam -C'. 8973 [Bodo Moeller] 8974 8975 *) The second argument to set_label in perlasm was already being used 8976 so couldn't be used as a "file scope" flag. Moved to third argument 8977 which was free. 8978 [Steve Henson] 8979 8980 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes 8981 instead of RAND_bytes for encryption IVs and salts. 8982 [Bodo Moeller] 8983 8984 *) Include RAND_status() into RAND_METHOD instead of implementing 8985 it only for md_rand.c Otherwise replacing the PRNG by calling 8986 RAND_set_rand_method would be impossible. 8987 [Bodo Moeller] 8988 8989 *) Don't let DSA_generate_key() enter an infinite loop if the random 8990 number generation fails. 8991 [Bodo Moeller] 8992 8993 *) New 'rand' application for creating pseudo-random output. 8994 [Bodo Moeller] 8995 8996 *) Added configuration support for Linux/IA64 8997 [Rolf Haberrecker <rolf@suse.de>] 8998 8999 *) Assembler module support for Mingw32. 9000 [Ulf M��ller] 9001 9002 *) Shared library support for HPUX (in shlib/). 9003 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous] 9004 9005 *) Shared library support for Solaris gcc. 9006 [Lutz Behnke <behnke@trustcenter.de>] 9007 9008 Changes between 0.9.4 and 0.9.5 [28 Feb 2000] 9009 9010 *) PKCS7_encrypt() was adding text MIME headers twice because they 9011 were added manually and by SMIME_crlf_copy(). 9012 [Steve Henson] 9013 9014 *) In bntest.c don't call BN_rand with zero bits argument. 9015 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>] 9016 9017 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n] 9018 case was implemented. This caused BN_div_recp() to fail occasionally. 9019 [Ulf M��ller] 9020 9021 *) Add an optional second argument to the set_label() in the perl 9022 assembly language builder. If this argument exists and is set 9023 to 1 it signals that the assembler should use a symbol whose 9024 scope is the entire file, not just the current function. This 9025 is needed with MASM which uses the format label:: for this scope. 9026 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>] 9027 9028 *) Change the ASN1 types so they are typedefs by default. Before 9029 almost all types were #define'd to ASN1_STRING which was causing 9030 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING) 9031 for example. 9032 [Steve Henson] 9033 9034 *) Change names of new functions to the new get1/get0 naming 9035 convention: After 'get1', the caller owns a reference count 9036 and has to call ..._free; 'get0' returns a pointer to some 9037 data structure without incrementing reference counters. 9038 (Some of the existing 'get' functions increment a reference 9039 counter, some don't.) 9040 Similarly, 'set1' and 'add1' functions increase reference 9041 counters or duplicate objects. 9042 [Steve Henson] 9043 9044 *) Allow for the possibility of temp RSA key generation failure: 9045 the code used to assume it always worked and crashed on failure. 9046 [Steve Henson] 9047 9048 *) Fix potential buffer overrun problem in BIO_printf(). 9049 [Ulf M��ller, using public domain code by Patrick Powell; problem 9050 pointed out by David Sacerdote <das33@cornell.edu>] 9051 9052 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions 9053 RAND_egd() and RAND_status(). In the command line application, 9054 the EGD socket can be specified like a seed file using RANDFILE 9055 or -rand. 9056 [Ulf M��ller] 9057 9058 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures. 9059 Some CAs (e.g. Verisign) distribute certificates in this form. 9060 [Steve Henson] 9061 9062 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher 9063 list to exclude them. This means that no special compilation option 9064 is needed to use anonymous DH: it just needs to be included in the 9065 cipher list. 9066 [Steve Henson] 9067 9068 *) Change the EVP_MD_CTX_type macro so its meaning consistent with 9069 EVP_MD_type. The old functionality is available in a new macro called 9070 EVP_MD_md(). Change code that uses it and update docs. 9071 [Steve Henson] 9072 9073 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions 9074 where the 'void *' argument is replaced by a function pointer argument. 9075 Previously 'void *' was abused to point to functions, which works on 9076 many platforms, but is not correct. As these functions are usually 9077 called by macros defined in OpenSSL header files, most source code 9078 should work without changes. 9079 [Richard Levitte] 9080 9081 *) <openssl/opensslconf.h> (which is created by Configure) now contains 9082 sections with information on -D... compiler switches used for 9083 compiling the library so that applications can see them. To enable 9084 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES 9085 must be defined. E.g., 9086 #define OPENSSL_ALGORITHM_DEFINES 9087 #include <openssl/opensslconf.h> 9088 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc. 9089 [Richard Levitte, Ulf and Bodo M��ller] 9090 9091 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS 9092 record layer. 9093 [Bodo Moeller] 9094 9095 *) Change the 'other' type in certificate aux info to a STACK_OF 9096 X509_ALGOR. Although not an AlgorithmIdentifier as such it has 9097 the required ASN1 format: arbitrary types determined by an OID. 9098 [Steve Henson] 9099 9100 *) Add some PEM_write_X509_REQ_NEW() functions and a command line 9101 argument to 'req'. This is not because the function is newer or 9102 better than others it just uses the work 'NEW' in the certificate 9103 request header lines. Some software needs this. 9104 [Steve Henson] 9105 9106 *) Reorganise password command line arguments: now passwords can be 9107 obtained from various sources. Delete the PEM_cb function and make 9108 it the default behaviour: i.e. if the callback is NULL and the 9109 usrdata argument is not NULL interpret it as a null terminated pass 9110 phrase. If usrdata and the callback are NULL then the pass phrase 9111 is prompted for as usual. 9112 [Steve Henson] 9113 9114 *) Add support for the Compaq Atalla crypto accelerator. If it is installed, 9115 the support is automatically enabled. The resulting binaries will 9116 autodetect the card and use it if present. 9117 [Ben Laurie and Compaq Inc.] 9118 9119 *) Work around for Netscape hang bug. This sends certificate request 9120 and server done in one record. Since this is perfectly legal in the 9121 SSL/TLS protocol it isn't a "bug" option and is on by default. See 9122 the bugs/SSLv3 entry for more info. 9123 [Steve Henson] 9124 9125 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround. 9126 [Andy Polyakov] 9127 9128 *) Add -rand argument to smime and pkcs12 applications and read/write 9129 of seed file. 9130 [Steve Henson] 9131 9132 *) New 'passwd' tool for crypt(3) and apr1 password hashes. 9133 [Bodo Moeller] 9134 9135 *) Add command line password options to the remaining applications. 9136 [Steve Henson] 9137 9138 *) Bug fix for BN_div_recp() for numerators with an even number of 9139 bits. 9140 [Ulf M��ller] 9141 9142 *) More tests in bntest.c, and changed test_bn output. 9143 [Ulf M��ller] 9144 9145 *) ./config recognizes MacOS X now. 9146 [Andy Polyakov] 9147 9148 *) Bug fix for BN_div() when the first words of num and divsor are 9149 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0). 9150 [Ulf M��ller] 9151 9152 *) Add support for various broken PKCS#8 formats, and command line 9153 options to produce them. 9154 [Steve Henson] 9155 9156 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to 9157 get temporary BIGNUMs from a BN_CTX. 9158 [Ulf M��ller] 9159 9160 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont() 9161 for p == 0. 9162 [Ulf M��ller] 9163 9164 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and 9165 include a #define from the old name to the new. The original intent 9166 was that statically linked binaries could for example just call 9167 SSLeay_add_all_ciphers() to just add ciphers to the table and not 9168 link with digests. This never worked becayse SSLeay_add_all_digests() 9169 and SSLeay_add_all_ciphers() were in the same source file so calling 9170 one would link with the other. They are now in separate source files. 9171 [Steve Henson] 9172 9173 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'. 9174 [Steve Henson] 9175 9176 *) Use a less unusual form of the Miller-Rabin primality test (it used 9177 a binary algorithm for exponentiation integrated into the Miller-Rabin 9178 loop, our standard modexp algorithms are faster). 9179 [Bodo Moeller] 9180 9181 *) Support for the EBCDIC character set completed. 9182 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>] 9183 9184 *) Source code cleanups: use const where appropriate, eliminate casts, 9185 use void * instead of char * in lhash. 9186 [Ulf M��ller] 9187 9188 *) Bugfix: ssl3_send_server_key_exchange was not restartable 9189 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of 9190 this the server could overwrite ephemeral keys that the client 9191 has already seen). 9192 [Bodo Moeller] 9193 9194 *) Turn DSA_is_prime into a macro that calls BN_is_prime, 9195 using 50 iterations of the Rabin-Miller test. 9196 9197 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50 9198 iterations of the Rabin-Miller test as required by the appendix 9199 to FIPS PUB 186[-1]) instead of DSA_is_prime. 9200 As BN_is_prime_fasttest includes trial division, DSA parameter 9201 generation becomes much faster. 9202 9203 This implies a change for the callback functions in DSA_is_prime 9204 and DSA_generate_parameters: The callback function is called once 9205 for each positive witness in the Rabin-Miller test, not just 9206 occasionally in the inner loop; and the parameters to the 9207 callback function now provide an iteration count for the outer 9208 loop rather than for the current invocation of the inner loop. 9209 DSA_generate_parameters additionally can call the callback 9210 function with an 'iteration count' of -1, meaning that a 9211 candidate has passed the trial division test (when q is generated 9212 from an application-provided seed, trial division is skipped). 9213 [Bodo Moeller] 9214 9215 *) New function BN_is_prime_fasttest that optionally does trial 9216 division before starting the Rabin-Miller test and has 9217 an additional BN_CTX * argument (whereas BN_is_prime always 9218 has to allocate at least one BN_CTX). 9219 'callback(1, -1, cb_arg)' is called when a number has passed the 9220 trial division stage. 9221 [Bodo Moeller] 9222 9223 *) Fix for bug in CRL encoding. The validity dates weren't being handled 9224 as ASN1_TIME. 9225 [Steve Henson] 9226 9227 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file. 9228 [Steve Henson] 9229 9230 *) New function BN_pseudo_rand(). 9231 [Ulf M��ller] 9232 9233 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable) 9234 bignum version of BN_from_montgomery() with the working code from 9235 SSLeay 0.9.0 (the word based version is faster anyway), and clean up 9236 the comments. 9237 [Ulf M��ller] 9238 9239 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that 9240 made it impossible to use the same SSL_SESSION data structure in 9241 SSL2 clients in multiple threads. 9242 [Bodo Moeller] 9243 9244 *) The return value of RAND_load_file() no longer counts bytes obtained 9245 by stat(). RAND_load_file(..., -1) is new and uses the complete file 9246 to seed the PRNG (previously an explicit byte count was required). 9247 [Ulf M��ller, Bodo M��ller] 9248 9249 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes 9250 used (char *) instead of (void *) and had casts all over the place. 9251 [Steve Henson] 9252 9253 *) Make BN_generate_prime() return NULL on error if ret!=NULL. 9254 [Ulf M��ller] 9255 9256 *) Retain source code compatibility for BN_prime_checks macro: 9257 BN_is_prime(..., BN_prime_checks, ...) now uses 9258 BN_prime_checks_for_size to determine the appropriate number of 9259 Rabin-Miller iterations. 9260 [Ulf M��ller] 9261 9262 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to 9263 DH_CHECK_P_NOT_SAFE_PRIME. 9264 (Check if this is true? OpenPGP calls them "strong".) 9265 [Ulf M��ller] 9266 9267 *) Merge the functionality of "dh" and "gendh" programs into a new program 9268 "dhparam". The old programs are retained for now but will handle DH keys 9269 (instead of parameters) in future. 9270 [Steve Henson] 9271 9272 *) Make the ciphers, s_server and s_client programs check the return values 9273 when a new cipher list is set. 9274 [Steve Henson] 9275 9276 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit 9277 ciphers. Before when the 56bit ciphers were enabled the sorting was 9278 wrong. 9279 9280 The syntax for the cipher sorting has been extended to support sorting by 9281 cipher-strength (using the strength_bits hard coded in the tables). 9282 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod). 9283 9284 Fix a bug in the cipher-command parser: when supplying a cipher command 9285 string with an "undefined" symbol (neither command nor alphanumeric 9286 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now 9287 an error is flagged. 9288 9289 Due to the strength-sorting extension, the code of the 9290 ssl_create_cipher_list() function was completely rearranged. I hope that 9291 the readability was also increased :-) 9292 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 9293 9294 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1 9295 for the first serial number and places 2 in the serial number file. This 9296 avoids problems when the root CA is created with serial number zero and 9297 the first user certificate has the same issuer name and serial number 9298 as the root CA. 9299 [Steve Henson] 9300 9301 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses 9302 the new code. Add documentation for this stuff. 9303 [Steve Henson] 9304 9305 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from 9306 X509_*() to X509at_*() on the grounds that they don't handle X509 9307 structures and behave in an analagous way to the X509v3 functions: 9308 they shouldn't be called directly but wrapper functions should be used 9309 instead. 9310 9311 So we also now have some wrapper functions that call the X509at functions 9312 when passed certificate requests. (TO DO: similar things can be done with 9313 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other 9314 things. Some of these need some d2i or i2d and print functionality 9315 because they handle more complex structures.) 9316 [Steve Henson] 9317 9318 *) Add missing #ifndefs that caused missing symbols when building libssl 9319 as a shared library without RSA. Use #ifndef NO_SSL2 instead of 9320 NO_RSA in ssl/s2*.c. 9321 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf M��ller] 9322 9323 *) Precautions against using the PRNG uninitialized: RAND_bytes() now 9324 has a return value which indicates the quality of the random data 9325 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's 9326 error queue. New function RAND_pseudo_bytes() generates output that is 9327 guaranteed to be unique but not unpredictable. RAND_add is like 9328 RAND_seed, but takes an extra argument for an entropy estimate 9329 (RAND_seed always assumes full entropy). 9330 [Ulf M��ller] 9331 9332 *) Do more iterations of Rabin-Miller probable prime test (specifically, 9333 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes 9334 instead of only 2 for all lengths; see BN_prime_checks_for_size definition 9335 in crypto/bn/bn_prime.c for the complete table). This guarantees a 9336 false-positive rate of at most 2^-80 for random input. 9337 [Bodo Moeller] 9338 9339 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs. 9340 [Bodo Moeller] 9341 9342 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain 9343 in the 0.9.5 release), this returns the chain 9344 from an X509_CTX structure with a dup of the stack and all 9345 the X509 reference counts upped: so the stack will exist 9346 after X509_CTX_cleanup() has been called. Modify pkcs12.c 9347 to use this. 9348 9349 Also make SSL_SESSION_print() print out the verify return 9350 code. 9351 [Steve Henson] 9352 9353 *) Add manpage for the pkcs12 command. Also change the default 9354 behaviour so MAC iteration counts are used unless the new 9355 -nomaciter option is used. This improves file security and 9356 only older versions of MSIE (4.0 for example) need it. 9357 [Steve Henson] 9358 9359 *) Honor the no-xxx Configure options when creating .DEF files. 9360 [Ulf M��ller] 9361 9362 *) Add PKCS#10 attributes to field table: challengePassword, 9363 unstructuredName and unstructuredAddress. These are taken from 9364 draft PKCS#9 v2.0 but are compatible with v1.2 provided no 9365 international characters are used. 9366 9367 More changes to X509_ATTRIBUTE code: allow the setting of types 9368 based on strings. Remove the 'loc' parameter when adding 9369 attributes because these will be a SET OF encoding which is sorted 9370 in ASN1 order. 9371 [Steve Henson] 9372 9373 *) Initial changes to the 'req' utility to allow request generation 9374 automation. This will allow an application to just generate a template 9375 file containing all the field values and have req construct the 9376 request. 9377 9378 Initial support for X509_ATTRIBUTE handling. Stacks of these are 9379 used all over the place including certificate requests and PKCS#7 9380 structures. They are currently handled manually where necessary with 9381 some primitive wrappers for PKCS#7. The new functions behave in a 9382 manner analogous to the X509 extension functions: they allow 9383 attributes to be looked up by NID and added. 9384 9385 Later something similar to the X509V3 code would be desirable to 9386 automatically handle the encoding, decoding and printing of the 9387 more complex types. The string types like challengePassword can 9388 be handled by the string table functions. 9389 9390 Also modified the multi byte string table handling. Now there is 9391 a 'global mask' which masks out certain types. The table itself 9392 can use the flag STABLE_NO_MASK to ignore the mask setting: this 9393 is useful when for example there is only one permissible type 9394 (as in countryName) and using the mask might result in no valid 9395 types at all. 9396 [Steve Henson] 9397 9398 *) Clean up 'Finished' handling, and add functions SSL_get_finished and 9399 SSL_get_peer_finished to allow applications to obtain the latest 9400 Finished messages sent to the peer or expected from the peer, 9401 respectively. (SSL_get_peer_finished is usually the Finished message 9402 actually received from the peer, otherwise the protocol will be aborted.) 9403 9404 As the Finished message are message digests of the complete handshake 9405 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can 9406 be used for external authentication procedures when the authentication 9407 provided by SSL/TLS is not desired or is not enough. 9408 [Bodo Moeller] 9409 9410 *) Enhanced support for Alpha Linux is added. Now ./config checks if 9411 the host supports BWX extension and if Compaq C is present on the 9412 $PATH. Just exploiting of the BWX extension results in 20-30% 9413 performance kick for some algorithms, e.g. DES and RC4 to mention 9414 a couple. Compaq C in turn generates ~20% faster code for MD5 and 9415 SHA1. 9416 [Andy Polyakov] 9417 9418 *) Add support for MS "fast SGC". This is arguably a violation of the 9419 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with 9420 weak crypto and after checking the certificate is SGC a second one 9421 with strong crypto. MS SGC stops the first handshake after receiving 9422 the server certificate message and sends a second client hello. Since 9423 a server will typically do all the time consuming operations before 9424 expecting any further messages from the client (server key exchange 9425 is the most expensive) there is little difference between the two. 9426 9427 To get OpenSSL to support MS SGC we have to permit a second client 9428 hello message after we have sent server done. In addition we have to 9429 reset the MAC if we do get this second client hello. 9430 [Steve Henson] 9431 9432 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide 9433 if a DER encoded private key is RSA or DSA traditional format. Changed 9434 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional" 9435 format DER encoded private key. Newer code should use PKCS#8 format which 9436 has the key type encoded in the ASN1 structure. Added DER private key 9437 support to pkcs8 application. 9438 [Steve Henson] 9439 9440 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous 9441 ciphersuites has been selected (as required by the SSL 3/TLS 1 9442 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT 9443 is set, we interpret this as a request to violate the specification 9444 (the worst that can happen is a handshake failure, and 'correct' 9445 behaviour would result in a handshake failure anyway). 9446 [Bodo Moeller] 9447 9448 *) In SSL_CTX_add_session, take into account that there might be multiple 9449 SSL_SESSION structures with the same session ID (e.g. when two threads 9450 concurrently obtain them from an external cache). 9451 The internal cache can handle only one SSL_SESSION with a given ID, 9452 so if there's a conflict, we now throw out the old one to achieve 9453 consistency. 9454 [Bodo Moeller] 9455 9456 *) Add OIDs for idea and blowfish in CBC mode. This will allow both 9457 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to 9458 some routines that use cipher OIDs: some ciphers do not have OIDs 9459 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for 9460 example. 9461 [Steve Henson] 9462 9463 *) Simplify the trust setting structure and code. Now we just have 9464 two sequences of OIDs for trusted and rejected settings. These will 9465 typically have values the same as the extended key usage extension 9466 and any application specific purposes. 9467 9468 The trust checking code now has a default behaviour: it will just 9469 check for an object with the same NID as the passed id. Functions can 9470 be provided to override either the default behaviour or the behaviour 9471 for a given id. SSL client, server and email already have functions 9472 in place for compatibility: they check the NID and also return "trusted" 9473 if the certificate is self signed. 9474 [Steve Henson] 9475 9476 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the 9477 traditional format into an EVP_PKEY structure. 9478 [Steve Henson] 9479 9480 *) Add a password callback function PEM_cb() which either prompts for 9481 a password if usr_data is NULL or otherwise assumes it is a null 9482 terminated password. Allow passwords to be passed on command line 9483 environment or config files in a few more utilities. 9484 [Steve Henson] 9485 9486 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private 9487 keys. Add some short names for PKCS#8 PBE algorithms and allow them 9488 to be specified on the command line for the pkcs8 and pkcs12 utilities. 9489 Update documentation. 9490 [Steve Henson] 9491 9492 *) Support for ASN1 "NULL" type. This could be handled before by using 9493 ASN1_TYPE but there wasn't any function that would try to read a NULL 9494 and produce an error if it couldn't. For compatibility we also have 9495 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and 9496 don't allocate anything because they don't need to. 9497 [Steve Henson] 9498 9499 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS 9500 for details. 9501 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>] 9502 9503 *) Rebuild of the memory allocation routines used by OpenSSL code and 9504 possibly others as well. The purpose is to make an interface that 9505 provide hooks so anyone can build a separate set of allocation and 9506 deallocation routines to be used by OpenSSL, for example memory 9507 pool implementations, or something else, which was previously hard 9508 since Malloc(), Realloc() and Free() were defined as macros having 9509 the values malloc, realloc and free, respectively (except for Win32 9510 compilations). The same is provided for memory debugging code. 9511 OpenSSL already comes with functionality to find memory leaks, but 9512 this gives people a chance to debug other memory problems. 9513 9514 With these changes, a new set of functions and macros have appeared: 9515 9516 CRYPTO_set_mem_debug_functions() [F] 9517 CRYPTO_get_mem_debug_functions() [F] 9518 CRYPTO_dbg_set_options() [F] 9519 CRYPTO_dbg_get_options() [F] 9520 CRYPTO_malloc_debug_init() [M] 9521 9522 The memory debug functions are NULL by default, unless the library 9523 is compiled with CRYPTO_MDEBUG or friends is defined. If someone 9524 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which 9525 gives the standard debugging functions that come with OpenSSL) or 9526 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions 9527 provided by the library user) must be used. When the standard 9528 debugging functions are used, CRYPTO_dbg_set_options can be used to 9529 request additional information: 9530 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting 9531 the CRYPTO_MDEBUG_xxx macro when compiling the library. 9532 9533 Also, things like CRYPTO_set_mem_functions will always give the 9534 expected result (the new set of functions is used for allocation 9535 and deallocation) at all times, regardless of platform and compiler 9536 options. 9537 9538 To finish it up, some functions that were never use in any other 9539 way than through macros have a new API and new semantic: 9540 9541 CRYPTO_dbg_malloc() 9542 CRYPTO_dbg_realloc() 9543 CRYPTO_dbg_free() 9544 9545 All macros of value have retained their old syntax. 9546 [Richard Levitte and Bodo Moeller] 9547 9548 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the 9549 ordering of SMIMECapabilities wasn't in "strength order" and there 9550 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature 9551 algorithm. 9552 [Steve Henson] 9553 9554 *) Some ASN1 types with illegal zero length encoding (INTEGER, 9555 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines. 9556 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson] 9557 9558 *) Merge in my S/MIME library for OpenSSL. This provides a simple 9559 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough 9560 functionality to handle multipart/signed properly) and a utility 9561 called 'smime' to call all this stuff. This is based on code I 9562 originally wrote for Celo who have kindly allowed it to be 9563 included in OpenSSL. 9564 [Steve Henson] 9565 9566 *) Add variants des_set_key_checked and des_set_key_unchecked of 9567 des_set_key (aka des_key_sched). Global variable des_check_key 9568 decides which of these is called by des_set_key; this way 9569 des_check_key behaves as it always did, but applications and 9570 the library itself, which was buggy for des_check_key == 1, 9571 have a cleaner way to pick the version they need. 9572 [Bodo Moeller] 9573 9574 *) New function PKCS12_newpass() which changes the password of a 9575 PKCS12 structure. 9576 [Steve Henson] 9577 9578 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and 9579 dynamic mix. In both cases the ids can be used as an index into the 9580 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add() 9581 functions so they accept a list of the field values and the 9582 application doesn't need to directly manipulate the X509_TRUST 9583 structure. 9584 [Steve Henson] 9585 9586 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't 9587 need initialising. 9588 [Steve Henson] 9589 9590 *) Modify the way the V3 extension code looks up extensions. This now 9591 works in a similar way to the object code: we have some "standard" 9592 extensions in a static table which is searched with OBJ_bsearch() 9593 and the application can add dynamic ones if needed. The file 9594 crypto/x509v3/ext_dat.h now has the info: this file needs to be 9595 updated whenever a new extension is added to the core code and kept 9596 in ext_nid order. There is a simple program 'tabtest.c' which checks 9597 this. New extensions are not added too often so this file can readily 9598 be maintained manually. 9599 9600 There are two big advantages in doing things this way. The extensions 9601 can be looked up immediately and no longer need to be "added" using 9602 X509V3_add_standard_extensions(): this function now does nothing. 9603 [Side note: I get *lots* of email saying the extension code doesn't 9604 work because people forget to call this function] 9605 Also no dynamic allocation is done unless new extensions are added: 9606 so if we don't add custom extensions there is no need to call 9607 X509V3_EXT_cleanup(). 9608 [Steve Henson] 9609 9610 *) Modify enc utility's salting as follows: make salting the default. Add a 9611 magic header, so unsalted files fail gracefully instead of just decrypting 9612 to garbage. This is because not salting is a big security hole, so people 9613 should be discouraged from doing it. 9614 [Ben Laurie] 9615 9616 *) Fixes and enhancements to the 'x509' utility. It allowed a message 9617 digest to be passed on the command line but it only used this 9618 parameter when signing a certificate. Modified so all relevant 9619 operations are affected by the digest parameter including the 9620 -fingerprint and -x509toreq options. Also -x509toreq choked if a 9621 DSA key was used because it didn't fix the digest. 9622 [Steve Henson] 9623 9624 *) Initial certificate chain verify code. Currently tests the untrusted 9625 certificates for consistency with the verify purpose (which is set 9626 when the X509_STORE_CTX structure is set up) and checks the pathlength. 9627 9628 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour: 9629 this is because it will reject chains with invalid extensions whereas 9630 every previous version of OpenSSL and SSLeay made no checks at all. 9631 9632 Trust code: checks the root CA for the relevant trust settings. Trust 9633 settings have an initial value consistent with the verify purpose: e.g. 9634 if the verify purpose is for SSL client use it expects the CA to be 9635 trusted for SSL client use. However the default value can be changed to 9636 permit custom trust settings: one example of this would be to only trust 9637 certificates from a specific "secure" set of CAs. 9638 9639 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions 9640 which should be used for version portability: especially since the 9641 verify structure is likely to change more often now. 9642 9643 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions 9644 to set them. If not set then assume SSL clients will verify SSL servers 9645 and vice versa. 9646 9647 Two new options to the verify program: -untrusted allows a set of 9648 untrusted certificates to be passed in and -purpose which sets the 9649 intended purpose of the certificate. If a purpose is set then the 9650 new chain verify code is used to check extension consistency. 9651 [Steve Henson] 9652 9653 *) Support for the authority information access extension. 9654 [Steve Henson] 9655 9656 *) Modify RSA and DSA PEM read routines to transparently handle 9657 PKCS#8 format private keys. New *_PUBKEY_* functions that handle 9658 public keys in a format compatible with certificate 9659 SubjectPublicKeyInfo structures. Unfortunately there were already 9660 functions called *_PublicKey_* which used various odd formats so 9661 these are retained for compatibility: however the DSA variants were 9662 never in a public release so they have been deleted. Changed dsa/rsa 9663 utilities to handle the new format: note no releases ever handled public 9664 keys so we should be OK. 9665 9666 The primary motivation for this change is to avoid the same fiasco 9667 that dogs private keys: there are several incompatible private key 9668 formats some of which are standard and some OpenSSL specific and 9669 require various evil hacks to allow partial transparent handling and 9670 even then it doesn't work with DER formats. Given the option anything 9671 other than PKCS#8 should be dumped: but the other formats have to 9672 stay in the name of compatibility. 9673 9674 With public keys and the benefit of hindsight one standard format 9675 is used which works with EVP_PKEY, RSA or DSA structures: though 9676 it clearly returns an error if you try to read the wrong kind of key. 9677 9678 Added a -pubkey option to the 'x509' utility to output the public key. 9679 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*() 9680 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add 9681 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*()) 9682 that do the same as the EVP_PKEY_assign_*() except they up the 9683 reference count of the added key (they don't "swallow" the 9684 supplied key). 9685 [Steve Henson] 9686 9687 *) Fixes to crypto/x509/by_file.c the code to read in certificates and 9688 CRLs would fail if the file contained no certificates or no CRLs: 9689 added a new function to read in both types and return the number 9690 read: this means that if none are read it will be an error. The 9691 DER versions of the certificate and CRL reader would always fail 9692 because it isn't possible to mix certificates and CRLs in DER format 9693 without choking one or the other routine. Changed this to just read 9694 a certificate: this is the best we can do. Also modified the code 9695 in apps/verify.c to take notice of return codes: it was previously 9696 attempting to read in certificates from NULL pointers and ignoring 9697 any errors: this is one reason why the cert and CRL reader seemed 9698 to work. It doesn't check return codes from the default certificate 9699 routines: these may well fail if the certificates aren't installed. 9700 [Steve Henson] 9701 9702 *) Code to support otherName option in GeneralName. 9703 [Steve Henson] 9704 9705 *) First update to verify code. Change the verify utility 9706 so it warns if it is passed a self signed certificate: 9707 for consistency with the normal behaviour. X509_verify 9708 has been modified to it will now verify a self signed 9709 certificate if *exactly* the same certificate appears 9710 in the store: it was previously impossible to trust a 9711 single self signed certificate. This means that: 9712 openssl verify ss.pem 9713 now gives a warning about a self signed certificate but 9714 openssl verify -CAfile ss.pem ss.pem 9715 is OK. 9716 [Steve Henson] 9717 9718 *) For servers, store verify_result in SSL_SESSION data structure 9719 (and add it to external session representation). 9720 This is needed when client certificate verifications fails, 9721 but an application-provided verification callback (set by 9722 SSL_CTX_set_cert_verify_callback) allows accepting the session 9723 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK 9724 but returns 1): When the session is reused, we have to set 9725 ssl->verify_result to the appropriate error code to avoid 9726 security holes. 9727 [Bodo Moeller, problem pointed out by Lutz Jaenicke] 9728 9729 *) Fix a bug in the new PKCS#7 code: it didn't consider the 9730 case in PKCS7_dataInit() where the signed PKCS7 structure 9731 didn't contain any existing data because it was being created. 9732 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson] 9733 9734 *) Add a salt to the key derivation routines in enc.c. This 9735 forms the first 8 bytes of the encrypted file. Also add a 9736 -S option to allow a salt to be input on the command line. 9737 [Steve Henson] 9738 9739 *) New function X509_cmp(). Oddly enough there wasn't a function 9740 to compare two certificates. We do this by working out the SHA1 9741 hash and comparing that. X509_cmp() will be needed by the trust 9742 code. 9743 [Steve Henson] 9744 9745 *) SSL_get1_session() is like SSL_get_session(), but increments 9746 the reference count in the SSL_SESSION returned. 9747 [Geoff Thorpe <geoff@eu.c2.net>] 9748 9749 *) Fix for 'req': it was adding a null to request attributes. 9750 Also change the X509_LOOKUP and X509_INFO code to handle 9751 certificate auxiliary information. 9752 [Steve Henson] 9753 9754 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document 9755 the 'enc' command. 9756 [Steve Henson] 9757 9758 *) Add the possibility to add extra information to the memory leak 9759 detecting output, to form tracebacks, showing from where each 9760 allocation was originated: CRYPTO_push_info("constant string") adds 9761 the string plus current file name and line number to a per-thread 9762 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info() 9763 is like calling CYRPTO_pop_info() until the stack is empty. 9764 Also updated memory leak detection code to be multi-thread-safe. 9765 [Richard Levitte] 9766 9767 *) Add options -text and -noout to pkcs7 utility and delete the 9768 encryption options which never did anything. Update docs. 9769 [Steve Henson] 9770 9771 *) Add options to some of the utilities to allow the pass phrase 9772 to be included on either the command line (not recommended on 9773 OSes like Unix) or read from the environment. Update the 9774 manpages and fix a few bugs. 9775 [Steve Henson] 9776 9777 *) Add a few manpages for some of the openssl commands. 9778 [Steve Henson] 9779 9780 *) Fix the -revoke option in ca. It was freeing up memory twice, 9781 leaking and not finding already revoked certificates. 9782 [Steve Henson] 9783 9784 *) Extensive changes to support certificate auxiliary information. 9785 This involves the use of X509_CERT_AUX structure and X509_AUX 9786 functions. An X509_AUX function such as PEM_read_X509_AUX() 9787 can still read in a certificate file in the usual way but it 9788 will also read in any additional "auxiliary information". By 9789 doing things this way a fair degree of compatibility can be 9790 retained: existing certificates can have this information added 9791 using the new 'x509' options. 9792 9793 Current auxiliary information includes an "alias" and some trust 9794 settings. The trust settings will ultimately be used in enhanced 9795 certificate chain verification routines: currently a certificate 9796 can only be trusted if it is self signed and then it is trusted 9797 for all purposes. 9798 [Steve Henson] 9799 9800 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD). 9801 The problem was that one of the replacement routines had not been working 9802 since SSLeay releases. For now the offending routine has been replaced 9803 with non-optimised assembler. Even so, this now gives around 95% 9804 performance improvement for 1024 bit RSA signs. 9805 [Mark Cox] 9806 9807 *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2 9808 handling. Most clients have the effective key size in bits equal to 9809 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key. 9810 A few however don't do this and instead use the size of the decrypted key 9811 to determine the RC2 key length and the AlgorithmIdentifier to determine 9812 the effective key length. In this case the effective key length can still 9813 be 40 bits but the key length can be 168 bits for example. This is fixed 9814 by manually forcing an RC2 key into the EVP_PKEY structure because the 9815 EVP code can't currently handle unusual RC2 key sizes: it always assumes 9816 the key length and effective key length are equal. 9817 [Steve Henson] 9818 9819 *) Add a bunch of functions that should simplify the creation of 9820 X509_NAME structures. Now you should be able to do: 9821 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0); 9822 and have it automatically work out the correct field type and fill in 9823 the structures. The more adventurous can try: 9824 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0); 9825 and it will (hopefully) work out the correct multibyte encoding. 9826 [Steve Henson] 9827 9828 *) Change the 'req' utility to use the new field handling and multibyte 9829 copy routines. Before the DN field creation was handled in an ad hoc 9830 way in req, ca, and x509 which was rather broken and didn't support 9831 BMPStrings or UTF8Strings. Since some software doesn't implement 9832 BMPStrings or UTF8Strings yet, they can be enabled using the config file 9833 using the dirstring_type option. See the new comment in the default 9834 openssl.cnf for more info. 9835 [Steve Henson] 9836 9837 *) Make crypto/rand/md_rand.c more robust: 9838 - Assure unique random numbers after fork(). 9839 - Make sure that concurrent threads access the global counter and 9840 md serializably so that we never lose entropy in them 9841 or use exactly the same state in multiple threads. 9842 Access to the large state is not always serializable because 9843 the additional locking could be a performance killer, and 9844 md should be large enough anyway. 9845 [Bodo Moeller] 9846 9847 *) New file apps/app_rand.c with commonly needed functionality 9848 for handling the random seed file. 9849 9850 Use the random seed file in some applications that previously did not: 9851 ca, 9852 dsaparam -genkey (which also ignored its '-rand' option), 9853 s_client, 9854 s_server, 9855 x509 (when signing). 9856 Except on systems with /dev/urandom, it is crucial to have a random 9857 seed file at least for key creation, DSA signing, and for DH exchanges; 9858 for RSA signatures we could do without one. 9859 9860 gendh and gendsa (unlike genrsa) used to read only the first byte 9861 of each file listed in the '-rand' option. The function as previously 9862 found in genrsa is now in app_rand.c and is used by all programs 9863 that support '-rand'. 9864 [Bodo Moeller] 9865 9866 *) In RAND_write_file, use mode 0600 for creating files; 9867 don't just chmod when it may be too late. 9868 [Bodo Moeller] 9869 9870 *) Report an error from X509_STORE_load_locations 9871 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed. 9872 [Bill Perry] 9873 9874 *) New function ASN1_mbstring_copy() this copies a string in either 9875 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format 9876 into an ASN1_STRING type. A mask of permissible types is passed 9877 and it chooses the "minimal" type to use or an error if not type 9878 is suitable. 9879 [Steve Henson] 9880 9881 *) Add function equivalents to the various macros in asn1.h. The old 9882 macros are retained with an M_ prefix. Code inside the library can 9883 use the M_ macros. External code (including the openssl utility) 9884 should *NOT* in order to be "shared library friendly". 9885 [Steve Henson] 9886 9887 *) Add various functions that can check a certificate's extensions 9888 to see if it usable for various purposes such as SSL client, 9889 server or S/MIME and CAs of these types. This is currently 9890 VERY EXPERIMENTAL but will ultimately be used for certificate chain 9891 verification. Also added a -purpose flag to x509 utility to 9892 print out all the purposes. 9893 [Steve Henson] 9894 9895 *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated 9896 functions. 9897 [Steve Henson] 9898 9899 *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search 9900 for, obtain and decode and extension and obtain its critical flag. 9901 This allows all the necessary extension code to be handled in a 9902 single function call. 9903 [Steve Henson] 9904 9905 *) RC4 tune-up featuring 30-40% performance improvement on most RISC 9906 platforms. See crypto/rc4/rc4_enc.c for further details. 9907 [Andy Polyakov] 9908 9909 *) New -noout option to asn1parse. This causes no output to be produced 9910 its main use is when combined with -strparse and -out to extract data 9911 from a file (which may not be in ASN.1 format). 9912 [Steve Henson] 9913 9914 *) Fix for pkcs12 program. It was hashing an invalid certificate pointer 9915 when producing the local key id. 9916 [Richard Levitte <levitte@stacken.kth.se>] 9917 9918 *) New option -dhparam in s_server. This allows a DH parameter file to be 9919 stated explicitly. If it is not stated then it tries the first server 9920 certificate file. The previous behaviour hard coded the filename 9921 "server.pem". 9922 [Steve Henson] 9923 9924 *) Add -pubin and -pubout options to the rsa and dsa commands. These allow 9925 a public key to be input or output. For example: 9926 openssl rsa -in key.pem -pubout -out pubkey.pem 9927 Also added necessary DSA public key functions to handle this. 9928 [Steve Henson] 9929 9930 *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained 9931 in the message. This was handled by allowing 9932 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it. 9933 [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>] 9934 9935 *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null 9936 to the end of the strings whereas this didn't. This would cause problems 9937 if strings read with d2i_ASN1_bytes() were later modified. 9938 [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>] 9939 9940 *) Fix for base64 decode bug. When a base64 bio reads only one line of 9941 data and it contains EOF it will end up returning an error. This is 9942 caused by input 46 bytes long. The cause is due to the way base64 9943 BIOs find the start of base64 encoded data. They do this by trying a 9944 trial decode on each line until they find one that works. When they 9945 do a flag is set and it starts again knowing it can pass all the 9946 data directly through the decoder. Unfortunately it doesn't reset 9947 the context it uses. This means that if EOF is reached an attempt 9948 is made to pass two EOFs through the context and this causes the 9949 resulting error. This can also cause other problems as well. As is 9950 usual with these problems it takes *ages* to find and the fix is 9951 trivial: move one line. 9952 [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ] 9953 9954 *) Ugly workaround to get s_client and s_server working under Windows. The 9955 old code wouldn't work because it needed to select() on sockets and the 9956 tty (for keypresses and to see if data could be written). Win32 only 9957 supports select() on sockets so we select() with a 1s timeout on the 9958 sockets and then see if any characters are waiting to be read, if none 9959 are present then we retry, we also assume we can always write data to 9960 the tty. This isn't nice because the code then blocks until we've 9961 received a complete line of data and it is effectively polling the 9962 keyboard at 1s intervals: however it's quite a bit better than not 9963 working at all :-) A dedicated Windows application might handle this 9964 with an event loop for example. 9965 [Steve Henson] 9966 9967 *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign 9968 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions 9969 will be called when RSA_sign() and RSA_verify() are used. This is useful 9970 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available. 9971 For this to work properly RSA_public_decrypt() and RSA_private_encrypt() 9972 should *not* be used: RSA_sign() and RSA_verify() must be used instead. 9973 This necessitated the support of an extra signature type NID_md5_sha1 9974 for SSL signatures and modifications to the SSL library to use it instead 9975 of calling RSA_public_decrypt() and RSA_private_encrypt(). 9976 [Steve Henson] 9977 9978 *) Add new -verify -CAfile and -CApath options to the crl program, these 9979 will lookup a CRL issuers certificate and verify the signature in a 9980 similar way to the verify program. Tidy up the crl program so it 9981 no longer accesses structures directly. Make the ASN1 CRL parsing a bit 9982 less strict. It will now permit CRL extensions even if it is not 9983 a V2 CRL: this will allow it to tolerate some broken CRLs. 9984 [Steve Henson] 9985 9986 *) Initialize all non-automatic variables each time one of the openssl 9987 sub-programs is started (this is necessary as they may be started 9988 multiple times from the "OpenSSL>" prompt). 9989 [Lennart Bang, Bodo Moeller] 9990 9991 *) Preliminary compilation option RSA_NULL which disables RSA crypto without 9992 removing all other RSA functionality (this is what NO_RSA does). This 9993 is so (for example) those in the US can disable those operations covered 9994 by the RSA patent while allowing storage and parsing of RSA keys and RSA 9995 key generation. 9996 [Steve Henson] 9997 9998 *) Non-copying interface to BIO pairs. 9999 (still largely untested) 10000 [Bodo Moeller] 10001 10002 *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive 10003 ASCII string. This was handled independently in various places before. 10004 [Steve Henson] 10005 10006 *) New functions UTF8_getc() and UTF8_putc() that parse and generate 10007 UTF8 strings a character at a time. 10008 [Steve Henson] 10009 10010 *) Use client_version from client hello to select the protocol 10011 (s23_srvr.c) and for RSA client key exchange verification 10012 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications. 10013 [Bodo Moeller] 10014 10015 *) Add various utility functions to handle SPKACs, these were previously 10016 handled by poking round in the structure internals. Added new function 10017 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to 10018 print, verify and generate SPKACs. Based on an original idea from 10019 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified. 10020 [Steve Henson] 10021 10022 *) RIPEMD160 is operational on all platforms and is back in 'make test'. 10023 [Andy Polyakov] 10024 10025 *) Allow the config file extension section to be overwritten on the 10026 command line. Based on an original idea from Massimiliano Pala 10027 <madwolf@comune.modena.it>. The new option is called -extensions 10028 and can be applied to ca, req and x509. Also -reqexts to override 10029 the request extensions in req and -crlexts to override the crl extensions 10030 in ca. 10031 [Steve Henson] 10032 10033 *) Add new feature to the SPKAC handling in ca. Now you can include 10034 the same field multiple times by preceding it by "XXXX." for example: 10035 1.OU="Unit name 1" 10036 2.OU="Unit name 2" 10037 this is the same syntax as used in the req config file. 10038 [Steve Henson] 10039 10040 *) Allow certificate extensions to be added to certificate requests. These 10041 are specified in a 'req_extensions' option of the req section of the 10042 config file. They can be printed out with the -text option to req but 10043 are otherwise ignored at present. 10044 [Steve Henson] 10045 10046 *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first 10047 data read consists of only the final block it would not decrypted because 10048 EVP_CipherUpdate() would correctly report zero bytes had been decrypted. 10049 A misplaced 'break' also meant the decrypted final block might not be 10050 copied until the next read. 10051 [Steve Henson] 10052 10053 *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added 10054 a few extra parameters to the DH structure: these will be useful if 10055 for example we want the value of 'q' or implement X9.42 DH. 10056 [Steve Henson] 10057 10058 *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and 10059 provides hooks that allow the default DSA functions or functions on a 10060 "per key" basis to be replaced. This allows hardware acceleration and 10061 hardware key storage to be handled without major modification to the 10062 library. Also added low level modexp hooks and CRYPTO_EX structure and 10063 associated functions. 10064 [Steve Henson] 10065 10066 *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO 10067 as "read only": it can't be written to and the buffer it points to will 10068 not be freed. Reading from a read only BIO is much more efficient than 10069 a normal memory BIO. This was added because there are several times when 10070 an area of memory needs to be read from a BIO. The previous method was 10071 to create a memory BIO and write the data to it, this results in two 10072 copies of the data and an O(n^2) reading algorithm. There is a new 10073 function BIO_new_mem_buf() which creates a read only memory BIO from 10074 an area of memory. Also modified the PKCS#7 routines to use read only 10075 memory BIOs. 10076 [Steve Henson] 10077 10078 *) Bugfix: ssl23_get_client_hello did not work properly when called in 10079 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of 10080 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read, 10081 but a retry condition occured while trying to read the rest. 10082 [Bodo Moeller] 10083 10084 *) The PKCS7_ENC_CONTENT_new() function was setting the content type as 10085 NID_pkcs7_encrypted by default: this was wrong since this should almost 10086 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle 10087 the encrypted data type: this is a more sensible place to put it and it 10088 allows the PKCS#12 code to be tidied up that duplicated this 10089 functionality. 10090 [Steve Henson] 10091 10092 *) Changed obj_dat.pl script so it takes its input and output files on 10093 the command line. This should avoid shell escape redirection problems 10094 under Win32. 10095 [Steve Henson] 10096 10097 *) Initial support for certificate extension requests, these are included 10098 in things like Xenroll certificate requests. Included functions to allow 10099 extensions to be obtained and added. 10100 [Steve Henson] 10101 10102 *) -crlf option to s_client and s_server for sending newlines as 10103 CRLF (as required by many protocols). 10104 [Bodo Moeller] 10105 10106 Changes between 0.9.3a and 0.9.4 [09 Aug 1999] 10107 10108 *) Install libRSAglue.a when OpenSSL is built with RSAref. 10109 [Ralf S. Engelschall] 10110 10111 *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency. 10112 [Andrija Antonijevic <TheAntony2@bigfoot.com>] 10113 10114 *) Fix -startdate and -enddate (which was missing) arguments to 'ca' 10115 program. 10116 [Steve Henson] 10117 10118 *) New function DSA_dup_DH, which duplicates DSA parameters/keys as 10119 DH parameters/keys (q is lost during that conversion, but the resulting 10120 DH parameters contain its length). 10121 10122 For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is 10123 much faster than DH_generate_parameters (which creates parameters 10124 where p = 2*q + 1), and also the smaller q makes DH computations 10125 much more efficient (160-bit exponentiation instead of 1024-bit 10126 exponentiation); so this provides a convenient way to support DHE 10127 ciphersuites in SSL/TLS servers (see ssl/ssltest.c). It is of 10128 utter importance to use 10129 SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 10130 or 10131 SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 10132 when such DH parameters are used, because otherwise small subgroup 10133 attacks may become possible! 10134 [Bodo Moeller] 10135 10136 *) Avoid memory leak in i2d_DHparams. 10137 [Bodo Moeller] 10138 10139 *) Allow the -k option to be used more than once in the enc program: 10140 this allows the same encrypted message to be read by multiple recipients. 10141 [Steve Henson] 10142 10143 *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts 10144 an ASN1_OBJECT to a text string. If the "no_name" parameter is set then 10145 it will always use the numerical form of the OID, even if it has a short 10146 or long name. 10147 [Steve Henson] 10148 10149 *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp 10150 method only got called if p,q,dmp1,dmq1,iqmp components were present, 10151 otherwise bn_mod_exp was called. In the case of hardware keys for example 10152 no private key components need be present and it might store extra data 10153 in the RSA structure, which cannot be accessed from bn_mod_exp. 10154 By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for 10155 private key operations. 10156 [Steve Henson] 10157 10158 *) Added support for SPARC Linux. 10159 [Andy Polyakov] 10160 10161 *) pem_password_cb function type incompatibly changed from 10162 typedef int pem_password_cb(char *buf, int size, int rwflag); 10163 to 10164 ....(char *buf, int size, int rwflag, void *userdata); 10165 so that applications can pass data to their callbacks: 10166 The PEM[_ASN1]_{read,write}... functions and macros now take an 10167 additional void * argument, which is just handed through whenever 10168 the password callback is called. 10169 [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller] 10170 10171 New function SSL_CTX_set_default_passwd_cb_userdata. 10172 10173 Compatibility note: As many C implementations push function arguments 10174 onto the stack in reverse order, the new library version is likely to 10175 interoperate with programs that have been compiled with the old 10176 pem_password_cb definition (PEM_whatever takes some data that 10177 happens to be on the stack as its last argument, and the callback 10178 just ignores this garbage); but there is no guarantee whatsoever that 10179 this will work. 10180 10181 *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=... 10182 (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused 10183 problems not only on Windows, but also on some Unix platforms. 10184 To avoid problematic command lines, these definitions are now in an 10185 auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl 10186 for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds). 10187 [Bodo Moeller] 10188 10189 *) MIPS III/IV assembler module is reimplemented. 10190 [Andy Polyakov] 10191 10192 *) More DES library cleanups: remove references to srand/rand and 10193 delete an unused file. 10194 [Ulf M��ller] 10195 10196 *) Add support for the the free Netwide assembler (NASM) under Win32, 10197 since not many people have MASM (ml) and it can be hard to obtain. 10198 This is currently experimental but it seems to work OK and pass all 10199 the tests. Check out INSTALL.W32 for info. 10200 [Steve Henson] 10201 10202 *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections 10203 without temporary keys kept an extra copy of the server key, 10204 and connections with temporary keys did not free everything in case 10205 of an error. 10206 [Bodo Moeller] 10207 10208 *) New function RSA_check_key and new openssl rsa option -check 10209 for verifying the consistency of RSA keys. 10210 [Ulf Moeller, Bodo Moeller] 10211 10212 *) Various changes to make Win32 compile work: 10213 1. Casts to avoid "loss of data" warnings in p5_crpt2.c 10214 2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned 10215 comparison" warnings. 10216 3. Add sk_<TYPE>_sort to DEF file generator and do make update. 10217 [Steve Henson] 10218 10219 *) Add a debugging option to PKCS#5 v2 key generation function: when 10220 you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and 10221 derived keys are printed to stderr. 10222 [Steve Henson] 10223 10224 *) Copy the flags in ASN1_STRING_dup(). 10225 [Roman E. Pavlov <pre@mo.msk.ru>] 10226 10227 *) The x509 application mishandled signing requests containing DSA 10228 keys when the signing key was also DSA and the parameters didn't match. 10229 10230 It was supposed to omit the parameters when they matched the signing key: 10231 the verifying software was then supposed to automatically use the CA's 10232 parameters if they were absent from the end user certificate. 10233 10234 Omitting parameters is no longer recommended. The test was also 10235 the wrong way round! This was probably due to unusual behaviour in 10236 EVP_cmp_parameters() which returns 1 if the parameters match. 10237 This meant that parameters were omitted when they *didn't* match and 10238 the certificate was useless. Certificates signed with 'ca' didn't have 10239 this bug. 10240 [Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>] 10241 10242 *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems. 10243 The interface is as follows: 10244 Applications can use 10245 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(), 10246 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop(); 10247 "off" is now the default. 10248 The library internally uses 10249 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(), 10250 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on() 10251 to disable memory-checking temporarily. 10252 10253 Some inconsistent states that previously were possible (and were 10254 even the default) are now avoided. 10255 10256 -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time 10257 with each memory chunk allocated; this is occasionally more helpful 10258 than just having a counter. 10259 10260 -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID. 10261 10262 -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future 10263 extensions. 10264 [Bodo Moeller] 10265 10266 *) Introduce "mode" for SSL structures (with defaults in SSL_CTX), 10267 which largely parallels "options", but is for changing API behaviour, 10268 whereas "options" are about protocol behaviour. 10269 Initial "mode" flags are: 10270 10271 SSL_MODE_ENABLE_PARTIAL_WRITE Allow SSL_write to report success when 10272 a single record has been written. 10273 SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER Don't insist that SSL_write 10274 retries use the same buffer location. 10275 (But all of the contents must be 10276 copied!) 10277 [Bodo Moeller] 10278 10279 *) Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options 10280 worked. 10281 10282 *) Fix problems with no-hmac etc. 10283 [Ulf M��ller, pointed out by Brian Wellington <bwelling@tislabs.com>] 10284 10285 *) New functions RSA_get_default_method(), RSA_set_method() and 10286 RSA_get_method(). These allows replacement of RSA_METHODs without having 10287 to mess around with the internals of an RSA structure. 10288 [Steve Henson] 10289 10290 *) Fix memory leaks in DSA_do_sign and DSA_is_prime. 10291 Also really enable memory leak checks in openssl.c and in some 10292 test programs. 10293 [Chad C. Mulligan, Bodo Moeller] 10294 10295 *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess 10296 up the length of negative integers. This has now been simplified to just 10297 store the length when it is first determined and use it later, rather 10298 than trying to keep track of where data is copied and updating it to 10299 point to the end. 10300 [Steve Henson, reported by Brien Wheeler 10301 <bwheeler@authentica-security.com>] 10302 10303 *) Add a new function PKCS7_signatureVerify. This allows the verification 10304 of a PKCS#7 signature but with the signing certificate passed to the 10305 function itself. This contrasts with PKCS7_dataVerify which assumes the 10306 certificate is present in the PKCS#7 structure. This isn't always the 10307 case: certificates can be omitted from a PKCS#7 structure and be 10308 distributed by "out of band" means (such as a certificate database). 10309 [Steve Henson] 10310 10311 *) Complete the PEM_* macros with DECLARE_PEM versions to replace the 10312 function prototypes in pem.h, also change util/mkdef.pl to add the 10313 necessary function names. 10314 [Steve Henson] 10315 10316 *) mk1mf.pl (used by Windows builds) did not properly read the 10317 options set by Configure in the top level Makefile, and Configure 10318 was not even able to write more than one option correctly. 10319 Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended. 10320 [Bodo Moeller] 10321 10322 *) New functions CONF_load_bio() and CONF_load_fp() to allow a config 10323 file to be loaded from a BIO or FILE pointer. The BIO version will 10324 for example allow memory BIOs to contain config info. 10325 [Steve Henson] 10326 10327 *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS. 10328 Whoever hopes to achieve shared-library compatibility across versions 10329 must use this, not the compile-time macro. 10330 (Exercise 0.9.4: Which is the minimum library version required by 10331 such programs?) 10332 Note: All this applies only to multi-threaded programs, others don't 10333 need locks. 10334 [Bodo Moeller] 10335 10336 *) Add missing case to s3_clnt.c state machine -- one of the new SSL tests 10337 through a BIO pair triggered the default case, i.e. 10338 SSLerr(...,SSL_R_UNKNOWN_STATE). 10339 [Bodo Moeller] 10340 10341 *) New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications 10342 can use the SSL library even if none of the specific BIOs is 10343 appropriate. 10344 [Bodo Moeller] 10345 10346 *) Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value 10347 for the encoded length. 10348 [Jeon KyoungHo <khjeon@sds.samsung.co.kr>] 10349 10350 *) Add initial documentation of the X509V3 functions. 10351 [Steve Henson] 10352 10353 *) Add a new pair of functions PEM_write_PKCS8PrivateKey() and 10354 PEM_write_bio_PKCS8PrivateKey() that are equivalent to 10355 PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more 10356 secure PKCS#8 private key format with a high iteration count. 10357 [Steve Henson] 10358 10359 *) Fix determination of Perl interpreter: A perl or perl5 10360 _directory_ in $PATH was also accepted as the interpreter. 10361 [Ralf S. Engelschall] 10362 10363 *) Fix demos/sign/sign.c: well there wasn't anything strictly speaking 10364 wrong with it but it was very old and did things like calling 10365 PEM_ASN1_read() directly and used MD5 for the hash not to mention some 10366 unusual formatting. 10367 [Steve Henson] 10368 10369 *) Fix demos/selfsign.c: it used obsolete and deleted functions, changed 10370 to use the new extension code. 10371 [Steve Henson] 10372 10373 *) Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c 10374 with macros. This should make it easier to change their form, add extra 10375 arguments etc. Fix a few PEM prototypes which didn't have cipher as a 10376 constant. 10377 [Steve Henson] 10378 10379 *) Add to configuration table a new entry that can specify an alternative 10380 name for unistd.h (for pre-POSIX systems); we need this for NeXTstep, 10381 according to Mark Crispin <MRC@Panda.COM>. 10382 [Bodo Moeller] 10383 10384#if 0 10385 *) DES CBC did not update the IV. Weird. 10386 [Ben Laurie] 10387#else 10388 des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does. 10389 Changing the behaviour of the former might break existing programs -- 10390 where IV updating is needed, des_ncbc_encrypt can be used. 10391#endif 10392 10393 *) When bntest is run from "make test" it drives bc to check its 10394 calculations, as well as internally checking them. If an internal check 10395 fails, it needs to cause bc to give a non-zero result or make test carries 10396 on without noticing the failure. Fixed. 10397 [Ben Laurie] 10398 10399 *) DES library cleanups. 10400 [Ulf M��ller] 10401 10402 *) Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be 10403 used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit 10404 ciphers. NOTE: although the key derivation function has been verified 10405 against some published test vectors it has not been extensively tested 10406 yet. Added a -v2 "cipher" option to pkcs8 application to allow the use 10407 of v2.0. 10408 [Steve Henson] 10409 10410 *) Instead of "mkdir -p", which is not fully portable, use new 10411 Perl script "util/mkdir-p.pl". 10412 [Bodo Moeller] 10413 10414 *) Rewrite the way password based encryption (PBE) is handled. It used to 10415 assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter 10416 structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms 10417 but doesn't apply to PKCS#5 v2.0 where it can be something else. Now 10418 the 'parameter' field of the AlgorithmIdentifier is passed to the 10419 underlying key generation function so it must do its own ASN1 parsing. 10420 This has also changed the EVP_PBE_CipherInit() function which now has a 10421 'parameter' argument instead of literal salt and iteration count values 10422 and the function EVP_PBE_ALGOR_CipherInit() has been deleted. 10423 [Steve Henson] 10424 10425 *) Support for PKCS#5 v1.5 compatible password based encryption algorithms 10426 and PKCS#8 functionality. New 'pkcs8' application linked to openssl. 10427 Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE 10428 KEY" because this clashed with PKCS#8 unencrypted string. Since this 10429 value was just used as a "magic string" and not used directly its 10430 value doesn't matter. 10431 [Steve Henson] 10432 10433 *) Introduce some semblance of const correctness to BN. Shame C doesn't 10434 support mutable. 10435 [Ben Laurie] 10436 10437 *) "linux-sparc64" configuration (ultrapenguin). 10438 [Ray Miller <ray.miller@oucs.ox.ac.uk>] 10439 "linux-sparc" configuration. 10440 [Christian Forster <fo@hawo.stw.uni-erlangen.de>] 10441 10442 *) config now generates no-xxx options for missing ciphers. 10443 [Ulf M��ller] 10444 10445 *) Support the EBCDIC character set (work in progress). 10446 File ebcdic.c not yet included because it has a different license. 10447 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 10448 10449 *) Support BS2000/OSD-POSIX. 10450 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 10451 10452 *) Make callbacks for key generation use void * instead of char *. 10453 [Ben Laurie] 10454 10455 *) Make S/MIME samples compile (not yet tested). 10456 [Ben Laurie] 10457 10458 *) Additional typesafe stacks. 10459 [Ben Laurie] 10460 10461 *) New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x). 10462 [Bodo Moeller] 10463 10464 10465 Changes between 0.9.3 and 0.9.3a [29 May 1999] 10466 10467 *) New configuration variant "sco5-gcc". 10468 10469 *) Updated some demos. 10470 [Sean O Riordain, Wade Scholine] 10471 10472 *) Add missing BIO_free at exit of pkcs12 application. 10473 [Wu Zhigang] 10474 10475 *) Fix memory leak in conf.c. 10476 [Steve Henson] 10477 10478 *) Updates for Win32 to assembler version of MD5. 10479 [Steve Henson] 10480 10481 *) Set #! path to perl in apps/der_chop to where we found it 10482 instead of using a fixed path. 10483 [Bodo Moeller] 10484 10485 *) SHA library changes for irix64-mips4-cc. 10486 [Andy Polyakov] 10487 10488 *) Improvements for VMS support. 10489 [Richard Levitte] 10490 10491 10492 Changes between 0.9.2b and 0.9.3 [24 May 1999] 10493 10494 *) Bignum library bug fix. IRIX 6 passes "make test" now! 10495 This also avoids the problems with SC4.2 and unpatched SC5. 10496 [Andy Polyakov <appro@fy.chalmers.se>] 10497 10498 *) New functions sk_num, sk_value and sk_set to replace the previous macros. 10499 These are required because of the typesafe stack would otherwise break 10500 existing code. If old code used a structure member which used to be STACK 10501 and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with 10502 sk_num or sk_value it would produce an error because the num, data members 10503 are not present in STACK_OF. Now it just produces a warning. sk_set 10504 replaces the old method of assigning a value to sk_value 10505 (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code 10506 that does this will no longer work (and should use sk_set instead) but 10507 this could be regarded as a "questionable" behaviour anyway. 10508 [Steve Henson] 10509 10510 *) Fix most of the other PKCS#7 bugs. The "experimental" code can now 10511 correctly handle encrypted S/MIME data. 10512 [Steve Henson] 10513 10514 *) Change type of various DES function arguments from des_cblock 10515 (which means, in function argument declarations, pointer to char) 10516 to des_cblock * (meaning pointer to array with 8 char elements), 10517 which allows the compiler to do more typechecking; it was like 10518 that back in SSLeay, but with lots of ugly casts. 10519 10520 Introduce new type const_des_cblock. 10521 [Bodo Moeller] 10522 10523 *) Reorganise the PKCS#7 library and get rid of some of the more obvious 10524 problems: find RecipientInfo structure that matches recipient certificate 10525 and initialise the ASN1 structures properly based on passed cipher. 10526 [Steve Henson] 10527 10528 *) Belatedly make the BN tests actually check the results. 10529 [Ben Laurie] 10530 10531 *) Fix the encoding and decoding of negative ASN1 INTEGERS and conversion 10532 to and from BNs: it was completely broken. New compilation option 10533 NEG_PUBKEY_BUG to allow for some broken certificates that encode public 10534 key elements as negative integers. 10535 [Steve Henson] 10536 10537 *) Reorganize and speed up MD5. 10538 [Andy Polyakov <appro@fy.chalmers.se>] 10539 10540 *) VMS support. 10541 [Richard Levitte <richard@levitte.org>] 10542 10543 *) New option -out to asn1parse to allow the parsed structure to be 10544 output to a file. This is most useful when combined with the -strparse 10545 option to examine the output of things like OCTET STRINGS. 10546 [Steve Henson] 10547 10548 *) Make SSL library a little more fool-proof by not requiring any longer 10549 that SSL_set_{accept,connect}_state be called before 10550 SSL_{accept,connect} may be used (SSL_set_..._state is omitted 10551 in many applications because usually everything *appeared* to work as 10552 intended anyway -- now it really works as intended). 10553 [Bodo Moeller] 10554 10555 *) Move openssl.cnf out of lib/. 10556 [Ulf M��ller] 10557 10558 *) Fix various things to let OpenSSL even pass ``egcc -pipe -O2 -Wall 10559 -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes 10560 -Wmissing-declarations -Wnested-externs -Winline'' with EGCS 1.1.2+ 10561 [Ralf S. Engelschall] 10562 10563 *) Various fixes to the EVP and PKCS#7 code. It may now be able to 10564 handle PKCS#7 enveloped data properly. 10565 [Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve] 10566 10567 *) Create a duplicate of the SSL_CTX's CERT in SSL_new instead of 10568 copying pointers. The cert_st handling is changed by this in 10569 various ways (and thus what used to be known as ctx->default_cert 10570 is now called ctx->cert, since we don't resort to s->ctx->[default_]cert 10571 any longer when s->cert does not give us what we need). 10572 ssl_cert_instantiate becomes obsolete by this change. 10573 As soon as we've got the new code right (possibly it already is?), 10574 we have solved a couple of bugs of the earlier code where s->cert 10575 was used as if it could not have been shared with other SSL structures. 10576 10577 Note that using the SSL API in certain dirty ways now will result 10578 in different behaviour than observed with earlier library versions: 10579 Changing settings for an SSL_CTX *ctx after having done s = SSL_new(ctx) 10580 does not influence s as it used to. 10581 10582 In order to clean up things more thoroughly, inside SSL_SESSION 10583 we don't use CERT any longer, but a new structure SESS_CERT 10584 that holds per-session data (if available); currently, this is 10585 the peer's certificate chain and, for clients, the server's certificate 10586 and temporary key. CERT holds only those values that can have 10587 meaningful defaults in an SSL_CTX. 10588 [Bodo Moeller] 10589 10590 *) New function X509V3_EXT_i2d() to create an X509_EXTENSION structure 10591 from the internal representation. Various PKCS#7 fixes: remove some 10592 evil casts and set the enc_dig_alg field properly based on the signing 10593 key type. 10594 [Steve Henson] 10595 10596 *) Allow PKCS#12 password to be set from the command line or the 10597 environment. Let 'ca' get its config file name from the environment 10598 variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req' 10599 and 'x509'). 10600 [Steve Henson] 10601 10602 *) Allow certificate policies extension to use an IA5STRING for the 10603 organization field. This is contrary to the PKIX definition but 10604 VeriSign uses it and IE5 only recognises this form. Document 'x509' 10605 extension option. 10606 [Steve Henson] 10607 10608 *) Add PEDANTIC compiler flag to allow compilation with gcc -pedantic, 10609 without disallowing inline assembler and the like for non-pedantic builds. 10610 [Ben Laurie] 10611 10612 *) Support Borland C++ builder. 10613 [Janez Jere <jj@void.si>, modified by Ulf M��ller] 10614 10615 *) Support Mingw32. 10616 [Ulf M��ller] 10617 10618 *) SHA-1 cleanups and performance enhancements. 10619 [Andy Polyakov <appro@fy.chalmers.se>] 10620 10621 *) Sparc v8plus assembler for the bignum library. 10622 [Andy Polyakov <appro@fy.chalmers.se>] 10623 10624 *) Accept any -xxx and +xxx compiler options in Configure. 10625 [Ulf M��ller] 10626 10627 *) Update HPUX configuration. 10628 [Anonymous] 10629 10630 *) Add missing sk_<type>_unshift() function to safestack.h 10631 [Ralf S. Engelschall] 10632 10633 *) New function SSL_CTX_use_certificate_chain_file that sets the 10634 "extra_cert"s in addition to the certificate. (This makes sense 10635 only for "PEM" format files, as chains as a whole are not 10636 DER-encoded.) 10637 [Bodo Moeller] 10638 10639 *) Support verify_depth from the SSL API. 10640 x509_vfy.c had what can be considered an off-by-one-error: 10641 Its depth (which was not part of the external interface) 10642 was actually counting the number of certificates in a chain; 10643 now it really counts the depth. 10644 [Bodo Moeller] 10645 10646 *) Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used 10647 instead of X509err, which often resulted in confusing error 10648 messages since the error codes are not globally unique 10649 (e.g. an alleged error in ssl3_accept when a certificate 10650 didn't match the private key). 10651 10652 *) New function SSL_CTX_set_session_id_context that allows to set a default 10653 value (so that you don't need SSL_set_session_id_context for each 10654 connection using the SSL_CTX). 10655 [Bodo Moeller] 10656 10657 *) OAEP decoding bug fix. 10658 [Ulf M��ller] 10659 10660 *) Support INSTALL_PREFIX for package builders, as proposed by 10661 David Harris. 10662 [Bodo Moeller] 10663 10664 *) New Configure options "threads" and "no-threads". For systems 10665 where the proper compiler options are known (currently Solaris 10666 and Linux), "threads" is the default. 10667 [Bodo Moeller] 10668 10669 *) New script util/mklink.pl as a faster substitute for util/mklink.sh. 10670 [Bodo Moeller] 10671 10672 *) Install various scripts to $(OPENSSLDIR)/misc, not to 10673 $(INSTALLTOP)/bin -- they shouldn't clutter directories 10674 such as /usr/local/bin. 10675 [Bodo Moeller] 10676 10677 *) "make linux-shared" to build shared libraries. 10678 [Niels Poppe <niels@netbox.org>] 10679 10680 *) New Configure option no-<cipher> (rsa, idea, rc5, ...). 10681 [Ulf M��ller] 10682 10683 *) Add the PKCS#12 API documentation to openssl.txt. Preliminary support for 10684 extension adding in x509 utility. 10685 [Steve Henson] 10686 10687 *) Remove NOPROTO sections and error code comments. 10688 [Ulf M��ller] 10689 10690 *) Partial rewrite of the DEF file generator to now parse the ANSI 10691 prototypes. 10692 [Steve Henson] 10693 10694 *) New Configure options --prefix=DIR and --openssldir=DIR. 10695 [Ulf M��ller] 10696 10697 *) Complete rewrite of the error code script(s). It is all now handled 10698 by one script at the top level which handles error code gathering, 10699 header rewriting and C source file generation. It should be much better 10700 than the old method: it now uses a modified version of Ulf's parser to 10701 read the ANSI prototypes in all header files (thus the old K&R definitions 10702 aren't needed for error creation any more) and do a better job of 10703 translating function codes into names. The old 'ASN1 error code imbedded 10704 in a comment' is no longer necessary and it doesn't use .err files which 10705 have now been deleted. Also the error code call doesn't have to appear all 10706 on one line (which resulted in some large lines...). 10707 [Steve Henson] 10708 10709 *) Change #include filenames from <foo.h> to <openssl/foo.h>. 10710 [Bodo Moeller] 10711 10712 *) Change behaviour of ssl2_read when facing length-0 packets: Don't return 10713 0 (which usually indicates a closed connection), but continue reading. 10714 [Bodo Moeller] 10715 10716 *) Fix some race conditions. 10717 [Bodo Moeller] 10718 10719 *) Add support for CRL distribution points extension. Add Certificate 10720 Policies and CRL distribution points documentation. 10721 [Steve Henson] 10722 10723 *) Move the autogenerated header file parts to crypto/opensslconf.h. 10724 [Ulf M��ller] 10725 10726 *) Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of 10727 8 of keying material. Merlin has also confirmed interop with this fix 10728 between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0. 10729 [Merlin Hughes <merlin@baltimore.ie>] 10730 10731 *) Fix lots of warnings. 10732 [Richard Levitte <levitte@stacken.kth.se>] 10733 10734 *) In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if 10735 the directory spec didn't end with a LIST_SEPARATOR_CHAR. 10736 [Richard Levitte <levitte@stacken.kth.se>] 10737 10738 *) Fix problems with sizeof(long) == 8. 10739 [Andy Polyakov <appro@fy.chalmers.se>] 10740 10741 *) Change functions to ANSI C. 10742 [Ulf M��ller] 10743 10744 *) Fix typos in error codes. 10745 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf M��ller] 10746 10747 *) Remove defunct assembler files from Configure. 10748 [Ulf M��ller] 10749 10750 *) SPARC v8 assembler BIGNUM implementation. 10751 [Andy Polyakov <appro@fy.chalmers.se>] 10752 10753 *) Support for Certificate Policies extension: both print and set. 10754 Various additions to support the r2i method this uses. 10755 [Steve Henson] 10756 10757 *) A lot of constification, and fix a bug in X509_NAME_oneline() that could 10758 return a const string when you are expecting an allocated buffer. 10759 [Ben Laurie] 10760 10761 *) Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE 10762 types DirectoryString and DisplayText. 10763 [Steve Henson] 10764 10765 *) Add code to allow r2i extensions to access the configuration database, 10766 add an LHASH database driver and add several ctx helper functions. 10767 [Steve Henson] 10768 10769 *) Fix an evil bug in bn_expand2() which caused various BN functions to 10770 fail when they extended the size of a BIGNUM. 10771 [Steve Henson] 10772 10773 *) Various utility functions to handle SXNet extension. Modify mkdef.pl to 10774 support typesafe stack. 10775 [Steve Henson] 10776 10777 *) Fix typo in SSL_[gs]et_options(). 10778 [Nils Frostberg <nils@medcom.se>] 10779 10780 *) Delete various functions and files that belonged to the (now obsolete) 10781 old X509V3 handling code. 10782 [Steve Henson] 10783 10784 *) New Configure option "rsaref". 10785 [Ulf M��ller] 10786 10787 *) Don't auto-generate pem.h. 10788 [Bodo Moeller] 10789 10790 *) Introduce type-safe ASN.1 SETs. 10791 [Ben Laurie] 10792 10793 *) Convert various additional casted stacks to type-safe STACK_OF() variants. 10794 [Ben Laurie, Ralf S. Engelschall, Steve Henson] 10795 10796 *) Introduce type-safe STACKs. This will almost certainly break lots of code 10797 that links with OpenSSL (well at least cause lots of warnings), but fear 10798 not: the conversion is trivial, and it eliminates loads of evil casts. A 10799 few STACKed things have been converted already. Feel free to convert more. 10800 In the fullness of time, I'll do away with the STACK type altogether. 10801 [Ben Laurie] 10802 10803 *) Add `openssl ca -revoke <certfile>' facility which revokes a certificate 10804 specified in <certfile> by updating the entry in the index.txt file. 10805 This way one no longer has to edit the index.txt file manually for 10806 revoking a certificate. The -revoke option does the gory details now. 10807 [Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall] 10808 10809 *) Fix `openssl crl -noout -text' combination where `-noout' killed the 10810 `-text' option at all and this way the `-noout -text' combination was 10811 inconsistent in `openssl crl' with the friends in `openssl x509|rsa|dsa'. 10812 [Ralf S. Engelschall] 10813 10814 *) Make sure a corresponding plain text error message exists for the 10815 X509_V_ERR_CERT_REVOKED/23 error number which can occur when a 10816 verify callback function determined that a certificate was revoked. 10817 [Ralf S. Engelschall] 10818 10819 *) Bugfix: In test/testenc, don't test "openssl <cipher>" for 10820 ciphers that were excluded, e.g. by -DNO_IDEA. Also, test 10821 all available cipers including rc5, which was forgotten until now. 10822 In order to let the testing shell script know which algorithms 10823 are available, a new (up to now undocumented) command 10824 "openssl list-cipher-commands" is used. 10825 [Bodo Moeller] 10826 10827 *) Bugfix: s_client occasionally would sleep in select() when 10828 it should have checked SSL_pending() first. 10829 [Bodo Moeller] 10830 10831 *) New functions DSA_do_sign and DSA_do_verify to provide access to 10832 the raw DSA values prior to ASN.1 encoding. 10833 [Ulf M��ller] 10834 10835 *) Tweaks to Configure 10836 [Niels Poppe <niels@netbox.org>] 10837 10838 *) Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support, 10839 yet... 10840 [Steve Henson] 10841 10842 *) New variables $(RANLIB) and $(PERL) in the Makefiles. 10843 [Ulf M��ller] 10844 10845 *) New config option to avoid instructions that are illegal on the 80386. 10846 The default code is faster, but requires at least a 486. 10847 [Ulf M��ller] 10848 10849 *) Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and 10850 SSL2_SERVER_VERSION (not used at all) macros, which are now the 10851 same as SSL2_VERSION anyway. 10852 [Bodo Moeller] 10853 10854 *) New "-showcerts" option for s_client. 10855 [Bodo Moeller] 10856 10857 *) Still more PKCS#12 integration. Add pkcs12 application to openssl 10858 application. Various cleanups and fixes. 10859 [Steve Henson] 10860 10861 *) More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and 10862 modify error routines to work internally. Add error codes and PBE init 10863 to library startup routines. 10864 [Steve Henson] 10865 10866 *) Further PKCS#12 integration. Added password based encryption, PKCS#8 and 10867 packing functions to asn1 and evp. Changed function names and error 10868 codes along the way. 10869 [Steve Henson] 10870 10871 *) PKCS12 integration: and so it begins... First of several patches to 10872 slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12 10873 objects to objects.h 10874 [Steve Henson] 10875 10876 *) Add a new 'indent' option to some X509V3 extension code. Initial ASN1 10877 and display support for Thawte strong extranet extension. 10878 [Steve Henson] 10879 10880 *) Add LinuxPPC support. 10881 [Jeff Dubrule <igor@pobox.org>] 10882 10883 *) Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to 10884 bn_div_words in alpha.s. 10885 [Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie] 10886 10887 *) Make sure the RSA OAEP test is skipped under -DRSAref because 10888 OAEP isn't supported when OpenSSL is built with RSAref. 10889 [Ulf Moeller <ulf@fitug.de>] 10890 10891 *) Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h 10892 so they no longer are missing under -DNOPROTO. 10893 [Soren S. Jorvang <soren@t.dk>] 10894 10895 10896 Changes between 0.9.1c and 0.9.2b [22 Mar 1999] 10897 10898 *) Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still 10899 doesn't work when the session is reused. Coming soon! 10900 [Ben Laurie] 10901 10902 *) Fix a security hole, that allows sessions to be reused in the wrong 10903 context thus bypassing client cert protection! All software that uses 10904 client certs and session caches in multiple contexts NEEDS PATCHING to 10905 allow session reuse! A fuller solution is in the works. 10906 [Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)] 10907 10908 *) Some more source tree cleanups (removed obsolete files 10909 crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed 10910 permission on "config" script to be executable) and a fix for the INSTALL 10911 document. 10912 [Ulf Moeller <ulf@fitug.de>] 10913 10914 *) Remove some legacy and erroneous uses of malloc, free instead of 10915 Malloc, Free. 10916 [Lennart Bang <lob@netstream.se>, with minor changes by Steve] 10917 10918 *) Make rsa_oaep_test return non-zero on error. 10919 [Ulf Moeller <ulf@fitug.de>] 10920 10921 *) Add support for native Solaris shared libraries. Configure 10922 solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice 10923 if someone would make that last step automatic. 10924 [Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>] 10925 10926 *) ctx_size was not built with the right compiler during "make links". Fixed. 10927 [Ben Laurie] 10928 10929 *) Change the meaning of 'ALL' in the cipher list. It now means "everything 10930 except NULL ciphers". This means the default cipher list will no longer 10931 enable NULL ciphers. They need to be specifically enabled e.g. with 10932 the string "DEFAULT:eNULL". 10933 [Steve Henson] 10934 10935 *) Fix to RSA private encryption routines: if p < q then it would 10936 occasionally produce an invalid result. This will only happen with 10937 externally generated keys because OpenSSL (and SSLeay) ensure p > q. 10938 [Steve Henson] 10939 10940 *) Be less restrictive and allow also `perl util/perlpath.pl 10941 /path/to/bin/perl' in addition to `perl util/perlpath.pl /path/to/bin', 10942 because this way one can also use an interpreter named `perl5' (which is 10943 usually the name of Perl 5.xxx on platforms where an Perl 4.x is still 10944 installed as `perl'). 10945 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 10946 10947 *) Let util/clean-depend.pl work also with older Perl 5.00x versions. 10948 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 10949 10950 *) Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add 10951 advapi32.lib to Win32 build and change the pem test comparision 10952 to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the 10953 suggestion). Fix misplaced ASNI prototypes and declarations in evp.h 10954 and crypto/des/ede_cbcm_enc.c. 10955 [Steve Henson] 10956 10957 *) DES quad checksum was broken on big-endian architectures. Fixed. 10958 [Ben Laurie] 10959 10960 *) Comment out two functions in bio.h that aren't implemented. Fix up the 10961 Win32 test batch file so it (might) work again. The Win32 test batch file 10962 is horrible: I feel ill.... 10963 [Steve Henson] 10964 10965 *) Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected 10966 in e_os.h. Audit of header files to check ANSI and non ANSI 10967 sections: 10 functions were absent from non ANSI section and not exported 10968 from Windows DLLs. Fixed up libeay.num for new functions. 10969 [Steve Henson] 10970 10971 *) Make `openssl version' output lines consistent. 10972 [Ralf S. Engelschall] 10973 10974 *) Fix Win32 symbol export lists for BIO functions: Added 10975 BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data 10976 to ms/libeay{16,32}.def. 10977 [Ralf S. Engelschall] 10978 10979 *) Second round of fixing the OpenSSL perl/ stuff. It now at least compiled 10980 fine under Unix and passes some trivial tests I've now added. But the 10981 whole stuff is horribly incomplete, so a README.1ST with a disclaimer was 10982 added to make sure no one expects that this stuff really works in the 10983 OpenSSL 0.9.2 release. Additionally I've started to clean the XS sources 10984 up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and 10985 openssl_bio.xs. 10986 [Ralf S. Engelschall] 10987 10988 *) Fix the generation of two part addresses in perl. 10989 [Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie] 10990 10991 *) Add config entry for Linux on MIPS. 10992 [John Tobey <jtobey@channel1.com>] 10993 10994 *) Make links whenever Configure is run, unless we are on Windoze. 10995 [Ben Laurie] 10996 10997 *) Permit extensions to be added to CRLs using crl_section in openssl.cnf. 10998 Currently only issuerAltName and AuthorityKeyIdentifier make any sense 10999 in CRLs. 11000 [Steve Henson] 11001 11002 *) Add a useful kludge to allow package maintainers to specify compiler and 11003 other platforms details on the command line without having to patch the 11004 Configure script everytime: One now can use ``perl Configure 11005 <id>:<details>'', i.e. platform ids are allowed to have details appended 11006 to them (seperated by colons). This is treated as there would be a static 11007 pre-configured entry in Configure's %table under key <id> with value 11008 <details> and ``perl Configure <id>'' is called. So, when you want to 11009 perform a quick test-compile under FreeBSD 3.1 with pgcc and without 11010 assembler stuff you can use ``perl Configure "FreeBSD-elf:pgcc:-O6:::"'' 11011 now, which overrides the FreeBSD-elf entry on-the-fly. 11012 [Ralf S. Engelschall] 11013 11014 *) Disable new TLS1 ciphersuites by default: they aren't official yet. 11015 [Ben Laurie] 11016 11017 *) Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified 11018 on the `perl Configure ...' command line. This way one can compile 11019 OpenSSL libraries with Position Independent Code (PIC) which is needed 11020 for linking it into DSOs. 11021 [Ralf S. Engelschall] 11022 11023 *) Remarkably, export ciphers were totally broken and no-one had noticed! 11024 Fixed. 11025 [Ben Laurie] 11026 11027 *) Cleaned up the LICENSE document: The official contact for any license 11028 questions now is the OpenSSL core team under openssl-core@openssl.org. 11029 And add a paragraph about the dual-license situation to make sure people 11030 recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply 11031 to the OpenSSL toolkit. 11032 [Ralf S. Engelschall] 11033 11034 *) General source tree makefile cleanups: Made `making xxx in yyy...' 11035 display consistent in the source tree and replaced `/bin/rm' by `rm'. 11036 Additonally cleaned up the `make links' target: Remove unnecessary 11037 semicolons, subsequent redundant removes, inline point.sh into mklink.sh 11038 to speed processing and no longer clutter the display with confusing 11039 stuff. Instead only the actually done links are displayed. 11040 [Ralf S. Engelschall] 11041 11042 *) Permit null encryption ciphersuites, used for authentication only. It used 11043 to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this. 11044 It is now necessary to set SSL_FORBID_ENULL to prevent the use of null 11045 encryption. 11046 [Ben Laurie] 11047 11048 *) Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder 11049 signed attributes when verifying signatures (this would break them), 11050 the detached data encoding was wrong and public keys obtained using 11051 X509_get_pubkey() weren't freed. 11052 [Steve Henson] 11053 11054 *) Add text documentation for the BUFFER functions. Also added a work around 11055 to a Win95 console bug. This was triggered by the password read stuff: the 11056 last character typed gets carried over to the next fread(). If you were 11057 generating a new cert request using 'req' for example then the last 11058 character of the passphrase would be CR which would then enter the first 11059 field as blank. 11060 [Steve Henson] 11061 11062 *) Added the new `Includes OpenSSL Cryptography Software' button as 11063 doc/openssl_button.{gif,html} which is similar in style to the old SSLeay 11064 button and can be used by applications based on OpenSSL to show the 11065 relationship to the OpenSSL project. 11066 [Ralf S. Engelschall] 11067 11068 *) Remove confusing variables in function signatures in files 11069 ssl/ssl_lib.c and ssl/ssl.h. 11070 [Lennart Bong <lob@kulthea.stacken.kth.se>] 11071 11072 *) Don't install bss_file.c under PREFIX/include/ 11073 [Lennart Bong <lob@kulthea.stacken.kth.se>] 11074 11075 *) Get the Win32 compile working again. Modify mkdef.pl so it can handle 11076 functions that return function pointers and has support for NT specific 11077 stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various 11078 #ifdef WIN32 and WINNTs sprinkled about the place and some changes from 11079 unsigned to signed types: this was killing the Win32 compile. 11080 [Steve Henson] 11081 11082 *) Add new certificate file to stack functions, 11083 SSL_add_dir_cert_subjects_to_stack() and 11084 SSL_add_file_cert_subjects_to_stack(). These largely supplant 11085 SSL_load_client_CA_file(), and can be used to add multiple certs easily 11086 to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()). 11087 This means that Apache-SSL and similar packages don't have to mess around 11088 to add as many CAs as they want to the preferred list. 11089 [Ben Laurie] 11090 11091 *) Experiment with doxygen documentation. Currently only partially applied to 11092 ssl/ssl_lib.c. 11093 See http://www.stack.nl/~dimitri/doxygen/index.html, and run doxygen with 11094 openssl.doxy as the configuration file. 11095 [Ben Laurie] 11096 11097 *) Get rid of remaining C++-style comments which strict C compilers hate. 11098 [Ralf S. Engelschall, pointed out by Carlos Amengual] 11099 11100 *) Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not 11101 compiled in by default: it has problems with large keys. 11102 [Steve Henson] 11103 11104 *) Add a bunch of SSL_xxx() functions for configuring the temporary RSA and 11105 DH private keys and/or callback functions which directly correspond to 11106 their SSL_CTX_xxx() counterparts but work on a per-connection basis. This 11107 is needed for applications which have to configure certificates on a 11108 per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis 11109 (e.g. s_server). 11110 For the RSA certificate situation is makes no difference, but 11111 for the DSA certificate situation this fixes the "no shared cipher" 11112 problem where the OpenSSL cipher selection procedure failed because the 11113 temporary keys were not overtaken from the context and the API provided 11114 no way to reconfigure them. 11115 The new functions now let applications reconfigure the stuff and they 11116 are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh, 11117 SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback. Additionally a new 11118 non-public-API function ssl_cert_instantiate() is used as a helper 11119 function and also to reduce code redundancy inside ssl_rsa.c. 11120 [Ralf S. Engelschall] 11121 11122 *) Move s_server -dcert and -dkey options out of the undocumented feature 11123 area because they are useful for the DSA situation and should be 11124 recognized by the users. 11125 [Ralf S. Engelschall] 11126 11127 *) Fix the cipher decision scheme for export ciphers: the export bits are 11128 *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within 11129 SSL_EXP_MASK. So, the original variable has to be used instead of the 11130 already masked variable. 11131 [Richard Levitte <levitte@stacken.kth.se>] 11132 11133 *) Fix 'port' variable from `int' to `unsigned int' in crypto/bio/b_sock.c 11134 [Richard Levitte <levitte@stacken.kth.se>] 11135 11136 *) Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal() 11137 from `int' to `unsigned int' because it's a length and initialized by 11138 EVP_DigestFinal() which expects an `unsigned int *'. 11139 [Richard Levitte <levitte@stacken.kth.se>] 11140 11141 *) Don't hard-code path to Perl interpreter on shebang line of Configure 11142 script. Instead use the usual Shell->Perl transition trick. 11143 [Ralf S. Engelschall] 11144 11145 *) Make `openssl x509 -noout -modulus' functional also for DSA certificates 11146 (in addition to RSA certificates) to match the behaviour of `openssl dsa 11147 -noout -modulus' as it's already the case for `openssl rsa -noout 11148 -modulus'. For RSA the -modulus is the real "modulus" while for DSA 11149 currently the public key is printed (a decision which was already done by 11150 `openssl dsa -modulus' in the past) which serves a similar purpose. 11151 Additionally the NO_RSA no longer completely removes the whole -modulus 11152 option; it now only avoids using the RSA stuff. Same applies to NO_DSA 11153 now, too. 11154 [Ralf S. Engelschall] 11155 11156 *) Add Arne Ansper's reliable BIO - this is an encrypted, block-digested 11157 BIO. See the source (crypto/evp/bio_ok.c) for more info. 11158 [Arne Ansper <arne@ats.cyber.ee>] 11159 11160 *) Dump the old yucky req code that tried (and failed) to allow raw OIDs 11161 to be added. Now both 'req' and 'ca' can use new objects defined in the 11162 config file. 11163 [Steve Henson] 11164 11165 *) Add cool BIO that does syslog (or event log on NT). 11166 [Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie] 11167 11168 *) Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5, 11169 TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and 11170 TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher 11171 Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt. 11172 [Ben Laurie] 11173 11174 *) Add preliminary config info for new extension code. 11175 [Steve Henson] 11176 11177 *) Make RSA_NO_PADDING really use no padding. 11178 [Ulf Moeller <ulf@fitug.de>] 11179 11180 *) Generate errors when private/public key check is done. 11181 [Ben Laurie] 11182 11183 *) Overhaul for 'crl' utility. New function X509_CRL_print. Partial support 11184 for some CRL extensions and new objects added. 11185 [Steve Henson] 11186 11187 *) Really fix the ASN1 IMPLICIT bug this time... Partial support for private 11188 key usage extension and fuller support for authority key id. 11189 [Steve Henson] 11190 11191 *) Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved 11192 padding method for RSA, which is recommended for new applications in PKCS 11193 #1 v2.0 (RFC 2437, October 1998). 11194 OAEP (Optimal Asymmetric Encryption Padding) has better theoretical 11195 foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure 11196 against Bleichbacher's attack on RSA. 11197 [Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by 11198 Ben Laurie] 11199 11200 *) Updates to the new SSL compression code 11201 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 11202 11203 *) Fix so that the version number in the master secret, when passed 11204 via RSA, checks that if TLS was proposed, but we roll back to SSLv3 11205 (because the server will not accept higher), that the version number 11206 is 0x03,0x01, not 0x03,0x00 11207 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 11208 11209 *) Run extensive memory leak checks on SSL apps. Fixed *lots* of memory 11210 leaks in ssl/ relating to new X509_get_pubkey() behaviour. Also fixes 11211 in apps/ and an unrelated leak in crypto/dsa/dsa_vrf.c 11212 [Steve Henson] 11213 11214 *) Support for RAW extensions where an arbitrary extension can be 11215 created by including its DER encoding. See apps/openssl.cnf for 11216 an example. 11217 [Steve Henson] 11218 11219 *) Make sure latest Perl versions don't interpret some generated C array 11220 code as Perl array code in the crypto/err/err_genc.pl script. 11221 [Lars Weber <3weber@informatik.uni-hamburg.de>] 11222 11223 *) Modify ms/do_ms.bat to not generate assembly language makefiles since 11224 not many people have the assembler. Various Win32 compilation fixes and 11225 update to the INSTALL.W32 file with (hopefully) more accurate Win32 11226 build instructions. 11227 [Steve Henson] 11228 11229 *) Modify configure script 'Configure' to automatically create crypto/date.h 11230 file under Win32 and also build pem.h from pem.org. New script 11231 util/mkfiles.pl to create the MINFO file on environments that can't do a 11232 'make files': perl util/mkfiles.pl >MINFO should work. 11233 [Steve Henson] 11234 11235 *) Major rework of DES function declarations, in the pursuit of correctness 11236 and purity. As a result, many evil casts evaporated, and some weirdness, 11237 too. You may find this causes warnings in your code. Zapping your evil 11238 casts will probably fix them. Mostly. 11239 [Ben Laurie] 11240 11241 *) Fix for a typo in asn1.h. Bug fix to object creation script 11242 obj_dat.pl. It considered a zero in an object definition to mean 11243 "end of object": none of the objects in objects.h have any zeros 11244 so it wasn't spotted. 11245 [Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>] 11246 11247 *) Add support for Triple DES Cipher Block Chaining with Output Feedback 11248 Masking (CBCM). In the absence of test vectors, the best I have been able 11249 to do is check that the decrypt undoes the encrypt, so far. Send me test 11250 vectors if you have them. 11251 [Ben Laurie] 11252 11253 *) Correct calculation of key length for export ciphers (too much space was 11254 allocated for null ciphers). This has not been tested! 11255 [Ben Laurie] 11256 11257 *) Modifications to the mkdef.pl for Win32 DEF file creation. The usage 11258 message is now correct (it understands "crypto" and "ssl" on its 11259 command line). There is also now an "update" option. This will update 11260 the util/ssleay.num and util/libeay.num files with any new functions. 11261 If you do a: 11262 perl util/mkdef.pl crypto ssl update 11263 it will update them. 11264 [Steve Henson] 11265 11266 *) Overhauled the Perl interface (perl/*): 11267 - ported BN stuff to OpenSSL's different BN library 11268 - made the perl/ source tree CVS-aware 11269 - renamed the package from SSLeay to OpenSSL (the files still contain 11270 their history because I've copied them in the repository) 11271 - removed obsolete files (the test scripts will be replaced 11272 by better Test::Harness variants in the future) 11273 [Ralf S. Engelschall] 11274 11275 *) First cut for a very conservative source tree cleanup: 11276 1. merge various obsolete readme texts into doc/ssleay.txt 11277 where we collect the old documents and readme texts. 11278 2. remove the first part of files where I'm already sure that we no 11279 longer need them because of three reasons: either they are just temporary 11280 files which were left by Eric or they are preserved original files where 11281 I've verified that the diff is also available in the CVS via "cvs diff 11282 -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for 11283 the crypto/md/ stuff). 11284 [Ralf S. Engelschall] 11285 11286 *) More extension code. Incomplete support for subject and issuer alt 11287 name, issuer and authority key id. Change the i2v function parameters 11288 and add an extra 'crl' parameter in the X509V3_CTX structure: guess 11289 what that's for :-) Fix to ASN1 macro which messed up 11290 IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED. 11291 [Steve Henson] 11292 11293 *) Preliminary support for ENUMERATED type. This is largely copied from the 11294 INTEGER code. 11295 [Steve Henson] 11296 11297 *) Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy. 11298 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 11299 11300 *) Make sure `make rehash' target really finds the `openssl' program. 11301 [Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 11302 11303 *) Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd 11304 like to hear about it if this slows down other processors. 11305 [Ben Laurie] 11306 11307 *) Add CygWin32 platform information to Configure script. 11308 [Alan Batie <batie@aahz.jf.intel.com>] 11309 11310 *) Fixed ms/32all.bat script: `no_asm' -> `no-asm' 11311 [Rainer W. Gerling <gerling@mpg-gv.mpg.de>] 11312 11313 *) New program nseq to manipulate netscape certificate sequences 11314 [Steve Henson] 11315 11316 *) Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a 11317 few typos. 11318 [Steve Henson] 11319 11320 *) Fixes to BN code. Previously the default was to define BN_RECURSION 11321 but the BN code had some problems that would cause failures when 11322 doing certificate verification and some other functions. 11323 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 11324 11325 *) Add ASN1 and PEM code to support netscape certificate sequences. 11326 [Steve Henson] 11327 11328 *) Add ASN1 and PEM code to support netscape certificate sequences. 11329 [Steve Henson] 11330 11331 *) Add several PKIX and private extended key usage OIDs. 11332 [Steve Henson] 11333 11334 *) Modify the 'ca' program to handle the new extension code. Modify 11335 openssl.cnf for new extension format, add comments. 11336 [Steve Henson] 11337 11338 *) More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req' 11339 and add a sample to openssl.cnf so req -x509 now adds appropriate 11340 CA extensions. 11341 [Steve Henson] 11342 11343 *) Continued X509 V3 changes. Add to other makefiles, integrate with the 11344 error code, add initial support to X509_print() and x509 application. 11345 [Steve Henson] 11346 11347 *) Takes a deep breath and start addding X509 V3 extension support code. Add 11348 files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this 11349 stuff is currently isolated and isn't even compiled yet. 11350 [Steve Henson] 11351 11352 *) Continuing patches for GeneralizedTime. Fix up certificate and CRL 11353 ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print. 11354 Removed the versions check from X509 routines when loading extensions: 11355 this allows certain broken certificates that don't set the version 11356 properly to be processed. 11357 [Steve Henson] 11358 11359 *) Deal with irritating shit to do with dependencies, in YAAHW (Yet Another 11360 Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which 11361 can still be regenerated with "make depend". 11362 [Ben Laurie] 11363 11364 *) Spelling mistake in C version of CAST-128. 11365 [Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>] 11366 11367 *) Changes to the error generation code. The perl script err-code.pl 11368 now reads in the old error codes and retains the old numbers, only 11369 adding new ones if necessary. It also only changes the .err files if new 11370 codes are added. The makefiles have been modified to only insert errors 11371 when needed (to avoid needlessly modifying header files). This is done 11372 by only inserting errors if the .err file is newer than the auto generated 11373 C file. To rebuild all the error codes from scratch (the old behaviour) 11374 either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl 11375 or delete all the .err files. 11376 [Steve Henson] 11377 11378 *) CAST-128 was incorrectly implemented for short keys. The C version has 11379 been fixed, but is untested. The assembler versions are also fixed, but 11380 new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing 11381 to regenerate it if needed. 11382 [Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun 11383 Hagino <itojun@kame.net>] 11384 11385 *) File was opened incorrectly in randfile.c. 11386 [Ulf M��ller <ulf@fitug.de>] 11387 11388 *) Beginning of support for GeneralizedTime. d2i, i2d, check and print 11389 functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or 11390 GeneralizedTime. ASN1_TIME is the proper type used in certificates et 11391 al: it's just almost always a UTCTime. Note this patch adds new error 11392 codes so do a "make errors" if there are problems. 11393 [Steve Henson] 11394 11395 *) Correct Linux 1 recognition in config. 11396 [Ulf M��ller <ulf@fitug.de>] 11397 11398 *) Remove pointless MD5 hash when using DSA keys in ca. 11399 [Anonymous <nobody@replay.com>] 11400 11401 *) Generate an error if given an empty string as a cert directory. Also 11402 generate an error if handed NULL (previously returned 0 to indicate an 11403 error, but didn't set one). 11404 [Ben Laurie, reported by Anonymous <nobody@replay.com>] 11405 11406 *) Add prototypes to SSL methods. Make SSL_write's buffer const, at last. 11407 [Ben Laurie] 11408 11409 *) Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct 11410 parameters. This was causing a warning which killed off the Win32 compile. 11411 [Steve Henson] 11412 11413 *) Remove C++ style comments from crypto/bn/bn_local.h. 11414 [Neil Costigan <neil.costigan@celocom.com>] 11415 11416 *) The function OBJ_txt2nid was broken. It was supposed to return a nid 11417 based on a text string, looking up short and long names and finally 11418 "dot" format. The "dot" format stuff didn't work. Added new function 11419 OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote 11420 OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the 11421 OID is not part of the table. 11422 [Steve Henson] 11423 11424 *) Add prototypes to X509 lookup/verify methods, fixing a bug in 11425 X509_LOOKUP_by_alias(). 11426 [Ben Laurie] 11427 11428 *) Sort openssl functions by name. 11429 [Ben Laurie] 11430 11431 *) Get the gendsa program working (hopefully) and add it to app list. Remove 11432 encryption from sample DSA keys (in case anyone is interested the password 11433 was "1234"). 11434 [Steve Henson] 11435 11436 *) Make _all_ *_free functions accept a NULL pointer. 11437 [Frans Heymans <fheymans@isaserver.be>] 11438 11439 *) If a DH key is generated in s3_srvr.c, don't blow it by trying to use 11440 NULL pointers. 11441 [Anonymous <nobody@replay.com>] 11442 11443 *) s_server should send the CAfile as acceptable CAs, not its own cert. 11444 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 11445 11446 *) Don't blow it for numeric -newkey arguments to apps/req. 11447 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 11448 11449 *) Temp key "for export" tests were wrong in s3_srvr.c. 11450 [Anonymous <nobody@replay.com>] 11451 11452 *) Add prototype for temp key callback functions 11453 SSL_CTX_set_tmp_{rsa,dh}_callback(). 11454 [Ben Laurie] 11455 11456 *) Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and 11457 DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey(). 11458 [Steve Henson] 11459 11460 *) X509_name_add_entry() freed the wrong thing after an error. 11461 [Arne Ansper <arne@ats.cyber.ee>] 11462 11463 *) rsa_eay.c would attempt to free a NULL context. 11464 [Arne Ansper <arne@ats.cyber.ee>] 11465 11466 *) BIO_s_socket() had a broken should_retry() on Windoze. 11467 [Arne Ansper <arne@ats.cyber.ee>] 11468 11469 *) BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH. 11470 [Arne Ansper <arne@ats.cyber.ee>] 11471 11472 *) Make sure the already existing X509_STORE->depth variable is initialized 11473 in X509_STORE_new(), but document the fact that this variable is still 11474 unused in the certificate verification process. 11475 [Ralf S. Engelschall] 11476 11477 *) Fix the various library and apps files to free up pkeys obtained from 11478 X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions. 11479 [Steve Henson] 11480 11481 *) Fix reference counting in X509_PUBKEY_get(). This makes 11482 demos/maurice/example2.c work, amongst others, probably. 11483 [Steve Henson and Ben Laurie] 11484 11485 *) First cut of a cleanup for apps/. First the `ssleay' program is now named 11486 `openssl' and second, the shortcut symlinks for the `openssl <command>' 11487 are no longer created. This way we have a single and consistent command 11488 line interface `openssl <command>', similar to `cvs <command>'. 11489 [Ralf S. Engelschall, Paul Sutton and Ben Laurie] 11490 11491 *) ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey 11492 BIT STRING wrapper always have zero unused bits. 11493 [Steve Henson] 11494 11495 *) Add CA.pl, perl version of CA.sh, add extended key usage OID. 11496 [Steve Henson] 11497 11498 *) Make the top-level INSTALL documentation easier to understand. 11499 [Paul Sutton] 11500 11501 *) Makefiles updated to exit if an error occurs in a sub-directory 11502 make (including if user presses ^C) [Paul Sutton] 11503 11504 *) Make Montgomery context stuff explicit in RSA data structure. 11505 [Ben Laurie] 11506 11507 *) Fix build order of pem and err to allow for generated pem.h. 11508 [Ben Laurie] 11509 11510 *) Fix renumbering bug in X509_NAME_delete_entry(). 11511 [Ben Laurie] 11512 11513 *) Enhanced the err-ins.pl script so it makes the error library number 11514 global and can add a library name. This is needed for external ASN1 and 11515 other error libraries. 11516 [Steve Henson] 11517 11518 *) Fixed sk_insert which never worked properly. 11519 [Steve Henson] 11520 11521 *) Fix ASN1 macros so they can handle indefinite length construted 11522 EXPLICIT tags. Some non standard certificates use these: they can now 11523 be read in. 11524 [Steve Henson] 11525 11526 *) Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc) 11527 into a single doc/ssleay.txt bundle. This way the information is still 11528 preserved but no longer messes up this directory. Now it's new room for 11529 the new set of documenation files. 11530 [Ralf S. Engelschall] 11531 11532 *) SETs were incorrectly DER encoded. This was a major pain, because they 11533 shared code with SEQUENCEs, which aren't coded the same. This means that 11534 almost everything to do with SETs or SEQUENCEs has either changed name or 11535 number of arguments. 11536 [Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>] 11537 11538 *) Fix test data to work with the above. 11539 [Ben Laurie] 11540 11541 *) Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but 11542 was already fixed by Eric for 0.9.1 it seems. 11543 [Ben Laurie - pointed out by Ulf M��ller <ulf@fitug.de>] 11544 11545 *) Autodetect FreeBSD3. 11546 [Ben Laurie] 11547 11548 *) Fix various bugs in Configure. This affects the following platforms: 11549 nextstep 11550 ncr-scde 11551 unixware-2.0 11552 unixware-2.0-pentium 11553 sco5-cc. 11554 [Ben Laurie] 11555 11556 *) Eliminate generated files from CVS. Reorder tests to regenerate files 11557 before they are needed. 11558 [Ben Laurie] 11559 11560 *) Generate Makefile.ssl from Makefile.org (to keep CVS happy). 11561 [Ben Laurie] 11562 11563 11564 Changes between 0.9.1b and 0.9.1c [23-Dec-1998] 11565 11566 *) Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and 11567 changed SSLeay to OpenSSL in version strings. 11568 [Ralf S. Engelschall] 11569 11570 *) Some fixups to the top-level documents. 11571 [Paul Sutton] 11572 11573 *) Fixed the nasty bug where rsaref.h was not found under compile-time 11574 because the symlink to include/ was missing. 11575 [Ralf S. Engelschall] 11576 11577 *) Incorporated the popular no-RSA/DSA-only patches 11578 which allow to compile a RSA-free SSLeay. 11579 [Andrew Cooke / Interrader Ldt., Ralf S. Engelschall] 11580 11581 *) Fixed nasty rehash problem under `make -f Makefile.ssl links' 11582 when "ssleay" is still not found. 11583 [Ralf S. Engelschall] 11584 11585 *) Added more platforms to Configure: Cray T3E, HPUX 11, 11586 [Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>] 11587 11588 *) Updated the README file. 11589 [Ralf S. Engelschall] 11590 11591 *) Added various .cvsignore files in the CVS repository subdirs 11592 to make a "cvs update" really silent. 11593 [Ralf S. Engelschall] 11594 11595 *) Recompiled the error-definition header files and added 11596 missing symbols to the Win32 linker tables. 11597 [Ralf S. Engelschall] 11598 11599 *) Cleaned up the top-level documents; 11600 o new files: CHANGES and LICENSE 11601 o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay 11602 o merged COPYRIGHT into LICENSE 11603 o removed obsolete TODO file 11604 o renamed MICROSOFT to INSTALL.W32 11605 [Ralf S. Engelschall] 11606 11607 *) Removed dummy files from the 0.9.1b source tree: 11608 crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi 11609 crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f 11610 crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f 11611 crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f 11612 util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f 11613 [Ralf S. Engelschall] 11614 11615 *) Added various platform portability fixes. 11616 [Mark J. Cox] 11617 11618 *) The Genesis of the OpenSSL rpject: 11619 We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A. 11620 Young and Tim J. Hudson created while they were working for C2Net until 11621 summer 1998. 11622 [The OpenSSL Project] 11623 11624 11625 Changes between 0.9.0b and 0.9.1b [not released] 11626 11627 *) Updated a few CA certificates under certs/ 11628 [Eric A. Young] 11629 11630 *) Changed some BIGNUM api stuff. 11631 [Eric A. Young] 11632 11633 *) Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD, 11634 DGUX x86, Linux Alpha, etc. 11635 [Eric A. Young] 11636 11637 *) New COMP library [crypto/comp/] for SSL Record Layer Compression: 11638 RLE (dummy implemented) and ZLIB (really implemented when ZLIB is 11639 available). 11640 [Eric A. Young] 11641 11642 *) Add -strparse option to asn1pars program which parses nested 11643 binary structures 11644 [Dr Stephen Henson <shenson@bigfoot.com>] 11645 11646 *) Added "oid_file" to ssleay.cnf for "ca" and "req" programs. 11647 [Eric A. Young] 11648 11649 *) DSA fix for "ca" program. 11650 [Eric A. Young] 11651 11652 *) Added "-genkey" option to "dsaparam" program. 11653 [Eric A. Young] 11654 11655 *) Added RIPE MD160 (rmd160) message digest. 11656 [Eric A. Young] 11657 11658 *) Added -a (all) option to "ssleay version" command. 11659 [Eric A. Young] 11660 11661 *) Added PLATFORM define which is the id given to Configure. 11662 [Eric A. Young] 11663 11664 *) Added MemCheck_XXXX functions to crypto/mem.c for memory checking. 11665 [Eric A. Young] 11666 11667 *) Extended the ASN.1 parser routines. 11668 [Eric A. Young] 11669 11670 *) Extended BIO routines to support REUSEADDR, seek, tell, etc. 11671 [Eric A. Young] 11672 11673 *) Added a BN_CTX to the BN library. 11674 [Eric A. Young] 11675 11676 *) Fixed the weak key values in DES library 11677 [Eric A. Young] 11678 11679 *) Changed API in EVP library for cipher aliases. 11680 [Eric A. Young] 11681 11682 *) Added support for RC2/64bit cipher. 11683 [Eric A. Young] 11684 11685 *) Converted the lhash library to the crypto/mem.c functions. 11686 [Eric A. Young] 11687 11688 *) Added more recognized ASN.1 object ids. 11689 [Eric A. Young] 11690 11691 *) Added more RSA padding checks for SSL/TLS. 11692 [Eric A. Young] 11693 11694 *) Added BIO proxy/filter functionality. 11695 [Eric A. Young] 11696 11697 *) Added extra_certs to SSL_CTX which can be used 11698 send extra CA certificates to the client in the CA cert chain sending 11699 process. It can be configured with SSL_CTX_add_extra_chain_cert(). 11700 [Eric A. Young] 11701 11702 *) Now Fortezza is denied in the authentication phase because 11703 this is key exchange mechanism is not supported by SSLeay at all. 11704 [Eric A. Young] 11705 11706 *) Additional PKCS1 checks. 11707 [Eric A. Young] 11708 11709 *) Support the string "TLSv1" for all TLS v1 ciphers. 11710 [Eric A. Young] 11711 11712 *) Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the 11713 ex_data index of the SSL context in the X509_STORE_CTX ex_data. 11714 [Eric A. Young] 11715 11716 *) Fixed a few memory leaks. 11717 [Eric A. Young] 11718 11719 *) Fixed various code and comment typos. 11720 [Eric A. Young] 11721 11722 *) A minor bug in ssl/s3_clnt.c where there would always be 4 0 11723 bytes sent in the client random. 11724 [Edward Bishop <ebishop@spyglass.com>] 11725 11726