CHANGES revision 306195
1 2 OpenSSL CHANGES 3 _______________ 4 5 Changes between 1.0.2h and 1.0.2i [22 Sep 2016] 6 7 *) OCSP Status Request extension unbounded memory growth 8 9 A malicious client can send an excessively large OCSP Status Request 10 extension. If that client continually requests renegotiation, sending a 11 large OCSP Status Request extension each time, then there will be unbounded 12 memory growth on the server. This will eventually lead to a Denial Of 13 Service attack through memory exhaustion. Servers with a default 14 configuration are vulnerable even if they do not support OCSP. Builds using 15 the "no-ocsp" build time option are not affected. 16 17 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 18 (CVE-2016-6304) 19 [Matt Caswell] 20 21 *) In order to mitigate the SWEET32 attack, the DES ciphers were moved from 22 HIGH to MEDIUM. 23 24 This issue was reported to OpenSSL Karthikeyan Bhargavan and Gaetan 25 Leurent (INRIA) 26 (CVE-2016-2183) 27 [Rich Salz] 28 29 *) OOB write in MDC2_Update() 30 31 An overflow can occur in MDC2_Update() either if called directly or 32 through the EVP_DigestUpdate() function using MDC2. If an attacker 33 is able to supply very large amounts of input data after a previous 34 call to EVP_EncryptUpdate() with a partial block then a length check 35 can overflow resulting in a heap corruption. 36 37 The amount of data needed is comparable to SIZE_MAX which is impractical 38 on most platforms. 39 40 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 41 (CVE-2016-6303) 42 [Stephen Henson] 43 44 *) Malformed SHA512 ticket DoS 45 46 If a server uses SHA512 for TLS session ticket HMAC it is vulnerable to a 47 DoS attack where a malformed ticket will result in an OOB read which will 48 ultimately crash. 49 50 The use of SHA512 in TLS session tickets is comparatively rare as it requires 51 a custom server callback and ticket lookup mechanism. 52 53 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 54 (CVE-2016-6302) 55 [Stephen Henson] 56 57 *) OOB write in BN_bn2dec() 58 59 The function BN_bn2dec() does not check the return value of BN_div_word(). 60 This can cause an OOB write if an application uses this function with an 61 overly large BIGNUM. This could be a problem if an overly large certificate 62 or CRL is printed out from an untrusted source. TLS is not affected because 63 record limits will reject an oversized certificate before it is parsed. 64 65 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 66 (CVE-2016-2182) 67 [Stephen Henson] 68 69 *) OOB read in TS_OBJ_print_bio() 70 71 The function TS_OBJ_print_bio() misuses OBJ_obj2txt(): the return value is 72 the total length the OID text representation would use and not the amount 73 of data written. This will result in OOB reads when large OIDs are 74 presented. 75 76 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 77 (CVE-2016-2180) 78 [Stephen Henson] 79 80 *) Pointer arithmetic undefined behaviour 81 82 Avoid some undefined pointer arithmetic 83 84 A common idiom in the codebase is to check limits in the following manner: 85 "p + len > limit" 86 87 Where "p" points to some malloc'd data of SIZE bytes and 88 limit == p + SIZE 89 90 "len" here could be from some externally supplied data (e.g. from a TLS 91 message). 92 93 The rules of C pointer arithmetic are such that "p + len" is only well 94 defined where len <= SIZE. Therefore the above idiom is actually 95 undefined behaviour. 96 97 For example this could cause problems if some malloc implementation 98 provides an address for "p" such that "p + len" actually overflows for 99 values of len that are too big and therefore p + len < limit. 100 101 This issue was reported to OpenSSL by Guido Vranken 102 (CVE-2016-2177) 103 [Matt Caswell] 104 105 *) Constant time flag not preserved in DSA signing 106 107 Operations in the DSA signing algorithm should run in constant time in 108 order to avoid side channel attacks. A flaw in the OpenSSL DSA 109 implementation means that a non-constant time codepath is followed for 110 certain operations. This has been demonstrated through a cache-timing 111 attack to be sufficient for an attacker to recover the private DSA key. 112 113 This issue was reported by C��sar Pereida (Aalto University), Billy Brumley 114 (Tampere University of Technology), and Yuval Yarom (The University of 115 Adelaide and NICTA). 116 (CVE-2016-2178) 117 [C��sar Pereida] 118 119 *) DTLS buffered message DoS 120 121 In a DTLS connection where handshake messages are delivered out-of-order 122 those messages that OpenSSL is not yet ready to process will be buffered 123 for later use. Under certain circumstances, a flaw in the logic means that 124 those messages do not get removed from the buffer even though the handshake 125 has been completed. An attacker could force up to approx. 15 messages to 126 remain in the buffer when they are no longer required. These messages will 127 be cleared when the DTLS connection is closed. The default maximum size for 128 a message is 100k. Therefore the attacker could force an additional 1500k 129 to be consumed per connection. By opening many simulataneous connections an 130 attacker could cause a DoS attack through memory exhaustion. 131 132 This issue was reported to OpenSSL by Quan Luo. 133 (CVE-2016-2179) 134 [Matt Caswell] 135 136 *) DTLS replay protection DoS 137 138 A flaw in the DTLS replay attack protection mechanism means that records 139 that arrive for future epochs update the replay protection "window" before 140 the MAC for the record has been validated. This could be exploited by an 141 attacker by sending a record for the next epoch (which does not have to 142 decrypt or have a valid MAC), with a very large sequence number. This means 143 that all subsequent legitimate packets are dropped causing a denial of 144 service for a specific DTLS connection. 145 146 This issue was reported to OpenSSL by the OCAP audit team. 147 (CVE-2016-2181) 148 [Matt Caswell] 149 150 *) Certificate message OOB reads 151 152 In OpenSSL 1.0.2 and earlier some missing message length checks can result 153 in OOB reads of up to 2 bytes beyond an allocated buffer. There is a 154 theoretical DoS risk but this has not been observed in practice on common 155 platforms. 156 157 The messages affected are client certificate, client certificate request 158 and server certificate. As a result the attack can only be performed 159 against a client or a server which enables client authentication. 160 161 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 162 (CVE-2016-6306) 163 [Stephen Henson] 164 165 Changes between 1.0.2g and 1.0.2h [3 May 2016] 166 167 *) Prevent padding oracle in AES-NI CBC MAC check 168 169 A MITM attacker can use a padding oracle attack to decrypt traffic 170 when the connection uses an AES CBC cipher and the server support 171 AES-NI. 172 173 This issue was introduced as part of the fix for Lucky 13 padding 174 attack (CVE-2013-0169). The padding check was rewritten to be in 175 constant time by making sure that always the same bytes are read and 176 compared against either the MAC or padding bytes. But it no longer 177 checked that there was enough data to have both the MAC and padding 178 bytes. 179 180 This issue was reported by Juraj Somorovsky using TLS-Attacker. 181 (CVE-2016-2107) 182 [Kurt Roeckx] 183 184 *) Fix EVP_EncodeUpdate overflow 185 186 An overflow can occur in the EVP_EncodeUpdate() function which is used for 187 Base64 encoding of binary data. If an attacker is able to supply very large 188 amounts of input data then a length check can overflow resulting in a heap 189 corruption. 190 191 Internally to OpenSSL the EVP_EncodeUpdate() function is primarly used by 192 the PEM_write_bio* family of functions. These are mainly used within the 193 OpenSSL command line applications, so any application which processes data 194 from an untrusted source and outputs it as a PEM file should be considered 195 vulnerable to this issue. User applications that call these APIs directly 196 with large amounts of untrusted data may also be vulnerable. 197 198 This issue was reported by Guido Vranken. 199 (CVE-2016-2105) 200 [Matt Caswell] 201 202 *) Fix EVP_EncryptUpdate overflow 203 204 An overflow can occur in the EVP_EncryptUpdate() function. If an attacker 205 is able to supply very large amounts of input data after a previous call to 206 EVP_EncryptUpdate() with a partial block then a length check can overflow 207 resulting in a heap corruption. Following an analysis of all OpenSSL 208 internal usage of the EVP_EncryptUpdate() function all usage is one of two 209 forms. The first form is where the EVP_EncryptUpdate() call is known to be 210 the first called function after an EVP_EncryptInit(), and therefore that 211 specific call must be safe. The second form is where the length passed to 212 EVP_EncryptUpdate() can be seen from the code to be some small value and 213 therefore there is no possibility of an overflow. Since all instances are 214 one of these two forms, it is believed that there can be no overflows in 215 internal code due to this problem. It should be noted that 216 EVP_DecryptUpdate() can call EVP_EncryptUpdate() in certain code paths. 217 Also EVP_CipherUpdate() is a synonym for EVP_EncryptUpdate(). All instances 218 of these calls have also been analysed too and it is believed there are no 219 instances in internal usage where an overflow could occur. 220 221 This issue was reported by Guido Vranken. 222 (CVE-2016-2106) 223 [Matt Caswell] 224 225 *) Prevent ASN.1 BIO excessive memory allocation 226 227 When ASN.1 data is read from a BIO using functions such as d2i_CMS_bio() 228 a short invalid encoding can casuse allocation of large amounts of memory 229 potentially consuming excessive resources or exhausting memory. 230 231 Any application parsing untrusted data through d2i BIO functions is 232 affected. The memory based functions such as d2i_X509() are *not* affected. 233 Since the memory based functions are used by the TLS library, TLS 234 applications are not affected. 235 236 This issue was reported by Brian Carpenter. 237 (CVE-2016-2109) 238 [Stephen Henson] 239 240 *) EBCDIC overread 241 242 ASN1 Strings that are over 1024 bytes can cause an overread in applications 243 using the X509_NAME_oneline() function on EBCDIC systems. This could result 244 in arbitrary stack data being returned in the buffer. 245 246 This issue was reported by Guido Vranken. 247 (CVE-2016-2176) 248 [Matt Caswell] 249 250 *) Modify behavior of ALPN to invoke callback after SNI/servername 251 callback, such that updates to the SSL_CTX affect ALPN. 252 [Todd Short] 253 254 *) Remove LOW from the DEFAULT cipher list. This removes singles DES from the 255 default. 256 [Kurt Roeckx] 257 258 *) Only remove the SSLv2 methods with the no-ssl2-method option. When the 259 methods are enabled and ssl2 is disabled the methods return NULL. 260 [Kurt Roeckx] 261 262 Changes between 1.0.2f and 1.0.2g [1 Mar 2016] 263 264 * Disable weak ciphers in SSLv3 and up in default builds of OpenSSL. 265 Builds that are not configured with "enable-weak-ssl-ciphers" will not 266 provide any "EXPORT" or "LOW" strength ciphers. 267 [Viktor Dukhovni] 268 269 * Disable SSLv2 default build, default negotiation and weak ciphers. SSLv2 270 is by default disabled at build-time. Builds that are not configured with 271 "enable-ssl2" will not support SSLv2. Even if "enable-ssl2" is used, 272 users who want to negotiate SSLv2 via the version-flexible SSLv23_method() 273 will need to explicitly call either of: 274 275 SSL_CTX_clear_options(ctx, SSL_OP_NO_SSLv2); 276 or 277 SSL_clear_options(ssl, SSL_OP_NO_SSLv2); 278 279 as appropriate. Even if either of those is used, or the application 280 explicitly uses the version-specific SSLv2_method() or its client and 281 server variants, SSLv2 ciphers vulnerable to exhaustive search key 282 recovery have been removed. Specifically, the SSLv2 40-bit EXPORT 283 ciphers, and SSLv2 56-bit DES are no longer available. 284 (CVE-2016-0800) 285 [Viktor Dukhovni] 286 287 *) Fix a double-free in DSA code 288 289 A double free bug was discovered when OpenSSL parses malformed DSA private 290 keys and could lead to a DoS attack or memory corruption for applications 291 that receive DSA private keys from untrusted sources. This scenario is 292 considered rare. 293 294 This issue was reported to OpenSSL by Adam Langley(Google/BoringSSL) using 295 libFuzzer. 296 (CVE-2016-0705) 297 [Stephen Henson] 298 299 *) Disable SRP fake user seed to address a server memory leak. 300 301 Add a new method SRP_VBASE_get1_by_user that handles the seed properly. 302 303 SRP_VBASE_get_by_user had inconsistent memory management behaviour. 304 In order to fix an unavoidable memory leak, SRP_VBASE_get_by_user 305 was changed to ignore the "fake user" SRP seed, even if the seed 306 is configured. 307 308 Users should use SRP_VBASE_get1_by_user instead. Note that in 309 SRP_VBASE_get1_by_user, caller must free the returned value. Note 310 also that even though configuring the SRP seed attempts to hide 311 invalid usernames by continuing the handshake with fake 312 credentials, this behaviour is not constant time and no strong 313 guarantees are made that the handshake is indistinguishable from 314 that of a valid user. 315 (CVE-2016-0798) 316 [Emilia K��sper] 317 318 *) Fix BN_hex2bn/BN_dec2bn NULL pointer deref/heap corruption 319 320 In the BN_hex2bn function the number of hex digits is calculated using an 321 int value |i|. Later |bn_expand| is called with a value of |i * 4|. For 322 large values of |i| this can result in |bn_expand| not allocating any 323 memory because |i * 4| is negative. This can leave the internal BIGNUM data 324 field as NULL leading to a subsequent NULL ptr deref. For very large values 325 of |i|, the calculation |i * 4| could be a positive value smaller than |i|. 326 In this case memory is allocated to the internal BIGNUM data field, but it 327 is insufficiently sized leading to heap corruption. A similar issue exists 328 in BN_dec2bn. This could have security consequences if BN_hex2bn/BN_dec2bn 329 is ever called by user applications with very large untrusted hex/dec data. 330 This is anticipated to be a rare occurrence. 331 332 All OpenSSL internal usage of these functions use data that is not expected 333 to be untrusted, e.g. config file data or application command line 334 arguments. If user developed applications generate config file data based 335 on untrusted data then it is possible that this could also lead to security 336 consequences. This is also anticipated to be rare. 337 338 This issue was reported to OpenSSL by Guido Vranken. 339 (CVE-2016-0797) 340 [Matt Caswell] 341 342 *) Fix memory issues in BIO_*printf functions 343 344 The internal |fmtstr| function used in processing a "%s" format string in 345 the BIO_*printf functions could overflow while calculating the length of a 346 string and cause an OOB read when printing very long strings. 347 348 Additionally the internal |doapr_outch| function can attempt to write to an 349 OOB memory location (at an offset from the NULL pointer) in the event of a 350 memory allocation failure. In 1.0.2 and below this could be caused where 351 the size of a buffer to be allocated is greater than INT_MAX. E.g. this 352 could be in processing a very long "%s" format string. Memory leaks can 353 also occur. 354 355 The first issue may mask the second issue dependent on compiler behaviour. 356 These problems could enable attacks where large amounts of untrusted data 357 is passed to the BIO_*printf functions. If applications use these functions 358 in this way then they could be vulnerable. OpenSSL itself uses these 359 functions when printing out human-readable dumps of ASN.1 data. Therefore 360 applications that print this data could be vulnerable if the data is from 361 untrusted sources. OpenSSL command line applications could also be 362 vulnerable where they print out ASN.1 data, or if untrusted data is passed 363 as command line arguments. 364 365 Libssl is not considered directly vulnerable. Additionally certificates etc 366 received via remote connections via libssl are also unlikely to be able to 367 trigger these issues because of message size limits enforced within libssl. 368 369 This issue was reported to OpenSSL Guido Vranken. 370 (CVE-2016-0799) 371 [Matt Caswell] 372 373 *) Side channel attack on modular exponentiation 374 375 A side-channel attack was found which makes use of cache-bank conflicts on 376 the Intel Sandy-Bridge microarchitecture which could lead to the recovery 377 of RSA keys. The ability to exploit this issue is limited as it relies on 378 an attacker who has control of code in a thread running on the same 379 hyper-threaded core as the victim thread which is performing decryptions. 380 381 This issue was reported to OpenSSL by Yuval Yarom, The University of 382 Adelaide and NICTA, Daniel Genkin, Technion and Tel Aviv University, and 383 Nadia Heninger, University of Pennsylvania with more information at 384 http://cachebleed.info. 385 (CVE-2016-0702) 386 [Andy Polyakov] 387 388 *) Change the req app to generate a 2048-bit RSA/DSA key by default, 389 if no keysize is specified with default_bits. This fixes an 390 omission in an earlier change that changed all RSA/DSA key generation 391 apps to use 2048 bits by default. 392 [Emilia K��sper] 393 394 Changes between 1.0.2e and 1.0.2f [28 Jan 2016] 395 396 *) DH small subgroups 397 398 Historically OpenSSL only ever generated DH parameters based on "safe" 399 primes. More recently (in version 1.0.2) support was provided for 400 generating X9.42 style parameter files such as those required for RFC 5114 401 support. The primes used in such files may not be "safe". Where an 402 application is using DH configured with parameters based on primes that are 403 not "safe" then an attacker could use this fact to find a peer's private 404 DH exponent. This attack requires that the attacker complete multiple 405 handshakes in which the peer uses the same private DH exponent. For example 406 this could be used to discover a TLS server's private DH exponent if it's 407 reusing the private DH exponent or it's using a static DH ciphersuite. 408 409 OpenSSL provides the option SSL_OP_SINGLE_DH_USE for ephemeral DH (DHE) in 410 TLS. It is not on by default. If the option is not set then the server 411 reuses the same private DH exponent for the life of the server process and 412 would be vulnerable to this attack. It is believed that many popular 413 applications do set this option and would therefore not be at risk. 414 415 The fix for this issue adds an additional check where a "q" parameter is 416 available (as is the case in X9.42 based parameters). This detects the 417 only known attack, and is the only possible defense for static DH 418 ciphersuites. This could have some performance impact. 419 420 Additionally the SSL_OP_SINGLE_DH_USE option has been switched on by 421 default and cannot be disabled. This could have some performance impact. 422 423 This issue was reported to OpenSSL by Antonio Sanso (Adobe). 424 (CVE-2016-0701) 425 [Matt Caswell] 426 427 *) SSLv2 doesn't block disabled ciphers 428 429 A malicious client can negotiate SSLv2 ciphers that have been disabled on 430 the server and complete SSLv2 handshakes even if all SSLv2 ciphers have 431 been disabled, provided that the SSLv2 protocol was not also disabled via 432 SSL_OP_NO_SSLv2. 433 434 This issue was reported to OpenSSL on 26th December 2015 by Nimrod Aviram 435 and Sebastian Schinzel. 436 (CVE-2015-3197) 437 [Viktor Dukhovni] 438 439 *) Reject DH handshakes with parameters shorter than 1024 bits. 440 [Kurt Roeckx] 441 442 Changes between 1.0.2d and 1.0.2e [3 Dec 2015] 443 444 *) BN_mod_exp may produce incorrect results on x86_64 445 446 There is a carry propagating bug in the x86_64 Montgomery squaring 447 procedure. No EC algorithms are affected. Analysis suggests that attacks 448 against RSA and DSA as a result of this defect would be very difficult to 449 perform and are not believed likely. Attacks against DH are considered just 450 feasible (although very difficult) because most of the work necessary to 451 deduce information about a private key may be performed offline. The amount 452 of resources required for such an attack would be very significant and 453 likely only accessible to a limited number of attackers. An attacker would 454 additionally need online access to an unpatched system using the target 455 private key in a scenario with persistent DH parameters and a private 456 key that is shared between multiple clients. For example this can occur by 457 default in OpenSSL DHE based SSL/TLS ciphersuites. 458 459 This issue was reported to OpenSSL by Hanno B��ck. 460 (CVE-2015-3193) 461 [Andy Polyakov] 462 463 *) Certificate verify crash with missing PSS parameter 464 465 The signature verification routines will crash with a NULL pointer 466 dereference if presented with an ASN.1 signature using the RSA PSS 467 algorithm and absent mask generation function parameter. Since these 468 routines are used to verify certificate signature algorithms this can be 469 used to crash any certificate verification operation and exploited in a 470 DoS attack. Any application which performs certificate verification is 471 vulnerable including OpenSSL clients and servers which enable client 472 authentication. 473 474 This issue was reported to OpenSSL by Lo��c Jonas Etienne (Qnective AG). 475 (CVE-2015-3194) 476 [Stephen Henson] 477 478 *) X509_ATTRIBUTE memory leak 479 480 When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak 481 memory. This structure is used by the PKCS#7 and CMS routines so any 482 application which reads PKCS#7 or CMS data from untrusted sources is 483 affected. SSL/TLS is not affected. 484 485 This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using 486 libFuzzer. 487 (CVE-2015-3195) 488 [Stephen Henson] 489 490 *) Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs. 491 This changes the decoding behaviour for some invalid messages, 492 though the change is mostly in the more lenient direction, and 493 legacy behaviour is preserved as much as possible. 494 [Emilia K��sper] 495 496 *) In DSA_generate_parameters_ex, if the provided seed is too short, 497 use a random seed, as already documented. 498 [Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>] 499 500 Changes between 1.0.2c and 1.0.2d [9 Jul 2015] 501 502 *) Alternate chains certificate forgery 503 504 During certificate verfification, OpenSSL will attempt to find an 505 alternative certificate chain if the first attempt to build such a chain 506 fails. An error in the implementation of this logic can mean that an 507 attacker could cause certain checks on untrusted certificates to be 508 bypassed, such as the CA flag, enabling them to use a valid leaf 509 certificate to act as a CA and "issue" an invalid certificate. 510 511 This issue was reported to OpenSSL by Adam Langley/David Benjamin 512 (Google/BoringSSL). 513 (CVE-2015-1793) 514 [Matt Caswell] 515 516 *) Race condition handling PSK identify hint 517 518 If PSK identity hints are received by a multi-threaded client then 519 the values are wrongly updated in the parent SSL_CTX structure. This can 520 result in a race condition potentially leading to a double free of the 521 identify hint data. 522 (CVE-2015-3196) 523 [Stephen Henson] 524 525 Changes between 1.0.2b and 1.0.2c [12 Jun 2015] 526 527 *) Fix HMAC ABI incompatibility. The previous version introduced an ABI 528 incompatibility in the handling of HMAC. The previous ABI has now been 529 restored. 530 531 Changes between 1.0.2a and 1.0.2b [11 Jun 2015] 532 533 *) Malformed ECParameters causes infinite loop 534 535 When processing an ECParameters structure OpenSSL enters an infinite loop 536 if the curve specified is over a specially malformed binary polynomial 537 field. 538 539 This can be used to perform denial of service against any 540 system which processes public keys, certificate requests or 541 certificates. This includes TLS clients and TLS servers with 542 client authentication enabled. 543 544 This issue was reported to OpenSSL by Joseph Barr-Pixton. 545 (CVE-2015-1788) 546 [Andy Polyakov] 547 548 *) Exploitable out-of-bounds read in X509_cmp_time 549 550 X509_cmp_time does not properly check the length of the ASN1_TIME 551 string and can read a few bytes out of bounds. In addition, 552 X509_cmp_time accepts an arbitrary number of fractional seconds in the 553 time string. 554 555 An attacker can use this to craft malformed certificates and CRLs of 556 various sizes and potentially cause a segmentation fault, resulting in 557 a DoS on applications that verify certificates or CRLs. TLS clients 558 that verify CRLs are affected. TLS clients and servers with client 559 authentication enabled may be affected if they use custom verification 560 callbacks. 561 562 This issue was reported to OpenSSL by Robert Swiecki (Google), and 563 independently by Hanno B��ck. 564 (CVE-2015-1789) 565 [Emilia K��sper] 566 567 *) PKCS7 crash with missing EnvelopedContent 568 569 The PKCS#7 parsing code does not handle missing inner EncryptedContent 570 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs 571 with missing content and trigger a NULL pointer dereference on parsing. 572 573 Applications that decrypt PKCS#7 data or otherwise parse PKCS#7 574 structures from untrusted sources are affected. OpenSSL clients and 575 servers are not affected. 576 577 This issue was reported to OpenSSL by Michal Zalewski (Google). 578 (CVE-2015-1790) 579 [Emilia K��sper] 580 581 *) CMS verify infinite loop with unknown hash function 582 583 When verifying a signedData message the CMS code can enter an infinite loop 584 if presented with an unknown hash function OID. This can be used to perform 585 denial of service against any system which verifies signedData messages using 586 the CMS code. 587 This issue was reported to OpenSSL by Johannes Bauer. 588 (CVE-2015-1792) 589 [Stephen Henson] 590 591 *) Race condition handling NewSessionTicket 592 593 If a NewSessionTicket is received by a multi-threaded client when attempting to 594 reuse a previous ticket then a race condition can occur potentially leading to 595 a double free of the ticket data. 596 (CVE-2015-1791) 597 [Matt Caswell] 598 599 *) Removed support for the two export grade static DH ciphersuites 600 EXP-DH-RSA-DES-CBC-SHA and EXP-DH-DSS-DES-CBC-SHA. These two ciphersuites 601 were newly added (along with a number of other static DH ciphersuites) to 602 1.0.2. However the two export ones have *never* worked since they were 603 introduced. It seems strange in any case to be adding new export 604 ciphersuites, and given "logjam" it also does not seem correct to fix them. 605 [Matt Caswell] 606 607 *) Only support 256-bit or stronger elliptic curves with the 608 'ecdh_auto' setting (server) or by default (client). Of supported 609 curves, prefer P-256 (both). 610 [Emilia Kasper] 611 612 *) Reject DH handshakes with parameters shorter than 768 bits. 613 [Kurt Roeckx and Emilia Kasper] 614 615 Changes between 1.0.2 and 1.0.2a [19 Mar 2015] 616 617 *) ClientHello sigalgs DoS fix 618 619 If a client connects to an OpenSSL 1.0.2 server and renegotiates with an 620 invalid signature algorithms extension a NULL pointer dereference will 621 occur. This can be exploited in a DoS attack against the server. 622 623 This issue was was reported to OpenSSL by David Ramos of Stanford 624 University. 625 (CVE-2015-0291) 626 [Stephen Henson and Matt Caswell] 627 628 *) Multiblock corrupted pointer fix 629 630 OpenSSL 1.0.2 introduced the "multiblock" performance improvement. This 631 feature only applies on 64 bit x86 architecture platforms that support AES 632 NI instructions. A defect in the implementation of "multiblock" can cause 633 OpenSSL's internal write buffer to become incorrectly set to NULL when 634 using non-blocking IO. Typically, when the user application is using a 635 socket BIO for writing, this will only result in a failed connection. 636 However if some other BIO is used then it is likely that a segmentation 637 fault will be triggered, thus enabling a potential DoS attack. 638 639 This issue was reported to OpenSSL by Daniel Danner and Rainer Mueller. 640 (CVE-2015-0290) 641 [Matt Caswell] 642 643 *) Segmentation fault in DTLSv1_listen fix 644 645 The DTLSv1_listen function is intended to be stateless and processes the 646 initial ClientHello from many peers. It is common for user code to loop 647 over the call to DTLSv1_listen until a valid ClientHello is received with 648 an associated cookie. A defect in the implementation of DTLSv1_listen means 649 that state is preserved in the SSL object from one invocation to the next 650 that can lead to a segmentation fault. Errors processing the initial 651 ClientHello can trigger this scenario. An example of such an error could be 652 that a DTLS1.0 only client is attempting to connect to a DTLS1.2 only 653 server. 654 655 This issue was reported to OpenSSL by Per Allansson. 656 (CVE-2015-0207) 657 [Matt Caswell] 658 659 *) Segmentation fault in ASN1_TYPE_cmp fix 660 661 The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is 662 made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check 663 certificate signature algorithm consistency this can be used to crash any 664 certificate verification operation and exploited in a DoS attack. Any 665 application which performs certificate verification is vulnerable including 666 OpenSSL clients and servers which enable client authentication. 667 (CVE-2015-0286) 668 [Stephen Henson] 669 670 *) Segmentation fault for invalid PSS parameters fix 671 672 The signature verification routines will crash with a NULL pointer 673 dereference if presented with an ASN.1 signature using the RSA PSS 674 algorithm and invalid parameters. Since these routines are used to verify 675 certificate signature algorithms this can be used to crash any 676 certificate verification operation and exploited in a DoS attack. Any 677 application which performs certificate verification is vulnerable including 678 OpenSSL clients and servers which enable client authentication. 679 680 This issue was was reported to OpenSSL by Brian Carpenter. 681 (CVE-2015-0208) 682 [Stephen Henson] 683 684 *) ASN.1 structure reuse memory corruption fix 685 686 Reusing a structure in ASN.1 parsing may allow an attacker to cause 687 memory corruption via an invalid write. Such reuse is and has been 688 strongly discouraged and is believed to be rare. 689 690 Applications that parse structures containing CHOICE or ANY DEFINED BY 691 components may be affected. Certificate parsing (d2i_X509 and related 692 functions) are however not affected. OpenSSL clients and servers are 693 not affected. 694 (CVE-2015-0287) 695 [Stephen Henson] 696 697 *) PKCS7 NULL pointer dereferences fix 698 699 The PKCS#7 parsing code does not handle missing outer ContentInfo 700 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with 701 missing content and trigger a NULL pointer dereference on parsing. 702 703 Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or 704 otherwise parse PKCS#7 structures from untrusted sources are 705 affected. OpenSSL clients and servers are not affected. 706 707 This issue was reported to OpenSSL by Michal Zalewski (Google). 708 (CVE-2015-0289) 709 [Emilia K��sper] 710 711 *) DoS via reachable assert in SSLv2 servers fix 712 713 A malicious client can trigger an OPENSSL_assert (i.e., an abort) in 714 servers that both support SSLv2 and enable export cipher suites by sending 715 a specially crafted SSLv2 CLIENT-MASTER-KEY message. 716 717 This issue was discovered by Sean Burford (Google) and Emilia K��sper 718 (OpenSSL development team). 719 (CVE-2015-0293) 720 [Emilia K��sper] 721 722 *) Empty CKE with client auth and DHE fix 723 724 If client auth is used then a server can seg fault in the event of a DHE 725 ciphersuite being selected and a zero length ClientKeyExchange message 726 being sent by the client. This could be exploited in a DoS attack. 727 (CVE-2015-1787) 728 [Matt Caswell] 729 730 *) Handshake with unseeded PRNG fix 731 732 Under certain conditions an OpenSSL 1.0.2 client can complete a handshake 733 with an unseeded PRNG. The conditions are: 734 - The client is on a platform where the PRNG has not been seeded 735 automatically, and the user has not seeded manually 736 - A protocol specific client method version has been used (i.e. not 737 SSL_client_methodv23) 738 - A ciphersuite is used that does not require additional random data from 739 the PRNG beyond the initial ClientHello client random (e.g. PSK-RC4-SHA). 740 741 If the handshake succeeds then the client random that has been used will 742 have been generated from a PRNG with insufficient entropy and therefore the 743 output may be predictable. 744 745 For example using the following command with an unseeded openssl will 746 succeed on an unpatched platform: 747 748 openssl s_client -psk 1a2b3c4d -tls1_2 -cipher PSK-RC4-SHA 749 (CVE-2015-0285) 750 [Matt Caswell] 751 752 *) Use After Free following d2i_ECPrivatekey error fix 753 754 A malformed EC private key file consumed via the d2i_ECPrivateKey function 755 could cause a use after free condition. This, in turn, could cause a double 756 free in several private key parsing functions (such as d2i_PrivateKey 757 or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption 758 for applications that receive EC private keys from untrusted 759 sources. This scenario is considered rare. 760 761 This issue was discovered by the BoringSSL project and fixed in their 762 commit 517073cd4b. 763 (CVE-2015-0209) 764 [Matt Caswell] 765 766 *) X509_to_X509_REQ NULL pointer deref fix 767 768 The function X509_to_X509_REQ will crash with a NULL pointer dereference if 769 the certificate key is invalid. This function is rarely used in practice. 770 771 This issue was discovered by Brian Carpenter. 772 (CVE-2015-0288) 773 [Stephen Henson] 774 775 *) Removed the export ciphers from the DEFAULT ciphers 776 [Kurt Roeckx] 777 778 Changes between 1.0.1l and 1.0.2 [22 Jan 2015] 779 780 *) Change RSA and DH/DSA key generation apps to generate 2048-bit 781 keys by default. 782 [Kurt Roeckx] 783 784 *) Facilitate "universal" ARM builds targeting range of ARM ISAs, e.g. 785 ARMv5 through ARMv8, as opposite to "locking" it to single one. 786 So far those who have to target multiple plaforms would compromise 787 and argue that binary targeting say ARMv5 would still execute on 788 ARMv8. "Universal" build resolves this compromise by providing 789 near-optimal performance even on newer platforms. 790 [Andy Polyakov] 791 792 *) Accelerated NIST P-256 elliptic curve implementation for x86_64 793 (other platforms pending). 794 [Shay Gueron & Vlad Krasnov (Intel Corp), Andy Polyakov] 795 796 *) Add support for the SignedCertificateTimestampList certificate and 797 OCSP response extensions from RFC6962. 798 [Rob Stradling] 799 800 *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.) 801 for corner cases. (Certain input points at infinity could lead to 802 bogus results, with non-infinity inputs mapped to infinity too.) 803 [Bodo Moeller] 804 805 *) Initial support for PowerISA 2.0.7, first implemented in POWER8. 806 This covers AES, SHA256/512 and GHASH. "Initial" means that most 807 common cases are optimized and there still is room for further 808 improvements. Vector Permutation AES for Altivec is also added. 809 [Andy Polyakov] 810 811 *) Add support for little-endian ppc64 Linux target. 812 [Marcelo Cerri (IBM)] 813 814 *) Initial support for AMRv8 ISA crypto extensions. This covers AES, 815 SHA1, SHA256 and GHASH. "Initial" means that most common cases 816 are optimized and there still is room for further improvements. 817 Both 32- and 64-bit modes are supported. 818 [Andy Polyakov, Ard Biesheuvel (Linaro)] 819 820 *) Improved ARMv7 NEON support. 821 [Andy Polyakov] 822 823 *) Support for SPARC Architecture 2011 crypto extensions, first 824 implemented in SPARC T4. This covers AES, DES, Camellia, SHA1, 825 SHA256/512, MD5, GHASH and modular exponentiation. 826 [Andy Polyakov, David Miller] 827 828 *) Accelerated modular exponentiation for Intel processors, a.k.a. 829 RSAZ. 830 [Shay Gueron & Vlad Krasnov (Intel Corp)] 831 832 *) Support for new and upcoming Intel processors, including AVX2, 833 BMI and SHA ISA extensions. This includes additional "stitched" 834 implementations, AESNI-SHA256 and GCM, and multi-buffer support 835 for TLS encrypt. 836 837 This work was sponsored by Intel Corp. 838 [Andy Polyakov] 839 840 *) Support for DTLS 1.2. This adds two sets of DTLS methods: DTLS_*_method() 841 supports both DTLS 1.2 and 1.0 and should use whatever version the peer 842 supports and DTLSv1_2_*_method() which supports DTLS 1.2 only. 843 [Steve Henson] 844 845 *) Use algorithm specific chains in SSL_CTX_use_certificate_chain_file(): 846 this fixes a limiation in previous versions of OpenSSL. 847 [Steve Henson] 848 849 *) Extended RSA OAEP support via EVP_PKEY API. Options to specify digest, 850 MGF1 digest and OAEP label. 851 [Steve Henson] 852 853 *) Add EVP support for key wrapping algorithms, to avoid problems with 854 existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in 855 the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap 856 algorithms and include tests cases. 857 [Steve Henson] 858 859 *) Add functions to allocate and set the fields of an ECDSA_METHOD 860 structure. 861 [Douglas E. Engert, Steve Henson] 862 863 *) New functions OPENSSL_gmtime_diff and ASN1_TIME_diff to find the 864 difference in days and seconds between two tm or ASN1_TIME structures. 865 [Steve Henson] 866 867 *) Add -rev test option to s_server to just reverse order of characters 868 received by client and send back to server. Also prints an abbreviated 869 summary of the connection parameters. 870 [Steve Henson] 871 872 *) New option -brief for s_client and s_server to print out a brief summary 873 of connection parameters. 874 [Steve Henson] 875 876 *) Add callbacks for arbitrary TLS extensions. 877 [Trevor Perrin <trevp@trevp.net> and Ben Laurie] 878 879 *) New option -crl_download in several openssl utilities to download CRLs 880 from CRLDP extension in certificates. 881 [Steve Henson] 882 883 *) New options -CRL and -CRLform for s_client and s_server for CRLs. 884 [Steve Henson] 885 886 *) New function X509_CRL_diff to generate a delta CRL from the difference 887 of two full CRLs. Add support to "crl" utility. 888 [Steve Henson] 889 890 *) New functions to set lookup_crls function and to retrieve 891 X509_STORE from X509_STORE_CTX. 892 [Steve Henson] 893 894 *) Print out deprecated issuer and subject unique ID fields in 895 certificates. 896 [Steve Henson] 897 898 *) Extend OCSP I/O functions so they can be used for simple general purpose 899 HTTP as well as OCSP. New wrapper function which can be used to download 900 CRLs using the OCSP API. 901 [Steve Henson] 902 903 *) Delegate command line handling in s_client/s_server to SSL_CONF APIs. 904 [Steve Henson] 905 906 *) SSL_CONF* functions. These provide a common framework for application 907 configuration using configuration files or command lines. 908 [Steve Henson] 909 910 *) SSL/TLS tracing code. This parses out SSL/TLS records using the 911 message callback and prints the results. Needs compile time option 912 "enable-ssl-trace". New options to s_client and s_server to enable 913 tracing. 914 [Steve Henson] 915 916 *) New ctrl and macro to retrieve supported points extensions. 917 Print out extension in s_server and s_client. 918 [Steve Henson] 919 920 *) New functions to retrieve certificate signature and signature 921 OID NID. 922 [Steve Henson] 923 924 *) Add functions to retrieve and manipulate the raw cipherlist sent by a 925 client to OpenSSL. 926 [Steve Henson] 927 928 *) New Suite B modes for TLS code. These use and enforce the requirements 929 of RFC6460: restrict ciphersuites, only permit Suite B algorithms and 930 only use Suite B curves. The Suite B modes can be set by using the 931 strings "SUITEB128", "SUITEB192" or "SUITEB128ONLY" for the cipherstring. 932 [Steve Henson] 933 934 *) New chain verification flags for Suite B levels of security. Check 935 algorithms are acceptable when flags are set in X509_verify_cert. 936 [Steve Henson] 937 938 *) Make tls1_check_chain return a set of flags indicating checks passed 939 by a certificate chain. Add additional tests to handle client 940 certificates: checks for matching certificate type and issuer name 941 comparison. 942 [Steve Henson] 943 944 *) If an attempt is made to use a signature algorithm not in the peer 945 preference list abort the handshake. If client has no suitable 946 signature algorithms in response to a certificate request do not 947 use the certificate. 948 [Steve Henson] 949 950 *) If server EC tmp key is not in client preference list abort handshake. 951 [Steve Henson] 952 953 *) Add support for certificate stores in CERT structure. This makes it 954 possible to have different stores per SSL structure or one store in 955 the parent SSL_CTX. Include distint stores for certificate chain 956 verification and chain building. New ctrl SSL_CTRL_BUILD_CERT_CHAIN 957 to build and store a certificate chain in CERT structure: returing 958 an error if the chain cannot be built: this will allow applications 959 to test if a chain is correctly configured. 960 961 Note: if the CERT based stores are not set then the parent SSL_CTX 962 store is used to retain compatibility with existing behaviour. 963 964 [Steve Henson] 965 966 *) New function ssl_set_client_disabled to set a ciphersuite disabled 967 mask based on the current session, check mask when sending client 968 hello and checking the requested ciphersuite. 969 [Steve Henson] 970 971 *) New ctrls to retrieve and set certificate types in a certificate 972 request message. Print out received values in s_client. If certificate 973 types is not set with custom values set sensible values based on 974 supported signature algorithms. 975 [Steve Henson] 976 977 *) Support for distinct client and server supported signature algorithms. 978 [Steve Henson] 979 980 *) Add certificate callback. If set this is called whenever a certificate 981 is required by client or server. An application can decide which 982 certificate chain to present based on arbitrary criteria: for example 983 supported signature algorithms. Add very simple example to s_server. 984 This fixes many of the problems and restrictions of the existing client 985 certificate callback: for example you can now clear an existing 986 certificate and specify the whole chain. 987 [Steve Henson] 988 989 *) Add new "valid_flags" field to CERT_PKEY structure which determines what 990 the certificate can be used for (if anything). Set valid_flags field 991 in new tls1_check_chain function. Simplify ssl_set_cert_masks which used 992 to have similar checks in it. 993 994 Add new "cert_flags" field to CERT structure and include a "strict mode". 995 This enforces some TLS certificate requirements (such as only permitting 996 certificate signature algorithms contained in the supported algorithms 997 extension) which some implementations ignore: this option should be used 998 with caution as it could cause interoperability issues. 999 [Steve Henson] 1000 1001 *) Update and tidy signature algorithm extension processing. Work out 1002 shared signature algorithms based on preferences and peer algorithms 1003 and print them out in s_client and s_server. Abort handshake if no 1004 shared signature algorithms. 1005 [Steve Henson] 1006 1007 *) Add new functions to allow customised supported signature algorithms 1008 for SSL and SSL_CTX structures. Add options to s_client and s_server 1009 to support them. 1010 [Steve Henson] 1011 1012 *) New function SSL_certs_clear() to delete all references to certificates 1013 from an SSL structure. Before this once a certificate had been added 1014 it couldn't be removed. 1015 [Steve Henson] 1016 1017 *) Integrate hostname, email address and IP address checking with certificate 1018 verification. New verify options supporting checking in opensl utility. 1019 [Steve Henson] 1020 1021 *) Fixes and wildcard matching support to hostname and email checking 1022 functions. Add manual page. 1023 [Florian Weimer (Red Hat Product Security Team)] 1024 1025 *) New functions to check a hostname email or IP address against a 1026 certificate. Add options x509 utility to print results of checks against 1027 a certificate. 1028 [Steve Henson] 1029 1030 *) Fix OCSP checking. 1031 [Rob Stradling <rob.stradling@comodo.com> and Ben Laurie] 1032 1033 *) Initial experimental support for explicitly trusted non-root CAs. 1034 OpenSSL still tries to build a complete chain to a root but if an 1035 intermediate CA has a trust setting included that is used. The first 1036 setting is used: whether to trust (e.g., -addtrust option to the x509 1037 utility) or reject. 1038 [Steve Henson] 1039 1040 *) Add -trusted_first option which attempts to find certificates in the 1041 trusted store even if an untrusted chain is also supplied. 1042 [Steve Henson] 1043 1044 *) MIPS assembly pack updates: support for MIPS32r2 and SmartMIPS ASE, 1045 platform support for Linux and Android. 1046 [Andy Polyakov] 1047 1048 *) Support for linux-x32, ILP32 environment in x86_64 framework. 1049 [Andy Polyakov] 1050 1051 *) Experimental multi-implementation support for FIPS capable OpenSSL. 1052 When in FIPS mode the approved implementations are used as normal, 1053 when not in FIPS mode the internal unapproved versions are used instead. 1054 This means that the FIPS capable OpenSSL isn't forced to use the 1055 (often lower perfomance) FIPS implementations outside FIPS mode. 1056 [Steve Henson] 1057 1058 *) Transparently support X9.42 DH parameters when calling 1059 PEM_read_bio_DHparameters. This means existing applications can handle 1060 the new parameter format automatically. 1061 [Steve Henson] 1062 1063 *) Initial experimental support for X9.42 DH parameter format: mainly 1064 to support use of 'q' parameter for RFC5114 parameters. 1065 [Steve Henson] 1066 1067 *) Add DH parameters from RFC5114 including test data to dhtest. 1068 [Steve Henson] 1069 1070 *) Support for automatic EC temporary key parameter selection. If enabled 1071 the most preferred EC parameters are automatically used instead of 1072 hardcoded fixed parameters. Now a server just has to call: 1073 SSL_CTX_set_ecdh_auto(ctx, 1) and the server will automatically 1074 support ECDH and use the most appropriate parameters. 1075 [Steve Henson] 1076 1077 *) Enhance and tidy EC curve and point format TLS extension code. Use 1078 static structures instead of allocation if default values are used. 1079 New ctrls to set curves we wish to support and to retrieve shared curves. 1080 Print out shared curves in s_server. New options to s_server and s_client 1081 to set list of supported curves. 1082 [Steve Henson] 1083 1084 *) New ctrls to retrieve supported signature algorithms and 1085 supported curve values as an array of NIDs. Extend openssl utility 1086 to print out received values. 1087 [Steve Henson] 1088 1089 *) Add new APIs EC_curve_nist2nid and EC_curve_nid2nist which convert 1090 between NIDs and the more common NIST names such as "P-256". Enhance 1091 ecparam utility and ECC method to recognise the NIST names for curves. 1092 [Steve Henson] 1093 1094 *) Enhance SSL/TLS certificate chain handling to support different 1095 chains for each certificate instead of one chain in the parent SSL_CTX. 1096 [Steve Henson] 1097 1098 *) Support for fixed DH ciphersuite client authentication: where both 1099 server and client use DH certificates with common parameters. 1100 [Steve Henson] 1101 1102 *) Support for fixed DH ciphersuites: those requiring DH server 1103 certificates. 1104 [Steve Henson] 1105 1106 *) New function i2d_re_X509_tbs for re-encoding the TBS portion of 1107 the certificate. 1108 Note: Related 1.0.2-beta specific macros X509_get_cert_info, 1109 X509_CINF_set_modified, X509_CINF_get_issuer, X509_CINF_get_extensions and 1110 X509_CINF_get_signature were reverted post internal team review. 1111 1112 Changes between 1.0.1k and 1.0.1l [15 Jan 2015] 1113 1114 *) Build fixes for the Windows and OpenVMS platforms 1115 [Matt Caswell and Richard Levitte] 1116 1117 Changes between 1.0.1j and 1.0.1k [8 Jan 2015] 1118 1119 *) Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS 1120 message can cause a segmentation fault in OpenSSL due to a NULL pointer 1121 dereference. This could lead to a Denial Of Service attack. Thanks to 1122 Markus Stenberg of Cisco Systems, Inc. for reporting this issue. 1123 (CVE-2014-3571) 1124 [Steve Henson] 1125 1126 *) Fix DTLS memory leak in dtls1_buffer_record. A memory leak can occur in the 1127 dtls1_buffer_record function under certain conditions. In particular this 1128 could occur if an attacker sent repeated DTLS records with the same 1129 sequence number but for the next epoch. The memory leak could be exploited 1130 by an attacker in a Denial of Service attack through memory exhaustion. 1131 Thanks to Chris Mueller for reporting this issue. 1132 (CVE-2015-0206) 1133 [Matt Caswell] 1134 1135 *) Fix issue where no-ssl3 configuration sets method to NULL. When openssl is 1136 built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl 1137 method would be set to NULL which could later result in a NULL pointer 1138 dereference. Thanks to Frank Schmirler for reporting this issue. 1139 (CVE-2014-3569) 1140 [Kurt Roeckx] 1141 1142 *) Abort handshake if server key exchange message is omitted for ephemeral 1143 ECDH ciphersuites. 1144 1145 Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for 1146 reporting this issue. 1147 (CVE-2014-3572) 1148 [Steve Henson] 1149 1150 *) Remove non-export ephemeral RSA code on client and server. This code 1151 violated the TLS standard by allowing the use of temporary RSA keys in 1152 non-export ciphersuites and could be used by a server to effectively 1153 downgrade the RSA key length used to a value smaller than the server 1154 certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at 1155 INRIA or reporting this issue. 1156 (CVE-2015-0204) 1157 [Steve Henson] 1158 1159 *) Fixed issue where DH client certificates are accepted without verification. 1160 An OpenSSL server will accept a DH certificate for client authentication 1161 without the certificate verify message. This effectively allows a client to 1162 authenticate without the use of a private key. This only affects servers 1163 which trust a client certificate authority which issues certificates 1164 containing DH keys: these are extremely rare and hardly ever encountered. 1165 Thanks for Karthikeyan Bhargavan of the PROSECCO team at INRIA or reporting 1166 this issue. 1167 (CVE-2015-0205) 1168 [Steve Henson] 1169 1170 *) Ensure that the session ID context of an SSL is updated when its 1171 SSL_CTX is updated via SSL_set_SSL_CTX. 1172 1173 The session ID context is typically set from the parent SSL_CTX, 1174 and can vary with the CTX. 1175 [Adam Langley] 1176 1177 *) Fix various certificate fingerprint issues. 1178 1179 By using non-DER or invalid encodings outside the signed portion of a 1180 certificate the fingerprint can be changed without breaking the signature. 1181 Although no details of the signed portion of the certificate can be changed 1182 this can cause problems with some applications: e.g. those using the 1183 certificate fingerprint for blacklists. 1184 1185 1. Reject signatures with non zero unused bits. 1186 1187 If the BIT STRING containing the signature has non zero unused bits reject 1188 the signature. All current signature algorithms require zero unused bits. 1189 1190 2. Check certificate algorithm consistency. 1191 1192 Check the AlgorithmIdentifier inside TBS matches the one in the 1193 certificate signature. NB: this will result in signature failure 1194 errors for some broken certificates. 1195 1196 Thanks to Konrad Kraszewski from Google for reporting this issue. 1197 1198 3. Check DSA/ECDSA signatures use DER. 1199 1200 Reencode DSA/ECDSA signatures and compare with the original received 1201 signature. Return an error if there is a mismatch. 1202 1203 This will reject various cases including garbage after signature 1204 (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS 1205 program for discovering this case) and use of BER or invalid ASN.1 INTEGERs 1206 (negative or with leading zeroes). 1207 1208 Further analysis was conducted and fixes were developed by Stephen Henson 1209 of the OpenSSL core team. 1210 1211 (CVE-2014-8275) 1212 [Steve Henson] 1213 1214 *) Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect 1215 results on some platforms, including x86_64. This bug occurs at random 1216 with a very low probability, and is not known to be exploitable in any 1217 way, though its exact impact is difficult to determine. Thanks to Pieter 1218 Wuille (Blockstream) who reported this issue and also suggested an initial 1219 fix. Further analysis was conducted by the OpenSSL development team and 1220 Adam Langley of Google. The final fix was developed by Andy Polyakov of 1221 the OpenSSL core team. 1222 (CVE-2014-3570) 1223 [Andy Polyakov] 1224 1225 *) Do not resume sessions on the server if the negotiated protocol 1226 version does not match the session's version. Resuming with a different 1227 version, while not strictly forbidden by the RFC, is of questionable 1228 sanity and breaks all known clients. 1229 [David Benjamin, Emilia K��sper] 1230 1231 *) Tighten handling of the ChangeCipherSpec (CCS) message: reject 1232 early CCS messages during renegotiation. (Note that because 1233 renegotiation is encrypted, this early CCS was not exploitable.) 1234 [Emilia K��sper] 1235 1236 *) Tighten client-side session ticket handling during renegotiation: 1237 ensure that the client only accepts a session ticket if the server sends 1238 the extension anew in the ServerHello. Previously, a TLS client would 1239 reuse the old extension state and thus accept a session ticket if one was 1240 announced in the initial ServerHello. 1241 1242 Similarly, ensure that the client requires a session ticket if one 1243 was advertised in the ServerHello. Previously, a TLS client would 1244 ignore a missing NewSessionTicket message. 1245 [Emilia K��sper] 1246 1247 Changes between 1.0.1i and 1.0.1j [15 Oct 2014] 1248 1249 *) SRTP Memory Leak. 1250 1251 A flaw in the DTLS SRTP extension parsing code allows an attacker, who 1252 sends a carefully crafted handshake message, to cause OpenSSL to fail 1253 to free up to 64k of memory causing a memory leak. This could be 1254 exploited in a Denial Of Service attack. This issue affects OpenSSL 1255 1.0.1 server implementations for both SSL/TLS and DTLS regardless of 1256 whether SRTP is used or configured. Implementations of OpenSSL that 1257 have been compiled with OPENSSL_NO_SRTP defined are not affected. 1258 1259 The fix was developed by the OpenSSL team. 1260 (CVE-2014-3513) 1261 [OpenSSL team] 1262 1263 *) Session Ticket Memory Leak. 1264 1265 When an OpenSSL SSL/TLS/DTLS server receives a session ticket the 1266 integrity of that ticket is first verified. In the event of a session 1267 ticket integrity check failing, OpenSSL will fail to free memory 1268 causing a memory leak. By sending a large number of invalid session 1269 tickets an attacker could exploit this issue in a Denial Of Service 1270 attack. 1271 (CVE-2014-3567) 1272 [Steve Henson] 1273 1274 *) Build option no-ssl3 is incomplete. 1275 1276 When OpenSSL is configured with "no-ssl3" as a build option, servers 1277 could accept and complete a SSL 3.0 handshake, and clients could be 1278 configured to send them. 1279 (CVE-2014-3568) 1280 [Akamai and the OpenSSL team] 1281 1282 *) Add support for TLS_FALLBACK_SCSV. 1283 Client applications doing fallback retries should call 1284 SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV). 1285 (CVE-2014-3566) 1286 [Adam Langley, Bodo Moeller] 1287 1288 *) Add additional DigestInfo checks. 1289 1290 Reencode DigestInto in DER and check against the original when 1291 verifying RSA signature: this will reject any improperly encoded 1292 DigestInfo structures. 1293 1294 Note: this is a precautionary measure and no attacks are currently known. 1295 1296 [Steve Henson] 1297 1298 Changes between 1.0.1h and 1.0.1i [6 Aug 2014] 1299 1300 *) Fix SRP buffer overrun vulnerability. Invalid parameters passed to the 1301 SRP code can be overrun an internal buffer. Add sanity check that 1302 g, A, B < N to SRP code. 1303 1304 Thanks to Sean Devlin and Watson Ladd of Cryptography Services, NCC 1305 Group for discovering this issue. 1306 (CVE-2014-3512) 1307 [Steve Henson] 1308 1309 *) A flaw in the OpenSSL SSL/TLS server code causes the server to negotiate 1310 TLS 1.0 instead of higher protocol versions when the ClientHello message 1311 is badly fragmented. This allows a man-in-the-middle attacker to force a 1312 downgrade to TLS 1.0 even if both the server and the client support a 1313 higher protocol version, by modifying the client's TLS records. 1314 1315 Thanks to David Benjamin and Adam Langley (Google) for discovering and 1316 researching this issue. 1317 (CVE-2014-3511) 1318 [David Benjamin] 1319 1320 *) OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject 1321 to a denial of service attack. A malicious server can crash the client 1322 with a null pointer dereference (read) by specifying an anonymous (EC)DH 1323 ciphersuite and sending carefully crafted handshake messages. 1324 1325 Thanks to Felix Gr��bert (Google) for discovering and researching this 1326 issue. 1327 (CVE-2014-3510) 1328 [Emilia K��sper] 1329 1330 *) By sending carefully crafted DTLS packets an attacker could cause openssl 1331 to leak memory. This can be exploited through a Denial of Service attack. 1332 Thanks to Adam Langley for discovering and researching this issue. 1333 (CVE-2014-3507) 1334 [Adam Langley] 1335 1336 *) An attacker can force openssl to consume large amounts of memory whilst 1337 processing DTLS handshake messages. This can be exploited through a 1338 Denial of Service attack. 1339 Thanks to Adam Langley for discovering and researching this issue. 1340 (CVE-2014-3506) 1341 [Adam Langley] 1342 1343 *) An attacker can force an error condition which causes openssl to crash 1344 whilst processing DTLS packets due to memory being freed twice. This 1345 can be exploited through a Denial of Service attack. 1346 Thanks to Adam Langley and Wan-Teh Chang for discovering and researching 1347 this issue. 1348 (CVE-2014-3505) 1349 [Adam Langley] 1350 1351 *) If a multithreaded client connects to a malicious server using a resumed 1352 session and the server sends an ec point format extension it could write 1353 up to 255 bytes to freed memory. 1354 1355 Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this 1356 issue. 1357 (CVE-2014-3509) 1358 [Gabor Tyukasz] 1359 1360 *) A malicious server can crash an OpenSSL client with a null pointer 1361 dereference (read) by specifying an SRP ciphersuite even though it was not 1362 properly negotiated with the client. This can be exploited through a 1363 Denial of Service attack. 1364 1365 Thanks to Joonas Kuorilehto and Riku Hietam��ki (Codenomicon) for 1366 discovering and researching this issue. 1367 (CVE-2014-5139) 1368 [Steve Henson] 1369 1370 *) A flaw in OBJ_obj2txt may cause pretty printing functions such as 1371 X509_name_oneline, X509_name_print_ex et al. to leak some information 1372 from the stack. Applications may be affected if they echo pretty printing 1373 output to the attacker. 1374 1375 Thanks to Ivan Fratric (Google) for discovering this issue. 1376 (CVE-2014-3508) 1377 [Emilia K��sper, and Steve Henson] 1378 1379 *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.) 1380 for corner cases. (Certain input points at infinity could lead to 1381 bogus results, with non-infinity inputs mapped to infinity too.) 1382 [Bodo Moeller] 1383 1384 Changes between 1.0.1g and 1.0.1h [5 Jun 2014] 1385 1386 *) Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted 1387 handshake can force the use of weak keying material in OpenSSL 1388 SSL/TLS clients and servers. 1389 1390 Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and 1391 researching this issue. (CVE-2014-0224) 1392 [KIKUCHI Masashi, Steve Henson] 1393 1394 *) Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an 1395 OpenSSL DTLS client the code can be made to recurse eventually crashing 1396 in a DoS attack. 1397 1398 Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue. 1399 (CVE-2014-0221) 1400 [Imre Rad, Steve Henson] 1401 1402 *) Fix DTLS invalid fragment vulnerability. A buffer overrun attack can 1403 be triggered by sending invalid DTLS fragments to an OpenSSL DTLS 1404 client or server. This is potentially exploitable to run arbitrary 1405 code on a vulnerable client or server. 1406 1407 Thanks to J��ri Aedla for reporting this issue. (CVE-2014-0195) 1408 [J��ri Aedla, Steve Henson] 1409 1410 *) Fix bug in TLS code where clients enable anonymous ECDH ciphersuites 1411 are subject to a denial of service attack. 1412 1413 Thanks to Felix Gr��bert and Ivan Fratric at Google for discovering 1414 this issue. (CVE-2014-3470) 1415 [Felix Gr��bert, Ivan Fratric, Steve Henson] 1416 1417 *) Harmonize version and its documentation. -f flag is used to display 1418 compilation flags. 1419 [mancha <mancha1@zoho.com>] 1420 1421 *) Fix eckey_priv_encode so it immediately returns an error upon a failure 1422 in i2d_ECPrivateKey. Thanks to Ted Unangst for feedback on this issue. 1423 [mancha <mancha1@zoho.com>] 1424 1425 *) Fix some double frees. These are not thought to be exploitable. 1426 [mancha <mancha1@zoho.com>] 1427 1428 Changes between 1.0.1f and 1.0.1g [7 Apr 2014] 1429 1430 *) A missing bounds check in the handling of the TLS heartbeat extension 1431 can be used to reveal up to 64k of memory to a connected client or 1432 server. 1433 1434 Thanks for Neel Mehta of Google Security for discovering this bug and to 1435 Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for 1436 preparing the fix (CVE-2014-0160) 1437 [Adam Langley, Bodo Moeller] 1438 1439 *) Fix for the attack described in the paper "Recovering OpenSSL 1440 ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack" 1441 by Yuval Yarom and Naomi Benger. Details can be obtained from: 1442 http://eprint.iacr.org/2014/140 1443 1444 Thanks to Yuval Yarom and Naomi Benger for discovering this 1445 flaw and to Yuval Yarom for supplying a fix (CVE-2014-0076) 1446 [Yuval Yarom and Naomi Benger] 1447 1448 *) TLS pad extension: draft-agl-tls-padding-03 1449 1450 Workaround for the "TLS hang bug" (see FAQ and PR#2771): if the 1451 TLS client Hello record length value would otherwise be > 255 and 1452 less that 512 pad with a dummy extension containing zeroes so it 1453 is at least 512 bytes long. 1454 1455 [Adam Langley, Steve Henson] 1456 1457 Changes between 1.0.1e and 1.0.1f [6 Jan 2014] 1458 1459 *) Fix for TLS record tampering bug. A carefully crafted invalid 1460 handshake could crash OpenSSL with a NULL pointer exception. 1461 Thanks to Anton Johansson for reporting this issues. 1462 (CVE-2013-4353) 1463 1464 *) Keep original DTLS digest and encryption contexts in retransmission 1465 structures so we can use the previous session parameters if they need 1466 to be resent. (CVE-2013-6450) 1467 [Steve Henson] 1468 1469 *) Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which 1470 avoids preferring ECDHE-ECDSA ciphers when the client appears to be 1471 Safari on OS X. Safari on OS X 10.8..10.8.3 advertises support for 1472 several ECDHE-ECDSA ciphers, but fails to negotiate them. The bug 1473 is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing 1474 10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer. 1475 [Rob Stradling, Adam Langley] 1476 1477 Changes between 1.0.1d and 1.0.1e [11 Feb 2013] 1478 1479 *) Correct fix for CVE-2013-0169. The original didn't work on AES-NI 1480 supporting platforms or when small records were transferred. 1481 [Andy Polyakov, Steve Henson] 1482 1483 Changes between 1.0.1c and 1.0.1d [5 Feb 2013] 1484 1485 *) Make the decoding of SSLv3, TLS and DTLS CBC records constant time. 1486 1487 This addresses the flaw in CBC record processing discovered by 1488 Nadhem Alfardan and Kenny Paterson. Details of this attack can be found 1489 at: http://www.isg.rhul.ac.uk/tls/ 1490 1491 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 1492 Security Group at Royal Holloway, University of London 1493 (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and 1494 Emilia K��sper for the initial patch. 1495 (CVE-2013-0169) 1496 [Emilia K��sper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson] 1497 1498 *) Fix flaw in AESNI handling of TLS 1.2 and 1.1 records for CBC mode 1499 ciphersuites which can be exploited in a denial of service attack. 1500 Thanks go to and to Adam Langley <agl@chromium.org> for discovering 1501 and detecting this bug and to Wolfgang Ettlinger 1502 <wolfgang.ettlinger@gmail.com> for independently discovering this issue. 1503 (CVE-2012-2686) 1504 [Adam Langley] 1505 1506 *) Return an error when checking OCSP signatures when key is NULL. 1507 This fixes a DoS attack. (CVE-2013-0166) 1508 [Steve Henson] 1509 1510 *) Make openssl verify return errors. 1511 [Chris Palmer <palmer@google.com> and Ben Laurie] 1512 1513 *) Call OCSP Stapling callback after ciphersuite has been chosen, so 1514 the right response is stapled. Also change SSL_get_certificate() 1515 so it returns the certificate actually sent. 1516 See http://rt.openssl.org/Ticket/Display.html?id=2836. 1517 [Rob Stradling <rob.stradling@comodo.com>] 1518 1519 *) Fix possible deadlock when decoding public keys. 1520 [Steve Henson] 1521 1522 *) Don't use TLS 1.0 record version number in initial client hello 1523 if renegotiating. 1524 [Steve Henson] 1525 1526 Changes between 1.0.1b and 1.0.1c [10 May 2012] 1527 1528 *) Sanity check record length before skipping explicit IV in TLS 1529 1.2, 1.1 and DTLS to fix DoS attack. 1530 1531 Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic 1532 fuzzing as a service testing platform. 1533 (CVE-2012-2333) 1534 [Steve Henson] 1535 1536 *) Initialise tkeylen properly when encrypting CMS messages. 1537 Thanks to Solar Designer of Openwall for reporting this issue. 1538 [Steve Henson] 1539 1540 *) In FIPS mode don't try to use composite ciphers as they are not 1541 approved. 1542 [Steve Henson] 1543 1544 Changes between 1.0.1a and 1.0.1b [26 Apr 2012] 1545 1546 *) OpenSSL 1.0.0 sets SSL_OP_ALL to 0x80000FFFL and OpenSSL 1.0.1 and 1547 1.0.1a set SSL_OP_NO_TLSv1_1 to 0x00000400L which would unfortunately 1548 mean any application compiled against OpenSSL 1.0.0 headers setting 1549 SSL_OP_ALL would also set SSL_OP_NO_TLSv1_1, unintentionally disablng 1550 TLS 1.1 also. Fix this by changing the value of SSL_OP_NO_TLSv1_1 to 1551 0x10000000L Any application which was previously compiled against 1552 OpenSSL 1.0.1 or 1.0.1a headers and which cares about SSL_OP_NO_TLSv1_1 1553 will need to be recompiled as a result. Letting be results in 1554 inability to disable specifically TLS 1.1 and in client context, 1555 in unlike event, limit maximum offered version to TLS 1.0 [see below]. 1556 [Steve Henson] 1557 1558 *) In order to ensure interoperabilty SSL_OP_NO_protocolX does not 1559 disable just protocol X, but all protocols above X *if* there are 1560 protocols *below* X still enabled. In more practical terms it means 1561 that if application wants to disable TLS1.0 in favor of TLS1.1 and 1562 above, it's not sufficient to pass SSL_OP_NO_TLSv1, one has to pass 1563 SSL_OP_NO_TLSv1|SSL_OP_NO_SSLv3|SSL_OP_NO_SSLv2. This applies to 1564 client side. 1565 [Andy Polyakov] 1566 1567 Changes between 1.0.1 and 1.0.1a [19 Apr 2012] 1568 1569 *) Check for potentially exploitable overflows in asn1_d2i_read_bio 1570 BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer 1571 in CRYPTO_realloc_clean. 1572 1573 Thanks to Tavis Ormandy, Google Security Team, for discovering this 1574 issue and to Adam Langley <agl@chromium.org> for fixing it. 1575 (CVE-2012-2110) 1576 [Adam Langley (Google), Tavis Ormandy, Google Security Team] 1577 1578 *) Don't allow TLS 1.2 SHA-256 ciphersuites in TLS 1.0, 1.1 connections. 1579 [Adam Langley] 1580 1581 *) Workarounds for some broken servers that "hang" if a client hello 1582 record length exceeds 255 bytes. 1583 1584 1. Do not use record version number > TLS 1.0 in initial client 1585 hello: some (but not all) hanging servers will now work. 1586 2. If we set OPENSSL_MAX_TLS1_2_CIPHER_LENGTH this will truncate 1587 the number of ciphers sent in the client hello. This should be 1588 set to an even number, such as 50, for example by passing: 1589 -DOPENSSL_MAX_TLS1_2_CIPHER_LENGTH=50 to config or Configure. 1590 Most broken servers should now work. 1591 3. If all else fails setting OPENSSL_NO_TLS1_2_CLIENT will disable 1592 TLS 1.2 client support entirely. 1593 [Steve Henson] 1594 1595 *) Fix SEGV in Vector Permutation AES module observed in OpenSSH. 1596 [Andy Polyakov] 1597 1598 Changes between 1.0.0h and 1.0.1 [14 Mar 2012] 1599 1600 *) Add compatibility with old MDC2 signatures which use an ASN1 OCTET 1601 STRING form instead of a DigestInfo. 1602 [Steve Henson] 1603 1604 *) The format used for MDC2 RSA signatures is inconsistent between EVP 1605 and the RSA_sign/RSA_verify functions. This was made more apparent when 1606 OpenSSL used RSA_sign/RSA_verify for some RSA signatures in particular 1607 those which went through EVP_PKEY_METHOD in 1.0.0 and later. Detect 1608 the correct format in RSA_verify so both forms transparently work. 1609 [Steve Henson] 1610 1611 *) Some servers which support TLS 1.0 can choke if we initially indicate 1612 support for TLS 1.2 and later renegotiate using TLS 1.0 in the RSA 1613 encrypted premaster secret. As a workaround use the maximum pemitted 1614 client version in client hello, this should keep such servers happy 1615 and still work with previous versions of OpenSSL. 1616 [Steve Henson] 1617 1618 *) Add support for TLS/DTLS heartbeats. 1619 [Robin Seggelmann <seggelmann@fh-muenster.de>] 1620 1621 *) Add support for SCTP. 1622 [Robin Seggelmann <seggelmann@fh-muenster.de>] 1623 1624 *) Improved PRNG seeding for VOS. 1625 [Paul Green <Paul.Green@stratus.com>] 1626 1627 *) Extensive assembler packs updates, most notably: 1628 1629 - x86[_64]: AES-NI, PCLMULQDQ, RDRAND support; 1630 - x86[_64]: SSSE3 support (SHA1, vector-permutation AES); 1631 - x86_64: bit-sliced AES implementation; 1632 - ARM: NEON support, contemporary platforms optimizations; 1633 - s390x: z196 support; 1634 - *: GHASH and GF(2^m) multiplication implementations; 1635 1636 [Andy Polyakov] 1637 1638 *) Make TLS-SRP code conformant with RFC 5054 API cleanup 1639 (removal of unnecessary code) 1640 [Peter Sylvester <peter.sylvester@edelweb.fr>] 1641 1642 *) Add TLS key material exporter from RFC 5705. 1643 [Eric Rescorla] 1644 1645 *) Add DTLS-SRTP negotiation from RFC 5764. 1646 [Eric Rescorla] 1647 1648 *) Add Next Protocol Negotiation, 1649 http://tools.ietf.org/html/draft-agl-tls-nextprotoneg-00. Can be 1650 disabled with a no-npn flag to config or Configure. Code donated 1651 by Google. 1652 [Adam Langley <agl@google.com> and Ben Laurie] 1653 1654 *) Add optional 64-bit optimized implementations of elliptic curves NIST-P224, 1655 NIST-P256, NIST-P521, with constant-time single point multiplication on 1656 typical inputs. Compiler support for the nonstandard type __uint128_t is 1657 required to use this (present in gcc 4.4 and later, for 64-bit builds). 1658 Code made available under Apache License version 2.0. 1659 1660 Specify "enable-ec_nistp_64_gcc_128" on the Configure (or config) command 1661 line to include this in your build of OpenSSL, and run "make depend" (or 1662 "make update"). This enables the following EC_METHODs: 1663 1664 EC_GFp_nistp224_method() 1665 EC_GFp_nistp256_method() 1666 EC_GFp_nistp521_method() 1667 1668 EC_GROUP_new_by_curve_name() will automatically use these (while 1669 EC_GROUP_new_curve_GFp() currently prefers the more flexible 1670 implementations). 1671 [Emilia K��sper, Adam Langley, Bodo Moeller (Google)] 1672 1673 *) Use type ossl_ssize_t instad of ssize_t which isn't available on 1674 all platforms. Move ssize_t definition from e_os.h to the public 1675 header file e_os2.h as it now appears in public header file cms.h 1676 [Steve Henson] 1677 1678 *) New -sigopt option to the ca, req and x509 utilities. Additional 1679 signature parameters can be passed using this option and in 1680 particular PSS. 1681 [Steve Henson] 1682 1683 *) Add RSA PSS signing function. This will generate and set the 1684 appropriate AlgorithmIdentifiers for PSS based on those in the 1685 corresponding EVP_MD_CTX structure. No application support yet. 1686 [Steve Henson] 1687 1688 *) Support for companion algorithm specific ASN1 signing routines. 1689 New function ASN1_item_sign_ctx() signs a pre-initialised 1690 EVP_MD_CTX structure and sets AlgorithmIdentifiers based on 1691 the appropriate parameters. 1692 [Steve Henson] 1693 1694 *) Add new algorithm specific ASN1 verification initialisation function 1695 to EVP_PKEY_ASN1_METHOD: this is not in EVP_PKEY_METHOD since the ASN1 1696 handling will be the same no matter what EVP_PKEY_METHOD is used. 1697 Add a PSS handler to support verification of PSS signatures: checked 1698 against a number of sample certificates. 1699 [Steve Henson] 1700 1701 *) Add signature printing for PSS. Add PSS OIDs. 1702 [Steve Henson, Martin Kaiser <lists@kaiser.cx>] 1703 1704 *) Add algorithm specific signature printing. An individual ASN1 method 1705 can now print out signatures instead of the standard hex dump. 1706 1707 More complex signatures (e.g. PSS) can print out more meaningful 1708 information. Include DSA version that prints out the signature 1709 parameters r, s. 1710 [Steve Henson] 1711 1712 *) Password based recipient info support for CMS library: implementing 1713 RFC3211. 1714 [Steve Henson] 1715 1716 *) Split password based encryption into PBES2 and PBKDF2 functions. This 1717 neatly separates the code into cipher and PBE sections and is required 1718 for some algorithms that split PBES2 into separate pieces (such as 1719 password based CMS). 1720 [Steve Henson] 1721 1722 *) Session-handling fixes: 1723 - Fix handling of connections that are resuming with a session ID, 1724 but also support Session Tickets. 1725 - Fix a bug that suppressed issuing of a new ticket if the client 1726 presented a ticket with an expired session. 1727 - Try to set the ticket lifetime hint to something reasonable. 1728 - Make tickets shorter by excluding irrelevant information. 1729 - On the client side, don't ignore renewed tickets. 1730 [Adam Langley, Bodo Moeller (Google)] 1731 1732 *) Fix PSK session representation. 1733 [Bodo Moeller] 1734 1735 *) Add RC4-MD5 and AESNI-SHA1 "stitched" implementations. 1736 1737 This work was sponsored by Intel. 1738 [Andy Polyakov] 1739 1740 *) Add GCM support to TLS library. Some custom code is needed to split 1741 the IV between the fixed (from PRF) and explicit (from TLS record) 1742 portions. This adds all GCM ciphersuites supported by RFC5288 and 1743 RFC5289. Generalise some AES* cipherstrings to inlclude GCM and 1744 add a special AESGCM string for GCM only. 1745 [Steve Henson] 1746 1747 *) Expand range of ctrls for AES GCM. Permit setting invocation 1748 field on decrypt and retrieval of invocation field only on encrypt. 1749 [Steve Henson] 1750 1751 *) Add HMAC ECC ciphersuites from RFC5289. Include SHA384 PRF support. 1752 As required by RFC5289 these ciphersuites cannot be used if for 1753 versions of TLS earlier than 1.2. 1754 [Steve Henson] 1755 1756 *) For FIPS capable OpenSSL interpret a NULL default public key method 1757 as unset and return the appopriate default but do *not* set the default. 1758 This means we can return the appopriate method in applications that 1759 swicth between FIPS and non-FIPS modes. 1760 [Steve Henson] 1761 1762 *) Redirect HMAC and CMAC operations to FIPS module in FIPS mode. If an 1763 ENGINE is used then we cannot handle that in the FIPS module so we 1764 keep original code iff non-FIPS operations are allowed. 1765 [Steve Henson] 1766 1767 *) Add -attime option to openssl utilities. 1768 [Peter Eckersley <pde@eff.org>, Ben Laurie and Steve Henson] 1769 1770 *) Redirect DSA and DH operations to FIPS module in FIPS mode. 1771 [Steve Henson] 1772 1773 *) Redirect ECDSA and ECDH operations to FIPS module in FIPS mode. Also use 1774 FIPS EC methods unconditionally for now. 1775 [Steve Henson] 1776 1777 *) New build option no-ec2m to disable characteristic 2 code. 1778 [Steve Henson] 1779 1780 *) Backport libcrypto audit of return value checking from 1.1.0-dev; not 1781 all cases can be covered as some introduce binary incompatibilities. 1782 [Steve Henson] 1783 1784 *) Redirect RSA operations to FIPS module including keygen, 1785 encrypt, decrypt, sign and verify. Block use of non FIPS RSA methods. 1786 [Steve Henson] 1787 1788 *) Add similar low level API blocking to ciphers. 1789 [Steve Henson] 1790 1791 *) Low level digest APIs are not approved in FIPS mode: any attempt 1792 to use these will cause a fatal error. Applications that *really* want 1793 to use them can use the private_* version instead. 1794 [Steve Henson] 1795 1796 *) Redirect cipher operations to FIPS module for FIPS builds. 1797 [Steve Henson] 1798 1799 *) Redirect digest operations to FIPS module for FIPS builds. 1800 [Steve Henson] 1801 1802 *) Update build system to add "fips" flag which will link in fipscanister.o 1803 for static and shared library builds embedding a signature if needed. 1804 [Steve Henson] 1805 1806 *) Output TLS supported curves in preference order instead of numerical 1807 order. This is currently hardcoded for the highest order curves first. 1808 This should be configurable so applications can judge speed vs strength. 1809 [Steve Henson] 1810 1811 *) Add TLS v1.2 server support for client authentication. 1812 [Steve Henson] 1813 1814 *) Add support for FIPS mode in ssl library: disable SSLv3, non-FIPS ciphers 1815 and enable MD5. 1816 [Steve Henson] 1817 1818 *) Functions FIPS_mode_set() and FIPS_mode() which call the underlying 1819 FIPS modules versions. 1820 [Steve Henson] 1821 1822 *) Add TLS v1.2 client side support for client authentication. Keep cache 1823 of handshake records longer as we don't know the hash algorithm to use 1824 until after the certificate request message is received. 1825 [Steve Henson] 1826 1827 *) Initial TLS v1.2 client support. Add a default signature algorithms 1828 extension including all the algorithms we support. Parse new signature 1829 format in client key exchange. Relax some ECC signing restrictions for 1830 TLS v1.2 as indicated in RFC5246. 1831 [Steve Henson] 1832 1833 *) Add server support for TLS v1.2 signature algorithms extension. Switch 1834 to new signature format when needed using client digest preference. 1835 All server ciphersuites should now work correctly in TLS v1.2. No client 1836 support yet and no support for client certificates. 1837 [Steve Henson] 1838 1839 *) Initial TLS v1.2 support. Add new SHA256 digest to ssl code, switch 1840 to SHA256 for PRF when using TLS v1.2 and later. Add new SHA256 based 1841 ciphersuites. At present only RSA key exchange ciphersuites work with 1842 TLS v1.2. Add new option for TLS v1.2 replacing the old and obsolete 1843 SSL_OP_PKCS1_CHECK flags with SSL_OP_NO_TLSv1_2. New TLSv1.2 methods 1844 and version checking. 1845 [Steve Henson] 1846 1847 *) New option OPENSSL_NO_SSL_INTERN. If an application can be compiled 1848 with this defined it will not be affected by any changes to ssl internal 1849 structures. Add several utility functions to allow openssl application 1850 to work with OPENSSL_NO_SSL_INTERN defined. 1851 [Steve Henson] 1852 1853 *) Add SRP support. 1854 [Tom Wu <tjw@cs.stanford.edu> and Ben Laurie] 1855 1856 *) Add functions to copy EVP_PKEY_METHOD and retrieve flags and id. 1857 [Steve Henson] 1858 1859 *) Permit abbreviated handshakes when renegotiating using the function 1860 SSL_renegotiate_abbreviated(). 1861 [Robin Seggelmann <seggelmann@fh-muenster.de>] 1862 1863 *) Add call to ENGINE_register_all_complete() to 1864 ENGINE_load_builtin_engines(), so some implementations get used 1865 automatically instead of needing explicit application support. 1866 [Steve Henson] 1867 1868 *) Add support for TLS key exporter as described in RFC5705. 1869 [Robin Seggelmann <seggelmann@fh-muenster.de>, Steve Henson] 1870 1871 *) Initial TLSv1.1 support. Since TLSv1.1 is very similar to TLS v1.0 only 1872 a few changes are required: 1873 1874 Add SSL_OP_NO_TLSv1_1 flag. 1875 Add TLSv1_1 methods. 1876 Update version checking logic to handle version 1.1. 1877 Add explicit IV handling (ported from DTLS code). 1878 Add command line options to s_client/s_server. 1879 [Steve Henson] 1880 1881 Changes between 1.0.0g and 1.0.0h [12 Mar 2012] 1882 1883 *) Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness 1884 in CMS and PKCS7 code. When RSA decryption fails use a random key for 1885 content decryption and always return the same error. Note: this attack 1886 needs on average 2^20 messages so it only affects automated senders. The 1887 old behaviour can be reenabled in the CMS code by setting the 1888 CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where 1889 an MMA defence is not necessary. 1890 Thanks to Ivan Nestlerode <inestlerode@us.ibm.com> for discovering 1891 this issue. (CVE-2012-0884) 1892 [Steve Henson] 1893 1894 *) Fix CVE-2011-4619: make sure we really are receiving a 1895 client hello before rejecting multiple SGC restarts. Thanks to 1896 Ivan Nestlerode <inestlerode@us.ibm.com> for discovering this bug. 1897 [Steve Henson] 1898 1899 Changes between 1.0.0f and 1.0.0g [18 Jan 2012] 1900 1901 *) Fix for DTLS DoS issue introduced by fix for CVE-2011-4109. 1902 Thanks to Antonio Martin, Enterprise Secure Access Research and 1903 Development, Cisco Systems, Inc. for discovering this bug and 1904 preparing a fix. (CVE-2012-0050) 1905 [Antonio Martin] 1906 1907 Changes between 1.0.0e and 1.0.0f [4 Jan 2012] 1908 1909 *) Nadhem Alfardan and Kenny Paterson have discovered an extension 1910 of the Vaudenay padding oracle attack on CBC mode encryption 1911 which enables an efficient plaintext recovery attack against 1912 the OpenSSL implementation of DTLS. Their attack exploits timing 1913 differences arising during decryption processing. A research 1914 paper describing this attack can be found at: 1915 http://www.isg.rhul.ac.uk/~kp/dtls.pdf 1916 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 1917 Security Group at Royal Holloway, University of London 1918 (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann 1919 <seggelmann@fh-muenster.de> and Michael Tuexen <tuexen@fh-muenster.de> 1920 for preparing the fix. (CVE-2011-4108) 1921 [Robin Seggelmann, Michael Tuexen] 1922 1923 *) Clear bytes used for block padding of SSL 3.0 records. 1924 (CVE-2011-4576) 1925 [Adam Langley (Google)] 1926 1927 *) Only allow one SGC handshake restart for SSL/TLS. Thanks to George 1928 Kadianakis <desnacked@gmail.com> for discovering this issue and 1929 Adam Langley for preparing the fix. (CVE-2011-4619) 1930 [Adam Langley (Google)] 1931 1932 *) Check parameters are not NULL in GOST ENGINE. (CVE-2012-0027) 1933 [Andrey Kulikov <amdeich@gmail.com>] 1934 1935 *) Prevent malformed RFC3779 data triggering an assertion failure. 1936 Thanks to Andrew Chi, BBN Technologies, for discovering the flaw 1937 and Rob Austein <sra@hactrn.net> for fixing it. (CVE-2011-4577) 1938 [Rob Austein <sra@hactrn.net>] 1939 1940 *) Improved PRNG seeding for VOS. 1941 [Paul Green <Paul.Green@stratus.com>] 1942 1943 *) Fix ssl_ciph.c set-up race. 1944 [Adam Langley (Google)] 1945 1946 *) Fix spurious failures in ecdsatest.c. 1947 [Emilia K��sper (Google)] 1948 1949 *) Fix the BIO_f_buffer() implementation (which was mixing different 1950 interpretations of the '..._len' fields). 1951 [Adam Langley (Google)] 1952 1953 *) Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than 1954 BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent 1955 threads won't reuse the same blinding coefficients. 1956 1957 This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING 1958 lock to call BN_BLINDING_invert_ex, and avoids one use of 1959 BN_BLINDING_update for each BN_BLINDING structure (previously, 1960 the last update always remained unused). 1961 [Emilia K��sper (Google)] 1962 1963 *) In ssl3_clear, preserve s3->init_extra along with s3->rbuf. 1964 [Bob Buckholz (Google)] 1965 1966 Changes between 1.0.0d and 1.0.0e [6 Sep 2011] 1967 1968 *) Fix bug where CRLs with nextUpdate in the past are sometimes accepted 1969 by initialising X509_STORE_CTX properly. (CVE-2011-3207) 1970 [Kaspar Brand <ossl@velox.ch>] 1971 1972 *) Fix SSL memory handling for (EC)DH ciphersuites, in particular 1973 for multi-threaded use of ECDH. (CVE-2011-3210) 1974 [Adam Langley (Google)] 1975 1976 *) Fix x509_name_ex_d2i memory leak on bad inputs. 1977 [Bodo Moeller] 1978 1979 *) Remove hard coded ecdsaWithSHA1 signature tests in ssl code and check 1980 signature public key algorithm by using OID xref utilities instead. 1981 Before this you could only use some ECC ciphersuites with SHA1 only. 1982 [Steve Henson] 1983 1984 *) Add protection against ECDSA timing attacks as mentioned in the paper 1985 by Billy Bob Brumley and Nicola Tuveri, see: 1986 1987 http://eprint.iacr.org/2011/232.pdf 1988 1989 [Billy Bob Brumley and Nicola Tuveri] 1990 1991 Changes between 1.0.0c and 1.0.0d [8 Feb 2011] 1992 1993 *) Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014 1994 [Neel Mehta, Adam Langley, Bodo Moeller (Google)] 1995 1996 *) Fix bug in string printing code: if *any* escaping is enabled we must 1997 escape the escape character (backslash) or the resulting string is 1998 ambiguous. 1999 [Steve Henson] 2000 2001 Changes between 1.0.0b and 1.0.0c [2 Dec 2010] 2002 2003 *) Disable code workaround for ancient and obsolete Netscape browsers 2004 and servers: an attacker can use it in a ciphersuite downgrade attack. 2005 Thanks to Martin Rex for discovering this bug. CVE-2010-4180 2006 [Steve Henson] 2007 2008 *) Fixed J-PAKE implementation error, originally discovered by 2009 Sebastien Martini, further info and confirmation from Stefan 2010 Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252 2011 [Ben Laurie] 2012 2013 Changes between 1.0.0a and 1.0.0b [16 Nov 2010] 2014 2015 *) Fix extension code to avoid race conditions which can result in a buffer 2016 overrun vulnerability: resumed sessions must not be modified as they can 2017 be shared by multiple threads. CVE-2010-3864 2018 [Steve Henson] 2019 2020 *) Fix WIN32 build system to correctly link an ENGINE directory into 2021 a DLL. 2022 [Steve Henson] 2023 2024 Changes between 1.0.0 and 1.0.0a [01 Jun 2010] 2025 2026 *) Check return value of int_rsa_verify in pkey_rsa_verifyrecover 2027 (CVE-2010-1633) 2028 [Steve Henson, Peter-Michael Hager <hager@dortmund.net>] 2029 2030 Changes between 0.9.8n and 1.0.0 [29 Mar 2010] 2031 2032 *) Add "missing" function EVP_CIPHER_CTX_copy(). This copies a cipher 2033 context. The operation can be customised via the ctrl mechanism in 2034 case ENGINEs want to include additional functionality. 2035 [Steve Henson] 2036 2037 *) Tolerate yet another broken PKCS#8 key format: private key value negative. 2038 [Steve Henson] 2039 2040 *) Add new -subject_hash_old and -issuer_hash_old options to x509 utility to 2041 output hashes compatible with older versions of OpenSSL. 2042 [Willy Weisz <weisz@vcpc.univie.ac.at>] 2043 2044 *) Fix compression algorithm handling: if resuming a session use the 2045 compression algorithm of the resumed session instead of determining 2046 it from client hello again. Don't allow server to change algorithm. 2047 [Steve Henson] 2048 2049 *) Add load_crls() function to apps tidying load_certs() too. Add option 2050 to verify utility to allow additional CRLs to be included. 2051 [Steve Henson] 2052 2053 *) Update OCSP request code to permit adding custom headers to the request: 2054 some responders need this. 2055 [Steve Henson] 2056 2057 *) The function EVP_PKEY_sign() returns <=0 on error: check return code 2058 correctly. 2059 [Julia Lawall <julia@diku.dk>] 2060 2061 *) Update verify callback code in apps/s_cb.c and apps/verify.c, it 2062 needlessly dereferenced structures, used obsolete functions and 2063 didn't handle all updated verify codes correctly. 2064 [Steve Henson] 2065 2066 *) Disable MD2 in the default configuration. 2067 [Steve Henson] 2068 2069 *) In BIO_pop() and BIO_push() use the ctrl argument (which was NULL) to 2070 indicate the initial BIO being pushed or popped. This makes it possible 2071 to determine whether the BIO is the one explicitly called or as a result 2072 of the ctrl being passed down the chain. Fix BIO_pop() and SSL BIOs so 2073 it handles reference counts correctly and doesn't zero out the I/O bio 2074 when it is not being explicitly popped. WARNING: applications which 2075 included workarounds for the old buggy behaviour will need to be modified 2076 or they could free up already freed BIOs. 2077 [Steve Henson] 2078 2079 *) Extend the uni2asc/asc2uni => OPENSSL_uni2asc/OPENSSL_asc2uni 2080 renaming to all platforms (within the 0.9.8 branch, this was 2081 done conditionally on Netware platforms to avoid a name clash). 2082 [Guenter <lists@gknw.net>] 2083 2084 *) Add ECDHE and PSK support to DTLS. 2085 [Michael Tuexen <tuexen@fh-muenster.de>] 2086 2087 *) Add CHECKED_STACK_OF macro to safestack.h, otherwise safestack can't 2088 be used on C++. 2089 [Steve Henson] 2090 2091 *) Add "missing" function EVP_MD_flags() (without this the only way to 2092 retrieve a digest flags is by accessing the structure directly. Update 2093 EVP_MD_do_all*() and EVP_CIPHER_do_all*() to include the name a digest 2094 or cipher is registered as in the "from" argument. Print out all 2095 registered digests in the dgst usage message instead of manually 2096 attempting to work them out. 2097 [Steve Henson] 2098 2099 *) If no SSLv2 ciphers are used don't use an SSLv2 compatible client hello: 2100 this allows the use of compression and extensions. Change default cipher 2101 string to remove SSLv2 ciphersuites. This effectively avoids ancient SSLv2 2102 by default unless an application cipher string requests it. 2103 [Steve Henson] 2104 2105 *) Alter match criteria in PKCS12_parse(). It used to try to use local 2106 key ids to find matching certificates and keys but some PKCS#12 files 2107 don't follow the (somewhat unwritten) rules and this strategy fails. 2108 Now just gather all certificates together and the first private key 2109 then look for the first certificate that matches the key. 2110 [Steve Henson] 2111 2112 *) Support use of registered digest and cipher names for dgst and cipher 2113 commands instead of having to add each one as a special case. So now 2114 you can do: 2115 2116 openssl sha256 foo 2117 2118 as well as: 2119 2120 openssl dgst -sha256 foo 2121 2122 and this works for ENGINE based algorithms too. 2123 2124 [Steve Henson] 2125 2126 *) Update Gost ENGINE to support parameter files. 2127 [Victor B. Wagner <vitus@cryptocom.ru>] 2128 2129 *) Support GeneralizedTime in ca utility. 2130 [Oliver Martin <oliver@volatilevoid.net>, Steve Henson] 2131 2132 *) Enhance the hash format used for certificate directory links. The new 2133 form uses the canonical encoding (meaning equivalent names will work 2134 even if they aren't identical) and uses SHA1 instead of MD5. This form 2135 is incompatible with the older format and as a result c_rehash should 2136 be used to rebuild symbolic links. 2137 [Steve Henson] 2138 2139 *) Make PKCS#8 the default write format for private keys, replacing the 2140 traditional format. This form is standardised, more secure and doesn't 2141 include an implicit MD5 dependency. 2142 [Steve Henson] 2143 2144 *) Add a $gcc_devteam_warn option to Configure. The idea is that any code 2145 committed to OpenSSL should pass this lot as a minimum. 2146 [Steve Henson] 2147 2148 *) Add session ticket override functionality for use by EAP-FAST. 2149 [Jouni Malinen <j@w1.fi>] 2150 2151 *) Modify HMAC functions to return a value. Since these can be implemented 2152 in an ENGINE errors can occur. 2153 [Steve Henson] 2154 2155 *) Type-checked OBJ_bsearch_ex. 2156 [Ben Laurie] 2157 2158 *) Type-checked OBJ_bsearch. Also some constification necessitated 2159 by type-checking. Still to come: TXT_DB, bsearch(?), 2160 OBJ_bsearch_ex, qsort, CRYPTO_EX_DATA, ASN1_VALUE, ASN1_STRING, 2161 CONF_VALUE. 2162 [Ben Laurie] 2163 2164 *) New function OPENSSL_gmtime_adj() to add a specific number of days and 2165 seconds to a tm structure directly, instead of going through OS 2166 specific date routines. This avoids any issues with OS routines such 2167 as the year 2038 bug. New *_adj() functions for ASN1 time structures 2168 and X509_time_adj_ex() to cover the extended range. The existing 2169 X509_time_adj() is still usable and will no longer have any date issues. 2170 [Steve Henson] 2171 2172 *) Delta CRL support. New use deltas option which will attempt to locate 2173 and search any appropriate delta CRLs available. 2174 2175 This work was sponsored by Google. 2176 [Steve Henson] 2177 2178 *) Support for CRLs partitioned by reason code. Reorganise CRL processing 2179 code and add additional score elements. Validate alternate CRL paths 2180 as part of the CRL checking and indicate a new error "CRL path validation 2181 error" in this case. Applications wanting additional details can use 2182 the verify callback and check the new "parent" field. If this is not 2183 NULL CRL path validation is taking place. Existing applications wont 2184 see this because it requires extended CRL support which is off by 2185 default. 2186 2187 This work was sponsored by Google. 2188 [Steve Henson] 2189 2190 *) Support for freshest CRL extension. 2191 2192 This work was sponsored by Google. 2193 [Steve Henson] 2194 2195 *) Initial indirect CRL support. Currently only supported in the CRLs 2196 passed directly and not via lookup. Process certificate issuer 2197 CRL entry extension and lookup CRL entries by bother issuer name 2198 and serial number. Check and process CRL issuer entry in IDP extension. 2199 2200 This work was sponsored by Google. 2201 [Steve Henson] 2202 2203 *) Add support for distinct certificate and CRL paths. The CRL issuer 2204 certificate is validated separately in this case. Only enabled if 2205 an extended CRL support flag is set: this flag will enable additional 2206 CRL functionality in future. 2207 2208 This work was sponsored by Google. 2209 [Steve Henson] 2210 2211 *) Add support for policy mappings extension. 2212 2213 This work was sponsored by Google. 2214 [Steve Henson] 2215 2216 *) Fixes to pathlength constraint, self issued certificate handling, 2217 policy processing to align with RFC3280 and PKITS tests. 2218 2219 This work was sponsored by Google. 2220 [Steve Henson] 2221 2222 *) Support for name constraints certificate extension. DN, email, DNS 2223 and URI types are currently supported. 2224 2225 This work was sponsored by Google. 2226 [Steve Henson] 2227 2228 *) To cater for systems that provide a pointer-based thread ID rather 2229 than numeric, deprecate the current numeric thread ID mechanism and 2230 replace it with a structure and associated callback type. This 2231 mechanism allows a numeric "hash" to be extracted from a thread ID in 2232 either case, and on platforms where pointers are larger than 'long', 2233 mixing is done to help ensure the numeric 'hash' is usable even if it 2234 can't be guaranteed unique. The default mechanism is to use "&errno" 2235 as a pointer-based thread ID to distinguish between threads. 2236 2237 Applications that want to provide their own thread IDs should now use 2238 CRYPTO_THREADID_set_callback() to register a callback that will call 2239 either CRYPTO_THREADID_set_numeric() or CRYPTO_THREADID_set_pointer(). 2240 2241 Note that ERR_remove_state() is now deprecated, because it is tied 2242 to the assumption that thread IDs are numeric. ERR_remove_state(0) 2243 to free the current thread's error state should be replaced by 2244 ERR_remove_thread_state(NULL). 2245 2246 (This new approach replaces the functions CRYPTO_set_idptr_callback(), 2247 CRYPTO_get_idptr_callback(), and CRYPTO_thread_idptr() that existed in 2248 OpenSSL 0.9.9-dev between June 2006 and August 2008. Also, if an 2249 application was previously providing a numeric thread callback that 2250 was inappropriate for distinguishing threads, then uniqueness might 2251 have been obtained with &errno that happened immediately in the 2252 intermediate development versions of OpenSSL; this is no longer the 2253 case, the numeric thread callback will now override the automatic use 2254 of &errno.) 2255 [Geoff Thorpe, with help from Bodo Moeller] 2256 2257 *) Initial support for different CRL issuing certificates. This covers a 2258 simple case where the self issued certificates in the chain exist and 2259 the real CRL issuer is higher in the existing chain. 2260 2261 This work was sponsored by Google. 2262 [Steve Henson] 2263 2264 *) Removed effectively defunct crypto/store from the build. 2265 [Ben Laurie] 2266 2267 *) Revamp of STACK to provide stronger type-checking. Still to come: 2268 TXT_DB, bsearch(?), OBJ_bsearch, qsort, CRYPTO_EX_DATA, ASN1_VALUE, 2269 ASN1_STRING, CONF_VALUE. 2270 [Ben Laurie] 2271 2272 *) Add a new SSL_MODE_RELEASE_BUFFERS mode flag to release unused buffer 2273 RAM on SSL connections. This option can save about 34k per idle SSL. 2274 [Nick Mathewson] 2275 2276 *) Revamp of LHASH to provide stronger type-checking. Still to come: 2277 STACK, TXT_DB, bsearch, qsort. 2278 [Ben Laurie] 2279 2280 *) Initial support for Cryptographic Message Syntax (aka CMS) based 2281 on RFC3850, RFC3851 and RFC3852. New cms directory and cms utility, 2282 support for data, signedData, compressedData, digestedData and 2283 encryptedData, envelopedData types included. Scripts to check against 2284 RFC4134 examples draft and interop and consistency checks of many 2285 content types and variants. 2286 [Steve Henson] 2287 2288 *) Add options to enc utility to support use of zlib compression BIO. 2289 [Steve Henson] 2290 2291 *) Extend mk1mf to support importing of options and assembly language 2292 files from Configure script, currently only included in VC-WIN32. 2293 The assembly language rules can now optionally generate the source 2294 files from the associated perl scripts. 2295 [Steve Henson] 2296 2297 *) Implement remaining functionality needed to support GOST ciphersuites. 2298 Interop testing has been performed using CryptoPro implementations. 2299 [Victor B. Wagner <vitus@cryptocom.ru>] 2300 2301 *) s390x assembler pack. 2302 [Andy Polyakov] 2303 2304 *) ARMv4 assembler pack. ARMv4 refers to v4 and later ISA, not CPU 2305 "family." 2306 [Andy Polyakov] 2307 2308 *) Implement Opaque PRF Input TLS extension as specified in 2309 draft-rescorla-tls-opaque-prf-input-00.txt. Since this is not an 2310 official specification yet and no extension type assignment by 2311 IANA exists, this extension (for now) will have to be explicitly 2312 enabled when building OpenSSL by providing the extension number 2313 to use. For example, specify an option 2314 2315 -DTLSEXT_TYPE_opaque_prf_input=0x9527 2316 2317 to the "config" or "Configure" script to enable the extension, 2318 assuming extension number 0x9527 (which is a completely arbitrary 2319 and unofficial assignment based on the MD5 hash of the Internet 2320 Draft). Note that by doing so, you potentially lose 2321 interoperability with other TLS implementations since these might 2322 be using the same extension number for other purposes. 2323 2324 SSL_set_tlsext_opaque_prf_input(ssl, src, len) is used to set the 2325 opaque PRF input value to use in the handshake. This will create 2326 an interal copy of the length-'len' string at 'src', and will 2327 return non-zero for success. 2328 2329 To get more control and flexibility, provide a callback function 2330 by using 2331 2332 SSL_CTX_set_tlsext_opaque_prf_input_callback(ctx, cb) 2333 SSL_CTX_set_tlsext_opaque_prf_input_callback_arg(ctx, arg) 2334 2335 where 2336 2337 int (*cb)(SSL *, void *peerinput, size_t len, void *arg); 2338 void *arg; 2339 2340 Callback function 'cb' will be called in handshakes, and is 2341 expected to use SSL_set_tlsext_opaque_prf_input() as appropriate. 2342 Argument 'arg' is for application purposes (the value as given to 2343 SSL_CTX_set_tlsext_opaque_prf_input_callback_arg() will directly 2344 be provided to the callback function). The callback function 2345 has to return non-zero to report success: usually 1 to use opaque 2346 PRF input just if possible, or 2 to enforce use of the opaque PRF 2347 input. In the latter case, the library will abort the handshake 2348 if opaque PRF input is not successfully negotiated. 2349 2350 Arguments 'peerinput' and 'len' given to the callback function 2351 will always be NULL and 0 in the case of a client. A server will 2352 see the client's opaque PRF input through these variables if 2353 available (NULL and 0 otherwise). Note that if the server 2354 provides an opaque PRF input, the length must be the same as the 2355 length of the client's opaque PRF input. 2356 2357 Note that the callback function will only be called when creating 2358 a new session (session resumption can resume whatever was 2359 previously negotiated), and will not be called in SSL 2.0 2360 handshakes; thus, SSL_CTX_set_options(ctx, SSL_OP_NO_SSLv2) or 2361 SSL_set_options(ssl, SSL_OP_NO_SSLv2) is especially recommended 2362 for applications that need to enforce opaque PRF input. 2363 2364 [Bodo Moeller] 2365 2366 *) Update ssl code to support digests other than SHA1+MD5 for handshake 2367 MAC. 2368 2369 [Victor B. Wagner <vitus@cryptocom.ru>] 2370 2371 *) Add RFC4507 support to OpenSSL. This includes the corrections in 2372 RFC4507bis. The encrypted ticket format is an encrypted encoded 2373 SSL_SESSION structure, that way new session features are automatically 2374 supported. 2375 2376 If a client application caches session in an SSL_SESSION structure 2377 support is transparent because tickets are now stored in the encoded 2378 SSL_SESSION. 2379 2380 The SSL_CTX structure automatically generates keys for ticket 2381 protection in servers so again support should be possible 2382 with no application modification. 2383 2384 If a client or server wishes to disable RFC4507 support then the option 2385 SSL_OP_NO_TICKET can be set. 2386 2387 Add a TLS extension debugging callback to allow the contents of any client 2388 or server extensions to be examined. 2389 2390 This work was sponsored by Google. 2391 [Steve Henson] 2392 2393 *) Final changes to avoid use of pointer pointer casts in OpenSSL. 2394 OpenSSL should now compile cleanly on gcc 4.2 2395 [Peter Hartley <pdh@utter.chaos.org.uk>, Steve Henson] 2396 2397 *) Update SSL library to use new EVP_PKEY MAC API. Include generic MAC 2398 support including streaming MAC support: this is required for GOST 2399 ciphersuite support. 2400 [Victor B. Wagner <vitus@cryptocom.ru>, Steve Henson] 2401 2402 *) Add option -stream to use PKCS#7 streaming in smime utility. New 2403 function i2d_PKCS7_bio_stream() and PEM_write_PKCS7_bio_stream() 2404 to output in BER and PEM format. 2405 [Steve Henson] 2406 2407 *) Experimental support for use of HMAC via EVP_PKEY interface. This 2408 allows HMAC to be handled via the EVP_DigestSign*() interface. The 2409 EVP_PKEY "key" in this case is the HMAC key, potentially allowing 2410 ENGINE support for HMAC keys which are unextractable. New -mac and 2411 -macopt options to dgst utility. 2412 [Steve Henson] 2413 2414 *) New option -sigopt to dgst utility. Update dgst to use 2415 EVP_Digest{Sign,Verify}*. These two changes make it possible to use 2416 alternative signing paramaters such as X9.31 or PSS in the dgst 2417 utility. 2418 [Steve Henson] 2419 2420 *) Change ssl_cipher_apply_rule(), the internal function that does 2421 the work each time a ciphersuite string requests enabling 2422 ("foo+bar"), moving ("+foo+bar"), disabling ("-foo+bar", or 2423 removing ("!foo+bar") a class of ciphersuites: Now it maintains 2424 the order of disabled ciphersuites such that those ciphersuites 2425 that most recently went from enabled to disabled not only stay 2426 in order with respect to each other, but also have higher priority 2427 than other disabled ciphersuites the next time ciphersuites are 2428 enabled again. 2429 2430 This means that you can now say, e.g., "PSK:-PSK:HIGH" to enable 2431 the same ciphersuites as with "HIGH" alone, but in a specific 2432 order where the PSK ciphersuites come first (since they are the 2433 most recently disabled ciphersuites when "HIGH" is parsed). 2434 2435 Also, change ssl_create_cipher_list() (using this new 2436 funcionality) such that between otherwise identical 2437 cihpersuites, ephemeral ECDH is preferred over ephemeral DH in 2438 the default order. 2439 [Bodo Moeller] 2440 2441 *) Change ssl_create_cipher_list() so that it automatically 2442 arranges the ciphersuites in reasonable order before starting 2443 to process the rule string. Thus, the definition for "DEFAULT" 2444 (SSL_DEFAULT_CIPHER_LIST) now is just "ALL:!aNULL:!eNULL", but 2445 remains equivalent to "AES:ALL:!aNULL:!eNULL:+aECDH:+kRSA:+RC4:@STRENGTH". 2446 This makes it much easier to arrive at a reasonable default order 2447 in applications for which anonymous ciphers are OK (meaning 2448 that you can't actually use DEFAULT). 2449 [Bodo Moeller; suggested by Victor Duchovni] 2450 2451 *) Split the SSL/TLS algorithm mask (as used for ciphersuite string 2452 processing) into multiple integers instead of setting 2453 "SSL_MKEY_MASK" bits, "SSL_AUTH_MASK" bits, "SSL_ENC_MASK", 2454 "SSL_MAC_MASK", and "SSL_SSL_MASK" bits all in a single integer. 2455 (These masks as well as the individual bit definitions are hidden 2456 away into the non-exported interface ssl/ssl_locl.h, so this 2457 change to the definition of the SSL_CIPHER structure shouldn't 2458 affect applications.) This give us more bits for each of these 2459 categories, so there is no longer a need to coagulate AES128 and 2460 AES256 into a single algorithm bit, and to coagulate Camellia128 2461 and Camellia256 into a single algorithm bit, which has led to all 2462 kinds of kludges. 2463 2464 Thus, among other things, the kludge introduced in 0.9.7m and 2465 0.9.8e for masking out AES256 independently of AES128 or masking 2466 out Camellia256 independently of AES256 is not needed here in 0.9.9. 2467 2468 With the change, we also introduce new ciphersuite aliases that 2469 so far were missing: "AES128", "AES256", "CAMELLIA128", and 2470 "CAMELLIA256". 2471 [Bodo Moeller] 2472 2473 *) Add support for dsa-with-SHA224 and dsa-with-SHA256. 2474 Use the leftmost N bytes of the signature input if the input is 2475 larger than the prime q (with N being the size in bytes of q). 2476 [Nils Larsch] 2477 2478 *) Very *very* experimental PKCS#7 streaming encoder support. Nothing uses 2479 it yet and it is largely untested. 2480 [Steve Henson] 2481 2482 *) Add support for the ecdsa-with-SHA224/256/384/512 signature types. 2483 [Nils Larsch] 2484 2485 *) Initial incomplete changes to avoid need for function casts in OpenSSL 2486 some compilers (gcc 4.2 and later) reject their use. Safestack is 2487 reimplemented. Update ASN1 to avoid use of legacy functions. 2488 [Steve Henson] 2489 2490 *) Win32/64 targets are linked with Winsock2. 2491 [Andy Polyakov] 2492 2493 *) Add an X509_CRL_METHOD structure to allow CRL processing to be redirected 2494 to external functions. This can be used to increase CRL handling 2495 efficiency especially when CRLs are very large by (for example) storing 2496 the CRL revoked certificates in a database. 2497 [Steve Henson] 2498 2499 *) Overhaul of by_dir code. Add support for dynamic loading of CRLs so 2500 new CRLs added to a directory can be used. New command line option 2501 -verify_return_error to s_client and s_server. This causes real errors 2502 to be returned by the verify callback instead of carrying on no matter 2503 what. This reflects the way a "real world" verify callback would behave. 2504 [Steve Henson] 2505 2506 *) GOST engine, supporting several GOST algorithms and public key formats. 2507 Kindly donated by Cryptocom. 2508 [Cryptocom] 2509 2510 *) Partial support for Issuing Distribution Point CRL extension. CRLs 2511 partitioned by DP are handled but no indirect CRL or reason partitioning 2512 (yet). Complete overhaul of CRL handling: now the most suitable CRL is 2513 selected via a scoring technique which handles IDP and AKID in CRLs. 2514 [Steve Henson] 2515 2516 *) New X509_STORE_CTX callbacks lookup_crls() and lookup_certs() which 2517 will ultimately be used for all verify operations: this will remove the 2518 X509_STORE dependency on certificate verification and allow alternative 2519 lookup methods. X509_STORE based implementations of these two callbacks. 2520 [Steve Henson] 2521 2522 *) Allow multiple CRLs to exist in an X509_STORE with matching issuer names. 2523 Modify get_crl() to find a valid (unexpired) CRL if possible. 2524 [Steve Henson] 2525 2526 *) New function X509_CRL_match() to check if two CRLs are identical. Normally 2527 this would be called X509_CRL_cmp() but that name is already used by 2528 a function that just compares CRL issuer names. Cache several CRL 2529 extensions in X509_CRL structure and cache CRLDP in X509. 2530 [Steve Henson] 2531 2532 *) Store a "canonical" representation of X509_NAME structure (ASN1 Name) 2533 this maps equivalent X509_NAME structures into a consistent structure. 2534 Name comparison can then be performed rapidly using memcmp(). 2535 [Steve Henson] 2536 2537 *) Non-blocking OCSP request processing. Add -timeout option to ocsp 2538 utility. 2539 [Steve Henson] 2540 2541 *) Allow digests to supply their own micalg string for S/MIME type using 2542 the ctrl EVP_MD_CTRL_MICALG. 2543 [Steve Henson] 2544 2545 *) During PKCS7 signing pass the PKCS7 SignerInfo structure to the 2546 EVP_PKEY_METHOD before and after signing via the EVP_PKEY_CTRL_PKCS7_SIGN 2547 ctrl. It can then customise the structure before and/or after signing 2548 if necessary. 2549 [Steve Henson] 2550 2551 *) New function OBJ_add_sigid() to allow application defined signature OIDs 2552 to be added to OpenSSLs internal tables. New function OBJ_sigid_free() 2553 to free up any added signature OIDs. 2554 [Steve Henson] 2555 2556 *) New functions EVP_CIPHER_do_all(), EVP_CIPHER_do_all_sorted(), 2557 EVP_MD_do_all() and EVP_MD_do_all_sorted() to enumerate internal 2558 digest and cipher tables. New options added to openssl utility: 2559 list-message-digest-algorithms and list-cipher-algorithms. 2560 [Steve Henson] 2561 2562 *) Change the array representation of binary polynomials: the list 2563 of degrees of non-zero coefficients is now terminated with -1. 2564 Previously it was terminated with 0, which was also part of the 2565 value; thus, the array representation was not applicable to 2566 polynomials where t^0 has coefficient zero. This change makes 2567 the array representation useful in a more general context. 2568 [Douglas Stebila] 2569 2570 *) Various modifications and fixes to SSL/TLS cipher string 2571 handling. For ECC, the code now distinguishes between fixed ECDH 2572 with RSA certificates on the one hand and with ECDSA certificates 2573 on the other hand, since these are separate ciphersuites. The 2574 unused code for Fortezza ciphersuites has been removed. 2575 2576 For consistency with EDH, ephemeral ECDH is now called "EECDH" 2577 (not "ECDHE"). For consistency with the code for DH 2578 certificates, use of ECDH certificates is now considered ECDH 2579 authentication, not RSA or ECDSA authentication (the latter is 2580 merely the CA's signing algorithm and not actively used in the 2581 protocol). 2582 2583 The temporary ciphersuite alias "ECCdraft" is no longer 2584 available, and ECC ciphersuites are no longer excluded from "ALL" 2585 and "DEFAULT". The following aliases now exist for RFC 4492 2586 ciphersuites, most of these by analogy with the DH case: 2587 2588 kECDHr - ECDH cert, signed with RSA 2589 kECDHe - ECDH cert, signed with ECDSA 2590 kECDH - ECDH cert (signed with either RSA or ECDSA) 2591 kEECDH - ephemeral ECDH 2592 ECDH - ECDH cert or ephemeral ECDH 2593 2594 aECDH - ECDH cert 2595 aECDSA - ECDSA cert 2596 ECDSA - ECDSA cert 2597 2598 AECDH - anonymous ECDH 2599 EECDH - non-anonymous ephemeral ECDH (equivalent to "kEECDH:-AECDH") 2600 2601 [Bodo Moeller] 2602 2603 *) Add additional S/MIME capabilities for AES and GOST ciphers if supported. 2604 Use correct micalg parameters depending on digest(s) in signed message. 2605 [Steve Henson] 2606 2607 *) Add engine support for EVP_PKEY_ASN1_METHOD. Add functions to process 2608 an ENGINE asn1 method. Support ENGINE lookups in the ASN1 code. 2609 [Steve Henson] 2610 2611 *) Initial engine support for EVP_PKEY_METHOD. New functions to permit 2612 an engine to register a method. Add ENGINE lookups for methods and 2613 functional reference processing. 2614 [Steve Henson] 2615 2616 *) New functions EVP_Digest{Sign,Verify)*. These are enchance versions of 2617 EVP_{Sign,Verify}* which allow an application to customise the signature 2618 process. 2619 [Steve Henson] 2620 2621 *) New -resign option to smime utility. This adds one or more signers 2622 to an existing PKCS#7 signedData structure. Also -md option to use an 2623 alternative message digest algorithm for signing. 2624 [Steve Henson] 2625 2626 *) Tidy up PKCS#7 routines and add new functions to make it easier to 2627 create PKCS7 structures containing multiple signers. Update smime 2628 application to support multiple signers. 2629 [Steve Henson] 2630 2631 *) New -macalg option to pkcs12 utility to allow setting of an alternative 2632 digest MAC. 2633 [Steve Henson] 2634 2635 *) Initial support for PKCS#5 v2.0 PRFs other than default SHA1 HMAC. 2636 Reorganize PBE internals to lookup from a static table using NIDs, 2637 add support for HMAC PBE OID translation. Add a EVP_CIPHER ctrl: 2638 EVP_CTRL_PBE_PRF_NID this allows a cipher to specify an alternative 2639 PRF which will be automatically used with PBES2. 2640 [Steve Henson] 2641 2642 *) Replace the algorithm specific calls to generate keys in "req" with the 2643 new API. 2644 [Steve Henson] 2645 2646 *) Update PKCS#7 enveloped data routines to use new API. This is now 2647 supported by any public key method supporting the encrypt operation. A 2648 ctrl is added to allow the public key algorithm to examine or modify 2649 the PKCS#7 RecipientInfo structure if it needs to: for RSA this is 2650 a no op. 2651 [Steve Henson] 2652 2653 *) Add a ctrl to asn1 method to allow a public key algorithm to express 2654 a default digest type to use. In most cases this will be SHA1 but some 2655 algorithms (such as GOST) need to specify an alternative digest. The 2656 return value indicates how strong the prefernce is 1 means optional and 2657 2 is mandatory (that is it is the only supported type). Modify 2658 ASN1_item_sign() to accept a NULL digest argument to indicate it should 2659 use the default md. Update openssl utilities to use the default digest 2660 type for signing if it is not explicitly indicated. 2661 [Steve Henson] 2662 2663 *) Use OID cross reference table in ASN1_sign() and ASN1_verify(). New 2664 EVP_MD flag EVP_MD_FLAG_PKEY_METHOD_SIGNATURE. This uses the relevant 2665 signing method from the key type. This effectively removes the link 2666 between digests and public key types. 2667 [Steve Henson] 2668 2669 *) Add an OID cross reference table and utility functions. Its purpose is to 2670 translate between signature OIDs such as SHA1WithrsaEncryption and SHA1, 2671 rsaEncryption. This will allow some of the algorithm specific hackery 2672 needed to use the correct OID to be removed. 2673 [Steve Henson] 2674 2675 *) Remove algorithm specific dependencies when setting PKCS7_SIGNER_INFO 2676 structures for PKCS7_sign(). They are now set up by the relevant public 2677 key ASN1 method. 2678 [Steve Henson] 2679 2680 *) Add provisional EC pkey method with support for ECDSA and ECDH. 2681 [Steve Henson] 2682 2683 *) Add support for key derivation (agreement) in the API, DH method and 2684 pkeyutl. 2685 [Steve Henson] 2686 2687 *) Add DSA pkey method and DH pkey methods, extend DH ASN1 method to support 2688 public and private key formats. As a side effect these add additional 2689 command line functionality not previously available: DSA signatures can be 2690 generated and verified using pkeyutl and DH key support and generation in 2691 pkey, genpkey. 2692 [Steve Henson] 2693 2694 *) BeOS support. 2695 [Oliver Tappe <zooey@hirschkaefer.de>] 2696 2697 *) New make target "install_html_docs" installs HTML renditions of the 2698 manual pages. 2699 [Oliver Tappe <zooey@hirschkaefer.de>] 2700 2701 *) New utility "genpkey" this is analagous to "genrsa" etc except it can 2702 generate keys for any algorithm. Extend and update EVP_PKEY_METHOD to 2703 support key and parameter generation and add initial key generation 2704 functionality for RSA. 2705 [Steve Henson] 2706 2707 *) Add functions for main EVP_PKEY_method operations. The undocumented 2708 functions EVP_PKEY_{encrypt,decrypt} have been renamed to 2709 EVP_PKEY_{encrypt,decrypt}_old. 2710 [Steve Henson] 2711 2712 *) Initial definitions for EVP_PKEY_METHOD. This will be a high level public 2713 key API, doesn't do much yet. 2714 [Steve Henson] 2715 2716 *) New function EVP_PKEY_asn1_get0_info() to retrieve information about 2717 public key algorithms. New option to openssl utility: 2718 "list-public-key-algorithms" to print out info. 2719 [Steve Henson] 2720 2721 *) Implement the Supported Elliptic Curves Extension for 2722 ECC ciphersuites from draft-ietf-tls-ecc-12.txt. 2723 [Douglas Stebila] 2724 2725 *) Don't free up OIDs in OBJ_cleanup() if they are in use by EVP_MD or 2726 EVP_CIPHER structures to avoid later problems in EVP_cleanup(). 2727 [Steve Henson] 2728 2729 *) New utilities pkey and pkeyparam. These are similar to algorithm specific 2730 utilities such as rsa, dsa, dsaparam etc except they process any key 2731 type. 2732 [Steve Henson] 2733 2734 *) Transfer public key printing routines to EVP_PKEY_ASN1_METHOD. New 2735 functions EVP_PKEY_print_public(), EVP_PKEY_print_private(), 2736 EVP_PKEY_print_param() to print public key data from an EVP_PKEY 2737 structure. 2738 [Steve Henson] 2739 2740 *) Initial support for pluggable public key ASN1. 2741 De-spaghettify the public key ASN1 handling. Move public and private 2742 key ASN1 handling to a new EVP_PKEY_ASN1_METHOD structure. Relocate 2743 algorithm specific handling to a single module within the relevant 2744 algorithm directory. Add functions to allow (near) opaque processing 2745 of public and private key structures. 2746 [Steve Henson] 2747 2748 *) Implement the Supported Point Formats Extension for 2749 ECC ciphersuites from draft-ietf-tls-ecc-12.txt. 2750 [Douglas Stebila] 2751 2752 *) Add initial support for RFC 4279 PSK TLS ciphersuites. Add members 2753 for the psk identity [hint] and the psk callback functions to the 2754 SSL_SESSION, SSL and SSL_CTX structure. 2755 2756 New ciphersuites: 2757 PSK-RC4-SHA, PSK-3DES-EDE-CBC-SHA, PSK-AES128-CBC-SHA, 2758 PSK-AES256-CBC-SHA 2759 2760 New functions: 2761 SSL_CTX_use_psk_identity_hint 2762 SSL_get_psk_identity_hint 2763 SSL_get_psk_identity 2764 SSL_use_psk_identity_hint 2765 2766 [Mika Kousa and Pasi Eronen of Nokia Corporation] 2767 2768 *) Add RFC 3161 compliant time stamp request creation, response generation 2769 and response verification functionality. 2770 [Zolt��n Gl��zik <zglozik@opentsa.org>, The OpenTSA Project] 2771 2772 *) Add initial support for TLS extensions, specifically for the server_name 2773 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now 2774 have new members for a host name. The SSL data structure has an 2775 additional member SSL_CTX *initial_ctx so that new sessions can be 2776 stored in that context to allow for session resumption, even after the 2777 SSL has been switched to a new SSL_CTX in reaction to a client's 2778 server_name extension. 2779 2780 New functions (subject to change): 2781 2782 SSL_get_servername() 2783 SSL_get_servername_type() 2784 SSL_set_SSL_CTX() 2785 2786 New CTRL codes and macros (subject to change): 2787 2788 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB 2789 - SSL_CTX_set_tlsext_servername_callback() 2790 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG 2791 - SSL_CTX_set_tlsext_servername_arg() 2792 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name() 2793 2794 openssl s_client has a new '-servername ...' option. 2795 2796 openssl s_server has new options '-servername_host ...', '-cert2 ...', 2797 '-key2 ...', '-servername_fatal' (subject to change). This allows 2798 testing the HostName extension for a specific single host name ('-cert' 2799 and '-key' remain fallbacks for handshakes without HostName 2800 negotiation). If the unrecogninzed_name alert has to be sent, this by 2801 default is a warning; it becomes fatal with the '-servername_fatal' 2802 option. 2803 2804 [Peter Sylvester, Remy Allais, Christophe Renou] 2805 2806 *) Whirlpool hash implementation is added. 2807 [Andy Polyakov] 2808 2809 *) BIGNUM code on 64-bit SPARCv9 targets is switched from bn(64,64) to 2810 bn(64,32). Because of instruction set limitations it doesn't have 2811 any negative impact on performance. This was done mostly in order 2812 to make it possible to share assembler modules, such as bn_mul_mont 2813 implementations, between 32- and 64-bit builds without hassle. 2814 [Andy Polyakov] 2815 2816 *) Move code previously exiled into file crypto/ec/ec2_smpt.c 2817 to ec2_smpl.c, and no longer require the OPENSSL_EC_BIN_PT_COMP 2818 macro. 2819 [Bodo Moeller] 2820 2821 *) New candidate for BIGNUM assembler implementation, bn_mul_mont, 2822 dedicated Montgomery multiplication procedure, is introduced. 2823 BN_MONT_CTX is modified to allow bn_mul_mont to reach for higher 2824 "64-bit" performance on certain 32-bit targets. 2825 [Andy Polyakov] 2826 2827 *) New option SSL_OP_NO_COMP to disable use of compression selectively 2828 in SSL structures. New SSL ctrl to set maximum send fragment size. 2829 Save memory by seeting the I/O buffer sizes dynamically instead of 2830 using the maximum available value. 2831 [Steve Henson] 2832 2833 *) New option -V for 'openssl ciphers'. This prints the ciphersuite code 2834 in addition to the text details. 2835 [Bodo Moeller] 2836 2837 *) Very, very preliminary EXPERIMENTAL support for printing of general 2838 ASN1 structures. This currently produces rather ugly output and doesn't 2839 handle several customised structures at all. 2840 [Steve Henson] 2841 2842 *) Integrated support for PVK file format and some related formats such 2843 as MS PUBLICKEYBLOB and PRIVATEKEYBLOB. Command line switches to support 2844 these in the 'rsa' and 'dsa' utilities. 2845 [Steve Henson] 2846 2847 *) Support for PKCS#1 RSAPublicKey format on rsa utility command line. 2848 [Steve Henson] 2849 2850 *) Remove the ancient ASN1_METHOD code. This was only ever used in one 2851 place for the (very old) "NETSCAPE" format certificates which are now 2852 handled using new ASN1 code equivalents. 2853 [Steve Henson] 2854 2855 *) Let the TLSv1_method() etc. functions return a 'const' SSL_METHOD 2856 pointer and make the SSL_METHOD parameter in SSL_CTX_new, 2857 SSL_CTX_set_ssl_version and SSL_set_ssl_method 'const'. 2858 [Nils Larsch] 2859 2860 *) Modify CRL distribution points extension code to print out previously 2861 unsupported fields. Enhance extension setting code to allow setting of 2862 all fields. 2863 [Steve Henson] 2864 2865 *) Add print and set support for Issuing Distribution Point CRL extension. 2866 [Steve Henson] 2867 2868 *) Change 'Configure' script to enable Camellia by default. 2869 [NTT] 2870 2871 Changes between 0.9.8m and 0.9.8n [24 Mar 2010] 2872 2873 *) When rejecting SSL/TLS records due to an incorrect version number, never 2874 update s->server with a new major version number. As of 2875 - OpenSSL 0.9.8m if 'short' is a 16-bit type, 2876 - OpenSSL 0.9.8f if 'short' is longer than 16 bits, 2877 the previous behavior could result in a read attempt at NULL when 2878 receiving specific incorrect SSL/TLS records once record payload 2879 protection is active. (CVE-2010-0740) 2880 [Bodo Moeller, Adam Langley <agl@chromium.org>] 2881 2882 *) Fix for CVE-2010-0433 where some kerberos enabled versions of OpenSSL 2883 could be crashed if the relevant tables were not present (e.g. chrooted). 2884 [Tomas Hoger <thoger@redhat.com>] 2885 2886 Changes between 0.9.8l and 0.9.8m [25 Feb 2010] 2887 2888 *) Always check bn_wexpend() return values for failure. (CVE-2009-3245) 2889 [Martin Olsson, Neel Mehta] 2890 2891 *) Fix X509_STORE locking: Every 'objs' access requires a lock (to 2892 accommodate for stack sorting, always a write lock!). 2893 [Bodo Moeller] 2894 2895 *) On some versions of WIN32 Heap32Next is very slow. This can cause 2896 excessive delays in the RAND_poll(): over a minute. As a workaround 2897 include a time check in the inner Heap32Next loop too. 2898 [Steve Henson] 2899 2900 *) The code that handled flushing of data in SSL/TLS originally used the 2901 BIO_CTRL_INFO ctrl to see if any data was pending first. This caused 2902 the problem outlined in PR#1949. The fix suggested there however can 2903 trigger problems with buggy BIO_CTRL_WPENDING (e.g. some versions 2904 of Apache). So instead simplify the code to flush unconditionally. 2905 This should be fine since flushing with no data to flush is a no op. 2906 [Steve Henson] 2907 2908 *) Handle TLS versions 2.0 and later properly and correctly use the 2909 highest version of TLS/SSL supported. Although TLS >= 2.0 is some way 2910 off ancient servers have a habit of sticking around for a while... 2911 [Steve Henson] 2912 2913 *) Modify compression code so it frees up structures without using the 2914 ex_data callbacks. This works around a problem where some applications 2915 call CRYPTO_cleanup_all_ex_data() before application exit (e.g. when 2916 restarting) then use compression (e.g. SSL with compression) later. 2917 This results in significant per-connection memory leaks and 2918 has caused some security issues including CVE-2008-1678 and 2919 CVE-2009-4355. 2920 [Steve Henson] 2921 2922 *) Constify crypto/cast (i.e., <openssl/cast.h>): a CAST_KEY doesn't 2923 change when encrypting or decrypting. 2924 [Bodo Moeller] 2925 2926 *) Add option SSL_OP_LEGACY_SERVER_CONNECT which will allow clients to 2927 connect and renegotiate with servers which do not support RI. 2928 Until RI is more widely deployed this option is enabled by default. 2929 [Steve Henson] 2930 2931 *) Add "missing" ssl ctrls to clear options and mode. 2932 [Steve Henson] 2933 2934 *) If client attempts to renegotiate and doesn't support RI respond with 2935 a no_renegotiation alert as required by RFC5746. Some renegotiating 2936 TLS clients will continue a connection gracefully when they receive 2937 the alert. Unfortunately OpenSSL mishandled this alert and would hang 2938 waiting for a server hello which it will never receive. Now we treat a 2939 received no_renegotiation alert as a fatal error. This is because 2940 applications requesting a renegotiation might well expect it to succeed 2941 and would have no code in place to handle the server denying it so the 2942 only safe thing to do is to terminate the connection. 2943 [Steve Henson] 2944 2945 *) Add ctrl macro SSL_get_secure_renegotiation_support() which returns 1 if 2946 peer supports secure renegotiation and 0 otherwise. Print out peer 2947 renegotiation support in s_client/s_server. 2948 [Steve Henson] 2949 2950 *) Replace the highly broken and deprecated SPKAC certification method with 2951 the updated NID creation version. This should correctly handle UTF8. 2952 [Steve Henson] 2953 2954 *) Implement RFC5746. Re-enable renegotiation but require the extension 2955 as needed. Unfortunately, SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION 2956 turns out to be a bad idea. It has been replaced by 2957 SSL_OP_ALLOW_UNSAFE_LEGACY_RENEGOTIATION which can be set with 2958 SSL_CTX_set_options(). This is really not recommended unless you 2959 know what you are doing. 2960 [Eric Rescorla <ekr@networkresonance.com>, Ben Laurie, Steve Henson] 2961 2962 *) Fixes to stateless session resumption handling. Use initial_ctx when 2963 issuing and attempting to decrypt tickets in case it has changed during 2964 servername handling. Use a non-zero length session ID when attempting 2965 stateless session resumption: this makes it possible to determine if 2966 a resumption has occurred immediately after receiving server hello 2967 (several places in OpenSSL subtly assume this) instead of later in 2968 the handshake. 2969 [Steve Henson] 2970 2971 *) The functions ENGINE_ctrl(), OPENSSL_isservice(), 2972 CMS_get1_RecipientRequest() and RAND_bytes() can return <=0 on error 2973 fixes for a few places where the return code is not checked 2974 correctly. 2975 [Julia Lawall <julia@diku.dk>] 2976 2977 *) Add --strict-warnings option to Configure script to include devteam 2978 warnings in other configurations. 2979 [Steve Henson] 2980 2981 *) Add support for --libdir option and LIBDIR variable in makefiles. This 2982 makes it possible to install openssl libraries in locations which 2983 have names other than "lib", for example "/usr/lib64" which some 2984 systems need. 2985 [Steve Henson, based on patch from Jeremy Utley] 2986 2987 *) Don't allow the use of leading 0x80 in OIDs. This is a violation of 2988 X690 8.9.12 and can produce some misleading textual output of OIDs. 2989 [Steve Henson, reported by Dan Kaminsky] 2990 2991 *) Delete MD2 from algorithm tables. This follows the recommendation in 2992 several standards that it is not used in new applications due to 2993 several cryptographic weaknesses. For binary compatibility reasons 2994 the MD2 API is still compiled in by default. 2995 [Steve Henson] 2996 2997 *) Add compression id to {d2i,i2d}_SSL_SESSION so it is correctly saved 2998 and restored. 2999 [Steve Henson] 3000 3001 *) Rename uni2asc and asc2uni functions to OPENSSL_uni2asc and 3002 OPENSSL_asc2uni conditionally on Netware platforms to avoid a name 3003 clash. 3004 [Guenter <lists@gknw.net>] 3005 3006 *) Fix the server certificate chain building code to use X509_verify_cert(), 3007 it used to have an ad-hoc builder which was unable to cope with anything 3008 other than a simple chain. 3009 [David Woodhouse <dwmw2@infradead.org>, Steve Henson] 3010 3011 *) Don't check self signed certificate signatures in X509_verify_cert() 3012 by default (a flag can override this): it just wastes time without 3013 adding any security. As a useful side effect self signed root CAs 3014 with non-FIPS digests are now usable in FIPS mode. 3015 [Steve Henson] 3016 3017 *) In dtls1_process_out_of_seq_message() the check if the current message 3018 is already buffered was missing. For every new message was memory 3019 allocated, allowing an attacker to perform an denial of service attack 3020 with sending out of seq handshake messages until there is no memory 3021 left. Additionally every future messege was buffered, even if the 3022 sequence number made no sense and would be part of another handshake. 3023 So only messages with sequence numbers less than 10 in advance will be 3024 buffered. (CVE-2009-1378) 3025 [Robin Seggelmann, discovered by Daniel Mentz] 3026 3027 *) Records are buffered if they arrive with a future epoch to be 3028 processed after finishing the corresponding handshake. There is 3029 currently no limitation to this buffer allowing an attacker to perform 3030 a DOS attack with sending records with future epochs until there is no 3031 memory left. This patch adds the pqueue_size() function to detemine 3032 the size of a buffer and limits the record buffer to 100 entries. 3033 (CVE-2009-1377) 3034 [Robin Seggelmann, discovered by Daniel Mentz] 3035 3036 *) Keep a copy of frag->msg_header.frag_len so it can be used after the 3037 parent structure is freed. (CVE-2009-1379) 3038 [Daniel Mentz] 3039 3040 *) Handle non-blocking I/O properly in SSL_shutdown() call. 3041 [Darryl Miles <darryl-mailinglists@netbauds.net>] 3042 3043 *) Add 2.5.4.* OIDs 3044 [Ilya O. <vrghost@gmail.com>] 3045 3046 Changes between 0.9.8k and 0.9.8l [5 Nov 2009] 3047 3048 *) Disable renegotiation completely - this fixes a severe security 3049 problem (CVE-2009-3555) at the cost of breaking all 3050 renegotiation. Renegotiation can be re-enabled by setting 3051 SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION in s3->flags at 3052 run-time. This is really not recommended unless you know what 3053 you're doing. 3054 [Ben Laurie] 3055 3056 Changes between 0.9.8j and 0.9.8k [25 Mar 2009] 3057 3058 *) Don't set val to NULL when freeing up structures, it is freed up by 3059 underlying code. If sizeof(void *) > sizeof(long) this can result in 3060 zeroing past the valid field. (CVE-2009-0789) 3061 [Paolo Ganci <Paolo.Ganci@AdNovum.CH>] 3062 3063 *) Fix bug where return value of CMS_SignerInfo_verify_content() was not 3064 checked correctly. This would allow some invalid signed attributes to 3065 appear to verify correctly. (CVE-2009-0591) 3066 [Ivan Nestlerode <inestlerode@us.ibm.com>] 3067 3068 *) Reject UniversalString and BMPString types with invalid lengths. This 3069 prevents a crash in ASN1_STRING_print_ex() which assumes the strings have 3070 a legal length. (CVE-2009-0590) 3071 [Steve Henson] 3072 3073 *) Set S/MIME signing as the default purpose rather than setting it 3074 unconditionally. This allows applications to override it at the store 3075 level. 3076 [Steve Henson] 3077 3078 *) Permit restricted recursion of ASN1 strings. This is needed in practice 3079 to handle some structures. 3080 [Steve Henson] 3081 3082 *) Improve efficiency of mem_gets: don't search whole buffer each time 3083 for a '\n' 3084 [Jeremy Shapiro <jnshapir@us.ibm.com>] 3085 3086 *) New -hex option for openssl rand. 3087 [Matthieu Herrb] 3088 3089 *) Print out UTF8String and NumericString when parsing ASN1. 3090 [Steve Henson] 3091 3092 *) Support NumericString type for name components. 3093 [Steve Henson] 3094 3095 *) Allow CC in the environment to override the automatically chosen 3096 compiler. Note that nothing is done to ensure flags work with the 3097 chosen compiler. 3098 [Ben Laurie] 3099 3100 Changes between 0.9.8i and 0.9.8j [07 Jan 2009] 3101 3102 *) Properly check EVP_VerifyFinal() and similar return values 3103 (CVE-2008-5077). 3104 [Ben Laurie, Bodo Moeller, Google Security Team] 3105 3106 *) Enable TLS extensions by default. 3107 [Ben Laurie] 3108 3109 *) Allow the CHIL engine to be loaded, whether the application is 3110 multithreaded or not. (This does not release the developer from the 3111 obligation to set up the dynamic locking callbacks.) 3112 [Sander Temme <sander@temme.net>] 3113 3114 *) Use correct exit code if there is an error in dgst command. 3115 [Steve Henson; problem pointed out by Roland Dirlewanger] 3116 3117 *) Tweak Configure so that you need to say "experimental-jpake" to enable 3118 JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications. 3119 [Bodo Moeller] 3120 3121 *) Add experimental JPAKE support, including demo authentication in 3122 s_client and s_server. 3123 [Ben Laurie] 3124 3125 *) Set the comparison function in v3_addr_canonize(). 3126 [Rob Austein <sra@hactrn.net>] 3127 3128 *) Add support for XMPP STARTTLS in s_client. 3129 [Philip Paeps <philip@freebsd.org>] 3130 3131 *) Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior 3132 to ensure that even with this option, only ciphersuites in the 3133 server's preference list will be accepted. (Note that the option 3134 applies only when resuming a session, so the earlier behavior was 3135 just about the algorithm choice for symmetric cryptography.) 3136 [Bodo Moeller] 3137 3138 Changes between 0.9.8h and 0.9.8i [15 Sep 2008] 3139 3140 *) Fix NULL pointer dereference if a DTLS server received 3141 ChangeCipherSpec as first record (CVE-2009-1386). 3142 [PR #1679] 3143 3144 *) Fix a state transitition in s3_srvr.c and d1_srvr.c 3145 (was using SSL3_ST_CW_CLNT_HELLO_B, should be ..._ST_SW_SRVR_...). 3146 [Nagendra Modadugu] 3147 3148 *) The fix in 0.9.8c that supposedly got rid of unsafe 3149 double-checked locking was incomplete for RSA blinding, 3150 addressing just one layer of what turns out to have been 3151 doubly unsafe triple-checked locking. 3152 3153 So now fix this for real by retiring the MONT_HELPER macro 3154 in crypto/rsa/rsa_eay.c. 3155 3156 [Bodo Moeller; problem pointed out by Marius Schilder] 3157 3158 *) Various precautionary measures: 3159 3160 - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h). 3161 3162 - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c). 3163 (NB: This would require knowledge of the secret session ticket key 3164 to exploit, in which case you'd be SOL either way.) 3165 3166 - Change bn_nist.c so that it will properly handle input BIGNUMs 3167 outside the expected range. 3168 3169 - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG 3170 builds. 3171 3172 [Neel Mehta, Bodo Moeller] 3173 3174 *) Allow engines to be "soft loaded" - i.e. optionally don't die if 3175 the load fails. Useful for distros. 3176 [Ben Laurie and the FreeBSD team] 3177 3178 *) Add support for Local Machine Keyset attribute in PKCS#12 files. 3179 [Steve Henson] 3180 3181 *) Fix BN_GF2m_mod_arr() top-bit cleanup code. 3182 [Huang Ying] 3183 3184 *) Expand ENGINE to support engine supplied SSL client certificate functions. 3185 3186 This work was sponsored by Logica. 3187 [Steve Henson] 3188 3189 *) Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows 3190 keystores. Support for SSL/TLS client authentication too. 3191 Not compiled unless enable-capieng specified to Configure. 3192 3193 This work was sponsored by Logica. 3194 [Steve Henson] 3195 3196 *) Fix bug in X509_ATTRIBUTE creation: dont set attribute using 3197 ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain 3198 attribute creation routines such as certifcate requests and PKCS#12 3199 files. 3200 [Steve Henson] 3201 3202 Changes between 0.9.8g and 0.9.8h [28 May 2008] 3203 3204 *) Fix flaw if 'Server Key exchange message' is omitted from a TLS 3205 handshake which could lead to a cilent crash as found using the 3206 Codenomicon TLS test suite (CVE-2008-1672) 3207 [Steve Henson, Mark Cox] 3208 3209 *) Fix double free in TLS server name extensions which could lead to 3210 a remote crash found by Codenomicon TLS test suite (CVE-2008-0891) 3211 [Joe Orton] 3212 3213 *) Clear error queue in SSL_CTX_use_certificate_chain_file() 3214 3215 Clear the error queue to ensure that error entries left from 3216 older function calls do not interfere with the correct operation. 3217 [Lutz Jaenicke, Erik de Castro Lopo] 3218 3219 *) Remove root CA certificates of commercial CAs: 3220 3221 The OpenSSL project does not recommend any specific CA and does not 3222 have any policy with respect to including or excluding any CA. 3223 Therefore it does not make any sense to ship an arbitrary selection 3224 of root CA certificates with the OpenSSL software. 3225 [Lutz Jaenicke] 3226 3227 *) RSA OAEP patches to fix two separate invalid memory reads. 3228 The first one involves inputs when 'lzero' is greater than 3229 'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes 3230 before the beginning of from). The second one involves inputs where 3231 the 'db' section contains nothing but zeroes (there is a one-byte 3232 invalid read after the end of 'db'). 3233 [Ivan Nestlerode <inestlerode@us.ibm.com>] 3234 3235 *) Partial backport from 0.9.9-dev: 3236 3237 Introduce bn_mul_mont (dedicated Montgomery multiplication 3238 procedure) as a candidate for BIGNUM assembler implementation. 3239 While 0.9.9-dev uses assembler for various architectures, only 3240 x86_64 is available by default here in the 0.9.8 branch, and 3241 32-bit x86 is available through a compile-time setting. 3242 3243 To try the 32-bit x86 assembler implementation, use Configure 3244 option "enable-montasm" (which exists only for this backport). 3245 3246 As "enable-montasm" for 32-bit x86 disclaims code stability 3247 anyway, in this constellation we activate additional code 3248 backported from 0.9.9-dev for further performance improvements, 3249 namely BN_from_montgomery_word. (To enable this otherwise, 3250 e.g. x86_64, try "-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD".) 3251 3252 [Andy Polyakov (backport partially by Bodo Moeller)] 3253 3254 *) Add TLS session ticket callback. This allows an application to set 3255 TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed 3256 values. This is useful for key rollover for example where several key 3257 sets may exist with different names. 3258 [Steve Henson] 3259 3260 *) Reverse ENGINE-internal logic for caching default ENGINE handles. 3261 This was broken until now in 0.9.8 releases, such that the only way 3262 a registered ENGINE could be used (assuming it initialises 3263 successfully on the host) was to explicitly set it as the default 3264 for the relevant algorithms. This is in contradiction with 0.9.7 3265 behaviour and the documentation. With this fix, when an ENGINE is 3266 registered into a given algorithm's table of implementations, the 3267 'uptodate' flag is reset so that auto-discovery will be used next 3268 time a new context for that algorithm attempts to select an 3269 implementation. 3270 [Ian Lister (tweaked by Geoff Thorpe)] 3271 3272 *) Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9 3273 implemention in the following ways: 3274 3275 Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be 3276 hard coded. 3277 3278 Lack of BER streaming support means one pass streaming processing is 3279 only supported if data is detached: setting the streaming flag is 3280 ignored for embedded content. 3281 3282 CMS support is disabled by default and must be explicitly enabled 3283 with the enable-cms configuration option. 3284 [Steve Henson] 3285 3286 *) Update the GMP engine glue to do direct copies between BIGNUM and 3287 mpz_t when openssl and GMP use the same limb size. Otherwise the 3288 existing "conversion via a text string export" trick is still used. 3289 [Paul Sheer <paulsheer@gmail.com>] 3290 3291 *) Zlib compression BIO. This is a filter BIO which compressed and 3292 uncompresses any data passed through it. 3293 [Steve Henson] 3294 3295 *) Add AES_wrap_key() and AES_unwrap_key() functions to implement 3296 RFC3394 compatible AES key wrapping. 3297 [Steve Henson] 3298 3299 *) Add utility functions to handle ASN1 structures. ASN1_STRING_set0(): 3300 sets string data without copying. X509_ALGOR_set0() and 3301 X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier) 3302 data. Attribute function X509at_get0_data_by_OBJ(): retrieves data 3303 from an X509_ATTRIBUTE structure optionally checking it occurs only 3304 once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied 3305 data. 3306 [Steve Henson] 3307 3308 *) Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set() 3309 to get the expected BN_FLG_CONSTTIME behavior. 3310 [Bodo Moeller (Google)] 3311 3312 *) Netware support: 3313 3314 - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets 3315 - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT) 3316 - added some more tests to do_tests.pl 3317 - fixed RunningProcess usage so that it works with newer LIBC NDKs too 3318 - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency 3319 - added new Configure targets netware-clib-bsdsock, netware-clib-gcc, 3320 netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc 3321 - various changes to netware.pl to enable gcc-cross builds on Win32 3322 platform 3323 - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD) 3324 - various changes to fix missing prototype warnings 3325 - fixed x86nasm.pl to create correct asm files for NASM COFF output 3326 - added AES, WHIRLPOOL and CPUID assembler code to build files 3327 - added missing AES assembler make rules to mk1mf.pl 3328 - fixed order of includes in apps/ocsp.c so that e_os.h settings apply 3329 [Guenter Knauf <eflash@gmx.net>] 3330 3331 *) Implement certificate status request TLS extension defined in RFC3546. 3332 A client can set the appropriate parameters and receive the encoded 3333 OCSP response via a callback. A server can query the supplied parameters 3334 and set the encoded OCSP response in the callback. Add simplified examples 3335 to s_client and s_server. 3336 [Steve Henson] 3337 3338 Changes between 0.9.8f and 0.9.8g [19 Oct 2007] 3339 3340 *) Fix various bugs: 3341 + Binary incompatibility of ssl_ctx_st structure 3342 + DTLS interoperation with non-compliant servers 3343 + Don't call get_session_cb() without proposed session 3344 + Fix ia64 assembler code 3345 [Andy Polyakov, Steve Henson] 3346 3347 Changes between 0.9.8e and 0.9.8f [11 Oct 2007] 3348 3349 *) DTLS Handshake overhaul. There were longstanding issues with 3350 OpenSSL DTLS implementation, which were making it impossible for 3351 RFC 4347 compliant client to communicate with OpenSSL server. 3352 Unfortunately just fixing these incompatibilities would "cut off" 3353 pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e 3354 server keeps tolerating non RFC compliant syntax. The opposite is 3355 not true, 0.9.8f client can not communicate with earlier server. 3356 This update even addresses CVE-2007-4995. 3357 [Andy Polyakov] 3358 3359 *) Changes to avoid need for function casts in OpenSSL: some compilers 3360 (gcc 4.2 and later) reject their use. 3361 [Kurt Roeckx <kurt@roeckx.be>, Peter Hartley <pdh@utter.chaos.org.uk>, 3362 Steve Henson] 3363 3364 *) Add RFC4507 support to OpenSSL. This includes the corrections in 3365 RFC4507bis. The encrypted ticket format is an encrypted encoded 3366 SSL_SESSION structure, that way new session features are automatically 3367 supported. 3368 3369 If a client application caches session in an SSL_SESSION structure 3370 support is transparent because tickets are now stored in the encoded 3371 SSL_SESSION. 3372 3373 The SSL_CTX structure automatically generates keys for ticket 3374 protection in servers so again support should be possible 3375 with no application modification. 3376 3377 If a client or server wishes to disable RFC4507 support then the option 3378 SSL_OP_NO_TICKET can be set. 3379 3380 Add a TLS extension debugging callback to allow the contents of any client 3381 or server extensions to be examined. 3382 3383 This work was sponsored by Google. 3384 [Steve Henson] 3385 3386 *) Add initial support for TLS extensions, specifically for the server_name 3387 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now 3388 have new members for a host name. The SSL data structure has an 3389 additional member SSL_CTX *initial_ctx so that new sessions can be 3390 stored in that context to allow for session resumption, even after the 3391 SSL has been switched to a new SSL_CTX in reaction to a client's 3392 server_name extension. 3393 3394 New functions (subject to change): 3395 3396 SSL_get_servername() 3397 SSL_get_servername_type() 3398 SSL_set_SSL_CTX() 3399 3400 New CTRL codes and macros (subject to change): 3401 3402 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB 3403 - SSL_CTX_set_tlsext_servername_callback() 3404 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG 3405 - SSL_CTX_set_tlsext_servername_arg() 3406 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name() 3407 3408 openssl s_client has a new '-servername ...' option. 3409 3410 openssl s_server has new options '-servername_host ...', '-cert2 ...', 3411 '-key2 ...', '-servername_fatal' (subject to change). This allows 3412 testing the HostName extension for a specific single host name ('-cert' 3413 and '-key' remain fallbacks for handshakes without HostName 3414 negotiation). If the unrecogninzed_name alert has to be sent, this by 3415 default is a warning; it becomes fatal with the '-servername_fatal' 3416 option. 3417 3418 [Peter Sylvester, Remy Allais, Christophe Renou, Steve Henson] 3419 3420 *) Add AES and SSE2 assembly language support to VC++ build. 3421 [Steve Henson] 3422 3423 *) Mitigate attack on final subtraction in Montgomery reduction. 3424 [Andy Polyakov] 3425 3426 *) Fix crypto/ec/ec_mult.c to work properly with scalars of value 0 3427 (which previously caused an internal error). 3428 [Bodo Moeller] 3429 3430 *) Squeeze another 10% out of IGE mode when in != out. 3431 [Ben Laurie] 3432 3433 *) AES IGE mode speedup. 3434 [Dean Gaudet (Google)] 3435 3436 *) Add the Korean symmetric 128-bit cipher SEED (see 3437 http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp) and 3438 add SEED ciphersuites from RFC 4162: 3439 3440 TLS_RSA_WITH_SEED_CBC_SHA = "SEED-SHA" 3441 TLS_DHE_DSS_WITH_SEED_CBC_SHA = "DHE-DSS-SEED-SHA" 3442 TLS_DHE_RSA_WITH_SEED_CBC_SHA = "DHE-RSA-SEED-SHA" 3443 TLS_DH_anon_WITH_SEED_CBC_SHA = "ADH-SEED-SHA" 3444 3445 To minimize changes between patchlevels in the OpenSSL 0.9.8 3446 series, SEED remains excluded from compilation unless OpenSSL 3447 is configured with 'enable-seed'. 3448 [KISA, Bodo Moeller] 3449 3450 *) Mitigate branch prediction attacks, which can be practical if a 3451 single processor is shared, allowing a spy process to extract 3452 information. For detailed background information, see 3453 http://eprint.iacr.org/2007/039 (O. Aciicmez, S. Gueron, 3454 J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL 3455 and Necessary Software Countermeasures"). The core of the change 3456 are new versions BN_div_no_branch() and 3457 BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(), 3458 respectively, which are slower, but avoid the security-relevant 3459 conditional branches. These are automatically called by BN_div() 3460 and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one 3461 of the input BIGNUMs. Also, BN_is_bit_set() has been changed to 3462 remove a conditional branch. 3463 3464 BN_FLG_CONSTTIME is the new name for the previous 3465 BN_FLG_EXP_CONSTTIME flag, since it now affects more than just 3466 modular exponentiation. (Since OpenSSL 0.9.7h, setting this flag 3467 in the exponent causes BN_mod_exp_mont() to use the alternative 3468 implementation in BN_mod_exp_mont_consttime().) The old name 3469 remains as a deprecated alias. 3470 3471 Similary, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general 3472 RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses 3473 constant-time implementations for more than just exponentiation. 3474 Here too the old name is kept as a deprecated alias. 3475 3476 BN_BLINDING_new() will now use BN_dup() for the modulus so that 3477 the BN_BLINDING structure gets an independent copy of the 3478 modulus. This means that the previous "BIGNUM *m" argument to 3479 BN_BLINDING_new() and to BN_BLINDING_create_param() now 3480 essentially becomes "const BIGNUM *m", although we can't actually 3481 change this in the header file before 0.9.9. It allows 3482 RSA_setup_blinding() to use BN_with_flags() on the modulus to 3483 enable BN_FLG_CONSTTIME. 3484 3485 [Matthew D Wood (Intel Corp)] 3486 3487 *) In the SSL/TLS server implementation, be strict about session ID 3488 context matching (which matters if an application uses a single 3489 external cache for different purposes). Previously, 3490 out-of-context reuse was forbidden only if SSL_VERIFY_PEER was 3491 set. This did ensure strict client verification, but meant that, 3492 with applications using a single external cache for quite 3493 different requirements, clients could circumvent ciphersuite 3494 restrictions for a given session ID context by starting a session 3495 in a different context. 3496 [Bodo Moeller] 3497 3498 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 3499 a ciphersuite string such as "DEFAULT:RSA" cannot enable 3500 authentication-only ciphersuites. 3501 [Bodo Moeller] 3502 3503 *) Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was 3504 not complete and could lead to a possible single byte overflow 3505 (CVE-2007-5135) [Ben Laurie] 3506 3507 Changes between 0.9.8d and 0.9.8e [23 Feb 2007] 3508 3509 *) Since AES128 and AES256 (and similarly Camellia128 and 3510 Camellia256) share a single mask bit in the logic of 3511 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 3512 kludge to work properly if AES128 is available and AES256 isn't 3513 (or if Camellia128 is available and Camellia256 isn't). 3514 [Victor Duchovni] 3515 3516 *) Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c 3517 (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters): 3518 When a point or a seed is encoded in a BIT STRING, we need to 3519 prevent the removal of trailing zero bits to get the proper DER 3520 encoding. (By default, crypto/asn1/a_bitstr.c assumes the case 3521 of a NamedBitList, for which trailing 0 bits need to be removed.) 3522 [Bodo Moeller] 3523 3524 *) Have SSL/TLS server implementation tolerate "mismatched" record 3525 protocol version while receiving ClientHello even if the 3526 ClientHello is fragmented. (The server can't insist on the 3527 particular protocol version it has chosen before the ServerHello 3528 message has informed the client about his choice.) 3529 [Bodo Moeller] 3530 3531 *) Add RFC 3779 support. 3532 [Rob Austein for ARIN, Ben Laurie] 3533 3534 *) Load error codes if they are not already present instead of using a 3535 static variable. This allows them to be cleanly unloaded and reloaded. 3536 Improve header file function name parsing. 3537 [Steve Henson] 3538 3539 *) extend SMTP and IMAP protocol emulation in s_client to use EHLO 3540 or CAPABILITY handshake as required by RFCs. 3541 [Goetz Babin-Ebell] 3542 3543 Changes between 0.9.8c and 0.9.8d [28 Sep 2006] 3544 3545 *) Introduce limits to prevent malicious keys being able to 3546 cause a denial of service. (CVE-2006-2940) 3547 [Steve Henson, Bodo Moeller] 3548 3549 *) Fix ASN.1 parsing of certain invalid structures that can result 3550 in a denial of service. (CVE-2006-2937) [Steve Henson] 3551 3552 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 3553 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 3554 3555 *) Fix SSL client code which could crash if connecting to a 3556 malicious SSLv2 server. (CVE-2006-4343) 3557 [Tavis Ormandy and Will Drewry, Google Security Team] 3558 3559 *) Since 0.9.8b, ciphersuite strings naming explicit ciphersuites 3560 match only those. Before that, "AES256-SHA" would be interpreted 3561 as a pattern and match "AES128-SHA" too (since AES128-SHA got 3562 the same strength classification in 0.9.7h) as we currently only 3563 have a single AES bit in the ciphersuite description bitmap. 3564 That change, however, also applied to ciphersuite strings such as 3565 "RC4-MD5" that intentionally matched multiple ciphersuites -- 3566 namely, SSL 2.0 ciphersuites in addition to the more common ones 3567 from SSL 3.0/TLS 1.0. 3568 3569 So we change the selection algorithm again: Naming an explicit 3570 ciphersuite selects this one ciphersuite, and any other similar 3571 ciphersuite (same bitmap) from *other* protocol versions. 3572 Thus, "RC4-MD5" again will properly select both the SSL 2.0 3573 ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite. 3574 3575 Since SSL 2.0 does not have any ciphersuites for which the 3576 128/256 bit distinction would be relevant, this works for now. 3577 The proper fix will be to use different bits for AES128 and 3578 AES256, which would have avoided the problems from the beginning; 3579 however, bits are scarce, so we can only do this in a new release 3580 (not just a patchlevel) when we can change the SSL_CIPHER 3581 definition to split the single 'unsigned long mask' bitmap into 3582 multiple values to extend the available space. 3583 3584 [Bodo Moeller] 3585 3586 Changes between 0.9.8b and 0.9.8c [05 Sep 2006] 3587 3588 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 3589 (CVE-2006-4339) [Ben Laurie and Google Security Team] 3590 3591 *) Add AES IGE and biIGE modes. 3592 [Ben Laurie] 3593 3594 *) Change the Unix randomness entropy gathering to use poll() when 3595 possible instead of select(), since the latter has some 3596 undesirable limitations. 3597 [Darryl Miles via Richard Levitte and Bodo Moeller] 3598 3599 *) Disable "ECCdraft" ciphersuites more thoroughly. Now special 3600 treatment in ssl/ssl_ciph.s makes sure that these ciphersuites 3601 cannot be implicitly activated as part of, e.g., the "AES" alias. 3602 However, please upgrade to OpenSSL 0.9.9[-dev] for 3603 non-experimental use of the ECC ciphersuites to get TLS extension 3604 support, which is required for curve and point format negotiation 3605 to avoid potential handshake problems. 3606 [Bodo Moeller] 3607 3608 *) Disable rogue ciphersuites: 3609 3610 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 3611 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 3612 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 3613 3614 The latter two were purportedly from 3615 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 3616 appear there. 3617 3618 Also deactivate the remaining ciphersuites from 3619 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 3620 unofficial, and the ID has long expired. 3621 [Bodo Moeller] 3622 3623 *) Fix RSA blinding Heisenbug (problems sometimes occured on 3624 dual-core machines) and other potential thread-safety issues. 3625 [Bodo Moeller] 3626 3627 *) Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key 3628 versions), which is now available for royalty-free use 3629 (see http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html). 3630 Also, add Camellia TLS ciphersuites from RFC 4132. 3631 3632 To minimize changes between patchlevels in the OpenSSL 0.9.8 3633 series, Camellia remains excluded from compilation unless OpenSSL 3634 is configured with 'enable-camellia'. 3635 [NTT] 3636 3637 *) Disable the padding bug check when compression is in use. The padding 3638 bug check assumes the first packet is of even length, this is not 3639 necessarily true if compresssion is enabled and can result in false 3640 positives causing handshake failure. The actual bug test is ancient 3641 code so it is hoped that implementations will either have fixed it by 3642 now or any which still have the bug do not support compression. 3643 [Steve Henson] 3644 3645 Changes between 0.9.8a and 0.9.8b [04 May 2006] 3646 3647 *) When applying a cipher rule check to see if string match is an explicit 3648 cipher suite and only match that one cipher suite if it is. 3649 [Steve Henson] 3650 3651 *) Link in manifests for VC++ if needed. 3652 [Austin Ziegler <halostatue@gmail.com>] 3653 3654 *) Update support for ECC-based TLS ciphersuites according to 3655 draft-ietf-tls-ecc-12.txt with proposed changes (but without 3656 TLS extensions, which are supported starting with the 0.9.9 3657 branch, not in the OpenSSL 0.9.8 branch). 3658 [Douglas Stebila] 3659 3660 *) New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support 3661 opaque EVP_CIPHER_CTX handling. 3662 [Steve Henson] 3663 3664 *) Fixes and enhancements to zlib compression code. We now only use 3665 "zlib1.dll" and use the default __cdecl calling convention on Win32 3666 to conform with the standards mentioned here: 3667 http://www.zlib.net/DLL_FAQ.txt 3668 Static zlib linking now works on Windows and the new --with-zlib-include 3669 --with-zlib-lib options to Configure can be used to supply the location 3670 of the headers and library. Gracefully handle case where zlib library 3671 can't be loaded. 3672 [Steve Henson] 3673 3674 *) Several fixes and enhancements to the OID generation code. The old code 3675 sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't 3676 handle numbers larger than ULONG_MAX, truncated printing and had a 3677 non standard OBJ_obj2txt() behaviour. 3678 [Steve Henson] 3679 3680 *) Add support for building of engines under engine/ as shared libraries 3681 under VC++ build system. 3682 [Steve Henson] 3683 3684 *) Corrected the numerous bugs in the Win32 path splitter in DSO. 3685 Hopefully, we will not see any false combination of paths any more. 3686 [Richard Levitte] 3687 3688 Changes between 0.9.8 and 0.9.8a [11 Oct 2005] 3689 3690 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 3691 (part of SSL_OP_ALL). This option used to disable the 3692 countermeasure against man-in-the-middle protocol-version 3693 rollback in the SSL 2.0 server implementation, which is a bad 3694 idea. (CVE-2005-2969) 3695 3696 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 3697 for Information Security, National Institute of Advanced Industrial 3698 Science and Technology [AIST], Japan)] 3699 3700 *) Add two function to clear and return the verify parameter flags. 3701 [Steve Henson] 3702 3703 *) Keep cipherlists sorted in the source instead of sorting them at 3704 runtime, thus removing the need for a lock. 3705 [Nils Larsch] 3706 3707 *) Avoid some small subgroup attacks in Diffie-Hellman. 3708 [Nick Mathewson and Ben Laurie] 3709 3710 *) Add functions for well-known primes. 3711 [Nick Mathewson] 3712 3713 *) Extended Windows CE support. 3714 [Satoshi Nakamura and Andy Polyakov] 3715 3716 *) Initialize SSL_METHOD structures at compile time instead of during 3717 runtime, thus removing the need for a lock. 3718 [Steve Henson] 3719 3720 *) Make PKCS7_decrypt() work even if no certificate is supplied by 3721 attempting to decrypt each encrypted key in turn. Add support to 3722 smime utility. 3723 [Steve Henson] 3724 3725 Changes between 0.9.7h and 0.9.8 [05 Jul 2005] 3726 3727 [NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after 3728 OpenSSL 0.9.8.] 3729 3730 *) Add libcrypto.pc and libssl.pc for those who feel they need them. 3731 [Richard Levitte] 3732 3733 *) Change CA.sh and CA.pl so they don't bundle the CSR and the private 3734 key into the same file any more. 3735 [Richard Levitte] 3736 3737 *) Add initial support for Win64, both IA64 and AMD64/x64 flavors. 3738 [Andy Polyakov] 3739 3740 *) Add -utf8 command line and config file option to 'ca'. 3741 [Stefan <stf@udoma.org] 3742 3743 *) Removed the macro des_crypt(), as it seems to conflict with some 3744 libraries. Use DES_crypt(). 3745 [Richard Levitte] 3746 3747 *) Correct naming of the 'chil' and '4758cca' ENGINEs. This 3748 involves renaming the source and generated shared-libs for 3749 both. The engines will accept the corrected or legacy ids 3750 ('ncipher' and '4758_cca' respectively) when binding. NB, 3751 this only applies when building 'shared'. 3752 [Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe] 3753 3754 *) Add attribute functions to EVP_PKEY structure. Modify 3755 PKCS12_create() to recognize a CSP name attribute and 3756 use it. Make -CSP option work again in pkcs12 utility. 3757 [Steve Henson] 3758 3759 *) Add new functionality to the bn blinding code: 3760 - automatic re-creation of the BN_BLINDING parameters after 3761 a fixed number of uses (currently 32) 3762 - add new function for parameter creation 3763 - introduce flags to control the update behaviour of the 3764 BN_BLINDING parameters 3765 - hide BN_BLINDING structure 3766 Add a second BN_BLINDING slot to the RSA structure to improve 3767 performance when a single RSA object is shared among several 3768 threads. 3769 [Nils Larsch] 3770 3771 *) Add support for DTLS. 3772 [Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie] 3773 3774 *) Add support for DER encoded private keys (SSL_FILETYPE_ASN1) 3775 to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file() 3776 [Walter Goulet] 3777 3778 *) Remove buggy and incompletet DH cert support from 3779 ssl/ssl_rsa.c and ssl/s3_both.c 3780 [Nils Larsch] 3781 3782 *) Use SHA-1 instead of MD5 as the default digest algorithm for 3783 the apps/openssl applications. 3784 [Nils Larsch] 3785 3786 *) Compile clean with "-Wall -Wmissing-prototypes 3787 -Wstrict-prototypes -Wmissing-declarations -Werror". Currently 3788 DEBUG_SAFESTACK must also be set. 3789 [Ben Laurie] 3790 3791 *) Change ./Configure so that certain algorithms can be disabled by default. 3792 The new counterpiece to "no-xxx" is "enable-xxx". 3793 3794 The patented RC5 and MDC2 algorithms will now be disabled unless 3795 "enable-rc5" and "enable-mdc2", respectively, are specified. 3796 3797 (IDEA remains enabled despite being patented. This is because IDEA 3798 is frequently required for interoperability, and there is no license 3799 fee for non-commercial use. As before, "no-idea" can be used to 3800 avoid this algorithm.) 3801 3802 [Bodo Moeller] 3803 3804 *) Add processing of proxy certificates (see RFC 3820). This work was 3805 sponsored by KTH (The Royal Institute of Technology in Stockholm) and 3806 EGEE (Enabling Grids for E-science in Europe). 3807 [Richard Levitte] 3808 3809 *) RC4 performance overhaul on modern architectures/implementations, such 3810 as Intel P4, IA-64 and AMD64. 3811 [Andy Polyakov] 3812 3813 *) New utility extract-section.pl. This can be used specify an alternative 3814 section number in a pod file instead of having to treat each file as 3815 a separate case in Makefile. This can be done by adding two lines to the 3816 pod file: 3817 3818 =for comment openssl_section:XXX 3819 3820 The blank line is mandatory. 3821 3822 [Steve Henson] 3823 3824 *) New arguments -certform, -keyform and -pass for s_client and s_server 3825 to allow alternative format key and certificate files and passphrase 3826 sources. 3827 [Steve Henson] 3828 3829 *) New structure X509_VERIFY_PARAM which combines current verify parameters, 3830 update associated structures and add various utility functions. 3831 3832 Add new policy related verify parameters, include policy checking in 3833 standard verify code. Enhance 'smime' application with extra parameters 3834 to support policy checking and print out. 3835 [Steve Henson] 3836 3837 *) Add a new engine to support VIA PadLock ACE extensions in the VIA C3 3838 Nehemiah processors. These extensions support AES encryption in hardware 3839 as well as RNG (though RNG support is currently disabled). 3840 [Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov] 3841 3842 *) Deprecate BN_[get|set]_params() functions (they were ignored internally). 3843 [Geoff Thorpe] 3844 3845 *) New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented. 3846 [Andy Polyakov and a number of other people] 3847 3848 *) Improved PowerPC platform support. Most notably BIGNUM assembler 3849 implementation contributed by IBM. 3850 [Suresh Chari, Peter Waltenberg, Andy Polyakov] 3851 3852 *) The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public 3853 exponent rather than 'unsigned long'. There is a corresponding change to 3854 the new 'rsa_keygen' element of the RSA_METHOD structure. 3855 [Jelte Jansen, Geoff Thorpe] 3856 3857 *) Functionality for creating the initial serial number file is now 3858 moved from CA.pl to the 'ca' utility with a new option -create_serial. 3859 3860 (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial 3861 number file to 1, which is bound to cause problems. To avoid 3862 the problems while respecting compatibility between different 0.9.7 3863 patchlevels, 0.9.7e employed 'openssl x509 -next_serial' in 3864 CA.pl for serial number initialization. With the new release 0.9.8, 3865 we can fix the problem directly in the 'ca' utility.) 3866 [Steve Henson] 3867 3868 *) Reduced header interdepencies by declaring more opaque objects in 3869 ossl_typ.h. As a consequence, including some headers (eg. engine.h) will 3870 give fewer recursive includes, which could break lazy source code - so 3871 this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always, 3872 developers should define this symbol when building and using openssl to 3873 ensure they track the recommended behaviour, interfaces, [etc], but 3874 backwards-compatible behaviour prevails when this isn't defined. 3875 [Geoff Thorpe] 3876 3877 *) New function X509_POLICY_NODE_print() which prints out policy nodes. 3878 [Steve Henson] 3879 3880 *) Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality. 3881 This will generate a random key of the appropriate length based on the 3882 cipher context. The EVP_CIPHER can provide its own random key generation 3883 routine to support keys of a specific form. This is used in the des and 3884 3des routines to generate a key of the correct parity. Update S/MIME 3885 code to use new functions and hence generate correct parity DES keys. 3886 Add EVP_CHECK_DES_KEY #define to return an error if the key is not 3887 valid (weak or incorrect parity). 3888 [Steve Henson] 3889 3890 *) Add a local set of CRLs that can be used by X509_verify_cert() as well 3891 as looking them up. This is useful when the verified structure may contain 3892 CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs 3893 present unless the new PKCS7_NO_CRL flag is asserted. 3894 [Steve Henson] 3895 3896 *) Extend ASN1 oid configuration module. It now additionally accepts the 3897 syntax: 3898 3899 shortName = some long name, 1.2.3.4 3900 [Steve Henson] 3901 3902 *) Reimplemented the BN_CTX implementation. There is now no more static 3903 limitation on the number of variables it can handle nor the depth of the 3904 "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack 3905 information can now expand as required, and rather than having a single 3906 static array of bignums, BN_CTX now uses a linked-list of such arrays 3907 allowing it to expand on demand whilst maintaining the usefulness of 3908 BN_CTX's "bundling". 3909 [Geoff Thorpe] 3910 3911 *) Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD 3912 to allow all RSA operations to function using a single BN_CTX. 3913 [Geoff Thorpe] 3914 3915 *) Preliminary support for certificate policy evaluation and checking. This 3916 is initially intended to pass the tests outlined in "Conformance Testing 3917 of Relying Party Client Certificate Path Processing Logic" v1.07. 3918 [Steve Henson] 3919 3920 *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and 3921 remained unused and not that useful. A variety of other little bignum 3922 tweaks and fixes have also been made continuing on from the audit (see 3923 below). 3924 [Geoff Thorpe] 3925 3926 *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with 3927 associated ASN1, EVP and SSL functions and old ASN1 macros. 3928 [Richard Levitte] 3929 3930 *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results, 3931 and this should never fail. So the return value from the use of 3932 BN_set_word() (which can fail due to needless expansion) is now deprecated; 3933 if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro. 3934 [Geoff Thorpe] 3935 3936 *) BN_CTX_get() should return zero-valued bignums, providing the same 3937 initialised value as BN_new(). 3938 [Geoff Thorpe, suggested by Ulf M��ller] 3939 3940 *) Support for inhibitAnyPolicy certificate extension. 3941 [Steve Henson] 3942 3943 *) An audit of the BIGNUM code is underway, for which debugging code is 3944 enabled when BN_DEBUG is defined. This makes stricter enforcements on what 3945 is considered valid when processing BIGNUMs, and causes execution to 3946 assert() when a problem is discovered. If BN_DEBUG_RAND is defined, 3947 further steps are taken to deliberately pollute unused data in BIGNUM 3948 structures to try and expose faulty code further on. For now, openssl will 3949 (in its default mode of operation) continue to tolerate the inconsistent 3950 forms that it has tolerated in the past, but authors and packagers should 3951 consider trying openssl and their own applications when compiled with 3952 these debugging symbols defined. It will help highlight potential bugs in 3953 their own code, and will improve the test coverage for OpenSSL itself. At 3954 some point, these tighter rules will become openssl's default to improve 3955 maintainability, though the assert()s and other overheads will remain only 3956 in debugging configurations. See bn.h for more details. 3957 [Geoff Thorpe, Nils Larsch, Ulf M��ller] 3958 3959 *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure 3960 that can only be obtained through BN_CTX_new() (which implicitly 3961 initialises it). The presence of this function only made it possible 3962 to overwrite an existing structure (and cause memory leaks). 3963 [Geoff Thorpe] 3964 3965 *) Because of the callback-based approach for implementing LHASH as a 3966 template type, lh_insert() adds opaque objects to hash-tables and 3967 lh_doall() or lh_doall_arg() are typically used with a destructor callback 3968 to clean up those corresponding objects before destroying the hash table 3969 (and losing the object pointers). So some over-zealous constifications in 3970 LHASH have been relaxed so that lh_insert() does not take (nor store) the 3971 objects as "const" and the lh_doall[_arg] callback wrappers are not 3972 prototyped to have "const" restrictions on the object pointers they are 3973 given (and so aren't required to cast them away any more). 3974 [Geoff Thorpe] 3975 3976 *) The tmdiff.h API was so ugly and minimal that our own timing utility 3977 (speed) prefers to use its own implementation. The two implementations 3978 haven't been consolidated as yet (volunteers?) but the tmdiff API has had 3979 its object type properly exposed (MS_TM) instead of casting to/from "char 3980 *". This may still change yet if someone realises MS_TM and "ms_time_***" 3981 aren't necessarily the greatest nomenclatures - but this is what was used 3982 internally to the implementation so I've used that for now. 3983 [Geoff Thorpe] 3984 3985 *) Ensure that deprecated functions do not get compiled when 3986 OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of 3987 the self-tests were still using deprecated key-generation functions so 3988 these have been updated also. 3989 [Geoff Thorpe] 3990 3991 *) Reorganise PKCS#7 code to separate the digest location functionality 3992 into PKCS7_find_digest(), digest addtion into PKCS7_bio_add_digest(). 3993 New function PKCS7_set_digest() to set the digest type for PKCS#7 3994 digestedData type. Add additional code to correctly generate the 3995 digestedData type and add support for this type in PKCS7 initialization 3996 functions. 3997 [Steve Henson] 3998 3999 *) New function PKCS7_set0_type_other() this initializes a PKCS7 4000 structure of type "other". 4001 [Steve Henson] 4002 4003 *) Fix prime generation loop in crypto/bn/bn_prime.pl by making 4004 sure the loop does correctly stop and breaking ("division by zero") 4005 modulus operations are not performed. The (pre-generated) prime 4006 table crypto/bn/bn_prime.h was already correct, but it could not be 4007 re-generated on some platforms because of the "division by zero" 4008 situation in the script. 4009 [Ralf S. Engelschall] 4010 4011 *) Update support for ECC-based TLS ciphersuites according to 4012 draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with 4013 SHA-1 now is only used for "small" curves (where the 4014 representation of a field element takes up to 24 bytes); for 4015 larger curves, the field element resulting from ECDH is directly 4016 used as premaster secret. 4017 [Douglas Stebila (Sun Microsystems Laboratories)] 4018 4019 *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2 4020 curve secp160r1 to the tests. 4021 [Douglas Stebila (Sun Microsystems Laboratories)] 4022 4023 *) Add the possibility to load symbols globally with DSO. 4024 [G��tz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte] 4025 4026 *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better 4027 control of the error stack. 4028 [Richard Levitte] 4029 4030 *) Add support for STORE in ENGINE. 4031 [Richard Levitte] 4032 4033 *) Add the STORE type. The intention is to provide a common interface 4034 to certificate and key stores, be they simple file-based stores, or 4035 HSM-type store, or LDAP stores, or... 4036 NOTE: The code is currently UNTESTED and isn't really used anywhere. 4037 [Richard Levitte] 4038 4039 *) Add a generic structure called OPENSSL_ITEM. This can be used to 4040 pass a list of arguments to any function as well as provide a way 4041 for a function to pass data back to the caller. 4042 [Richard Levitte] 4043 4044 *) Add the functions BUF_strndup() and BUF_memdup(). BUF_strndup() 4045 works like BUF_strdup() but can be used to duplicate a portion of 4046 a string. The copy gets NUL-terminated. BUF_memdup() duplicates 4047 a memory area. 4048 [Richard Levitte] 4049 4050 *) Add the function sk_find_ex() which works like sk_find(), but will 4051 return an index to an element even if an exact match couldn't be 4052 found. The index is guaranteed to point at the element where the 4053 searched-for key would be inserted to preserve sorting order. 4054 [Richard Levitte] 4055 4056 *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but 4057 takes an extra flags argument for optional functionality. Currently, 4058 the following flags are defined: 4059 4060 OBJ_BSEARCH_VALUE_ON_NOMATCH 4061 This one gets OBJ_bsearch_ex() to return a pointer to the first 4062 element where the comparing function returns a negative or zero 4063 number. 4064 4065 OBJ_BSEARCH_FIRST_VALUE_ON_MATCH 4066 This one gets OBJ_bsearch_ex() to return a pointer to the first 4067 element where the comparing function returns zero. This is useful 4068 if there are more than one element where the comparing function 4069 returns zero. 4070 [Richard Levitte] 4071 4072 *) Make it possible to create self-signed certificates with 'openssl ca' 4073 in such a way that the self-signed certificate becomes part of the 4074 CA database and uses the same mechanisms for serial number generation 4075 as all other certificate signing. The new flag '-selfsign' enables 4076 this functionality. Adapt CA.sh and CA.pl.in. 4077 [Richard Levitte] 4078 4079 *) Add functionality to check the public key of a certificate request 4080 against a given private. This is useful to check that a certificate 4081 request can be signed by that key (self-signing). 4082 [Richard Levitte] 4083 4084 *) Make it possible to have multiple active certificates with the same 4085 subject in the CA index file. This is done only if the keyword 4086 'unique_subject' is set to 'no' in the main CA section (default 4087 if 'CA_default') of the configuration file. The value is saved 4088 with the database itself in a separate index attribute file, 4089 named like the index file with '.attr' appended to the name. 4090 [Richard Levitte] 4091 4092 *) Generate muti valued AVAs using '+' notation in config files for 4093 req and dirName. 4094 [Steve Henson] 4095 4096 *) Support for nameConstraints certificate extension. 4097 [Steve Henson] 4098 4099 *) Support for policyConstraints certificate extension. 4100 [Steve Henson] 4101 4102 *) Support for policyMappings certificate extension. 4103 [Steve Henson] 4104 4105 *) Make sure the default DSA_METHOD implementation only uses its 4106 dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL, 4107 and change its own handlers to be NULL so as to remove unnecessary 4108 indirection. This lets alternative implementations fallback to the 4109 default implementation more easily. 4110 [Geoff Thorpe] 4111 4112 *) Support for directoryName in GeneralName related extensions 4113 in config files. 4114 [Steve Henson] 4115 4116 *) Make it possible to link applications using Makefile.shared. 4117 Make that possible even when linking against static libraries! 4118 [Richard Levitte] 4119 4120 *) Support for single pass processing for S/MIME signing. This now 4121 means that S/MIME signing can be done from a pipe, in addition 4122 cleartext signing (multipart/signed type) is effectively streaming 4123 and the signed data does not need to be all held in memory. 4124 4125 This is done with a new flag PKCS7_STREAM. When this flag is set 4126 PKCS7_sign() only initializes the PKCS7 structure and the actual signing 4127 is done after the data is output (and digests calculated) in 4128 SMIME_write_PKCS7(). 4129 [Steve Henson] 4130 4131 *) Add full support for -rpath/-R, both in shared libraries and 4132 applications, at least on the platforms where it's known how 4133 to do it. 4134 [Richard Levitte] 4135 4136 *) In crypto/ec/ec_mult.c, implement fast point multiplication with 4137 precomputation, based on wNAF splitting: EC_GROUP_precompute_mult() 4138 will now compute a table of multiples of the generator that 4139 makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul() 4140 faster (notably in the case of a single point multiplication, 4141 scalar * generator). 4142 [Nils Larsch, Bodo Moeller] 4143 4144 *) IPv6 support for certificate extensions. The various extensions 4145 which use the IP:a.b.c.d can now take IPv6 addresses using the 4146 formats of RFC1884 2.2 . IPv6 addresses are now also displayed 4147 correctly. 4148 [Steve Henson] 4149 4150 *) Added an ENGINE that implements RSA by performing private key 4151 exponentiations with the GMP library. The conversions to and from 4152 GMP's mpz_t format aren't optimised nor are any montgomery forms 4153 cached, and on x86 it appears OpenSSL's own performance has caught up. 4154 However there are likely to be other architectures where GMP could 4155 provide a boost. This ENGINE is not built in by default, but it can be 4156 specified at Configure time and should be accompanied by the necessary 4157 linker additions, eg; 4158 ./config -DOPENSSL_USE_GMP -lgmp 4159 [Geoff Thorpe] 4160 4161 *) "openssl engine" will not display ENGINE/DSO load failure errors when 4162 testing availability of engines with "-t" - the old behaviour is 4163 produced by increasing the feature's verbosity with "-tt". 4164 [Geoff Thorpe] 4165 4166 *) ECDSA routines: under certain error conditions uninitialized BN objects 4167 could be freed. Solution: make sure initialization is performed early 4168 enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de> 4169 via PR#459) 4170 [Lutz Jaenicke] 4171 4172 *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD 4173 and DH_METHOD (eg. by ENGINE implementations) to override the normal 4174 software implementations. For DSA and DH, parameter generation can 4175 also be overriden by providing the appropriate method callbacks. 4176 [Geoff Thorpe] 4177 4178 *) Change the "progress" mechanism used in key-generation and 4179 primality testing to functions that take a new BN_GENCB pointer in 4180 place of callback/argument pairs. The new API functions have "_ex" 4181 postfixes and the older functions are reimplemented as wrappers for 4182 the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide 4183 declarations of the old functions to help (graceful) attempts to 4184 migrate to the new functions. Also, the new key-generation API 4185 functions operate on a caller-supplied key-structure and return 4186 success/failure rather than returning a key or NULL - this is to 4187 help make "keygen" another member function of RSA_METHOD etc. 4188 4189 Example for using the new callback interface: 4190 4191 int (*my_callback)(int a, int b, BN_GENCB *cb) = ...; 4192 void *my_arg = ...; 4193 BN_GENCB my_cb; 4194 4195 BN_GENCB_set(&my_cb, my_callback, my_arg); 4196 4197 return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb); 4198 /* For the meaning of a, b in calls to my_callback(), see the 4199 * documentation of the function that calls the callback. 4200 * cb will point to my_cb; my_arg can be retrieved as cb->arg. 4201 * my_callback should return 1 if it wants BN_is_prime_ex() 4202 * to continue, or 0 to stop. 4203 */ 4204 4205 [Geoff Thorpe] 4206 4207 *) Change the ZLIB compression method to be stateful, and make it 4208 available to TLS with the number defined in 4209 draft-ietf-tls-compression-04.txt. 4210 [Richard Levitte] 4211 4212 *) Add the ASN.1 structures and functions for CertificatePair, which 4213 is defined as follows (according to X.509_4thEditionDraftV6.pdf): 4214 4215 CertificatePair ::= SEQUENCE { 4216 forward [0] Certificate OPTIONAL, 4217 reverse [1] Certificate OPTIONAL, 4218 -- at least one of the pair shall be present -- } 4219 4220 Also implement the PEM functions to read and write certificate 4221 pairs, and defined the PEM tag as "CERTIFICATE PAIR". 4222 4223 This needed to be defined, mostly for the sake of the LDAP 4224 attribute crossCertificatePair, but may prove useful elsewhere as 4225 well. 4226 [Richard Levitte] 4227 4228 *) Make it possible to inhibit symlinking of shared libraries in 4229 Makefile.shared, for Cygwin's sake. 4230 [Richard Levitte] 4231 4232 *) Extend the BIGNUM API by creating a function 4233 void BN_set_negative(BIGNUM *a, int neg); 4234 and a macro that behave like 4235 int BN_is_negative(const BIGNUM *a); 4236 4237 to avoid the need to access 'a->neg' directly in applications. 4238 [Nils Larsch] 4239 4240 *) Implement fast modular reduction for pseudo-Mersenne primes 4241 used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c). 4242 EC_GROUP_new_curve_GFp() will now automatically use this 4243 if applicable. 4244 [Nils Larsch <nla@trustcenter.de>] 4245 4246 *) Add new lock type (CRYPTO_LOCK_BN). 4247 [Bodo Moeller] 4248 4249 *) Change the ENGINE framework to automatically load engines 4250 dynamically from specific directories unless they could be 4251 found to already be built in or loaded. Move all the 4252 current engines except for the cryptodev one to a new 4253 directory engines/. 4254 The engines in engines/ are built as shared libraries if 4255 the "shared" options was given to ./Configure or ./config. 4256 Otherwise, they are inserted in libcrypto.a. 4257 /usr/local/ssl/engines is the default directory for dynamic 4258 engines, but that can be overriden at configure time through 4259 the usual use of --prefix and/or --openssldir, and at run 4260 time with the environment variable OPENSSL_ENGINES. 4261 [Geoff Thorpe and Richard Levitte] 4262 4263 *) Add Makefile.shared, a helper makefile to build shared 4264 libraries. Addapt Makefile.org. 4265 [Richard Levitte] 4266 4267 *) Add version info to Win32 DLLs. 4268 [Peter 'Luna' Runestig" <peter@runestig.com>] 4269 4270 *) Add new 'medium level' PKCS#12 API. Certificates and keys 4271 can be added using this API to created arbitrary PKCS#12 4272 files while avoiding the low level API. 4273 4274 New options to PKCS12_create(), key or cert can be NULL and 4275 will then be omitted from the output file. The encryption 4276 algorithm NIDs can be set to -1 for no encryption, the mac 4277 iteration count can be set to 0 to omit the mac. 4278 4279 Enhance pkcs12 utility by making the -nokeys and -nocerts 4280 options work when creating a PKCS#12 file. New option -nomac 4281 to omit the mac, NONE can be set for an encryption algorithm. 4282 New code is modified to use the enhanced PKCS12_create() 4283 instead of the low level API. 4284 [Steve Henson] 4285 4286 *) Extend ASN1 encoder to support indefinite length constructed 4287 encoding. This can output sequences tags and octet strings in 4288 this form. Modify pk7_asn1.c to support indefinite length 4289 encoding. This is experimental and needs additional code to 4290 be useful, such as an ASN1 bio and some enhanced streaming 4291 PKCS#7 code. 4292 4293 Extend template encode functionality so that tagging is passed 4294 down to the template encoder. 4295 [Steve Henson] 4296 4297 *) Let 'openssl req' fail if an argument to '-newkey' is not 4298 recognized instead of using RSA as a default. 4299 [Bodo Moeller] 4300 4301 *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt. 4302 As these are not official, they are not included in "ALL"; 4303 the "ECCdraft" ciphersuite group alias can be used to select them. 4304 [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)] 4305 4306 *) Add ECDH engine support. 4307 [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)] 4308 4309 *) Add ECDH in new directory crypto/ecdh/. 4310 [Douglas Stebila (Sun Microsystems Laboratories)] 4311 4312 *) Let BN_rand_range() abort with an error after 100 iterations 4313 without success (which indicates a broken PRNG). 4314 [Bodo Moeller] 4315 4316 *) Change BN_mod_sqrt() so that it verifies that the input value 4317 is really the square of the return value. (Previously, 4318 BN_mod_sqrt would show GIGO behaviour.) 4319 [Bodo Moeller] 4320 4321 *) Add named elliptic curves over binary fields from X9.62, SECG, 4322 and WAP/WTLS; add OIDs that were still missing. 4323 4324 [Sheueling Chang Shantz and Douglas Stebila 4325 (Sun Microsystems Laboratories)] 4326 4327 *) Extend the EC library for elliptic curves over binary fields 4328 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/). 4329 New EC_METHOD: 4330 4331 EC_GF2m_simple_method 4332 4333 New API functions: 4334 4335 EC_GROUP_new_curve_GF2m 4336 EC_GROUP_set_curve_GF2m 4337 EC_GROUP_get_curve_GF2m 4338 EC_POINT_set_affine_coordinates_GF2m 4339 EC_POINT_get_affine_coordinates_GF2m 4340 EC_POINT_set_compressed_coordinates_GF2m 4341 4342 Point compression for binary fields is disabled by default for 4343 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to 4344 enable it). 4345 4346 As binary polynomials are represented as BIGNUMs, various members 4347 of the EC_GROUP and EC_POINT data structures can be shared 4348 between the implementations for prime fields and binary fields; 4349 the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m) 4350 are essentially identical to their ..._GFp counterparts. 4351 (For simplicity, the '..._GFp' prefix has been dropped from 4352 various internal method names.) 4353 4354 An internal 'field_div' method (similar to 'field_mul' and 4355 'field_sqr') has been added; this is used only for binary fields. 4356 4357 [Sheueling Chang Shantz and Douglas Stebila 4358 (Sun Microsystems Laboratories)] 4359 4360 *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult() 4361 through methods ('mul', 'precompute_mult'). 4362 4363 The generic implementations (now internally called 'ec_wNAF_mul' 4364 and 'ec_wNAF_precomputed_mult') remain the default if these 4365 methods are undefined. 4366 4367 [Sheueling Chang Shantz and Douglas Stebila 4368 (Sun Microsystems Laboratories)] 4369 4370 *) New function EC_GROUP_get_degree, which is defined through 4371 EC_METHOD. For curves over prime fields, this returns the bit 4372 length of the modulus. 4373 4374 [Sheueling Chang Shantz and Douglas Stebila 4375 (Sun Microsystems Laboratories)] 4376 4377 *) New functions EC_GROUP_dup, EC_POINT_dup. 4378 (These simply call ..._new and ..._copy). 4379 4380 [Sheueling Chang Shantz and Douglas Stebila 4381 (Sun Microsystems Laboratories)] 4382 4383 *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c. 4384 Polynomials are represented as BIGNUMs (where the sign bit is not 4385 used) in the following functions [macros]: 4386 4387 BN_GF2m_add 4388 BN_GF2m_sub [= BN_GF2m_add] 4389 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr] 4390 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr] 4391 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr] 4392 BN_GF2m_mod_inv 4393 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr] 4394 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr] 4395 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr] 4396 BN_GF2m_cmp [= BN_ucmp] 4397 4398 (Note that only the 'mod' functions are actually for fields GF(2^m). 4399 BN_GF2m_add() is misnomer, but this is for the sake of consistency.) 4400 4401 For some functions, an the irreducible polynomial defining a 4402 field can be given as an 'unsigned int[]' with strictly 4403 decreasing elements giving the indices of those bits that are set; 4404 i.e., p[] represents the polynomial 4405 f(t) = t^p[0] + t^p[1] + ... + t^p[k] 4406 where 4407 p[0] > p[1] > ... > p[k] = 0. 4408 This applies to the following functions: 4409 4410 BN_GF2m_mod_arr 4411 BN_GF2m_mod_mul_arr 4412 BN_GF2m_mod_sqr_arr 4413 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv] 4414 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div] 4415 BN_GF2m_mod_exp_arr 4416 BN_GF2m_mod_sqrt_arr 4417 BN_GF2m_mod_solve_quad_arr 4418 BN_GF2m_poly2arr 4419 BN_GF2m_arr2poly 4420 4421 Conversion can be performed by the following functions: 4422 4423 BN_GF2m_poly2arr 4424 BN_GF2m_arr2poly 4425 4426 bntest.c has additional tests for binary polynomial arithmetic. 4427 4428 Two implementations for BN_GF2m_mod_div() are available. 4429 The default algorithm simply uses BN_GF2m_mod_inv() and 4430 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only 4431 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the 4432 copyright notice in crypto/bn/bn_gf2m.c before enabling it). 4433 4434 [Sheueling Chang Shantz and Douglas Stebila 4435 (Sun Microsystems Laboratories)] 4436 4437 *) Add new error code 'ERR_R_DISABLED' that can be used when some 4438 functionality is disabled at compile-time. 4439 [Douglas Stebila <douglas.stebila@sun.com>] 4440 4441 *) Change default behaviour of 'openssl asn1parse' so that more 4442 information is visible when viewing, e.g., a certificate: 4443 4444 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump' 4445 mode the content of non-printable OCTET STRINGs is output in a 4446 style similar to INTEGERs, but with '[HEX DUMP]' prepended to 4447 avoid the appearance of a printable string. 4448 [Nils Larsch <nla@trustcenter.de>] 4449 4450 *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access 4451 functions 4452 EC_GROUP_set_asn1_flag() 4453 EC_GROUP_get_asn1_flag() 4454 EC_GROUP_set_point_conversion_form() 4455 EC_GROUP_get_point_conversion_form() 4456 These control ASN1 encoding details: 4457 - Curves (i.e., groups) are encoded explicitly unless asn1_flag 4458 has been set to OPENSSL_EC_NAMED_CURVE. 4459 - Points are encoded in uncompressed form by default; options for 4460 asn1_for are as for point2oct, namely 4461 POINT_CONVERSION_COMPRESSED 4462 POINT_CONVERSION_UNCOMPRESSED 4463 POINT_CONVERSION_HYBRID 4464 4465 Also add 'seed' and 'seed_len' members to EC_GROUP with access 4466 functions 4467 EC_GROUP_set_seed() 4468 EC_GROUP_get0_seed() 4469 EC_GROUP_get_seed_len() 4470 This is used only for ASN1 purposes (so far). 4471 [Nils Larsch <nla@trustcenter.de>] 4472 4473 *) Add 'field_type' member to EC_METHOD, which holds the NID 4474 of the appropriate field type OID. The new function 4475 EC_METHOD_get_field_type() returns this value. 4476 [Nils Larsch <nla@trustcenter.de>] 4477 4478 *) Add functions 4479 EC_POINT_point2bn() 4480 EC_POINT_bn2point() 4481 EC_POINT_point2hex() 4482 EC_POINT_hex2point() 4483 providing useful interfaces to EC_POINT_point2oct() and 4484 EC_POINT_oct2point(). 4485 [Nils Larsch <nla@trustcenter.de>] 4486 4487 *) Change internals of the EC library so that the functions 4488 EC_GROUP_set_generator() 4489 EC_GROUP_get_generator() 4490 EC_GROUP_get_order() 4491 EC_GROUP_get_cofactor() 4492 are implemented directly in crypto/ec/ec_lib.c and not dispatched 4493 to methods, which would lead to unnecessary code duplication when 4494 adding different types of curves. 4495 [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller] 4496 4497 *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM 4498 arithmetic, and such that modified wNAFs are generated 4499 (which avoid length expansion in many cases). 4500 [Bodo Moeller] 4501 4502 *) Add a function EC_GROUP_check_discriminant() (defined via 4503 EC_METHOD) that verifies that the curve discriminant is non-zero. 4504 4505 Add a function EC_GROUP_check() that makes some sanity tests 4506 on a EC_GROUP, its generator and order. This includes 4507 EC_GROUP_check_discriminant(). 4508 [Nils Larsch <nla@trustcenter.de>] 4509 4510 *) Add ECDSA in new directory crypto/ecdsa/. 4511 4512 Add applications 'openssl ecparam' and 'openssl ecdsa' 4513 (these are based on 'openssl dsaparam' and 'openssl dsa'). 4514 4515 ECDSA support is also included in various other files across the 4516 library. Most notably, 4517 - 'openssl req' now has a '-newkey ecdsa:file' option; 4518 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA; 4519 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and 4520 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make 4521 them suitable for ECDSA where domain parameters must be 4522 extracted before the specific public key; 4523 - ECDSA engine support has been added. 4524 [Nils Larsch <nla@trustcenter.de>] 4525 4526 *) Include some named elliptic curves, and add OIDs from X9.62, 4527 SECG, and WAP/WTLS. Each curve can be obtained from the new 4528 function 4529 EC_GROUP_new_by_curve_name(), 4530 and the list of available named curves can be obtained with 4531 EC_get_builtin_curves(). 4532 Also add a 'curve_name' member to EC_GROUP objects, which can be 4533 accessed via 4534 EC_GROUP_set_curve_name() 4535 EC_GROUP_get_curve_name() 4536 [Nils Larsch <larsch@trustcenter.de, Bodo Moeller] 4537 4538 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 4539 was actually never needed) and in BN_mul(). The removal in BN_mul() 4540 required a small change in bn_mul_part_recursive() and the addition 4541 of the functions bn_cmp_part_words(), bn_sub_part_words() and 4542 bn_add_part_words(), which do the same thing as bn_cmp_words(), 4543 bn_sub_words() and bn_add_words() except they take arrays with 4544 differing sizes. 4545 [Richard Levitte] 4546 4547 Changes between 0.9.7l and 0.9.7m [23 Feb 2007] 4548 4549 *) Cleanse PEM buffers before freeing them since they may contain 4550 sensitive data. 4551 [Benjamin Bennett <ben@psc.edu>] 4552 4553 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 4554 a ciphersuite string such as "DEFAULT:RSA" cannot enable 4555 authentication-only ciphersuites. 4556 [Bodo Moeller] 4557 4558 *) Since AES128 and AES256 share a single mask bit in the logic of 4559 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 4560 kludge to work properly if AES128 is available and AES256 isn't. 4561 [Victor Duchovni] 4562 4563 *) Expand security boundary to match 1.1.1 module. 4564 [Steve Henson] 4565 4566 *) Remove redundant features: hash file source, editing of test vectors 4567 modify fipsld to use external fips_premain.c signature. 4568 [Steve Henson] 4569 4570 *) New perl script mkfipsscr.pl to create shell scripts or batch files to 4571 run algorithm test programs. 4572 [Steve Henson] 4573 4574 *) Make algorithm test programs more tolerant of whitespace. 4575 [Steve Henson] 4576 4577 *) Have SSL/TLS server implementation tolerate "mismatched" record 4578 protocol version while receiving ClientHello even if the 4579 ClientHello is fragmented. (The server can't insist on the 4580 particular protocol version it has chosen before the ServerHello 4581 message has informed the client about his choice.) 4582 [Bodo Moeller] 4583 4584 *) Load error codes if they are not already present instead of using a 4585 static variable. This allows them to be cleanly unloaded and reloaded. 4586 [Steve Henson] 4587 4588 Changes between 0.9.7k and 0.9.7l [28 Sep 2006] 4589 4590 *) Introduce limits to prevent malicious keys being able to 4591 cause a denial of service. (CVE-2006-2940) 4592 [Steve Henson, Bodo Moeller] 4593 4594 *) Fix ASN.1 parsing of certain invalid structures that can result 4595 in a denial of service. (CVE-2006-2937) [Steve Henson] 4596 4597 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 4598 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 4599 4600 *) Fix SSL client code which could crash if connecting to a 4601 malicious SSLv2 server. (CVE-2006-4343) 4602 [Tavis Ormandy and Will Drewry, Google Security Team] 4603 4604 *) Change ciphersuite string processing so that an explicit 4605 ciphersuite selects this one ciphersuite (so that "AES256-SHA" 4606 will no longer include "AES128-SHA"), and any other similar 4607 ciphersuite (same bitmap) from *other* protocol versions (so that 4608 "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the 4609 SSL 3.0/TLS 1.0 ciphersuite). This is a backport combining 4610 changes from 0.9.8b and 0.9.8d. 4611 [Bodo Moeller] 4612 4613 Changes between 0.9.7j and 0.9.7k [05 Sep 2006] 4614 4615 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 4616 (CVE-2006-4339) [Ben Laurie and Google Security Team] 4617 4618 *) Change the Unix randomness entropy gathering to use poll() when 4619 possible instead of select(), since the latter has some 4620 undesirable limitations. 4621 [Darryl Miles via Richard Levitte and Bodo Moeller] 4622 4623 *) Disable rogue ciphersuites: 4624 4625 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 4626 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 4627 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 4628 4629 The latter two were purportedly from 4630 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 4631 appear there. 4632 4633 Also deactive the remaining ciphersuites from 4634 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 4635 unofficial, and the ID has long expired. 4636 [Bodo Moeller] 4637 4638 *) Fix RSA blinding Heisenbug (problems sometimes occured on 4639 dual-core machines) and other potential thread-safety issues. 4640 [Bodo Moeller] 4641 4642 Changes between 0.9.7i and 0.9.7j [04 May 2006] 4643 4644 *) Adapt fipsld and the build system to link against the validated FIPS 4645 module in FIPS mode. 4646 [Steve Henson] 4647 4648 *) Fixes for VC++ 2005 build under Windows. 4649 [Steve Henson] 4650 4651 *) Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make 4652 from a Windows bash shell such as MSYS. It is autodetected from the 4653 "config" script when run from a VC++ environment. Modify standard VC++ 4654 build to use fipscanister.o from the GNU make build. 4655 [Steve Henson] 4656 4657 Changes between 0.9.7h and 0.9.7i [14 Oct 2005] 4658 4659 *) Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS. 4660 The value now differs depending on if you build for FIPS or not. 4661 BEWARE! A program linked with a shared FIPSed libcrypto can't be 4662 safely run with a non-FIPSed libcrypto, as it may crash because of 4663 the difference induced by this change. 4664 [Andy Polyakov] 4665 4666 Changes between 0.9.7g and 0.9.7h [11 Oct 2005] 4667 4668 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 4669 (part of SSL_OP_ALL). This option used to disable the 4670 countermeasure against man-in-the-middle protocol-version 4671 rollback in the SSL 2.0 server implementation, which is a bad 4672 idea. (CVE-2005-2969) 4673 4674 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 4675 for Information Security, National Institute of Advanced Industrial 4676 Science and Technology [AIST], Japan)] 4677 4678 *) Minimal support for X9.31 signatures and PSS padding modes. This is 4679 mainly for FIPS compliance and not fully integrated at this stage. 4680 [Steve Henson] 4681 4682 *) For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform 4683 the exponentiation using a fixed-length exponent. (Otherwise, 4684 the information leaked through timing could expose the secret key 4685 after many signatures; cf. Bleichenbacher's attack on DSA with 4686 biased k.) 4687 [Bodo Moeller] 4688 4689 *) Make a new fixed-window mod_exp implementation the default for 4690 RSA, DSA, and DH private-key operations so that the sequence of 4691 squares and multiplies and the memory access pattern are 4692 independent of the particular secret key. This will mitigate 4693 cache-timing and potential related attacks. 4694 4695 BN_mod_exp_mont_consttime() is the new exponentiation implementation, 4696 and this is automatically used by BN_mod_exp_mont() if the new flag 4697 BN_FLG_EXP_CONSTTIME is set for the exponent. RSA, DSA, and DH 4698 will use this BN flag for private exponents unless the flag 4699 RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or 4700 DH_FLAG_NO_EXP_CONSTTIME, respectively, is set. 4701 4702 [Matthew D Wood (Intel Corp), with some changes by Bodo Moeller] 4703 4704 *) Change the client implementation for SSLv23_method() and 4705 SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0 4706 Client Hello message format if the SSL_OP_NO_SSLv2 option is set. 4707 (Previously, the SSL 2.0 backwards compatible Client Hello 4708 message format would be used even with SSL_OP_NO_SSLv2.) 4709 [Bodo Moeller] 4710 4711 *) Add support for smime-type MIME parameter in S/MIME messages which some 4712 clients need. 4713 [Steve Henson] 4714 4715 *) New function BN_MONT_CTX_set_locked() to set montgomery parameters in 4716 a threadsafe manner. Modify rsa code to use new function and add calls 4717 to dsa and dh code (which had race conditions before). 4718 [Steve Henson] 4719 4720 *) Include the fixed error library code in the C error file definitions 4721 instead of fixing them up at runtime. This keeps the error code 4722 structures constant. 4723 [Steve Henson] 4724 4725 Changes between 0.9.7f and 0.9.7g [11 Apr 2005] 4726 4727 [NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after 4728 OpenSSL 0.9.8.] 4729 4730 *) Fixes for newer kerberos headers. NB: the casts are needed because 4731 the 'length' field is signed on one version and unsigned on another 4732 with no (?) obvious way to tell the difference, without these VC++ 4733 complains. Also the "definition" of FAR (blank) is no longer included 4734 nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up 4735 some needed definitions. 4736 [Steve Henson] 4737 4738 *) Undo Cygwin change. 4739 [Ulf M��ller] 4740 4741 *) Added support for proxy certificates according to RFC 3820. 4742 Because they may be a security thread to unaware applications, 4743 they must be explicitely allowed in run-time. See 4744 docs/HOWTO/proxy_certificates.txt for further information. 4745 [Richard Levitte] 4746 4747 Changes between 0.9.7e and 0.9.7f [22 Mar 2005] 4748 4749 *) Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating 4750 server and client random values. Previously 4751 (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in 4752 less random data when sizeof(time_t) > 4 (some 64 bit platforms). 4753 4754 This change has negligible security impact because: 4755 4756 1. Server and client random values still have 24 bytes of pseudo random 4757 data. 4758 4759 2. Server and client random values are sent in the clear in the initial 4760 handshake. 4761 4762 3. The master secret is derived using the premaster secret (48 bytes in 4763 size for static RSA ciphersuites) as well as client server and random 4764 values. 4765 4766 The OpenSSL team would like to thank the UK NISCC for bringing this issue 4767 to our attention. 4768 4769 [Stephen Henson, reported by UK NISCC] 4770 4771 *) Use Windows randomness collection on Cygwin. 4772 [Ulf M��ller] 4773 4774 *) Fix hang in EGD/PRNGD query when communication socket is closed 4775 prematurely by EGD/PRNGD. 4776 [Darren Tucker <dtucker@zip.com.au> via Lutz J��nicke, resolves #1014] 4777 4778 *) Prompt for pass phrases when appropriate for PKCS12 input format. 4779 [Steve Henson] 4780 4781 *) Back-port of selected performance improvements from development 4782 branch, as well as improved support for PowerPC platforms. 4783 [Andy Polyakov] 4784 4785 *) Add lots of checks for memory allocation failure, error codes to indicate 4786 failure and freeing up memory if a failure occurs. 4787 [Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson] 4788 4789 *) Add new -passin argument to dgst. 4790 [Steve Henson] 4791 4792 *) Perform some character comparisons of different types in X509_NAME_cmp: 4793 this is needed for some certificates that reencode DNs into UTF8Strings 4794 (in violation of RFC3280) and can't or wont issue name rollover 4795 certificates. 4796 [Steve Henson] 4797 4798 *) Make an explicit check during certificate validation to see that 4799 the CA setting in each certificate on the chain is correct. As a 4800 side effect always do the following basic checks on extensions, 4801 not just when there's an associated purpose to the check: 4802 4803 - if there is an unhandled critical extension (unless the user 4804 has chosen to ignore this fault) 4805 - if the path length has been exceeded (if one is set at all) 4806 - that certain extensions fit the associated purpose (if one has 4807 been given) 4808 [Richard Levitte] 4809 4810 Changes between 0.9.7d and 0.9.7e [25 Oct 2004] 4811 4812 *) Avoid a race condition when CRLs are checked in a multi threaded 4813 environment. This would happen due to the reordering of the revoked 4814 entries during signature checking and serial number lookup. Now the 4815 encoding is cached and the serial number sort performed under a lock. 4816 Add new STACK function sk_is_sorted(). 4817 [Steve Henson] 4818 4819 *) Add Delta CRL to the extension code. 4820 [Steve Henson] 4821 4822 *) Various fixes to s3_pkt.c so alerts are sent properly. 4823 [David Holmes <d.holmes@f5.com>] 4824 4825 *) Reduce the chances of duplicate issuer name and serial numbers (in 4826 violation of RFC3280) using the OpenSSL certificate creation utilities. 4827 This is done by creating a random 64 bit value for the initial serial 4828 number when a serial number file is created or when a self signed 4829 certificate is created using 'openssl req -x509'. The initial serial 4830 number file is created using 'openssl x509 -next_serial' in CA.pl 4831 rather than being initialized to 1. 4832 [Steve Henson] 4833 4834 Changes between 0.9.7c and 0.9.7d [17 Mar 2004] 4835 4836 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 4837 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 4838 [Joe Orton, Steve Henson] 4839 4840 *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites 4841 (CVE-2004-0112) 4842 [Joe Orton, Steve Henson] 4843 4844 *) Make it possible to have multiple active certificates with the same 4845 subject in the CA index file. This is done only if the keyword 4846 'unique_subject' is set to 'no' in the main CA section (default 4847 if 'CA_default') of the configuration file. The value is saved 4848 with the database itself in a separate index attribute file, 4849 named like the index file with '.attr' appended to the name. 4850 [Richard Levitte] 4851 4852 *) X509 verify fixes. Disable broken certificate workarounds when 4853 X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if 4854 keyUsage extension present. Don't accept CRLs with unhandled critical 4855 extensions: since verify currently doesn't process CRL extensions this 4856 rejects a CRL with *any* critical extensions. Add new verify error codes 4857 for these cases. 4858 [Steve Henson] 4859 4860 *) When creating an OCSP nonce use an OCTET STRING inside the extnValue. 4861 A clarification of RFC2560 will require the use of OCTET STRINGs and 4862 some implementations cannot handle the current raw format. Since OpenSSL 4863 copies and compares OCSP nonces as opaque blobs without any attempt at 4864 parsing them this should not create any compatibility issues. 4865 [Steve Henson] 4866 4867 *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when 4868 calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without 4869 this HMAC (and other) operations are several times slower than OpenSSL 4870 < 0.9.7. 4871 [Steve Henson] 4872 4873 *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex(). 4874 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 4875 4876 *) Use the correct content when signing type "other". 4877 [Steve Henson] 4878 4879 Changes between 0.9.7b and 0.9.7c [30 Sep 2003] 4880 4881 *) Fix various bugs revealed by running the NISCC test suite: 4882 4883 Stop out of bounds reads in the ASN1 code when presented with 4884 invalid tags (CVE-2003-0543 and CVE-2003-0544). 4885 4886 Free up ASN1_TYPE correctly if ANY type is invalid (CVE-2003-0545). 4887 4888 If verify callback ignores invalid public key errors don't try to check 4889 certificate signature with the NULL public key. 4890 4891 [Steve Henson] 4892 4893 *) New -ignore_err option in ocsp application to stop the server 4894 exiting on the first error in a request. 4895 [Steve Henson] 4896 4897 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 4898 if the server requested one: as stated in TLS 1.0 and SSL 3.0 4899 specifications. 4900 [Steve Henson] 4901 4902 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 4903 extra data after the compression methods not only for TLS 1.0 4904 but also for SSL 3.0 (as required by the specification). 4905 [Bodo Moeller; problem pointed out by Matthias Loepfe] 4906 4907 *) Change X509_certificate_type() to mark the key as exported/exportable 4908 when it's 512 *bits* long, not 512 bytes. 4909 [Richard Levitte] 4910 4911 *) Change AES_cbc_encrypt() so it outputs exact multiple of 4912 blocks during encryption. 4913 [Richard Levitte] 4914 4915 *) Various fixes to base64 BIO and non blocking I/O. On write 4916 flushes were not handled properly if the BIO retried. On read 4917 data was not being buffered properly and had various logic bugs. 4918 This also affects blocking I/O when the data being decoded is a 4919 certain size. 4920 [Steve Henson] 4921 4922 *) Various S/MIME bugfixes and compatibility changes: 4923 output correct application/pkcs7 MIME type if 4924 PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures. 4925 Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening 4926 of files as .eml work). Correctly handle very long lines in MIME 4927 parser. 4928 [Steve Henson] 4929 4930 Changes between 0.9.7a and 0.9.7b [10 Apr 2003] 4931 4932 *) Countermeasure against the Klima-Pokorny-Rosa extension of 4933 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 4934 a protocol version number mismatch like a decryption error 4935 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 4936 [Bodo Moeller] 4937 4938 *) Turn on RSA blinding by default in the default implementation 4939 to avoid a timing attack. Applications that don't want it can call 4940 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 4941 They would be ill-advised to do so in most cases. 4942 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 4943 4944 *) Change RSA blinding code so that it works when the PRNG is not 4945 seeded (in this case, the secret RSA exponent is abused as 4946 an unpredictable seed -- if it is not unpredictable, there 4947 is no point in blinding anyway). Make RSA blinding thread-safe 4948 by remembering the creator's thread ID in rsa->blinding and 4949 having all other threads use local one-time blinding factors 4950 (this requires more computation than sharing rsa->blinding, but 4951 avoids excessive locking; and if an RSA object is not shared 4952 between threads, blinding will still be very fast). 4953 [Bodo Moeller] 4954 4955 *) Fixed a typo bug that would cause ENGINE_set_default() to set an 4956 ENGINE as defaults for all supported algorithms irrespective of 4957 the 'flags' parameter. 'flags' is now honoured, so applications 4958 should make sure they are passing it correctly. 4959 [Geoff Thorpe] 4960 4961 *) Target "mingw" now allows native Windows code to be generated in 4962 the Cygwin environment as well as with the MinGW compiler. 4963 [Ulf Moeller] 4964 4965 Changes between 0.9.7 and 0.9.7a [19 Feb 2003] 4966 4967 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 4968 via timing by performing a MAC computation even if incorrrect 4969 block cipher padding has been found. This is a countermeasure 4970 against active attacks where the attacker has to distinguish 4971 between bad padding and a MAC verification error. (CVE-2003-0078) 4972 4973 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 4974 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 4975 Martin Vuagnoux (EPFL, Ilion)] 4976 4977 *) Make the no-err option work as intended. The intention with no-err 4978 is not to have the whole error stack handling routines removed from 4979 libcrypto, it's only intended to remove all the function name and 4980 reason texts, thereby removing some of the footprint that may not 4981 be interesting if those errors aren't displayed anyway. 4982 4983 NOTE: it's still possible for any application or module to have it's 4984 own set of error texts inserted. The routines are there, just not 4985 used by default when no-err is given. 4986 [Richard Levitte] 4987 4988 *) Add support for FreeBSD on IA64. 4989 [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454] 4990 4991 *) Adjust DES_cbc_cksum() so it returns the same value as the MIT 4992 Kerberos function mit_des_cbc_cksum(). Before this change, 4993 the value returned by DES_cbc_cksum() was like the one from 4994 mit_des_cbc_cksum(), except the bytes were swapped. 4995 [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte] 4996 4997 *) Allow an application to disable the automatic SSL chain building. 4998 Before this a rather primitive chain build was always performed in 4999 ssl3_output_cert_chain(): an application had no way to send the 5000 correct chain if the automatic operation produced an incorrect result. 5001 5002 Now the chain builder is disabled if either: 5003 5004 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert(). 5005 5006 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set. 5007 5008 The reasoning behind this is that an application would not want the 5009 auto chain building to take place if extra chain certificates are 5010 present and it might also want a means of sending no additional 5011 certificates (for example the chain has two certificates and the 5012 root is omitted). 5013 [Steve Henson] 5014 5015 *) Add the possibility to build without the ENGINE framework. 5016 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 5017 5018 *) Under Win32 gmtime() can return NULL: check return value in 5019 OPENSSL_gmtime(). Add error code for case where gmtime() fails. 5020 [Steve Henson] 5021 5022 *) DSA routines: under certain error conditions uninitialized BN objects 5023 could be freed. Solution: make sure initialization is performed early 5024 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>, 5025 Nils Larsch <nla@trustcenter.de> via PR#459) 5026 [Lutz Jaenicke] 5027 5028 *) Another fix for SSLv2 session ID handling: the session ID was incorrectly 5029 checked on reconnect on the client side, therefore session resumption 5030 could still fail with a "ssl session id is different" error. This 5031 behaviour is masked when SSL_OP_ALL is used due to 5032 SSL_OP_MICROSOFT_SESS_ID_BUG being set. 5033 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 5034 followup to PR #377. 5035 [Lutz Jaenicke] 5036 5037 *) IA-32 assembler support enhancements: unified ELF targets, support 5038 for SCO/Caldera platforms, fix for Cygwin shared build. 5039 [Andy Polyakov] 5040 5041 *) Add support for FreeBSD on sparc64. As a consequence, support for 5042 FreeBSD on non-x86 processors is separate from x86 processors on 5043 the config script, much like the NetBSD support. 5044 [Richard Levitte & Kris Kennaway <kris@obsecurity.org>] 5045 5046 Changes between 0.9.6h and 0.9.7 [31 Dec 2002] 5047 5048 [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after 5049 OpenSSL 0.9.7.] 5050 5051 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED 5052 code (06) was taken as the first octet of the session ID and the last 5053 octet was ignored consequently. As a result SSLv2 client side session 5054 caching could not have worked due to the session ID mismatch between 5055 client and server. 5056 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 5057 PR #377. 5058 [Lutz Jaenicke] 5059 5060 *) Change the declaration of needed Kerberos libraries to use EX_LIBS 5061 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is 5062 removed entirely. 5063 [Richard Levitte] 5064 5065 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it 5066 seems that in spite of existing for more than a year, many application 5067 author have done nothing to provide the necessary callbacks, which 5068 means that this particular engine will not work properly anywhere. 5069 This is a very unfortunate situation which forces us, in the name 5070 of usability, to give the hw_ncipher.c a static lock, which is part 5071 of libcrypto. 5072 NOTE: This is for the 0.9.7 series ONLY. This hack will never 5073 appear in 0.9.8 or later. We EXPECT application authors to have 5074 dealt properly with this when 0.9.8 is released (unless we actually 5075 make such changes in the libcrypto locking code that changes will 5076 have to be made anyway). 5077 [Richard Levitte] 5078 5079 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content 5080 octets have been read, EOF or an error occurs. Without this change 5081 some truncated ASN1 structures will not produce an error. 5082 [Steve Henson] 5083 5084 *) Disable Heimdal support, since it hasn't been fully implemented. 5085 Still give the possibility to force the use of Heimdal, but with 5086 warnings and a request that patches get sent to openssl-dev. 5087 [Richard Levitte] 5088 5089 *) Add the VC-CE target, introduce the WINCE sysname, and add 5090 INSTALL.WCE and appropriate conditionals to make it build. 5091 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 5092 5093 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and 5094 cygssl-x.y.z.dll, where x, y and z are the major, minor and 5095 edit numbers of the version. 5096 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 5097 5098 *) Introduce safe string copy and catenation functions 5099 (BUF_strlcpy() and BUF_strlcat()). 5100 [Ben Laurie (CHATS) and Richard Levitte] 5101 5102 *) Avoid using fixed-size buffers for one-line DNs. 5103 [Ben Laurie (CHATS)] 5104 5105 *) Add BUF_MEM_grow_clean() to avoid information leakage when 5106 resizing buffers containing secrets, and use where appropriate. 5107 [Ben Laurie (CHATS)] 5108 5109 *) Avoid using fixed size buffers for configuration file location. 5110 [Ben Laurie (CHATS)] 5111 5112 *) Avoid filename truncation for various CA files. 5113 [Ben Laurie (CHATS)] 5114 5115 *) Use sizeof in preference to magic numbers. 5116 [Ben Laurie (CHATS)] 5117 5118 *) Avoid filename truncation in cert requests. 5119 [Ben Laurie (CHATS)] 5120 5121 *) Add assertions to check for (supposedly impossible) buffer 5122 overflows. 5123 [Ben Laurie (CHATS)] 5124 5125 *) Don't cache truncated DNS entries in the local cache (this could 5126 potentially lead to a spoofing attack). 5127 [Ben Laurie (CHATS)] 5128 5129 *) Fix various buffers to be large enough for hex/decimal 5130 representations in a platform independent manner. 5131 [Ben Laurie (CHATS)] 5132 5133 *) Add CRYPTO_realloc_clean() to avoid information leakage when 5134 resizing buffers containing secrets, and use where appropriate. 5135 [Ben Laurie (CHATS)] 5136 5137 *) Add BIO_indent() to avoid much slightly worrying code to do 5138 indents. 5139 [Ben Laurie (CHATS)] 5140 5141 *) Convert sprintf()/BIO_puts() to BIO_printf(). 5142 [Ben Laurie (CHATS)] 5143 5144 *) buffer_gets() could terminate with the buffer only half 5145 full. Fixed. 5146 [Ben Laurie (CHATS)] 5147 5148 *) Add assertions to prevent user-supplied crypto functions from 5149 overflowing internal buffers by having large block sizes, etc. 5150 [Ben Laurie (CHATS)] 5151 5152 *) New OPENSSL_assert() macro (similar to assert(), but enabled 5153 unconditionally). 5154 [Ben Laurie (CHATS)] 5155 5156 *) Eliminate unused copy of key in RC4. 5157 [Ben Laurie (CHATS)] 5158 5159 *) Eliminate unused and incorrectly sized buffers for IV in pem.h. 5160 [Ben Laurie (CHATS)] 5161 5162 *) Fix off-by-one error in EGD path. 5163 [Ben Laurie (CHATS)] 5164 5165 *) If RANDFILE path is too long, ignore instead of truncating. 5166 [Ben Laurie (CHATS)] 5167 5168 *) Eliminate unused and incorrectly sized X.509 structure 5169 CBCParameter. 5170 [Ben Laurie (CHATS)] 5171 5172 *) Eliminate unused and dangerous function knumber(). 5173 [Ben Laurie (CHATS)] 5174 5175 *) Eliminate unused and dangerous structure, KSSL_ERR. 5176 [Ben Laurie (CHATS)] 5177 5178 *) Protect against overlong session ID context length in an encoded 5179 session object. Since these are local, this does not appear to be 5180 exploitable. 5181 [Ben Laurie (CHATS)] 5182 5183 *) Change from security patch (see 0.9.6e below) that did not affect 5184 the 0.9.6 release series: 5185 5186 Remote buffer overflow in SSL3 protocol - an attacker could 5187 supply an oversized master key in Kerberos-enabled versions. 5188 (CVE-2002-0657) 5189 [Ben Laurie (CHATS)] 5190 5191 *) Change the SSL kerb5 codes to match RFC 2712. 5192 [Richard Levitte] 5193 5194 *) Make -nameopt work fully for req and add -reqopt switch. 5195 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson] 5196 5197 *) The "block size" for block ciphers in CFB and OFB mode should be 1. 5198 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>] 5199 5200 *) Make sure tests can be performed even if the corresponding algorithms 5201 have been removed entirely. This was also the last step to make 5202 OpenSSL compilable with DJGPP under all reasonable conditions. 5203 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>] 5204 5205 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT 5206 to allow version independent disabling of normally unselected ciphers, 5207 which may be activated as a side-effect of selecting a single cipher. 5208 5209 (E.g., cipher list string "RSA" enables ciphersuites that are left 5210 out of "ALL" because they do not provide symmetric encryption. 5211 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.) 5212 [Lutz Jaenicke, Bodo Moeller] 5213 5214 *) Add appropriate support for separate platform-dependent build 5215 directories. The recommended way to make a platform-dependent 5216 build directory is the following (tested on Linux), maybe with 5217 some local tweaks: 5218 5219 # Place yourself outside of the OpenSSL source tree. In 5220 # this example, the environment variable OPENSSL_SOURCE 5221 # is assumed to contain the absolute OpenSSL source directory. 5222 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`" 5223 cd objtree/"`uname -s`-`uname -r`-`uname -m`" 5224 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do 5225 mkdir -p `dirname $F` 5226 ln -s $OPENSSL_SOURCE/$F $F 5227 done 5228 5229 To be absolutely sure not to disturb the source tree, a "make clean" 5230 is a good thing. If it isn't successfull, don't worry about it, 5231 it probably means the source directory is very clean. 5232 [Richard Levitte] 5233 5234 *) Make sure any ENGINE control commands make local copies of string 5235 pointers passed to them whenever necessary. Otherwise it is possible 5236 the caller may have overwritten (or deallocated) the original string 5237 data when a later ENGINE operation tries to use the stored values. 5238 [G��tz Babin-Ebell <babinebell@trustcenter.de>] 5239 5240 *) Improve diagnostics in file reading and command-line digests. 5241 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>] 5242 5243 *) Add AES modes CFB and OFB to the object database. Correct an 5244 error in AES-CFB decryption. 5245 [Richard Levitte] 5246 5247 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this 5248 allows existing EVP_CIPHER_CTX structures to be reused after 5249 calling EVP_*Final(). This behaviour is used by encryption 5250 BIOs and some applications. This has the side effect that 5251 applications must explicitly clean up cipher contexts with 5252 EVP_CIPHER_CTX_cleanup() or they will leak memory. 5253 [Steve Henson] 5254 5255 *) Check the values of dna and dnb in bn_mul_recursive before calling 5256 bn_mul_comba (a non zero value means the a or b arrays do not contain 5257 n2 elements) and fallback to bn_mul_normal if either is not zero. 5258 [Steve Henson] 5259 5260 *) Fix escaping of non-ASCII characters when using the -subj option 5261 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>) 5262 [Lutz Jaenicke] 5263 5264 *) Make object definitions compliant to LDAP (RFC2256): SN is the short 5265 form for "surname", serialNumber has no short form. 5266 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798; 5267 therefore remove "mail" short name for "internet 7". 5268 The OID for unique identifiers in X509 certificates is 5269 x500UniqueIdentifier, not uniqueIdentifier. 5270 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>) 5271 [Lutz Jaenicke] 5272 5273 *) Add an "init" command to the ENGINE config module and auto initialize 5274 ENGINEs. Without any "init" command the ENGINE will be initialized 5275 after all ctrl commands have been executed on it. If init=1 the 5276 ENGINE is initailized at that point (ctrls before that point are run 5277 on the uninitialized ENGINE and after on the initialized one). If 5278 init=0 then the ENGINE will not be iniatialized at all. 5279 [Steve Henson] 5280 5281 *) Fix the 'app_verify_callback' interface so that the user-defined 5282 argument is actually passed to the callback: In the 5283 SSL_CTX_set_cert_verify_callback() prototype, the callback 5284 declaration has been changed from 5285 int (*cb)() 5286 into 5287 int (*cb)(X509_STORE_CTX *,void *); 5288 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call 5289 i=s->ctx->app_verify_callback(&ctx) 5290 has been changed into 5291 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg). 5292 5293 To update applications using SSL_CTX_set_cert_verify_callback(), 5294 a dummy argument can be added to their callback functions. 5295 [D. K. Smetters <smetters@parc.xerox.com>] 5296 5297 *) Added the '4758cca' ENGINE to support IBM 4758 cards. 5298 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe] 5299 5300 *) Add and OPENSSL_LOAD_CONF define which will cause 5301 OpenSSL_add_all_algorithms() to load the openssl.cnf config file. 5302 This allows older applications to transparently support certain 5303 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading. 5304 Two new functions OPENSSL_add_all_algorithms_noconf() which will never 5305 load the config file and OPENSSL_add_all_algorithms_conf() which will 5306 always load it have also been added. 5307 [Steve Henson] 5308 5309 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES. 5310 Adjust NIDs and EVP layer. 5311 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 5312 5313 *) Config modules support in openssl utility. 5314 5315 Most commands now load modules from the config file, 5316 though in a few (such as version) this isn't done 5317 because it couldn't be used for anything. 5318 5319 In the case of ca and req the config file used is 5320 the same as the utility itself: that is the -config 5321 command line option can be used to specify an 5322 alternative file. 5323 [Steve Henson] 5324 5325 *) Move default behaviour from OPENSSL_config(). If appname is NULL 5326 use "openssl_conf" if filename is NULL use default openssl config file. 5327 [Steve Henson] 5328 5329 *) Add an argument to OPENSSL_config() to allow the use of an alternative 5330 config section name. Add a new flag to tolerate a missing config file 5331 and move code to CONF_modules_load_file(). 5332 [Steve Henson] 5333 5334 *) Support for crypto accelerator cards from Accelerated Encryption 5335 Processing, www.aep.ie. (Use engine 'aep') 5336 The support was copied from 0.9.6c [engine] and adapted/corrected 5337 to work with the new engine framework. 5338 [AEP Inc. and Richard Levitte] 5339 5340 *) Support for SureWare crypto accelerator cards from Baltimore 5341 Technologies. (Use engine 'sureware') 5342 The support was copied from 0.9.6c [engine] and adapted 5343 to work with the new engine framework. 5344 [Richard Levitte] 5345 5346 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually 5347 make the newer ENGINE framework commands for the CHIL engine work. 5348 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte] 5349 5350 *) Make it possible to produce shared libraries on ReliantUNIX. 5351 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte] 5352 5353 *) Add the configuration target debug-linux-ppro. 5354 Make 'openssl rsa' use the general key loading routines 5355 implemented in apps.c, and make those routines able to 5356 handle the key format FORMAT_NETSCAPE and the variant 5357 FORMAT_IISSGC. 5358 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 5359 5360 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 5361 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 5362 5363 *) Add -keyform to rsautl, and document -engine. 5364 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>] 5365 5366 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new 5367 BIO_R_NO_SUCH_FILE error code rather than the generic 5368 ERR_R_SYS_LIB error code if fopen() fails with ENOENT. 5369 [Ben Laurie] 5370 5371 *) Add new functions 5372 ERR_peek_last_error 5373 ERR_peek_last_error_line 5374 ERR_peek_last_error_line_data. 5375 These are similar to 5376 ERR_peek_error 5377 ERR_peek_error_line 5378 ERR_peek_error_line_data, 5379 but report on the latest error recorded rather than the first one 5380 still in the error queue. 5381 [Ben Laurie, Bodo Moeller] 5382 5383 *) default_algorithms option in ENGINE config module. This allows things 5384 like: 5385 default_algorithms = ALL 5386 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS 5387 [Steve Henson] 5388 5389 *) Prelminary ENGINE config module. 5390 [Steve Henson] 5391 5392 *) New experimental application configuration code. 5393 [Steve Henson] 5394 5395 *) Change the AES code to follow the same name structure as all other 5396 symmetric ciphers, and behave the same way. Move everything to 5397 the directory crypto/aes, thereby obsoleting crypto/rijndael. 5398 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 5399 5400 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c. 5401 [Ben Laurie and Theo de Raadt] 5402 5403 *) Add option to output public keys in req command. 5404 [Massimiliano Pala madwolf@openca.org] 5405 5406 *) Use wNAFs in EC_POINTs_mul() for improved efficiency 5407 (up to about 10% better than before for P-192 and P-224). 5408 [Bodo Moeller] 5409 5410 *) New functions/macros 5411 5412 SSL_CTX_set_msg_callback(ctx, cb) 5413 SSL_CTX_set_msg_callback_arg(ctx, arg) 5414 SSL_set_msg_callback(ssl, cb) 5415 SSL_set_msg_callback_arg(ssl, arg) 5416 5417 to request calling a callback function 5418 5419 void cb(int write_p, int version, int content_type, 5420 const void *buf, size_t len, SSL *ssl, void *arg) 5421 5422 whenever a protocol message has been completely received 5423 (write_p == 0) or sent (write_p == 1). Here 'version' is the 5424 protocol version according to which the SSL library interprets 5425 the current protocol message (SSL2_VERSION, SSL3_VERSION, or 5426 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or 5427 the content type as defined in the SSL 3.0/TLS 1.0 protocol 5428 specification (change_cipher_spec(20), alert(21), handshake(22)). 5429 'buf' and 'len' point to the actual message, 'ssl' to the 5430 SSL object, and 'arg' is the application-defined value set by 5431 SSL[_CTX]_set_msg_callback_arg(). 5432 5433 'openssl s_client' and 'openssl s_server' have new '-msg' options 5434 to enable a callback that displays all protocol messages. 5435 [Bodo Moeller] 5436 5437 *) Change the shared library support so shared libraries are built as 5438 soon as the corresponding static library is finished, and thereby get 5439 openssl and the test programs linked against the shared library. 5440 This still only happens when the keyword "shard" has been given to 5441 the configuration scripts. 5442 5443 NOTE: shared library support is still an experimental thing, and 5444 backward binary compatibility is still not guaranteed. 5445 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte] 5446 5447 *) Add support for Subject Information Access extension. 5448 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 5449 5450 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero 5451 additional bytes when new memory had to be allocated, not just 5452 when reusing an existing buffer. 5453 [Bodo Moeller] 5454 5455 *) New command line and configuration option 'utf8' for the req command. 5456 This allows field values to be specified as UTF8 strings. 5457 [Steve Henson] 5458 5459 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel 5460 runs for the former and machine-readable output for the latter. 5461 [Ben Laurie] 5462 5463 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion 5464 of the e-mail address in the DN (i.e., it will go into a certificate 5465 extension only). The new configuration file option 'email_in_dn = no' 5466 has the same effect. 5467 [Massimiliano Pala madwolf@openca.org] 5468 5469 *) Change all functions with names starting with des_ to be starting 5470 with DES_ instead. Add wrappers that are compatible with libdes, 5471 but are named _ossl_old_des_*. Finally, add macros that map the 5472 des_* symbols to the corresponding _ossl_old_des_* if libdes 5473 compatibility is desired. If OpenSSL 0.9.6c compatibility is 5474 desired, the des_* symbols will be mapped to DES_*, with one 5475 exception. 5476 5477 Since we provide two compatibility mappings, the user needs to 5478 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes 5479 compatibility is desired. The default (i.e., when that macro 5480 isn't defined) is OpenSSL 0.9.6c compatibility. 5481 5482 There are also macros that enable and disable the support of old 5483 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT 5484 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those 5485 are defined, the default will apply: to support the old des routines. 5486 5487 In either case, one must include openssl/des.h to get the correct 5488 definitions. Do not try to just include openssl/des_old.h, that 5489 won't work. 5490 5491 NOTE: This is a major break of an old API into a new one. Software 5492 authors are encouraged to switch to the DES_ style functions. Some 5493 time in the future, des_old.h and the libdes compatibility functions 5494 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the 5495 default), and then completely removed. 5496 [Richard Levitte] 5497 5498 *) Test for certificates which contain unsupported critical extensions. 5499 If such a certificate is found during a verify operation it is 5500 rejected by default: this behaviour can be overridden by either 5501 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or 5502 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function 5503 X509_supported_extension() has also been added which returns 1 if a 5504 particular extension is supported. 5505 [Steve Henson] 5506 5507 *) Modify the behaviour of EVP cipher functions in similar way to digests 5508 to retain compatibility with existing code. 5509 [Steve Henson] 5510 5511 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain 5512 compatibility with existing code. In particular the 'ctx' parameter does 5513 not have to be to be initialized before the call to EVP_DigestInit() and 5514 it is tidied up after a call to EVP_DigestFinal(). New function 5515 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function 5516 EVP_MD_CTX_copy() changed to not require the destination to be 5517 initialized valid and new function EVP_MD_CTX_copy_ex() added which 5518 requires the destination to be valid. 5519 5520 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(), 5521 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex(). 5522 [Steve Henson] 5523 5524 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it 5525 so that complete 'Handshake' protocol structures are kept in memory 5526 instead of overwriting 'msg_type' and 'length' with 'body' data. 5527 [Bodo Moeller] 5528 5529 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32. 5530 [Massimo Santin via Richard Levitte] 5531 5532 *) Major restructuring to the underlying ENGINE code. This includes 5533 reduction of linker bloat, separation of pure "ENGINE" manipulation 5534 (initialisation, etc) from functionality dealing with implementations 5535 of specific crypto iterfaces. This change also introduces integrated 5536 support for symmetric ciphers and digest implementations - so ENGINEs 5537 can now accelerate these by providing EVP_CIPHER and EVP_MD 5538 implementations of their own. This is detailed in crypto/engine/README 5539 as it couldn't be adequately described here. However, there are a few 5540 API changes worth noting - some RSA, DSA, DH, and RAND functions that 5541 were changed in the original introduction of ENGINE code have now 5542 reverted back - the hooking from this code to ENGINE is now a good 5543 deal more passive and at run-time, operations deal directly with 5544 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than 5545 dereferencing through an ENGINE pointer any more. Also, the ENGINE 5546 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed - 5547 they were not being used by the framework as there is no concept of a 5548 BIGNUM_METHOD and they could not be generalised to the new 5549 'ENGINE_TABLE' mechanism that underlies the new code. Similarly, 5550 ENGINE_cpy() has been removed as it cannot be consistently defined in 5551 the new code. 5552 [Geoff Thorpe] 5553 5554 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds. 5555 [Steve Henson] 5556 5557 *) Change mkdef.pl to sort symbols that get the same entry number, 5558 and make sure the automatically generated functions ERR_load_* 5559 become part of libeay.num as well. 5560 [Richard Levitte] 5561 5562 *) New function SSL_renegotiate_pending(). This returns true once 5563 renegotiation has been requested (either SSL_renegotiate() call 5564 or HelloRequest/ClientHello receveived from the peer) and becomes 5565 false once a handshake has been completed. 5566 (For servers, SSL_renegotiate() followed by SSL_do_handshake() 5567 sends a HelloRequest, but does not ensure that a handshake takes 5568 place. SSL_renegotiate_pending() is useful for checking if the 5569 client has followed the request.) 5570 [Bodo Moeller] 5571 5572 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION. 5573 By default, clients may request session resumption even during 5574 renegotiation (if session ID contexts permit); with this option, 5575 session resumption is possible only in the first handshake. 5576 5577 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes 5578 more bits available for options that should not be part of 5579 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION). 5580 [Bodo Moeller] 5581 5582 *) Add some demos for certificate and certificate request creation. 5583 [Steve Henson] 5584 5585 *) Make maximum certificate chain size accepted from the peer application 5586 settable (SSL*_get/set_max_cert_list()), as proposed by 5587 "Douglas E. Engert" <deengert@anl.gov>. 5588 [Lutz Jaenicke] 5589 5590 *) Add support for shared libraries for Unixware-7 5591 (Boyd Lynn Gerber <gerberb@zenez.com>). 5592 [Lutz Jaenicke] 5593 5594 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to 5595 be done prior to destruction. Use this to unload error strings from 5596 ENGINEs that load their own error strings. NB: This adds two new API 5597 functions to "get" and "set" this destroy handler in an ENGINE. 5598 [Geoff Thorpe] 5599 5600 *) Alter all existing ENGINE implementations (except "openssl" and 5601 "openbsd") to dynamically instantiate their own error strings. This 5602 makes them more flexible to be built both as statically-linked ENGINEs 5603 and self-contained shared-libraries loadable via the "dynamic" ENGINE. 5604 Also, add stub code to each that makes building them as self-contained 5605 shared-libraries easier (see README.ENGINE). 5606 [Geoff Thorpe] 5607 5608 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE 5609 implementations into applications that are completely implemented in 5610 self-contained shared-libraries. The "dynamic" ENGINE exposes control 5611 commands that can be used to configure what shared-library to load and 5612 to control aspects of the way it is handled. Also, made an update to 5613 the README.ENGINE file that brings its information up-to-date and 5614 provides some information and instructions on the "dynamic" ENGINE 5615 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc). 5616 [Geoff Thorpe] 5617 5618 *) Make it possible to unload ranges of ERR strings with a new 5619 "ERR_unload_strings" function. 5620 [Geoff Thorpe] 5621 5622 *) Add a copy() function to EVP_MD. 5623 [Ben Laurie] 5624 5625 *) Make EVP_MD routines take a context pointer instead of just the 5626 md_data void pointer. 5627 [Ben Laurie] 5628 5629 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates 5630 that the digest can only process a single chunk of data 5631 (typically because it is provided by a piece of 5632 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application 5633 is only going to provide a single chunk of data, and hence the 5634 framework needn't accumulate the data for oneshot drivers. 5635 [Ben Laurie] 5636 5637 *) As with "ERR", make it possible to replace the underlying "ex_data" 5638 functions. This change also alters the storage and management of global 5639 ex_data state - it's now all inside ex_data.c and all "class" code (eg. 5640 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class 5641 index counters. The API functions that use this state have been changed 5642 to take a "class_index" rather than pointers to the class's local STACK 5643 and counter, and there is now an API function to dynamically create new 5644 classes. This centralisation allows us to (a) plug a lot of the 5645 thread-safety problems that existed, and (b) makes it possible to clean 5646 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b) 5647 such data would previously have always leaked in application code and 5648 workarounds were in place to make the memory debugging turn a blind eye 5649 to it. Application code that doesn't use this new function will still 5650 leak as before, but their memory debugging output will announce it now 5651 rather than letting it slide. 5652 5653 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change 5654 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now 5655 has a return value to indicate success or failure. 5656 [Geoff Thorpe] 5657 5658 *) Make it possible to replace the underlying "ERR" functions such that the 5659 global state (2 LHASH tables and 2 locks) is only used by the "default" 5660 implementation. This change also adds two functions to "get" and "set" 5661 the implementation prior to it being automatically set the first time 5662 any other ERR function takes place. Ie. an application can call "get", 5663 pass the return value to a module it has just loaded, and that module 5664 can call its own "set" function using that value. This means the 5665 module's "ERR" operations will use (and modify) the error state in the 5666 application and not in its own statically linked copy of OpenSSL code. 5667 [Geoff Thorpe] 5668 5669 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment 5670 reference counts. This performs normal REF_PRINT/REF_CHECK macros on 5671 the operation, and provides a more encapsulated way for external code 5672 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code 5673 to use these functions rather than manually incrementing the counts. 5674 5675 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()". 5676 [Geoff Thorpe] 5677 5678 *) Add EVP test program. 5679 [Ben Laurie] 5680 5681 *) Add symmetric cipher support to ENGINE. Expect the API to change! 5682 [Ben Laurie] 5683 5684 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name() 5685 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(), 5686 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate(). 5687 These allow a CRL to be built without having to access X509_CRL fields 5688 directly. Modify 'ca' application to use new functions. 5689 [Steve Henson] 5690 5691 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended 5692 bug workarounds. Rollback attack detection is a security feature. 5693 The problem will only arise on OpenSSL servers when TLSv1 is not 5694 available (sslv3_server_method() or SSL_OP_NO_TLSv1). 5695 Software authors not wanting to support TLSv1 will have special reasons 5696 for their choice and can explicitly enable this option. 5697 [Bodo Moeller, Lutz Jaenicke] 5698 5699 *) Rationalise EVP so it can be extended: don't include a union of 5700 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX 5701 (similar to those existing for EVP_CIPHER_CTX). 5702 Usage example: 5703 5704 EVP_MD_CTX md; 5705 5706 EVP_MD_CTX_init(&md); /* new function call */ 5707 EVP_DigestInit(&md, EVP_sha1()); 5708 EVP_DigestUpdate(&md, in, len); 5709 EVP_DigestFinal(&md, out, NULL); 5710 EVP_MD_CTX_cleanup(&md); /* new function call */ 5711 5712 [Ben Laurie] 5713 5714 *) Make DES key schedule conform to the usual scheme, as well as 5715 correcting its structure. This means that calls to DES functions 5716 now have to pass a pointer to a des_key_schedule instead of a 5717 plain des_key_schedule (which was actually always a pointer 5718 anyway): E.g., 5719 5720 des_key_schedule ks; 5721 5722 des_set_key_checked(..., &ks); 5723 des_ncbc_encrypt(..., &ks, ...); 5724 5725 (Note that a later change renames 'des_...' into 'DES_...'.) 5726 [Ben Laurie] 5727 5728 *) Initial reduction of linker bloat: the use of some functions, such as 5729 PEM causes large amounts of unused functions to be linked in due to 5730 poor organisation. For example pem_all.c contains every PEM function 5731 which has a knock on effect of linking in large amounts of (unused) 5732 ASN1 code. Grouping together similar functions and splitting unrelated 5733 functions prevents this. 5734 [Steve Henson] 5735 5736 *) Cleanup of EVP macros. 5737 [Ben Laurie] 5738 5739 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the 5740 correct _ecb suffix. 5741 [Ben Laurie] 5742 5743 *) Add initial OCSP responder support to ocsp application. The 5744 revocation information is handled using the text based index 5745 use by the ca application. The responder can either handle 5746 requests generated internally, supplied in files (for example 5747 via a CGI script) or using an internal minimal server. 5748 [Steve Henson] 5749 5750 *) Add configuration choices to get zlib compression for TLS. 5751 [Richard Levitte] 5752 5753 *) Changes to Kerberos SSL for RFC 2712 compliance: 5754 1. Implemented real KerberosWrapper, instead of just using 5755 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>] 5756 2. Implemented optional authenticator field of KerberosWrapper. 5757 5758 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req, 5759 and authenticator structs; see crypto/krb5/. 5760 5761 Generalized Kerberos calls to support multiple Kerberos libraries. 5762 [Vern Staats <staatsvr@asc.hpc.mil>, 5763 Jeffrey Altman <jaltman@columbia.edu> 5764 via Richard Levitte] 5765 5766 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it 5767 already does with RSA. testdsa.h now has 'priv_key/pub_key' 5768 values for each of the key sizes rather than having just 5769 parameters (and 'speed' generating keys each time). 5770 [Geoff Thorpe] 5771 5772 *) Speed up EVP routines. 5773 Before: 5774encrypt 5775type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes 5776des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k 5777des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k 5778des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k 5779decrypt 5780des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k 5781des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k 5782des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k 5783 After: 5784encrypt 5785des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k 5786decrypt 5787des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k 5788 [Ben Laurie] 5789 5790 *) Added the OS2-EMX target. 5791 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte] 5792 5793 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions 5794 to support NCONF routines in extension code. New function CONF_set_nconf() 5795 to allow functions which take an NCONF to also handle the old LHASH 5796 structure: this means that the old CONF compatible routines can be 5797 retained (in particular wrt extensions) without having to duplicate the 5798 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack. 5799 [Steve Henson] 5800 5801 *) Enhance the general user interface with mechanisms for inner control 5802 and with possibilities to have yes/no kind of prompts. 5803 [Richard Levitte] 5804 5805 *) Change all calls to low level digest routines in the library and 5806 applications to use EVP. Add missing calls to HMAC_cleanup() and 5807 don't assume HMAC_CTX can be copied using memcpy(). 5808 [Verdon Walker <VWalker@novell.com>, Steve Henson] 5809 5810 *) Add the possibility to control engines through control names but with 5811 arbitrary arguments instead of just a string. 5812 Change the key loaders to take a UI_METHOD instead of a callback 5813 function pointer. NOTE: this breaks binary compatibility with earlier 5814 versions of OpenSSL [engine]. 5815 Adapt the nCipher code for these new conditions and add a card insertion 5816 callback. 5817 [Richard Levitte] 5818 5819 *) Enhance the general user interface with mechanisms to better support 5820 dialog box interfaces, application-defined prompts, the possibility 5821 to use defaults (for example default passwords from somewhere else) 5822 and interrupts/cancellations. 5823 [Richard Levitte] 5824 5825 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name 5826 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility. 5827 [Steve Henson] 5828 5829 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also 5830 tidy up some unnecessarily weird code in 'sk_new()'). 5831 [Geoff, reported by Diego Tartara <dtartara@novamens.com>] 5832 5833 *) Change the key loading routines for ENGINEs to use the same kind 5834 callback (pem_password_cb) as all other routines that need this 5835 kind of callback. 5836 [Richard Levitte] 5837 5838 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with 5839 256 bit (=32 byte) keys. Of course seeding with more entropy bytes 5840 than this minimum value is recommended. 5841 [Lutz Jaenicke] 5842 5843 *) New random seeder for OpenVMS, using the system process statistics 5844 that are easily reachable. 5845 [Richard Levitte] 5846 5847 *) Windows apparently can't transparently handle global 5848 variables defined in DLLs. Initialisations such as: 5849 5850 const ASN1_ITEM *it = &ASN1_INTEGER_it; 5851 5852 wont compile. This is used by the any applications that need to 5853 declare their own ASN1 modules. This was fixed by adding the option 5854 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly 5855 needed for static libraries under Win32. 5856 [Steve Henson] 5857 5858 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle 5859 setting of purpose and trust fields. New X509_STORE trust and 5860 purpose functions and tidy up setting in other SSL functions. 5861 [Steve Henson] 5862 5863 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE 5864 structure. These are inherited by X509_STORE_CTX when it is 5865 initialised. This allows various defaults to be set in the 5866 X509_STORE structure (such as flags for CRL checking and custom 5867 purpose or trust settings) for functions which only use X509_STORE_CTX 5868 internally such as S/MIME. 5869 5870 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and 5871 trust settings if they are not set in X509_STORE. This allows X509_STORE 5872 purposes and trust (in S/MIME for example) to override any set by default. 5873 5874 Add command line options for CRL checking to smime, s_client and s_server 5875 applications. 5876 [Steve Henson] 5877 5878 *) Initial CRL based revocation checking. If the CRL checking flag(s) 5879 are set then the CRL is looked up in the X509_STORE structure and 5880 its validity and signature checked, then if the certificate is found 5881 in the CRL the verify fails with a revoked error. 5882 5883 Various new CRL related callbacks added to X509_STORE_CTX structure. 5884 5885 Command line options added to 'verify' application to support this. 5886 5887 This needs some additional work, such as being able to handle multiple 5888 CRLs with different times, extension based lookup (rather than just 5889 by subject name) and ultimately more complete V2 CRL extension 5890 handling. 5891 [Steve Henson] 5892 5893 *) Add a general user interface API (crypto/ui/). This is designed 5894 to replace things like des_read_password and friends (backward 5895 compatibility functions using this new API are provided). 5896 The purpose is to remove prompting functions from the DES code 5897 section as well as provide for prompting through dialog boxes in 5898 a window system and the like. 5899 [Richard Levitte] 5900 5901 *) Add "ex_data" support to ENGINE so implementations can add state at a 5902 per-structure level rather than having to store it globally. 5903 [Geoff] 5904 5905 *) Make it possible for ENGINE structures to be copied when retrieved by 5906 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY. 5907 This causes the "original" ENGINE structure to act like a template, 5908 analogous to the RSA vs. RSA_METHOD type of separation. Because of this 5909 operational state can be localised to each ENGINE structure, despite the 5910 fact they all share the same "methods". New ENGINE structures returned in 5911 this case have no functional references and the return value is the single 5912 structural reference. This matches the single structural reference returned 5913 by ENGINE_by_id() normally, when it is incremented on the pre-existing 5914 ENGINE structure. 5915 [Geoff] 5916 5917 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this 5918 needs to match any other type at all we need to manually clear the 5919 tag cache. 5920 [Steve Henson] 5921 5922 *) Changes to the "openssl engine" utility to include; 5923 - verbosity levels ('-v', '-vv', and '-vvv') that provide information 5924 about an ENGINE's available control commands. 5925 - executing control commands from command line arguments using the 5926 '-pre' and '-post' switches. '-post' is only used if '-t' is 5927 specified and the ENGINE is successfully initialised. The syntax for 5928 the individual commands are colon-separated, for example; 5929 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so 5930 [Geoff] 5931 5932 *) New dynamic control command support for ENGINEs. ENGINEs can now 5933 declare their own commands (numbers), names (strings), descriptions, 5934 and input types for run-time discovery by calling applications. A 5935 subset of these commands are implicitly classed as "executable" 5936 depending on their input type, and only these can be invoked through 5937 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this 5938 can be based on user input, config files, etc). The distinction is 5939 that "executable" commands cannot return anything other than a boolean 5940 result and can only support numeric or string input, whereas some 5941 discoverable commands may only be for direct use through 5942 ENGINE_ctrl(), eg. supporting the exchange of binary data, function 5943 pointers, or other custom uses. The "executable" commands are to 5944 support parameterisations of ENGINE behaviour that can be 5945 unambiguously defined by ENGINEs and used consistently across any 5946 OpenSSL-based application. Commands have been added to all the 5947 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow 5948 control over shared-library paths without source code alterations. 5949 [Geoff] 5950 5951 *) Changed all ENGINE implementations to dynamically allocate their 5952 ENGINEs rather than declaring them statically. Apart from this being 5953 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction, 5954 this also allows the implementations to compile without using the 5955 internal engine_int.h header. 5956 [Geoff] 5957 5958 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a 5959 'const' value. Any code that should be able to modify a RAND_METHOD 5960 should already have non-const pointers to it (ie. they should only 5961 modify their own ones). 5962 [Geoff] 5963 5964 *) Made a variety of little tweaks to the ENGINE code. 5965 - "atalla" and "ubsec" string definitions were moved from header files 5966 to C code. "nuron" string definitions were placed in variables 5967 rather than hard-coded - allowing parameterisation of these values 5968 later on via ctrl() commands. 5969 - Removed unused "#if 0"'d code. 5970 - Fixed engine list iteration code so it uses ENGINE_free() to release 5971 structural references. 5972 - Constified the RAND_METHOD element of ENGINE structures. 5973 - Constified various get/set functions as appropriate and added 5974 missing functions (including a catch-all ENGINE_cpy that duplicates 5975 all ENGINE values onto a new ENGINE except reference counts/state). 5976 - Removed NULL parameter checks in get/set functions. Setting a method 5977 or function to NULL is a way of cancelling out a previously set 5978 value. Passing a NULL ENGINE parameter is just plain stupid anyway 5979 and doesn't justify the extra error symbols and code. 5980 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for 5981 flags from engine_int.h to engine.h. 5982 - Changed prototypes for ENGINE handler functions (init(), finish(), 5983 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter. 5984 [Geoff] 5985 5986 *) Implement binary inversion algorithm for BN_mod_inverse in addition 5987 to the algorithm using long division. The binary algorithm can be 5988 used only if the modulus is odd. On 32-bit systems, it is faster 5989 only for relatively small moduli (roughly 20-30% for 128-bit moduli, 5990 roughly 5-15% for 256-bit moduli), so we use it only for moduli 5991 up to 450 bits. In 64-bit environments, the binary algorithm 5992 appears to be advantageous for much longer moduli; here we use it 5993 for moduli up to 2048 bits. 5994 [Bodo Moeller] 5995 5996 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code 5997 could not support the combine flag in choice fields. 5998 [Steve Henson] 5999 6000 *) Add a 'copy_extensions' option to the 'ca' utility. This copies 6001 extensions from a certificate request to the certificate. 6002 [Steve Henson] 6003 6004 *) Allow multiple 'certopt' and 'nameopt' options to be separated 6005 by commas. Add 'namopt' and 'certopt' options to the 'ca' config 6006 file: this allows the display of the certificate about to be 6007 signed to be customised, to allow certain fields to be included 6008 or excluded and extension details. The old system didn't display 6009 multicharacter strings properly, omitted fields not in the policy 6010 and couldn't display additional details such as extensions. 6011 [Steve Henson] 6012 6013 *) Function EC_POINTs_mul for multiple scalar multiplication 6014 of an arbitrary number of elliptic curve points 6015 \sum scalars[i]*points[i], 6016 optionally including the generator defined for the EC_GROUP: 6017 scalar*generator + \sum scalars[i]*points[i]. 6018 6019 EC_POINT_mul is a simple wrapper function for the typical case 6020 that the point list has just one item (besides the optional 6021 generator). 6022 [Bodo Moeller] 6023 6024 *) First EC_METHODs for curves over GF(p): 6025 6026 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr 6027 operations and provides various method functions that can also 6028 operate with faster implementations of modular arithmetic. 6029 6030 EC_GFp_mont_method() reuses most functions that are part of 6031 EC_GFp_simple_method, but uses Montgomery arithmetic. 6032 6033 [Bodo Moeller; point addition and point doubling 6034 implementation directly derived from source code provided by 6035 Lenka Fibikova <fibikova@exp-math.uni-essen.de>] 6036 6037 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h, 6038 crypto/ec/ec_lib.c): 6039 6040 Curves are EC_GROUP objects (with an optional group generator) 6041 based on EC_METHODs that are built into the library. 6042 6043 Points are EC_POINT objects based on EC_GROUP objects. 6044 6045 Most of the framework would be able to handle curves over arbitrary 6046 finite fields, but as there are no obvious types for fields other 6047 than GF(p), some functions are limited to that for now. 6048 [Bodo Moeller] 6049 6050 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires 6051 that the file contains a complete HTTP response. 6052 [Richard Levitte] 6053 6054 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl 6055 change the def and num file printf format specifier from "%-40sXXX" 6056 to "%-39s XXX". The latter will always guarantee a space after the 6057 field while the former will cause them to run together if the field 6058 is 40 of more characters long. 6059 [Steve Henson] 6060 6061 *) Constify the cipher and digest 'method' functions and structures 6062 and modify related functions to take constant EVP_MD and EVP_CIPHER 6063 pointers. 6064 [Steve Henson] 6065 6066 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them 6067 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32. 6068 [Bodo Moeller] 6069 6070 *) Modify EVP_Digest*() routines so they now return values. Although the 6071 internal software routines can never fail additional hardware versions 6072 might. 6073 [Steve Henson] 6074 6075 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts: 6076 6077 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7 6078 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32. 6079 6080 ASN1 error codes 6081 ERR_R_NESTED_ASN1_ERROR 6082 ... 6083 ERR_R_MISSING_ASN1_EOS 6084 were 4 .. 9, conflicting with 6085 ERR_LIB_RSA (= ERR_R_RSA_LIB) 6086 ... 6087 ERR_LIB_PEM (= ERR_R_PEM_LIB). 6088 They are now 58 .. 63 (i.e., just below ERR_R_FATAL). 6089 6090 Add new error code 'ERR_R_INTERNAL_ERROR'. 6091 [Bodo Moeller] 6092 6093 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock 6094 suffices. 6095 [Bodo Moeller] 6096 6097 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This 6098 sets the subject name for a new request or supersedes the 6099 subject name in a given request. Formats that can be parsed are 6100 'CN=Some Name, OU=myOU, C=IT' 6101 and 6102 'CN=Some Name/OU=myOU/C=IT'. 6103 6104 Add options '-batch' and '-verbose' to 'openssl req'. 6105 [Massimiliano Pala <madwolf@hackmasters.net>] 6106 6107 *) Introduce the possibility to access global variables through 6108 functions on platform were that's the best way to handle exporting 6109 global variables in shared libraries. To enable this functionality, 6110 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro 6111 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter 6112 is normally done by Configure or something similar). 6113 6114 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL 6115 in the source file (foo.c) like this: 6116 6117 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1; 6118 OPENSSL_IMPLEMENT_GLOBAL(double,bar); 6119 6120 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL 6121 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this: 6122 6123 OPENSSL_DECLARE_GLOBAL(int,foo); 6124 #define foo OPENSSL_GLOBAL_REF(foo) 6125 OPENSSL_DECLARE_GLOBAL(double,bar); 6126 #define bar OPENSSL_GLOBAL_REF(bar) 6127 6128 The #defines are very important, and therefore so is including the 6129 header file everywhere where the defined globals are used. 6130 6131 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition 6132 of ASN.1 items, but that structure is a bit different. 6133 6134 The largest change is in util/mkdef.pl which has been enhanced with 6135 better and easier to understand logic to choose which symbols should 6136 go into the Windows .def files as well as a number of fixes and code 6137 cleanup (among others, algorithm keywords are now sorted 6138 lexicographically to avoid constant rewrites). 6139 [Richard Levitte] 6140 6141 *) In BN_div() keep a copy of the sign of 'num' before writing the 6142 result to 'rm' because if rm==num the value will be overwritten 6143 and produce the wrong result if 'num' is negative: this caused 6144 problems with BN_mod() and BN_nnmod(). 6145 [Steve Henson] 6146 6147 *) Function OCSP_request_verify(). This checks the signature on an 6148 OCSP request and verifies the signer certificate. The signer 6149 certificate is just checked for a generic purpose and OCSP request 6150 trust settings. 6151 [Steve Henson] 6152 6153 *) Add OCSP_check_validity() function to check the validity of OCSP 6154 responses. OCSP responses are prepared in real time and may only 6155 be a few seconds old. Simply checking that the current time lies 6156 between thisUpdate and nextUpdate max reject otherwise valid responses 6157 caused by either OCSP responder or client clock inaccuracy. Instead 6158 we allow thisUpdate and nextUpdate to fall within a certain period of 6159 the current time. The age of the response can also optionally be 6160 checked. Two new options -validity_period and -status_age added to 6161 ocsp utility. 6162 [Steve Henson] 6163 6164 *) If signature or public key algorithm is unrecognized print out its 6165 OID rather that just UNKNOWN. 6166 [Steve Henson] 6167 6168 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and 6169 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate 6170 ID to be generated from the issuer certificate alone which can then be 6171 passed to OCSP_id_issuer_cmp(). 6172 [Steve Henson] 6173 6174 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new 6175 ASN1 modules to export functions returning ASN1_ITEM pointers 6176 instead of the ASN1_ITEM structures themselves. This adds several 6177 new macros which allow the underlying ASN1 function/structure to 6178 be accessed transparently. As a result code should not use ASN1_ITEM 6179 references directly (such as &X509_it) but instead use the relevant 6180 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow 6181 use of the new ASN1 code on platforms where exporting structures 6182 is problematical (for example in shared libraries) but exporting 6183 functions returning pointers to structures is not. 6184 [Steve Henson] 6185 6186 *) Add support for overriding the generation of SSL/TLS session IDs. 6187 These callbacks can be registered either in an SSL_CTX or per SSL. 6188 The purpose of this is to allow applications to control, if they wish, 6189 the arbitrary values chosen for use as session IDs, particularly as it 6190 can be useful for session caching in multiple-server environments. A 6191 command-line switch for testing this (and any client code that wishes 6192 to use such a feature) has been added to "s_server". 6193 [Geoff Thorpe, Lutz Jaenicke] 6194 6195 *) Modify mkdef.pl to recognise and parse preprocessor conditionals 6196 of the form '#if defined(...) || defined(...) || ...' and 6197 '#if !defined(...) && !defined(...) && ...'. This also avoids 6198 the growing number of special cases it was previously handling. 6199 [Richard Levitte] 6200 6201 *) Make all configuration macros available for application by making 6202 sure they are available in opensslconf.h, by giving them names starting 6203 with "OPENSSL_" to avoid conflicts with other packages and by making 6204 sure e_os2.h will cover all platform-specific cases together with 6205 opensslconf.h. 6206 Additionally, it is now possible to define configuration/platform- 6207 specific names (called "system identities"). In the C code, these 6208 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another 6209 macro with the name beginning with "OPENSSL_SYS_", which is determined 6210 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on 6211 what is available. 6212 [Richard Levitte] 6213 6214 *) New option -set_serial to 'req' and 'x509' this allows the serial 6215 number to use to be specified on the command line. Previously self 6216 signed certificates were hard coded with serial number 0 and the 6217 CA options of 'x509' had to use a serial number in a file which was 6218 auto incremented. 6219 [Steve Henson] 6220 6221 *) New options to 'ca' utility to support V2 CRL entry extensions. 6222 Currently CRL reason, invalidity date and hold instruction are 6223 supported. Add new CRL extensions to V3 code and some new objects. 6224 [Steve Henson] 6225 6226 *) New function EVP_CIPHER_CTX_set_padding() this is used to 6227 disable standard block padding (aka PKCS#5 padding) in the EVP 6228 API, which was previously mandatory. This means that the data is 6229 not padded in any way and so the total length much be a multiple 6230 of the block size, otherwise an error occurs. 6231 [Steve Henson] 6232 6233 *) Initial (incomplete) OCSP SSL support. 6234 [Steve Henson] 6235 6236 *) New function OCSP_parse_url(). This splits up a URL into its host, 6237 port and path components: primarily to parse OCSP URLs. New -url 6238 option to ocsp utility. 6239 [Steve Henson] 6240 6241 *) New nonce behavior. The return value of OCSP_check_nonce() now 6242 reflects the various checks performed. Applications can decide 6243 whether to tolerate certain situations such as an absent nonce 6244 in a response when one was present in a request: the ocsp application 6245 just prints out a warning. New function OCSP_add1_basic_nonce() 6246 this is to allow responders to include a nonce in a response even if 6247 the request is nonce-less. 6248 [Steve Henson] 6249 6250 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are 6251 skipped when using openssl x509 multiple times on a single input file, 6252 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs". 6253 [Bodo Moeller] 6254 6255 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string() 6256 set string type: to handle setting ASN1_TIME structures. Fix ca 6257 utility to correctly initialize revocation date of CRLs. 6258 [Steve Henson] 6259 6260 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override 6261 the clients preferred ciphersuites and rather use its own preferences. 6262 Should help to work around M$ SGC (Server Gated Cryptography) bug in 6263 Internet Explorer by ensuring unchanged hash method during stepup. 6264 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.) 6265 [Lutz Jaenicke] 6266 6267 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael 6268 to aes and add a new 'exist' option to print out symbols that don't 6269 appear to exist. 6270 [Steve Henson] 6271 6272 *) Additional options to ocsp utility to allow flags to be set and 6273 additional certificates supplied. 6274 [Steve Henson] 6275 6276 *) Add the option -VAfile to 'openssl ocsp', so the user can give the 6277 OCSP client a number of certificate to only verify the response 6278 signature against. 6279 [Richard Levitte] 6280 6281 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to 6282 handle the new API. Currently only ECB, CBC modes supported. Add new 6283 AES OIDs. 6284 6285 Add TLS AES ciphersuites as described in RFC3268, "Advanced 6286 Encryption Standard (AES) Ciphersuites for Transport Layer 6287 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were 6288 not enabled by default and were not part of the "ALL" ciphersuite 6289 alias because they were not yet official; they could be 6290 explicitly requested by specifying the "AESdraft" ciphersuite 6291 group alias. In the final release of OpenSSL 0.9.7, the group 6292 alias is called "AES" and is part of "ALL".) 6293 [Ben Laurie, Steve Henson, Bodo Moeller] 6294 6295 *) New function OCSP_copy_nonce() to copy nonce value (if present) from 6296 request to response. 6297 [Steve Henson] 6298 6299 *) Functions for OCSP responders. OCSP_request_onereq_count(), 6300 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info() 6301 extract information from a certificate request. OCSP_response_create() 6302 creates a response and optionally adds a basic response structure. 6303 OCSP_basic_add1_status() adds a complete single response to a basic 6304 response and returns the OCSP_SINGLERESP structure just added (to allow 6305 extensions to be included for example). OCSP_basic_add1_cert() adds a 6306 certificate to a basic response and OCSP_basic_sign() signs a basic 6307 response with various flags. New helper functions ASN1_TIME_check() 6308 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime() 6309 (converts ASN1_TIME to GeneralizedTime). 6310 [Steve Henson] 6311 6312 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}() 6313 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key 6314 structure from a certificate. X509_pubkey_digest() digests the public_key 6315 contents: this is used in various key identifiers. 6316 [Steve Henson] 6317 6318 *) Make sk_sort() tolerate a NULL argument. 6319 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>] 6320 6321 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates 6322 passed by the function are trusted implicitly. If any of them signed the 6323 response then it is assumed to be valid and is not verified. 6324 [Steve Henson] 6325 6326 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT 6327 to data. This was previously part of the PKCS7 ASN1 code. This 6328 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures. 6329 [Steve Henson, reported by Kenneth R. Robinette 6330 <support@securenetterm.com>] 6331 6332 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1 6333 routines: without these tracing memory leaks is very painful. 6334 Fix leaks in PKCS12 and PKCS7 routines. 6335 [Steve Henson] 6336 6337 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new(). 6338 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which 6339 effectively meant GeneralizedTime would never be used. Now it 6340 is initialised to -1 but X509_time_adj() now has to check the value 6341 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or 6342 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime. 6343 [Steve Henson, reported by Kenneth R. Robinette 6344 <support@securenetterm.com>] 6345 6346 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously 6347 result in a zero length in the ASN1_INTEGER structure which was 6348 not consistent with the structure when d2i_ASN1_INTEGER() was used 6349 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER() 6350 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER() 6351 where it did not print out a minus for negative ASN1_INTEGER. 6352 [Steve Henson] 6353 6354 *) Add summary printout to ocsp utility. The various functions which 6355 convert status values to strings have been renamed to: 6356 OCSP_response_status_str(), OCSP_cert_status_str() and 6357 OCSP_crl_reason_str() and are no longer static. New options 6358 to verify nonce values and to disable verification. OCSP response 6359 printout format cleaned up. 6360 [Steve Henson] 6361 6362 *) Add additional OCSP certificate checks. These are those specified 6363 in RFC2560. This consists of two separate checks: the CA of the 6364 certificate being checked must either be the OCSP signer certificate 6365 or the issuer of the OCSP signer certificate. In the latter case the 6366 OCSP signer certificate must contain the OCSP signing extended key 6367 usage. This check is performed by attempting to match the OCSP 6368 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash 6369 in the OCSP_CERTID structures of the response. 6370 [Steve Henson] 6371 6372 *) Initial OCSP certificate verification added to OCSP_basic_verify() 6373 and related routines. This uses the standard OpenSSL certificate 6374 verify routines to perform initial checks (just CA validity) and 6375 to obtain the certificate chain. Then additional checks will be 6376 performed on the chain. Currently the root CA is checked to see 6377 if it is explicitly trusted for OCSP signing. This is used to set 6378 a root CA as a global signing root: that is any certificate that 6379 chains to that CA is an acceptable OCSP signing certificate. 6380 [Steve Henson] 6381 6382 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3 6383 extensions from a separate configuration file. 6384 As when reading extensions from the main configuration file, 6385 the '-extensions ...' option may be used for specifying the 6386 section to use. 6387 [Massimiliano Pala <madwolf@comune.modena.it>] 6388 6389 *) New OCSP utility. Allows OCSP requests to be generated or 6390 read. The request can be sent to a responder and the output 6391 parsed, outputed or printed in text form. Not complete yet: 6392 still needs to check the OCSP response validity. 6393 [Steve Henson] 6394 6395 *) New subcommands for 'openssl ca': 6396 'openssl ca -status <serial>' prints the status of the cert with 6397 the given serial number (according to the index file). 6398 'openssl ca -updatedb' updates the expiry status of certificates 6399 in the index file. 6400 [Massimiliano Pala <madwolf@comune.modena.it>] 6401 6402 *) New '-newreq-nodes' command option to CA.pl. This is like 6403 '-newreq', but calls 'openssl req' with the '-nodes' option 6404 so that the resulting key is not encrypted. 6405 [Damien Miller <djm@mindrot.org>] 6406 6407 *) New configuration for the GNU Hurd. 6408 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte] 6409 6410 *) Initial code to implement OCSP basic response verify. This 6411 is currently incomplete. Currently just finds the signer's 6412 certificate and verifies the signature on the response. 6413 [Steve Henson] 6414 6415 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in 6416 value of OPENSSLDIR. This is available via the new '-d' option 6417 to 'openssl version', and is also included in 'openssl version -a'. 6418 [Bodo Moeller] 6419 6420 *) Allowing defining memory allocation callbacks that will be given 6421 file name and line number information in additional arguments 6422 (a const char* and an int). The basic functionality remains, as 6423 well as the original possibility to just replace malloc(), 6424 realloc() and free() by functions that do not know about these 6425 additional arguments. To register and find out the current 6426 settings for extended allocation functions, the following 6427 functions are provided: 6428 6429 CRYPTO_set_mem_ex_functions 6430 CRYPTO_set_locked_mem_ex_functions 6431 CRYPTO_get_mem_ex_functions 6432 CRYPTO_get_locked_mem_ex_functions 6433 6434 These work the same way as CRYPTO_set_mem_functions and friends. 6435 CRYPTO_get_[locked_]mem_functions now writes 0 where such an 6436 extended allocation function is enabled. 6437 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where 6438 a conventional allocation function is enabled. 6439 [Richard Levitte, Bodo Moeller] 6440 6441 *) Finish off removing the remaining LHASH function pointer casts. 6442 There should no longer be any prototype-casting required when using 6443 the LHASH abstraction, and any casts that remain are "bugs". See 6444 the callback types and macros at the head of lhash.h for details 6445 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example). 6446 [Geoff Thorpe] 6447 6448 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant. 6449 If /dev/[u]random devices are not available or do not return enough 6450 entropy, EGD style sockets (served by EGD or PRNGD) will automatically 6451 be queried. 6452 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and 6453 /etc/entropy will be queried once each in this sequence, quering stops 6454 when enough entropy was collected without querying more sockets. 6455 [Lutz Jaenicke] 6456 6457 *) Change the Unix RAND_poll() variant to be able to poll several 6458 random devices, as specified by DEVRANDOM, until a sufficient amount 6459 of data has been collected. We spend at most 10 ms on each file 6460 (select timeout) and read in non-blocking mode. DEVRANDOM now 6461 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom" 6462 (previously it was just the string "/dev/urandom"), so on typical 6463 platforms the 10 ms delay will never occur. 6464 Also separate out the Unix variant to its own file, rand_unix.c. 6465 For VMS, there's a currently-empty rand_vms.c. 6466 [Richard Levitte] 6467 6468 *) Move OCSP client related routines to ocsp_cl.c. These 6469 provide utility functions which an application needing 6470 to issue a request to an OCSP responder and analyse the 6471 response will typically need: as opposed to those which an 6472 OCSP responder itself would need which will be added later. 6473 6474 OCSP_request_sign() signs an OCSP request with an API similar 6475 to PKCS7_sign(). OCSP_response_status() returns status of OCSP 6476 response. OCSP_response_get1_basic() extracts basic response 6477 from response. OCSP_resp_find_status(): finds and extracts status 6478 information from an OCSP_CERTID structure (which will be created 6479 when the request structure is built). These are built from lower 6480 level functions which work on OCSP_SINGLERESP structures but 6481 wont normally be used unless the application wishes to examine 6482 extensions in the OCSP response for example. 6483 6484 Replace nonce routines with a pair of functions. 6485 OCSP_request_add1_nonce() adds a nonce value and optionally 6486 generates a random value. OCSP_check_nonce() checks the 6487 validity of the nonce in an OCSP response. 6488 [Steve Henson] 6489 6490 *) Change function OCSP_request_add() to OCSP_request_add0_id(). 6491 This doesn't copy the supplied OCSP_CERTID and avoids the 6492 need to free up the newly created id. Change return type 6493 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure. 6494 This can then be used to add extensions to the request. 6495 Deleted OCSP_request_new(), since most of its functionality 6496 is now in OCSP_REQUEST_new() (and the case insensitive name 6497 clash) apart from the ability to set the request name which 6498 will be added elsewhere. 6499 [Steve Henson] 6500 6501 *) Update OCSP API. Remove obsolete extensions argument from 6502 various functions. Extensions are now handled using the new 6503 OCSP extension code. New simple OCSP HTTP function which 6504 can be used to send requests and parse the response. 6505 [Steve Henson] 6506 6507 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new 6508 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN 6509 uses the special reorder version of SET OF to sort the attributes 6510 and reorder them to match the encoded order. This resolves a long 6511 standing problem: a verify on a PKCS7 structure just after signing 6512 it used to fail because the attribute order did not match the 6513 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes: 6514 it uses the received order. This is necessary to tolerate some broken 6515 software that does not order SET OF. This is handled by encoding 6516 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class) 6517 to produce the required SET OF. 6518 [Steve Henson] 6519 6520 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and 6521 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header 6522 files to get correct declarations of the ASN.1 item variables. 6523 [Richard Levitte] 6524 6525 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many 6526 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs: 6527 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was 6528 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i(). 6529 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant 6530 ASN1_ITEM and no wrapper functions. 6531 [Steve Henson] 6532 6533 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These 6534 replace the old function pointer based I/O routines. Change most of 6535 the *_d2i_bio() and *_d2i_fp() functions to use these. 6536 [Steve Henson] 6537 6538 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor 6539 lines, recognice more "algorithms" that can be deselected, and make 6540 it complain about algorithm deselection that isn't recognised. 6541 [Richard Levitte] 6542 6543 *) New ASN1 functions to handle dup, sign, verify, digest, pack and 6544 unpack operations in terms of ASN1_ITEM. Modify existing wrappers 6545 to use new functions. Add NO_ASN1_OLD which can be set to remove 6546 some old style ASN1 functions: this can be used to determine if old 6547 code will still work when these eventually go away. 6548 [Steve Henson] 6549 6550 *) New extension functions for OCSP structures, these follow the 6551 same conventions as certificates and CRLs. 6552 [Steve Henson] 6553 6554 *) New function X509V3_add1_i2d(). This automatically encodes and 6555 adds an extension. Its behaviour can be customised with various 6556 flags to append, replace or delete. Various wrappers added for 6557 certifcates and CRLs. 6558 [Steve Henson] 6559 6560 *) Fix to avoid calling the underlying ASN1 print routine when 6561 an extension cannot be parsed. Correct a typo in the 6562 OCSP_SERVICELOC extension. Tidy up print OCSP format. 6563 [Steve Henson] 6564 6565 *) Make mkdef.pl parse some of the ASN1 macros and add apropriate 6566 entries for variables. 6567 [Steve Henson] 6568 6569 *) Add functionality to apps/openssl.c for detecting locking 6570 problems: As the program is single-threaded, all we have 6571 to do is register a locking callback using an array for 6572 storing which locks are currently held by the program. 6573 [Bodo Moeller] 6574 6575 *) Use a lock around the call to CRYPTO_get_ex_new_index() in 6576 SSL_get_ex_data_X509_STORE_idx(), which is used in 6577 ssl_verify_cert_chain() and thus can be called at any time 6578 during TLS/SSL handshakes so that thread-safety is essential. 6579 Unfortunately, the ex_data design is not at all suited 6580 for multi-threaded use, so it probably should be abolished. 6581 [Bodo Moeller] 6582 6583 *) Added Broadcom "ubsec" ENGINE to OpenSSL. 6584 [Broadcom, tweaked and integrated by Geoff Thorpe] 6585 6586 *) Move common extension printing code to new function 6587 X509V3_print_extensions(). Reorganise OCSP print routines and 6588 implement some needed OCSP ASN1 functions. Add OCSP extensions. 6589 [Steve Henson] 6590 6591 *) New function X509_signature_print() to remove duplication in some 6592 print routines. 6593 [Steve Henson] 6594 6595 *) Add a special meaning when SET OF and SEQUENCE OF flags are both 6596 set (this was treated exactly the same as SET OF previously). This 6597 is used to reorder the STACK representing the structure to match the 6598 encoding. This will be used to get round a problem where a PKCS7 6599 structure which was signed could not be verified because the STACK 6600 order did not reflect the encoded order. 6601 [Steve Henson] 6602 6603 *) Reimplement the OCSP ASN1 module using the new code. 6604 [Steve Henson] 6605 6606 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure 6607 for its ASN1 operations. The old style function pointers still exist 6608 for now but they will eventually go away. 6609 [Steve Henson] 6610 6611 *) Merge in replacement ASN1 code from the ASN1 branch. This almost 6612 completely replaces the old ASN1 functionality with a table driven 6613 encoder and decoder which interprets an ASN1_ITEM structure describing 6614 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is 6615 largely maintained. Almost all of the old asn1_mac.h macro based ASN1 6616 has also been converted to the new form. 6617 [Steve Henson] 6618 6619 *) Change BN_mod_exp_recp so that negative moduli are tolerated 6620 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set 6621 so that BN_mod_exp_mont and BN_mod_exp_mont_word work 6622 for negative moduli. 6623 [Bodo Moeller] 6624 6625 *) Fix BN_uadd and BN_usub: Always return non-negative results instead 6626 of not touching the result's sign bit. 6627 [Bodo Moeller] 6628 6629 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be 6630 set. 6631 [Bodo Moeller] 6632 6633 *) Changed the LHASH code to use prototypes for callbacks, and created 6634 macros to declare and implement thin (optionally static) functions 6635 that provide type-safety and avoid function pointer casting for the 6636 type-specific callbacks. 6637 [Geoff Thorpe] 6638 6639 *) Added Kerberos Cipher Suites to be used with TLS, as written in 6640 RFC 2712. 6641 [Veers Staats <staatsvr@asc.hpc.mil>, 6642 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte] 6643 6644 *) Reformat the FAQ so the different questions and answers can be divided 6645 in sections depending on the subject. 6646 [Richard Levitte] 6647 6648 *) Have the zlib compression code load ZLIB.DLL dynamically under 6649 Windows. 6650 [Richard Levitte] 6651 6652 *) New function BN_mod_sqrt for computing square roots modulo a prime 6653 (using the probabilistic Tonelli-Shanks algorithm unless 6654 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can 6655 be handled deterministically). 6656 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 6657 6658 *) Make BN_mod_inverse faster by explicitly handling small quotients 6659 in the Euclid loop. (Speed gain about 20% for small moduli [256 or 6660 512 bits], about 30% for larger ones [1024 or 2048 bits].) 6661 [Bodo Moeller] 6662 6663 *) New function BN_kronecker. 6664 [Bodo Moeller] 6665 6666 *) Fix BN_gcd so that it works on negative inputs; the result is 6667 positive unless both parameters are zero. 6668 Previously something reasonably close to an infinite loop was 6669 possible because numbers could be growing instead of shrinking 6670 in the implementation of Euclid's algorithm. 6671 [Bodo Moeller] 6672 6673 *) Fix BN_is_word() and BN_is_one() macros to take into account the 6674 sign of the number in question. 6675 6676 Fix BN_is_word(a,w) to work correctly for w == 0. 6677 6678 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w) 6679 because its test if the absolute value of 'a' equals 'w'. 6680 Note that BN_abs_is_word does *not* handle w == 0 reliably; 6681 it exists mostly for use in the implementations of BN_is_zero(), 6682 BN_is_one(), and BN_is_word(). 6683 [Bodo Moeller] 6684 6685 *) New function BN_swap. 6686 [Bodo Moeller] 6687 6688 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that 6689 the exponentiation functions are more likely to produce reasonable 6690 results on negative inputs. 6691 [Bodo Moeller] 6692 6693 *) Change BN_mod_mul so that the result is always non-negative. 6694 Previously, it could be negative if one of the factors was negative; 6695 I don't think anyone really wanted that behaviour. 6696 [Bodo Moeller] 6697 6698 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c 6699 (except for exponentiation, which stays in crypto/bn/bn_exp.c, 6700 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c) 6701 and add new functions: 6702 6703 BN_nnmod 6704 BN_mod_sqr 6705 BN_mod_add 6706 BN_mod_add_quick 6707 BN_mod_sub 6708 BN_mod_sub_quick 6709 BN_mod_lshift1 6710 BN_mod_lshift1_quick 6711 BN_mod_lshift 6712 BN_mod_lshift_quick 6713 6714 These functions always generate non-negative results. 6715 6716 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r 6717 such that |m| < r < 0, BN_nnmod will output rem + |m| instead). 6718 6719 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as 6720 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b] 6721 be reduced modulo m. 6722 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 6723 6724#if 0 6725 The following entry accidentily appeared in the CHANGES file 6726 distributed with OpenSSL 0.9.7. The modifications described in 6727 it do *not* apply to OpenSSL 0.9.7. 6728 6729 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 6730 was actually never needed) and in BN_mul(). The removal in BN_mul() 6731 required a small change in bn_mul_part_recursive() and the addition 6732 of the functions bn_cmp_part_words(), bn_sub_part_words() and 6733 bn_add_part_words(), which do the same thing as bn_cmp_words(), 6734 bn_sub_words() and bn_add_words() except they take arrays with 6735 differing sizes. 6736 [Richard Levitte] 6737#endif 6738 6739 *) In 'openssl passwd', verify passwords read from the terminal 6740 unless the '-salt' option is used (which usually means that 6741 verification would just waste user's time since the resulting 6742 hash is going to be compared with some given password hash) 6743 or the new '-noverify' option is used. 6744 6745 This is an incompatible change, but it does not affect 6746 non-interactive use of 'openssl passwd' (passwords on the command 6747 line, '-stdin' option, '-in ...' option) and thus should not 6748 cause any problems. 6749 [Bodo Moeller] 6750 6751 *) Remove all references to RSAref, since there's no more need for it. 6752 [Richard Levitte] 6753 6754 *) Make DSO load along a path given through an environment variable 6755 (SHLIB_PATH) with shl_load(). 6756 [Richard Levitte] 6757 6758 *) Constify the ENGINE code as a result of BIGNUM constification. 6759 Also constify the RSA code and most things related to it. In a 6760 few places, most notable in the depth of the ASN.1 code, ugly 6761 casts back to non-const were required (to be solved at a later 6762 time) 6763 [Richard Levitte] 6764 6765 *) Make it so the openssl application has all engines loaded by default. 6766 [Richard Levitte] 6767 6768 *) Constify the BIGNUM routines a little more. 6769 [Richard Levitte] 6770 6771 *) Add the following functions: 6772 6773 ENGINE_load_cswift() 6774 ENGINE_load_chil() 6775 ENGINE_load_atalla() 6776 ENGINE_load_nuron() 6777 ENGINE_load_builtin_engines() 6778 6779 That way, an application can itself choose if external engines that 6780 are built-in in OpenSSL shall ever be used or not. The benefit is 6781 that applications won't have to be linked with libdl or other dso 6782 libraries unless it's really needed. 6783 6784 Changed 'openssl engine' to load all engines on demand. 6785 Changed the engine header files to avoid the duplication of some 6786 declarations (they differed!). 6787 [Richard Levitte] 6788 6789 *) 'openssl engine' can now list capabilities. 6790 [Richard Levitte] 6791 6792 *) Better error reporting in 'openssl engine'. 6793 [Richard Levitte] 6794 6795 *) Never call load_dh_param(NULL) in s_server. 6796 [Bodo Moeller] 6797 6798 *) Add engine application. It can currently list engines by name and 6799 identity, and test if they are actually available. 6800 [Richard Levitte] 6801 6802 *) Improve RPM specification file by forcing symbolic linking and making 6803 sure the installed documentation is also owned by root.root. 6804 [Damien Miller <djm@mindrot.org>] 6805 6806 *) Give the OpenSSL applications more possibilities to make use of 6807 keys (public as well as private) handled by engines. 6808 [Richard Levitte] 6809 6810 *) Add OCSP code that comes from CertCo. 6811 [Richard Levitte] 6812 6813 *) Add VMS support for the Rijndael code. 6814 [Richard Levitte] 6815 6816 *) Added untested support for Nuron crypto accelerator. 6817 [Ben Laurie] 6818 6819 *) Add support for external cryptographic devices. This code was 6820 previously distributed separately as the "engine" branch. 6821 [Geoff Thorpe, Richard Levitte] 6822 6823 *) Rework the filename-translation in the DSO code. It is now possible to 6824 have far greater control over how a "name" is turned into a filename 6825 depending on the operating environment and any oddities about the 6826 different shared library filenames on each system. 6827 [Geoff Thorpe] 6828 6829 *) Support threads on FreeBSD-elf in Configure. 6830 [Richard Levitte] 6831 6832 *) Fix for SHA1 assembly problem with MASM: it produces 6833 warnings about corrupt line number information when assembling 6834 with debugging information. This is caused by the overlapping 6835 of two sections. 6836 [Bernd Matthes <mainbug@celocom.de>, Steve Henson] 6837 6838 *) NCONF changes. 6839 NCONF_get_number() has no error checking at all. As a replacement, 6840 NCONF_get_number_e() is defined (_e for "error checking") and is 6841 promoted strongly. The old NCONF_get_number is kept around for 6842 binary backward compatibility. 6843 Make it possible for methods to load from something other than a BIO, 6844 by providing a function pointer that is given a name instead of a BIO. 6845 For example, this could be used to load configuration data from an 6846 LDAP server. 6847 [Richard Levitte] 6848 6849 *) Fix for non blocking accept BIOs. Added new I/O special reason 6850 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs 6851 with non blocking I/O was not possible because no retry code was 6852 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover 6853 this case. 6854 [Steve Henson] 6855 6856 *) Added the beginnings of Rijndael support. 6857 [Ben Laurie] 6858 6859 *) Fix for bug in DirectoryString mask setting. Add support for 6860 X509_NAME_print_ex() in 'req' and X509_print_ex() function 6861 to allow certificate printing to more controllable, additional 6862 'certopt' option to 'x509' to allow new printing options to be 6863 set. 6864 [Steve Henson] 6865 6866 *) Clean old EAY MD5 hack from e_os.h. 6867 [Richard Levitte] 6868 6869 Changes between 0.9.6l and 0.9.6m [17 Mar 2004] 6870 6871 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 6872 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 6873 [Joe Orton, Steve Henson] 6874 6875 Changes between 0.9.6k and 0.9.6l [04 Nov 2003] 6876 6877 *) Fix additional bug revealed by the NISCC test suite: 6878 6879 Stop bug triggering large recursion when presented with 6880 certain ASN.1 tags (CVE-2003-0851) 6881 [Steve Henson] 6882 6883 Changes between 0.9.6j and 0.9.6k [30 Sep 2003] 6884 6885 *) Fix various bugs revealed by running the NISCC test suite: 6886 6887 Stop out of bounds reads in the ASN1 code when presented with 6888 invalid tags (CVE-2003-0543 and CVE-2003-0544). 6889 6890 If verify callback ignores invalid public key errors don't try to check 6891 certificate signature with the NULL public key. 6892 6893 [Steve Henson] 6894 6895 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 6896 if the server requested one: as stated in TLS 1.0 and SSL 3.0 6897 specifications. 6898 [Steve Henson] 6899 6900 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 6901 extra data after the compression methods not only for TLS 1.0 6902 but also for SSL 3.0 (as required by the specification). 6903 [Bodo Moeller; problem pointed out by Matthias Loepfe] 6904 6905 *) Change X509_certificate_type() to mark the key as exported/exportable 6906 when it's 512 *bits* long, not 512 bytes. 6907 [Richard Levitte] 6908 6909 Changes between 0.9.6i and 0.9.6j [10 Apr 2003] 6910 6911 *) Countermeasure against the Klima-Pokorny-Rosa extension of 6912 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 6913 a protocol version number mismatch like a decryption error 6914 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 6915 [Bodo Moeller] 6916 6917 *) Turn on RSA blinding by default in the default implementation 6918 to avoid a timing attack. Applications that don't want it can call 6919 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 6920 They would be ill-advised to do so in most cases. 6921 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 6922 6923 *) Change RSA blinding code so that it works when the PRNG is not 6924 seeded (in this case, the secret RSA exponent is abused as 6925 an unpredictable seed -- if it is not unpredictable, there 6926 is no point in blinding anyway). Make RSA blinding thread-safe 6927 by remembering the creator's thread ID in rsa->blinding and 6928 having all other threads use local one-time blinding factors 6929 (this requires more computation than sharing rsa->blinding, but 6930 avoids excessive locking; and if an RSA object is not shared 6931 between threads, blinding will still be very fast). 6932 [Bodo Moeller] 6933 6934 Changes between 0.9.6h and 0.9.6i [19 Feb 2003] 6935 6936 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 6937 via timing by performing a MAC computation even if incorrrect 6938 block cipher padding has been found. This is a countermeasure 6939 against active attacks where the attacker has to distinguish 6940 between bad padding and a MAC verification error. (CVE-2003-0078) 6941 6942 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 6943 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 6944 Martin Vuagnoux (EPFL, Ilion)] 6945 6946 Changes between 0.9.6g and 0.9.6h [5 Dec 2002] 6947 6948 *) New function OPENSSL_cleanse(), which is used to cleanse a section of 6949 memory from it's contents. This is done with a counter that will 6950 place alternating values in each byte. This can be used to solve 6951 two issues: 1) the removal of calls to memset() by highly optimizing 6952 compilers, and 2) cleansing with other values than 0, since those can 6953 be read through on certain media, for example a swap space on disk. 6954 [Geoff Thorpe] 6955 6956 *) Bugfix: client side session caching did not work with external caching, 6957 because the session->cipher setting was not restored when reloading 6958 from the external cache. This problem was masked, when 6959 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set. 6960 (Found by Steve Haslam <steve@araqnid.ddts.net>.) 6961 [Lutz Jaenicke] 6962 6963 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total 6964 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33. 6965 [Zeev Lieber <zeev-l@yahoo.com>] 6966 6967 *) Undo an undocumented change introduced in 0.9.6e which caused 6968 repeated calls to OpenSSL_add_all_ciphers() and 6969 OpenSSL_add_all_digests() to be ignored, even after calling 6970 EVP_cleanup(). 6971 [Richard Levitte] 6972 6973 *) Change the default configuration reader to deal with last line not 6974 being properly terminated. 6975 [Richard Levitte] 6976 6977 *) Change X509_NAME_cmp() so it applies the special rules on handling 6978 DN values that are of type PrintableString, as well as RDNs of type 6979 emailAddress where the value has the type ia5String. 6980 [stefank@valicert.com via Richard Levitte] 6981 6982 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half 6983 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently 6984 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be 6985 the bitwise-OR of the two for use by the majority of applications 6986 wanting this behaviour, and update the docs. The documented 6987 behaviour and actual behaviour were inconsistent and had been 6988 changing anyway, so this is more a bug-fix than a behavioural 6989 change. 6990 [Geoff Thorpe, diagnosed by Nadav Har'El] 6991 6992 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c 6993 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes). 6994 [Bodo Moeller] 6995 6996 *) Fix initialization code race conditions in 6997 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(), 6998 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(), 6999 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(), 7000 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(), 7001 ssl2_get_cipher_by_char(), 7002 ssl3_get_cipher_by_char(). 7003 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller] 7004 7005 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after 7006 the cached sessions are flushed, as the remove_cb() might use ex_data 7007 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com> 7008 (see [openssl.org #212]). 7009 [Geoff Thorpe, Lutz Jaenicke] 7010 7011 *) Fix typo in OBJ_txt2obj which incorrectly passed the content 7012 length, instead of the encoding length to d2i_ASN1_OBJECT. 7013 [Steve Henson] 7014 7015 Changes between 0.9.6f and 0.9.6g [9 Aug 2002] 7016 7017 *) [In 0.9.6g-engine release:] 7018 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall'). 7019 [Lynn Gazis <lgazis@rainbow.com>] 7020 7021 Changes between 0.9.6e and 0.9.6f [8 Aug 2002] 7022 7023 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX 7024 and get fix the header length calculation. 7025 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>, 7026 Alon Kantor <alonk@checkpoint.com> (and others), 7027 Steve Henson] 7028 7029 *) Use proper error handling instead of 'assertions' in buffer 7030 overflow checks added in 0.9.6e. This prevents DoS (the 7031 assertions could call abort()). 7032 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller] 7033 7034 Changes between 0.9.6d and 0.9.6e [30 Jul 2002] 7035 7036 *) Add various sanity checks to asn1_get_length() to reject 7037 the ASN1 length bytes if they exceed sizeof(long), will appear 7038 negative or the content length exceeds the length of the 7039 supplied buffer. 7040 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 7041 7042 *) Fix cipher selection routines: ciphers without encryption had no flags 7043 for the cipher strength set and where therefore not handled correctly 7044 by the selection routines (PR #130). 7045 [Lutz Jaenicke] 7046 7047 *) Fix EVP_dsa_sha macro. 7048 [Nils Larsch] 7049 7050 *) New option 7051 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS 7052 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure 7053 that was added in OpenSSL 0.9.6d. 7054 7055 As the countermeasure turned out to be incompatible with some 7056 broken SSL implementations, the new option is part of SSL_OP_ALL. 7057 SSL_OP_ALL is usually employed when compatibility with weird SSL 7058 implementations is desired (e.g. '-bugs' option to 's_client' and 7059 's_server'), so the new option is automatically set in many 7060 applications. 7061 [Bodo Moeller] 7062 7063 *) Changes in security patch: 7064 7065 Changes marked "(CHATS)" were sponsored by the Defense Advanced 7066 Research Projects Agency (DARPA) and Air Force Research Laboratory, 7067 Air Force Materiel Command, USAF, under agreement number 7068 F30602-01-2-0537. 7069 7070 *) Add various sanity checks to asn1_get_length() to reject 7071 the ASN1 length bytes if they exceed sizeof(long), will appear 7072 negative or the content length exceeds the length of the 7073 supplied buffer. (CVE-2002-0659) 7074 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 7075 7076 *) Assertions for various potential buffer overflows, not known to 7077 happen in practice. 7078 [Ben Laurie (CHATS)] 7079 7080 *) Various temporary buffers to hold ASCII versions of integers were 7081 too small for 64 bit platforms. (CVE-2002-0655) 7082 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)> 7083 7084 *) Remote buffer overflow in SSL3 protocol - an attacker could 7085 supply an oversized session ID to a client. (CVE-2002-0656) 7086 [Ben Laurie (CHATS)] 7087 7088 *) Remote buffer overflow in SSL2 protocol - an attacker could 7089 supply an oversized client master key. (CVE-2002-0656) 7090 [Ben Laurie (CHATS)] 7091 7092 Changes between 0.9.6c and 0.9.6d [9 May 2002] 7093 7094 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not 7095 encoded as NULL) with id-dsa-with-sha1. 7096 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller] 7097 7098 *) Check various X509_...() return values in apps/req.c. 7099 [Nils Larsch <nla@trustcenter.de>] 7100 7101 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines: 7102 an end-of-file condition would erronously be flagged, when the CRLF 7103 was just at the end of a processed block. The bug was discovered when 7104 processing data through a buffering memory BIO handing the data to a 7105 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov 7106 <ptsekov@syntrex.com> and Nedelcho Stanev. 7107 [Lutz Jaenicke] 7108 7109 *) Implement a countermeasure against a vulnerability recently found 7110 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment 7111 before application data chunks to avoid the use of known IVs 7112 with data potentially chosen by the attacker. 7113 [Bodo Moeller] 7114 7115 *) Fix length checks in ssl3_get_client_hello(). 7116 [Bodo Moeller] 7117 7118 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently 7119 to prevent ssl3_read_internal() from incorrectly assuming that 7120 ssl3_read_bytes() found application data while handshake 7121 processing was enabled when in fact s->s3->in_read_app_data was 7122 merely automatically cleared during the initial handshake. 7123 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>] 7124 7125 *) Fix object definitions for Private and Enterprise: they were not 7126 recognized in their shortname (=lowercase) representation. Extend 7127 obj_dat.pl to issue an error when using undefined keywords instead 7128 of silently ignoring the problem (Svenning Sorensen 7129 <sss@sss.dnsalias.net>). 7130 [Lutz Jaenicke] 7131 7132 *) Fix DH_generate_parameters() so that it works for 'non-standard' 7133 generators, i.e. generators other than 2 and 5. (Previously, the 7134 code did not properly initialise the 'add' and 'rem' values to 7135 BN_generate_prime().) 7136 7137 In the new general case, we do not insist that 'generator' is 7138 actually a primitive root: This requirement is rather pointless; 7139 a generator of the order-q subgroup is just as good, if not 7140 better. 7141 [Bodo Moeller] 7142 7143 *) Map new X509 verification errors to alerts. Discovered and submitted by 7144 Tom Wu <tom@arcot.com>. 7145 [Lutz Jaenicke] 7146 7147 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from 7148 returning non-zero before the data has been completely received 7149 when using non-blocking I/O. 7150 [Bodo Moeller; problem pointed out by John Hughes] 7151 7152 *) Some of the ciphers missed the strength entry (SSL_LOW etc). 7153 [Ben Laurie, Lutz Jaenicke] 7154 7155 *) Fix bug in SSL_clear(): bad sessions were not removed (found by 7156 Yoram Zahavi <YoramZ@gilian.com>). 7157 [Lutz Jaenicke] 7158 7159 *) Add information about CygWin 1.3 and on, and preserve proper 7160 configuration for the versions before that. 7161 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 7162 7163 *) Make removal from session cache (SSL_CTX_remove_session()) more robust: 7164 check whether we deal with a copy of a session and do not delete from 7165 the cache in this case. Problem reported by "Izhar Shoshani Levi" 7166 <izhar@checkpoint.com>. 7167 [Lutz Jaenicke] 7168 7169 *) Do not store session data into the internal session cache, if it 7170 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP 7171 flag is set). Proposed by Aslam <aslam@funk.com>. 7172 [Lutz Jaenicke] 7173 7174 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested 7175 value is 0. 7176 [Richard Levitte] 7177 7178 *) [In 0.9.6d-engine release:] 7179 Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 7180 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 7181 7182 *) Add the configuration target linux-s390x. 7183 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte] 7184 7185 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of 7186 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag 7187 variable as an indication that a ClientHello message has been 7188 received. As the flag value will be lost between multiple 7189 invocations of ssl3_accept when using non-blocking I/O, the 7190 function may not be aware that a handshake has actually taken 7191 place, thus preventing a new session from being added to the 7192 session cache. 7193 7194 To avoid this problem, we now set s->new_session to 2 instead of 7195 using a local variable. 7196 [Lutz Jaenicke, Bodo Moeller] 7197 7198 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c) 7199 if the SSL_R_LENGTH_MISMATCH error is detected. 7200 [Geoff Thorpe, Bodo Moeller] 7201 7202 *) New 'shared_ldflag' column in Configure platform table. 7203 [Richard Levitte] 7204 7205 *) Fix EVP_CIPHER_mode macro. 7206 ["Dan S. Camper" <dan@bti.net>] 7207 7208 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown 7209 type, we must throw them away by setting rr->length to 0. 7210 [D P Chang <dpc@qualys.com>] 7211 7212 Changes between 0.9.6b and 0.9.6c [21 dec 2001] 7213 7214 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl 7215 <Dominikus.Scherkl@biodata.com>. (The previous implementation 7216 worked incorrectly for those cases where range = 10..._2 and 7217 3*range is two bits longer than range.) 7218 [Bodo Moeller] 7219 7220 *) Only add signing time to PKCS7 structures if it is not already 7221 present. 7222 [Steve Henson] 7223 7224 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce", 7225 OBJ_ld_ce should be OBJ_id_ce. 7226 Also some ip-pda OIDs in crypto/objects/objects.txt were 7227 incorrect (cf. RFC 3039). 7228 [Matt Cooper, Frederic Giudicelli, Bodo Moeller] 7229 7230 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid() 7231 returns early because it has nothing to do. 7232 [Andy Schneider <andy.schneider@bjss.co.uk>] 7233 7234 *) [In 0.9.6c-engine release:] 7235 Fix mutex callback return values in crypto/engine/hw_ncipher.c. 7236 [Andy Schneider <andy.schneider@bjss.co.uk>] 7237 7238 *) [In 0.9.6c-engine release:] 7239 Add support for Cryptographic Appliance's keyserver technology. 7240 (Use engine 'keyclient') 7241 [Cryptographic Appliances and Geoff Thorpe] 7242 7243 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89' 7244 is called via tools/c89.sh because arguments have to be 7245 rearranged (all '-L' options must appear before the first object 7246 modules). 7247 [Richard Shapiro <rshapiro@abinitio.com>] 7248 7249 *) [In 0.9.6c-engine release:] 7250 Add support for Broadcom crypto accelerator cards, backported 7251 from 0.9.7. 7252 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox] 7253 7254 *) [In 0.9.6c-engine release:] 7255 Add support for SureWare crypto accelerator cards from 7256 Baltimore Technologies. (Use engine 'sureware') 7257 [Baltimore Technologies and Mark Cox] 7258 7259 *) [In 0.9.6c-engine release:] 7260 Add support for crypto accelerator cards from Accelerated 7261 Encryption Processing, www.aep.ie. (Use engine 'aep') 7262 [AEP Inc. and Mark Cox] 7263 7264 *) Add a configuration entry for gcc on UnixWare. 7265 [Gary Benson <gbenson@redhat.com>] 7266 7267 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake 7268 messages are stored in a single piece (fixed-length part and 7269 variable-length part combined) and fix various bugs found on the way. 7270 [Bodo Moeller] 7271 7272 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname() 7273 instead. BIO_gethostbyname() does not know what timeouts are 7274 appropriate, so entries would stay in cache even when they have 7275 become invalid. 7276 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com> 7277 7278 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when 7279 faced with a pathologically small ClientHello fragment that does 7280 not contain client_version: Instead of aborting with an error, 7281 simply choose the highest available protocol version (i.e., 7282 TLS 1.0 unless it is disabled). In practice, ClientHello 7283 messages are never sent like this, but this change gives us 7284 strictly correct behaviour at least for TLS. 7285 [Bodo Moeller] 7286 7287 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear() 7288 never resets s->method to s->ctx->method when called from within 7289 one of the SSL handshake functions. 7290 [Bodo Moeller; problem pointed out by Niko Baric] 7291 7292 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert 7293 (sent using the client's version number) if client_version is 7294 smaller than the protocol version in use. Also change 7295 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if 7296 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then 7297 the client will at least see that alert. 7298 [Bodo Moeller] 7299 7300 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation 7301 correctly. 7302 [Bodo Moeller] 7303 7304 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a 7305 client receives HelloRequest while in a handshake. 7306 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>] 7307 7308 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C 7309 should end in 'break', not 'goto end' which circuments various 7310 cleanups done in state SSL_ST_OK. But session related stuff 7311 must be disabled for SSL_ST_OK in the case that we just sent a 7312 HelloRequest. 7313 7314 Also avoid some overhead by not calling ssl_init_wbio_buffer() 7315 before just sending a HelloRequest. 7316 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>] 7317 7318 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't 7319 reveal whether illegal block cipher padding was found or a MAC 7320 verification error occured. (Neither SSLerr() codes nor alerts 7321 are directly visible to potential attackers, but the information 7322 may leak via logfiles.) 7323 7324 Similar changes are not required for the SSL 2.0 implementation 7325 because the number of padding bytes is sent in clear for SSL 2.0, 7326 and the extra bytes are just ignored. However ssl/s2_pkt.c 7327 failed to verify that the purported number of padding bytes is in 7328 the legal range. 7329 [Bodo Moeller] 7330 7331 *) Add OpenUNIX-8 support including shared libraries 7332 (Boyd Lynn Gerber <gerberb@zenez.com>). 7333 [Lutz Jaenicke] 7334 7335 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid 7336 'wristwatch attack' using huge encoding parameters (cf. 7337 James H. Manger's CRYPTO 2001 paper). Note that the 7338 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use 7339 encoding parameters and hence was not vulnerable. 7340 [Bodo Moeller] 7341 7342 *) BN_sqr() bug fix. 7343 [Ulf M��ller, reported by Jim Ellis <jim.ellis@cavium.com>] 7344 7345 *) Rabin-Miller test analyses assume uniformly distributed witnesses, 7346 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand() 7347 followed by modular reduction. 7348 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>] 7349 7350 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range() 7351 equivalent based on BN_pseudo_rand() instead of BN_rand(). 7352 [Bodo Moeller] 7353 7354 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB). 7355 This function was broken, as the check for a new client hello message 7356 to handle SGC did not allow these large messages. 7357 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.) 7358 [Lutz Jaenicke] 7359 7360 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long](). 7361 [Lutz Jaenicke] 7362 7363 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl() 7364 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>). 7365 [Lutz Jaenicke] 7366 7367 *) Rework the configuration and shared library support for Tru64 Unix. 7368 The configuration part makes use of modern compiler features and 7369 still retains old compiler behavior for those that run older versions 7370 of the OS. The shared library support part includes a variant that 7371 uses the RPATH feature, and is available through the special 7372 configuration target "alpha-cc-rpath", which will never be selected 7373 automatically. 7374 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte] 7375 7376 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message() 7377 with the same message size as in ssl3_get_certificate_request(). 7378 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest 7379 messages might inadvertently be reject as too long. 7380 [Petr Lampa <lampa@fee.vutbr.cz>] 7381 7382 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX). 7383 [Andy Polyakov] 7384 7385 *) Modified SSL library such that the verify_callback that has been set 7386 specificly for an SSL object with SSL_set_verify() is actually being 7387 used. Before the change, a verify_callback set with this function was 7388 ignored and the verify_callback() set in the SSL_CTX at the time of 7389 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced 7390 to allow the necessary settings. 7391 [Lutz Jaenicke] 7392 7393 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c 7394 explicitly to NULL, as at least on Solaris 8 this seems not always to be 7395 done automatically (in contradiction to the requirements of the C 7396 standard). This made problems when used from OpenSSH. 7397 [Lutz Jaenicke] 7398 7399 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored 7400 dh->length and always used 7401 7402 BN_rand_range(priv_key, dh->p). 7403 7404 BN_rand_range() is not necessary for Diffie-Hellman, and this 7405 specific range makes Diffie-Hellman unnecessarily inefficient if 7406 dh->length (recommended exponent length) is much smaller than the 7407 length of dh->p. We could use BN_rand_range() if the order of 7408 the subgroup was stored in the DH structure, but we only have 7409 dh->length. 7410 7411 So switch back to 7412 7413 BN_rand(priv_key, l, ...) 7414 7415 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1 7416 otherwise. 7417 [Bodo Moeller] 7418 7419 *) In 7420 7421 RSA_eay_public_encrypt 7422 RSA_eay_private_decrypt 7423 RSA_eay_private_encrypt (signing) 7424 RSA_eay_public_decrypt (signature verification) 7425 7426 (default implementations for RSA_public_encrypt, 7427 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt), 7428 always reject numbers >= n. 7429 [Bodo Moeller] 7430 7431 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2 7432 to synchronize access to 'locking_thread'. This is necessary on 7433 systems where access to 'locking_thread' (an 'unsigned long' 7434 variable) is not atomic. 7435 [Bodo Moeller] 7436 7437 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID 7438 *before* setting the 'crypto_lock_rand' flag. The previous code had 7439 a race condition if 0 is a valid thread ID. 7440 [Travis Vitek <vitek@roguewave.com>] 7441 7442 *) Add support for shared libraries under Irix. 7443 [Albert Chin-A-Young <china@thewrittenword.com>] 7444 7445 *) Add configuration option to build on Linux on both big-endian and 7446 little-endian MIPS. 7447 [Ralf Baechle <ralf@uni-koblenz.de>] 7448 7449 *) Add the possibility to create shared libraries on HP-UX. 7450 [Richard Levitte] 7451 7452 Changes between 0.9.6a and 0.9.6b [9 Jul 2001] 7453 7454 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c) 7455 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by 7456 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>: 7457 PRNG state recovery was possible based on the output of 7458 one PRNG request appropriately sized to gain knowledge on 7459 'md' followed by enough consecutive 1-byte PRNG requests 7460 to traverse all of 'state'. 7461 7462 1. When updating 'md_local' (the current thread's copy of 'md') 7463 during PRNG output generation, hash all of the previous 7464 'md_local' value, not just the half used for PRNG output. 7465 7466 2. Make the number of bytes from 'state' included into the hash 7467 independent from the number of PRNG bytes requested. 7468 7469 The first measure alone would be sufficient to avoid 7470 Markku-Juhani's attack. (Actually it had never occurred 7471 to me that the half of 'md_local' used for chaining was the 7472 half from which PRNG output bytes were taken -- I had always 7473 assumed that the secret half would be used.) The second 7474 measure makes sure that additional data from 'state' is never 7475 mixed into 'md_local' in small portions; this heuristically 7476 further strengthens the PRNG. 7477 [Bodo Moeller] 7478 7479 *) Fix crypto/bn/asm/mips3.s. 7480 [Andy Polyakov] 7481 7482 *) When only the key is given to "enc", the IV is undefined. Print out 7483 an error message in this case. 7484 [Lutz Jaenicke] 7485 7486 *) Handle special case when X509_NAME is empty in X509 printing routines. 7487 [Steve Henson] 7488 7489 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are 7490 positive and less than q. 7491 [Bodo Moeller] 7492 7493 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is 7494 used: it isn't thread safe and the add_lock_callback should handle 7495 that itself. 7496 [Paul Rose <Paul.Rose@bridge.com>] 7497 7498 *) Verify that incoming data obeys the block size in 7499 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c). 7500 [Bodo Moeller] 7501 7502 *) Fix OAEP check. 7503 [Ulf M��ller, Bodo M��ller] 7504 7505 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5 7506 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5 7507 when fixing the server behaviour for backwards-compatible 'client 7508 hello' messages. (Note that the attack is impractical against 7509 SSL 3.0 and TLS 1.0 anyway because length and version checking 7510 means that the probability of guessing a valid ciphertext is 7511 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98 7512 paper.) 7513 7514 Before 0.9.5, the countermeasure (hide the error by generating a 7515 random 'decryption result') did not work properly because 7516 ERR_clear_error() was missing, meaning that SSL_get_error() would 7517 detect the supposedly ignored error. 7518 7519 Both problems are now fixed. 7520 [Bodo Moeller] 7521 7522 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096 7523 (previously it was 1024). 7524 [Bodo Moeller] 7525 7526 *) Fix for compatibility mode trust settings: ignore trust settings 7527 unless some valid trust or reject settings are present. 7528 [Steve Henson] 7529 7530 *) Fix for blowfish EVP: its a variable length cipher. 7531 [Steve Henson] 7532 7533 *) Fix various bugs related to DSA S/MIME verification. Handle missing 7534 parameters in DSA public key structures and return an error in the 7535 DSA routines if parameters are absent. 7536 [Steve Henson] 7537 7538 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd" 7539 in the current directory if neither $RANDFILE nor $HOME was set. 7540 RAND_file_name() in 0.9.6a returned NULL in this case. This has 7541 caused some confusion to Windows users who haven't defined $HOME. 7542 Thus RAND_file_name() is changed again: e_os.h can define a 7543 DEFAULT_HOME, which will be used if $HOME is not set. 7544 For Windows, we use "C:"; on other platforms, we still require 7545 environment variables. 7546 7547 *) Move 'if (!initialized) RAND_poll()' into regions protected by 7548 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids 7549 having multiple threads call RAND_poll() concurrently. 7550 [Bodo Moeller] 7551 7552 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a 7553 combination of a flag and a thread ID variable. 7554 Otherwise while one thread is in ssleay_rand_bytes (which sets the 7555 flag), *other* threads can enter ssleay_add_bytes without obeying 7556 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock 7557 that they do not hold after the first thread unsets add_do_not_lock). 7558 [Bodo Moeller] 7559 7560 *) Change bctest again: '-x' expressions are not available in all 7561 versions of 'test'. 7562 [Bodo Moeller] 7563 7564 Changes between 0.9.6 and 0.9.6a [5 Apr 2001] 7565 7566 *) Fix a couple of memory leaks in PKCS7_dataDecode() 7567 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>] 7568 7569 *) Change Configure and Makefiles to provide EXE_EXT, which will contain 7570 the default extension for executables, if any. Also, make the perl 7571 scripts that use symlink() to test if it really exists and use "cp" 7572 if it doesn't. All this made OpenSSL compilable and installable in 7573 CygWin. 7574 [Richard Levitte] 7575 7576 *) Fix for asn1_GetSequence() for indefinite length constructed data. 7577 If SEQUENCE is length is indefinite just set c->slen to the total 7578 amount of data available. 7579 [Steve Henson, reported by shige@FreeBSD.org] 7580 [This change does not apply to 0.9.7.] 7581 7582 *) Change bctest to avoid here-documents inside command substitution 7583 (workaround for FreeBSD /bin/sh bug). 7584 For compatibility with Ultrix, avoid shell functions (introduced 7585 in the bctest version that searches along $PATH). 7586 [Bodo Moeller] 7587 7588 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes 7589 with des_encrypt() defined on some operating systems, like Solaris 7590 and UnixWare. 7591 [Richard Levitte] 7592 7593 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton: 7594 On the Importance of Eliminating Errors in Cryptographic 7595 Computations, J. Cryptology 14 (2001) 2, 101-119, 7596 http://theory.stanford.edu/~dabo/papers/faults.ps.gz). 7597 [Ulf Moeller] 7598 7599 *) MIPS assembler BIGNUM division bug fix. 7600 [Andy Polyakov] 7601 7602 *) Disabled incorrect Alpha assembler code. 7603 [Richard Levitte] 7604 7605 *) Fix PKCS#7 decode routines so they correctly update the length 7606 after reading an EOC for the EXPLICIT tag. 7607 [Steve Henson] 7608 [This change does not apply to 0.9.7.] 7609 7610 *) Fix bug in PKCS#12 key generation routines. This was triggered 7611 if a 3DES key was generated with a 0 initial byte. Include 7612 PKCS12_BROKEN_KEYGEN compilation option to retain the old 7613 (but broken) behaviour. 7614 [Steve Henson] 7615 7616 *) Enhance bctest to search for a working bc along $PATH and print 7617 it when found. 7618 [Tim Rice <tim@multitalents.net> via Richard Levitte] 7619 7620 *) Fix memory leaks in err.c: free err_data string if necessary; 7621 don't write to the wrong index in ERR_set_error_data. 7622 [Bodo Moeller] 7623 7624 *) Implement ssl23_peek (analogous to ssl23_read), which previously 7625 did not exist. 7626 [Bodo Moeller] 7627 7628 *) Replace rdtsc with _emit statements for VC++ version 5. 7629 [Jeremy Cooper <jeremy@baymoo.org>] 7630 7631 *) Make it possible to reuse SSLv2 sessions. 7632 [Richard Levitte] 7633 7634 *) In copy_email() check for >= 0 as a return value for 7635 X509_NAME_get_index_by_NID() since 0 is a valid index. 7636 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>] 7637 7638 *) Avoid coredump with unsupported or invalid public keys by checking if 7639 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when 7640 PKCS7_verify() fails with non detached data. 7641 [Steve Henson] 7642 7643 *) Don't use getenv in library functions when run as setuid/setgid. 7644 New function OPENSSL_issetugid(). 7645 [Ulf Moeller] 7646 7647 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c) 7648 due to incorrect handling of multi-threading: 7649 7650 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl(). 7651 7652 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on(). 7653 7654 3. Count how many times MemCheck_off() has been called so that 7655 nested use can be treated correctly. This also avoids 7656 inband-signalling in the previous code (which relied on the 7657 assumption that thread ID 0 is impossible). 7658 [Bodo Moeller] 7659 7660 *) Add "-rand" option also to s_client and s_server. 7661 [Lutz Jaenicke] 7662 7663 *) Fix CPU detection on Irix 6.x. 7664 [Kurt Hockenbury <khockenb@stevens-tech.edu> and 7665 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 7666 7667 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME 7668 was empty. 7669 [Steve Henson] 7670 [This change does not apply to 0.9.7.] 7671 7672 *) Use the cached encoding of an X509_NAME structure rather than 7673 copying it. This is apparently the reason for the libsafe "errors" 7674 but the code is actually correct. 7675 [Steve Henson] 7676 7677 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent 7678 Bleichenbacher's DSA attack. 7679 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits 7680 to be set and top=0 forces the highest bit to be set; top=-1 is new 7681 and leaves the highest bit random. 7682 [Ulf Moeller, Bodo Moeller] 7683 7684 *) In the NCONF_...-based implementations for CONF_... queries 7685 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using 7686 a temporary CONF structure with the data component set to NULL 7687 (which gives segmentation faults in lh_retrieve). 7688 Instead, use NULL for the CONF pointer in CONF_get_string and 7689 CONF_get_number (which may use environment variables) and directly 7690 return NULL from CONF_get_section. 7691 [Bodo Moeller] 7692 7693 *) Fix potential buffer overrun for EBCDIC. 7694 [Ulf Moeller] 7695 7696 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign 7697 keyUsage if basicConstraints absent for a CA. 7698 [Steve Henson] 7699 7700 *) Make SMIME_write_PKCS7() write mail header values with a format that 7701 is more generally accepted (no spaces before the semicolon), since 7702 some programs can't parse those values properly otherwise. Also make 7703 sure BIO's that break lines after each write do not create invalid 7704 headers. 7705 [Richard Levitte] 7706 7707 *) Make the CRL encoding routines work with empty SEQUENCE OF. The 7708 macros previously used would not encode an empty SEQUENCE OF 7709 and break the signature. 7710 [Steve Henson] 7711 [This change does not apply to 0.9.7.] 7712 7713 *) Zero the premaster secret after deriving the master secret in 7714 DH ciphersuites. 7715 [Steve Henson] 7716 7717 *) Add some EVP_add_digest_alias registrations (as found in 7718 OpenSSL_add_all_digests()) to SSL_library_init() 7719 aka OpenSSL_add_ssl_algorithms(). This provides improved 7720 compatibility with peers using X.509 certificates 7721 with unconventional AlgorithmIdentifier OIDs. 7722 [Bodo Moeller] 7723 7724 *) Fix for Irix with NO_ASM. 7725 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 7726 7727 *) ./config script fixes. 7728 [Ulf Moeller, Richard Levitte] 7729 7730 *) Fix 'openssl passwd -1'. 7731 [Bodo Moeller] 7732 7733 *) Change PKCS12_key_gen_asc() so it can cope with non null 7734 terminated strings whose length is passed in the passlen 7735 parameter, for example from PEM callbacks. This was done 7736 by adding an extra length parameter to asc2uni(). 7737 [Steve Henson, reported by <oddissey@samsung.co.kr>] 7738 7739 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn 7740 call failed, free the DSA structure. 7741 [Bodo Moeller] 7742 7743 *) Fix to uni2asc() to cope with zero length Unicode strings. 7744 These are present in some PKCS#12 files. 7745 [Steve Henson] 7746 7747 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c). 7748 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits 7749 when writing a 32767 byte record. 7750 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>] 7751 7752 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c), 7753 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}. 7754 7755 (RSA objects have a reference count access to which is protected 7756 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c], 7757 so they are meant to be shared between threads.) 7758 [Bodo Moeller, Geoff Thorpe; original patch submitted by 7759 "Reddie, Steven" <Steven.Reddie@ca.com>] 7760 7761 *) Fix a deadlock in CRYPTO_mem_leaks(). 7762 [Bodo Moeller] 7763 7764 *) Use better test patterns in bntest. 7765 [Ulf M��ller] 7766 7767 *) rand_win.c fix for Borland C. 7768 [Ulf M��ller] 7769 7770 *) BN_rshift bugfix for n == 0. 7771 [Bodo Moeller] 7772 7773 *) Add a 'bctest' script that checks for some known 'bc' bugs 7774 so that 'make test' does not abort just because 'bc' is broken. 7775 [Bodo Moeller] 7776 7777 *) Store verify_result within SSL_SESSION also for client side to 7778 avoid potential security hole. (Re-used sessions on the client side 7779 always resulted in verify_result==X509_V_OK, not using the original 7780 result of the server certificate verification.) 7781 [Lutz Jaenicke] 7782 7783 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type 7784 SSL3_RT_APPLICATION_DATA, return 0. 7785 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true. 7786 [Bodo Moeller] 7787 7788 *) Fix SSL_peek: 7789 Both ssl2_peek and ssl3_peek, which were totally broken in earlier 7790 releases, have been re-implemented by renaming the previous 7791 implementations of ssl2_read and ssl3_read to ssl2_read_internal 7792 and ssl3_read_internal, respectively, and adding 'peek' parameters 7793 to them. The new ssl[23]_{read,peek} functions are calls to 7794 ssl[23]_read_internal with the 'peek' flag set appropriately. 7795 A 'peek' parameter has also been added to ssl3_read_bytes, which 7796 does the actual work for ssl3_read_internal. 7797 [Bodo Moeller] 7798 7799 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling 7800 the method-specific "init()" handler. Also clean up ex_data after 7801 calling the method-specific "finish()" handler. Previously, this was 7802 happening the other way round. 7803 [Geoff Thorpe] 7804 7805 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16. 7806 The previous value, 12, was not always sufficient for BN_mod_exp(). 7807 [Bodo Moeller] 7808 7809 *) Make sure that shared libraries get the internal name engine with 7810 the full version number and not just 0. This should mark the 7811 shared libraries as not backward compatible. Of course, this should 7812 be changed again when we can guarantee backward binary compatibility. 7813 [Richard Levitte] 7814 7815 *) Fix typo in get_cert_by_subject() in by_dir.c 7816 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>] 7817 7818 *) Rework the system to generate shared libraries: 7819 7820 - Make note of the expected extension for the shared libraries and 7821 if there is a need for symbolic links from for example libcrypto.so.0 7822 to libcrypto.so.0.9.7. There is extended info in Configure for 7823 that. 7824 7825 - Make as few rebuilds of the shared libraries as possible. 7826 7827 - Still avoid linking the OpenSSL programs with the shared libraries. 7828 7829 - When installing, install the shared libraries separately from the 7830 static ones. 7831 [Richard Levitte] 7832 7833 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument. 7834 7835 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new 7836 and not in SSL_clear because the latter is also used by the 7837 accept/connect functions; previously, the settings made by 7838 SSL_set_read_ahead would be lost during the handshake. 7839 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>] 7840 7841 *) Correct util/mkdef.pl to be selective about disabled algorithms. 7842 Previously, it would create entries for disableed algorithms no 7843 matter what. 7844 [Richard Levitte] 7845 7846 *) Added several new manual pages for SSL_* function. 7847 [Lutz Jaenicke] 7848 7849 Changes between 0.9.5a and 0.9.6 [24 Sep 2000] 7850 7851 *) In ssl23_get_client_hello, generate an error message when faced 7852 with an initial SSL 3.0/TLS record that is too small to contain the 7853 first two bytes of the ClientHello message, i.e. client_version. 7854 (Note that this is a pathologic case that probably has never happened 7855 in real life.) The previous approach was to use the version number 7856 from the record header as a substitute; but our protocol choice 7857 should not depend on that one because it is not authenticated 7858 by the Finished messages. 7859 [Bodo Moeller] 7860 7861 *) More robust randomness gathering functions for Windows. 7862 [Jeffrey Altman <jaltman@columbia.edu>] 7863 7864 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is 7865 not set then we don't setup the error code for issuer check errors 7866 to avoid possibly overwriting other errors which the callback does 7867 handle. If an application does set the flag then we assume it knows 7868 what it is doing and can handle the new informational codes 7869 appropriately. 7870 [Steve Henson] 7871 7872 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for 7873 a general "ANY" type, as such it should be able to decode anything 7874 including tagged types. However it didn't check the class so it would 7875 wrongly interpret tagged types in the same way as their universal 7876 counterpart and unknown types were just rejected. Changed so that the 7877 tagged and unknown types are handled in the same way as a SEQUENCE: 7878 that is the encoding is stored intact. There is also a new type 7879 "V_ASN1_OTHER" which is used when the class is not universal, in this 7880 case we have no idea what the actual type is so we just lump them all 7881 together. 7882 [Steve Henson] 7883 7884 *) On VMS, stdout may very well lead to a file that is written to 7885 in a record-oriented fashion. That means that every write() will 7886 write a separate record, which will be read separately by the 7887 programs trying to read from it. This can be very confusing. 7888 7889 The solution is to put a BIO filter in the way that will buffer 7890 text until a linefeed is reached, and then write everything a 7891 line at a time, so every record written will be an actual line, 7892 not chunks of lines and not (usually doesn't happen, but I've 7893 seen it once) several lines in one record. BIO_f_linebuffer() is 7894 the answer. 7895 7896 Currently, it's a VMS-only method, because that's where it has 7897 been tested well enough. 7898 [Richard Levitte] 7899 7900 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery, 7901 it can return incorrect results. 7902 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a, 7903 but it was in 0.9.6-beta[12].) 7904 [Bodo Moeller] 7905 7906 *) Disable the check for content being present when verifying detached 7907 signatures in pk7_smime.c. Some versions of Netscape (wrongly) 7908 include zero length content when signing messages. 7909 [Steve Henson] 7910 7911 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR 7912 BIO_ctrl (for BIO pairs). 7913 [Bodo M��ller] 7914 7915 *) Add DSO method for VMS. 7916 [Richard Levitte] 7917 7918 *) Bug fix: Montgomery multiplication could produce results with the 7919 wrong sign. 7920 [Ulf M��ller] 7921 7922 *) Add RPM specification openssl.spec and modify it to build three 7923 packages. The default package contains applications, application 7924 documentation and run-time libraries. The devel package contains 7925 include files, static libraries and function documentation. The 7926 doc package contains the contents of the doc directory. The original 7927 openssl.spec was provided by Damien Miller <djm@mindrot.org>. 7928 [Richard Levitte] 7929 7930 *) Add a large number of documentation files for many SSL routines. 7931 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 7932 7933 *) Add a configuration entry for Sony News 4. 7934 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>] 7935 7936 *) Don't set the two most significant bits to one when generating a 7937 random number < q in the DSA library. 7938 [Ulf M��ller] 7939 7940 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default 7941 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if 7942 the underlying transport is blocking) if a handshake took place. 7943 (The default behaviour is needed by applications such as s_client 7944 and s_server that use select() to determine when to use SSL_read; 7945 but for applications that know in advance when to expect data, it 7946 just makes things more complicated.) 7947 [Bodo Moeller] 7948 7949 *) Add RAND_egd_bytes(), which gives control over the number of bytes read 7950 from EGD. 7951 [Ben Laurie] 7952 7953 *) Add a few more EBCDIC conditionals that make `req' and `x509' 7954 work better on such systems. 7955 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 7956 7957 *) Add two demo programs for PKCS12_parse() and PKCS12_create(). 7958 Update PKCS12_parse() so it copies the friendlyName and the 7959 keyid to the certificates aux info. 7960 [Steve Henson] 7961 7962 *) Fix bug in PKCS7_verify() which caused an infinite loop 7963 if there was more than one signature. 7964 [Sven Uszpelkat <su@celocom.de>] 7965 7966 *) Major change in util/mkdef.pl to include extra information 7967 about each symbol, as well as presentig variables as well 7968 as functions. This change means that there's n more need 7969 to rebuild the .num files when some algorithms are excluded. 7970 [Richard Levitte] 7971 7972 *) Allow the verify time to be set by an application, 7973 rather than always using the current time. 7974 [Steve Henson] 7975 7976 *) Phase 2 verify code reorganisation. The certificate 7977 verify code now looks up an issuer certificate by a 7978 number of criteria: subject name, authority key id 7979 and key usage. It also verifies self signed certificates 7980 by the same criteria. The main comparison function is 7981 X509_check_issued() which performs these checks. 7982 7983 Lot of changes were necessary in order to support this 7984 without completely rewriting the lookup code. 7985 7986 Authority and subject key identifier are now cached. 7987 7988 The LHASH 'certs' is X509_STORE has now been replaced 7989 by a STACK_OF(X509_OBJECT). This is mainly because an 7990 LHASH can't store or retrieve multiple objects with 7991 the same hash value. 7992 7993 As a result various functions (which were all internal 7994 use only) have changed to handle the new X509_STORE 7995 structure. This will break anything that messed round 7996 with X509_STORE internally. 7997 7998 The functions X509_STORE_add_cert() now checks for an 7999 exact match, rather than just subject name. 8000 8001 The X509_STORE API doesn't directly support the retrieval 8002 of multiple certificates matching a given criteria, however 8003 this can be worked round by performing a lookup first 8004 (which will fill the cache with candidate certificates) 8005 and then examining the cache for matches. This is probably 8006 the best we can do without throwing out X509_LOOKUP 8007 entirely (maybe later...). 8008 8009 The X509_VERIFY_CTX structure has been enhanced considerably. 8010 8011 All certificate lookup operations now go via a get_issuer() 8012 callback. Although this currently uses an X509_STORE it 8013 can be replaced by custom lookups. This is a simple way 8014 to bypass the X509_STORE hackery necessary to make this 8015 work and makes it possible to use more efficient techniques 8016 in future. A very simple version which uses a simple 8017 STACK for its trusted certificate store is also provided 8018 using X509_STORE_CTX_trusted_stack(). 8019 8020 The verify_cb() and verify() callbacks now have equivalents 8021 in the X509_STORE_CTX structure. 8022 8023 X509_STORE_CTX also has a 'flags' field which can be used 8024 to customise the verify behaviour. 8025 [Steve Henson] 8026 8027 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which 8028 excludes S/MIME capabilities. 8029 [Steve Henson] 8030 8031 *) When a certificate request is read in keep a copy of the 8032 original encoding of the signed data and use it when outputing 8033 again. Signatures then use the original encoding rather than 8034 a decoded, encoded version which may cause problems if the 8035 request is improperly encoded. 8036 [Steve Henson] 8037 8038 *) For consistency with other BIO_puts implementations, call 8039 buffer_write(b, ...) directly in buffer_puts instead of calling 8040 BIO_write(b, ...). 8041 8042 In BIO_puts, increment b->num_write as in BIO_write. 8043 [Peter.Sylvester@EdelWeb.fr] 8044 8045 *) Fix BN_mul_word for the case where the word is 0. (We have to use 8046 BN_zero, we may not return a BIGNUM with an array consisting of 8047 words set to zero.) 8048 [Bodo Moeller] 8049 8050 *) Avoid calling abort() from within the library when problems are 8051 detected, except if preprocessor symbols have been defined 8052 (such as REF_CHECK, BN_DEBUG etc.). 8053 [Bodo Moeller] 8054 8055 *) New openssl application 'rsautl'. This utility can be 8056 used for low level RSA operations. DER public key 8057 BIO/fp routines also added. 8058 [Steve Henson] 8059 8060 *) New Configure entry and patches for compiling on QNX 4. 8061 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>] 8062 8063 *) A demo state-machine implementation was sponsored by 8064 Nuron (http://www.nuron.com/) and is now available in 8065 demos/state_machine. 8066 [Ben Laurie] 8067 8068 *) New options added to the 'dgst' utility for signature 8069 generation and verification. 8070 [Steve Henson] 8071 8072 *) Unrecognized PKCS#7 content types are now handled via a 8073 catch all ASN1_TYPE structure. This allows unsupported 8074 types to be stored as a "blob" and an application can 8075 encode and decode it manually. 8076 [Steve Henson] 8077 8078 *) Fix various signed/unsigned issues to make a_strex.c 8079 compile under VC++. 8080 [Oscar Jacobsson <oscar.jacobsson@celocom.com>] 8081 8082 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct 8083 length if passed a buffer. ASN1_INTEGER_to_BN failed 8084 if passed a NULL BN and its argument was negative. 8085 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>] 8086 8087 *) Modification to PKCS#7 encoding routines to output definite 8088 length encoding. Since currently the whole structures are in 8089 memory there's not real point in using indefinite length 8090 constructed encoding. However if OpenSSL is compiled with 8091 the flag PKCS7_INDEFINITE_ENCODING the old form is used. 8092 [Steve Henson] 8093 8094 *) Added BIO_vprintf() and BIO_vsnprintf(). 8095 [Richard Levitte] 8096 8097 *) Added more prefixes to parse for in the the strings written 8098 through a logging bio, to cover all the levels that are available 8099 through syslog. The prefixes are now: 8100 8101 PANIC, EMERG, EMR => LOG_EMERG 8102 ALERT, ALR => LOG_ALERT 8103 CRIT, CRI => LOG_CRIT 8104 ERROR, ERR => LOG_ERR 8105 WARNING, WARN, WAR => LOG_WARNING 8106 NOTICE, NOTE, NOT => LOG_NOTICE 8107 INFO, INF => LOG_INFO 8108 DEBUG, DBG => LOG_DEBUG 8109 8110 and as before, if none of those prefixes are present at the 8111 beginning of the string, LOG_ERR is chosen. 8112 8113 On Win32, the LOG_* levels are mapped according to this: 8114 8115 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE 8116 LOG_WARNING => EVENTLOG_WARNING_TYPE 8117 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE 8118 8119 [Richard Levitte] 8120 8121 *) Made it possible to reconfigure with just the configuration 8122 argument "reconf" or "reconfigure". The command line arguments 8123 are stored in Makefile.ssl in the variable CONFIGURE_ARGS, 8124 and are retrieved from there when reconfiguring. 8125 [Richard Levitte] 8126 8127 *) MD4 implemented. 8128 [Assar Westerlund <assar@sics.se>, Richard Levitte] 8129 8130 *) Add the arguments -CAfile and -CApath to the pkcs12 utility. 8131 [Richard Levitte] 8132 8133 *) The obj_dat.pl script was messing up the sorting of object 8134 names. The reason was that it compared the quoted version 8135 of strings as a result "OCSP" > "OCSP Signing" because 8136 " > SPACE. Changed script to store unquoted versions of 8137 names and add quotes on output. It was also omitting some 8138 names from the lookup table if they were given a default 8139 value (that is if SN is missing it is given the same 8140 value as LN and vice versa), these are now added on the 8141 grounds that if an object has a name we should be able to 8142 look it up. Finally added warning output when duplicate 8143 short or long names are found. 8144 [Steve Henson] 8145 8146 *) Changes needed for Tandem NSK. 8147 [Scott Uroff <scott@xypro.com>] 8148 8149 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in 8150 RSA_padding_check_SSLv23(), special padding was never detected 8151 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol 8152 version rollback attacks was not effective. 8153 8154 In s23_clnt.c, don't use special rollback-attack detection padding 8155 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the 8156 client; similarly, in s23_srvr.c, don't do the rollback check if 8157 SSL 2.0 is the only protocol enabled in the server. 8158 [Bodo Moeller] 8159 8160 *) Make it possible to get hexdumps of unprintable data with 'openssl 8161 asn1parse'. By implication, the functions ASN1_parse_dump() and 8162 BIO_dump_indent() are added. 8163 [Richard Levitte] 8164 8165 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex() 8166 these print out strings and name structures based on various 8167 flags including RFC2253 support and proper handling of 8168 multibyte characters. Added options to the 'x509' utility 8169 to allow the various flags to be set. 8170 [Steve Henson] 8171 8172 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME. 8173 Also change the functions X509_cmp_current_time() and 8174 X509_gmtime_adj() work with an ASN1_TIME structure, 8175 this will enable certificates using GeneralizedTime in validity 8176 dates to be checked. 8177 [Steve Henson] 8178 8179 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid 8180 negative public key encodings) on by default, 8181 NO_NEG_PUBKEY_BUG can be set to disable it. 8182 [Steve Henson] 8183 8184 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT 8185 content octets. An i2c_ASN1_OBJECT is unnecessary because 8186 the encoding can be trivially obtained from the structure. 8187 [Steve Henson] 8188 8189 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock), 8190 not read locks (CRYPTO_r_[un]lock). 8191 [Bodo Moeller] 8192 8193 *) A first attempt at creating official support for shared 8194 libraries through configuration. I've kept it so the 8195 default is static libraries only, and the OpenSSL programs 8196 are always statically linked for now, but there are 8197 preparations for dynamic linking in place. 8198 This has been tested on Linux and Tru64. 8199 [Richard Levitte] 8200 8201 *) Randomness polling function for Win9x, as described in: 8202 Peter Gutmann, Software Generation of Practically Strong 8203 Random Numbers. 8204 [Ulf M��ller] 8205 8206 *) Fix so PRNG is seeded in req if using an already existing 8207 DSA key. 8208 [Steve Henson] 8209 8210 *) New options to smime application. -inform and -outform 8211 allow alternative formats for the S/MIME message including 8212 PEM and DER. The -content option allows the content to be 8213 specified separately. This should allow things like Netscape 8214 form signing output easier to verify. 8215 [Steve Henson] 8216 8217 *) Fix the ASN1 encoding of tags using the 'long form'. 8218 [Steve Henson] 8219 8220 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT 8221 STRING types. These convert content octets to and from the 8222 underlying type. The actual tag and length octets are 8223 already assumed to have been read in and checked. These 8224 are needed because all other string types have virtually 8225 identical handling apart from the tag. By having versions 8226 of the ASN1 functions that just operate on content octets 8227 IMPLICIT tagging can be handled properly. It also allows 8228 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED 8229 and ASN1_INTEGER are identical apart from the tag. 8230 [Steve Henson] 8231 8232 *) Change the handling of OID objects as follows: 8233 8234 - New object identifiers are inserted in objects.txt, following 8235 the syntax given in objects.README. 8236 - objects.pl is used to process obj_mac.num and create a new 8237 obj_mac.h. 8238 - obj_dat.pl is used to create a new obj_dat.h, using the data in 8239 obj_mac.h. 8240 8241 This is currently kind of a hack, and the perl code in objects.pl 8242 isn't very elegant, but it works as I intended. The simplest way 8243 to check that it worked correctly is to look in obj_dat.h and 8244 check the array nid_objs and make sure the objects haven't moved 8245 around (this is important!). Additions are OK, as well as 8246 consistent name changes. 8247 [Richard Levitte] 8248 8249 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1'). 8250 [Bodo Moeller] 8251 8252 *) Addition of the command line parameter '-rand file' to 'openssl req'. 8253 The given file adds to whatever has already been seeded into the 8254 random pool through the RANDFILE configuration file option or 8255 environment variable, or the default random state file. 8256 [Richard Levitte] 8257 8258 *) mkstack.pl now sorts each macro group into lexical order. 8259 Previously the output order depended on the order the files 8260 appeared in the directory, resulting in needless rewriting 8261 of safestack.h . 8262 [Steve Henson] 8263 8264 *) Patches to make OpenSSL compile under Win32 again. Mostly 8265 work arounds for the VC++ problem that it treats func() as 8266 func(void). Also stripped out the parts of mkdef.pl that 8267 added extra typesafe functions: these no longer exist. 8268 [Steve Henson] 8269 8270 *) Reorganisation of the stack code. The macros are now all 8271 collected in safestack.h . Each macro is defined in terms of 8272 a "stack macro" of the form SKM_<name>(type, a, b). The 8273 DEBUG_SAFESTACK is now handled in terms of function casts, 8274 this has the advantage of retaining type safety without the 8275 use of additional functions. If DEBUG_SAFESTACK is not defined 8276 then the non typesafe macros are used instead. Also modified the 8277 mkstack.pl script to handle the new form. Needs testing to see 8278 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK 8279 the default if no major problems. Similar behaviour for ASN1_SET_OF 8280 and PKCS12_STACK_OF. 8281 [Steve Henson] 8282 8283 *) When some versions of IIS use the 'NET' form of private key the 8284 key derivation algorithm is different. Normally MD5(password) is 8285 used as a 128 bit RC4 key. In the modified case 8286 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some 8287 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same 8288 as the old Netscape_RSA functions except they have an additional 8289 'sgckey' parameter which uses the modified algorithm. Also added 8290 an -sgckey command line option to the rsa utility. Thanks to 8291 Adrian Peck <bertie@ncipher.com> for posting details of the modified 8292 algorithm to openssl-dev. 8293 [Steve Henson] 8294 8295 *) The evp_local.h macros were using 'c.##kname' which resulted in 8296 invalid expansion on some systems (SCO 5.0.5 for example). 8297 Corrected to 'c.kname'. 8298 [Phillip Porch <root@theporch.com>] 8299 8300 *) New X509_get1_email() and X509_REQ_get1_email() functions that return 8301 a STACK of email addresses from a certificate or request, these look 8302 in the subject name and the subject alternative name extensions and 8303 omit any duplicate addresses. 8304 [Steve Henson] 8305 8306 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows. 8307 This makes DSA verification about 2 % faster. 8308 [Bodo Moeller] 8309 8310 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5 8311 (meaning that now 2^5 values will be precomputed, which is only 4 KB 8312 plus overhead for 1024 bit moduli). 8313 This makes exponentiations about 0.5 % faster for 1024 bit 8314 exponents (as measured by "openssl speed rsa2048"). 8315 [Bodo Moeller] 8316 8317 *) Rename memory handling macros to avoid conflicts with other 8318 software: 8319 Malloc => OPENSSL_malloc 8320 Malloc_locked => OPENSSL_malloc_locked 8321 Realloc => OPENSSL_realloc 8322 Free => OPENSSL_free 8323 [Richard Levitte] 8324 8325 *) New function BN_mod_exp_mont_word for small bases (roughly 15% 8326 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange). 8327 [Bodo Moeller] 8328 8329 *) CygWin32 support. 8330 [John Jarvie <jjarvie@newsguy.com>] 8331 8332 *) The type-safe stack code has been rejigged. It is now only compiled 8333 in when OpenSSL is configured with the DEBUG_SAFESTACK option and 8334 by default all type-specific stack functions are "#define"d back to 8335 standard stack functions. This results in more streamlined output 8336 but retains the type-safety checking possibilities of the original 8337 approach. 8338 [Geoff Thorpe] 8339 8340 *) The STACK code has been cleaned up, and certain type declarations 8341 that didn't make a lot of sense have been brought in line. This has 8342 also involved a cleanup of sorts in safestack.h to more correctly 8343 map type-safe stack functions onto their plain stack counterparts. 8344 This work has also resulted in a variety of "const"ifications of 8345 lots of the code, especially "_cmp" operations which should normally 8346 be prototyped with "const" parameters anyway. 8347 [Geoff Thorpe] 8348 8349 *) When generating bytes for the first time in md_rand.c, 'stir the pool' 8350 by seeding with STATE_SIZE dummy bytes (with zero entropy count). 8351 (The PRNG state consists of two parts, the large pool 'state' and 'md', 8352 where all of 'md' is used each time the PRNG is used, but 'state' 8353 is used only indexed by a cyclic counter. As entropy may not be 8354 well distributed from the beginning, 'md' is important as a 8355 chaining variable. However, the output function chains only half 8356 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains 8357 all of 'md', and seeding with STATE_SIZE dummy bytes will result 8358 in all of 'state' being rewritten, with the new values depending 8359 on virtually all of 'md'. This overcomes the 80 bit limitation.) 8360 [Bodo Moeller] 8361 8362 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when 8363 the handshake is continued after ssl_verify_cert_chain(); 8364 otherwise, if SSL_VERIFY_NONE is set, remaining error codes 8365 can lead to 'unexplainable' connection aborts later. 8366 [Bodo Moeller; problem tracked down by Lutz Jaenicke] 8367 8368 *) Major EVP API cipher revision. 8369 Add hooks for extra EVP features. This allows various cipher 8370 parameters to be set in the EVP interface. Support added for variable 8371 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and 8372 setting of RC2 and RC5 parameters. 8373 8374 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length 8375 ciphers. 8376 8377 Remove lots of duplicated code from the EVP library. For example *every* 8378 cipher init() function handles the 'iv' in the same way according to the 8379 cipher mode. They also all do nothing if the 'key' parameter is NULL and 8380 for CFB and OFB modes they zero ctx->num. 8381 8382 New functionality allows removal of S/MIME code RC2 hack. 8383 8384 Most of the routines have the same form and so can be declared in terms 8385 of macros. 8386 8387 By shifting this to the top level EVP_CipherInit() it can be removed from 8388 all individual ciphers. If the cipher wants to handle IVs or keys 8389 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT 8390 flags. 8391 8392 Change lots of functions like EVP_EncryptUpdate() to now return a 8393 value: although software versions of the algorithms cannot fail 8394 any installed hardware versions can. 8395 [Steve Henson] 8396 8397 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if 8398 this option is set, tolerate broken clients that send the negotiated 8399 protocol version number instead of the requested protocol version 8400 number. 8401 [Bodo Moeller] 8402 8403 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag; 8404 i.e. non-zero for export ciphersuites, zero otherwise. 8405 Previous versions had this flag inverted, inconsistent with 8406 rsa_tmp_cb (..._TMP_RSA_CB). 8407 [Bodo Moeller; problem reported by Amit Chopra] 8408 8409 *) Add missing DSA library text string. Work around for some IIS 8410 key files with invalid SEQUENCE encoding. 8411 [Steve Henson] 8412 8413 *) Add a document (doc/standards.txt) that list all kinds of standards 8414 and so on that are implemented in OpenSSL. 8415 [Richard Levitte] 8416 8417 *) Enhance c_rehash script. Old version would mishandle certificates 8418 with the same subject name hash and wouldn't handle CRLs at all. 8419 Added -fingerprint option to crl utility, to support new c_rehash 8420 features. 8421 [Steve Henson] 8422 8423 *) Eliminate non-ANSI declarations in crypto.h and stack.h. 8424 [Ulf M��ller] 8425 8426 *) Fix for SSL server purpose checking. Server checking was 8427 rejecting certificates which had extended key usage present 8428 but no ssl client purpose. 8429 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>] 8430 8431 *) Make PKCS#12 code work with no password. The PKCS#12 spec 8432 is a little unclear about how a blank password is handled. 8433 Since the password in encoded as a BMPString with terminating 8434 double NULL a zero length password would end up as just the 8435 double NULL. However no password at all is different and is 8436 handled differently in the PKCS#12 key generation code. NS 8437 treats a blank password as zero length. MSIE treats it as no 8438 password on export: but it will try both on import. We now do 8439 the same: PKCS12_parse() tries zero length and no password if 8440 the password is set to "" or NULL (NULL is now a valid password: 8441 it wasn't before) as does the pkcs12 application. 8442 [Steve Henson] 8443 8444 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use 8445 perror when PEM_read_bio_X509_REQ fails, the error message must 8446 be obtained from the error queue. 8447 [Bodo Moeller] 8448 8449 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing 8450 it in ERR_remove_state if appropriate, and change ERR_get_state 8451 accordingly to avoid race conditions (this is necessary because 8452 thread_hash is no longer constant once set). 8453 [Bodo Moeller] 8454 8455 *) Bugfix for linux-elf makefile.one. 8456 [Ulf M��ller] 8457 8458 *) RSA_get_default_method() will now cause a default 8459 RSA_METHOD to be chosen if one doesn't exist already. 8460 Previously this was only set during a call to RSA_new() 8461 or RSA_new_method(NULL) meaning it was possible for 8462 RSA_get_default_method() to return NULL. 8463 [Geoff Thorpe] 8464 8465 *) Added native name translation to the existing DSO code 8466 that will convert (if the flag to do so is set) filenames 8467 that are sufficiently small and have no path information 8468 into a canonical native form. Eg. "blah" converted to 8469 "libblah.so" or "blah.dll" etc. 8470 [Geoff Thorpe] 8471 8472 *) New function ERR_error_string_n(e, buf, len) which is like 8473 ERR_error_string(e, buf), but writes at most 'len' bytes 8474 including the 0 terminator. For ERR_error_string_n, 'buf' 8475 may not be NULL. 8476 [Damien Miller <djm@mindrot.org>, Bodo Moeller] 8477 8478 *) CONF library reworked to become more general. A new CONF 8479 configuration file reader "class" is implemented as well as a 8480 new functions (NCONF_*, for "New CONF") to handle it. The now 8481 old CONF_* functions are still there, but are reimplemented to 8482 work in terms of the new functions. Also, a set of functions 8483 to handle the internal storage of the configuration data is 8484 provided to make it easier to write new configuration file 8485 reader "classes" (I can definitely see something reading a 8486 configuration file in XML format, for example), called _CONF_*, 8487 or "the configuration storage API"... 8488 8489 The new configuration file reading functions are: 8490 8491 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio, 8492 NCONF_get_section, NCONF_get_string, NCONF_get_numbre 8493 8494 NCONF_default, NCONF_WIN32 8495 8496 NCONF_dump_fp, NCONF_dump_bio 8497 8498 NCONF_default and NCONF_WIN32 are method (or "class") choosers, 8499 NCONF_new creates a new CONF object. This works in the same way 8500 as other interfaces in OpenSSL, like the BIO interface. 8501 NCONF_dump_* dump the internal storage of the configuration file, 8502 which is useful for debugging. All other functions take the same 8503 arguments as the old CONF_* functions wth the exception of the 8504 first that must be a `CONF *' instead of a `LHASH *'. 8505 8506 To make it easer to use the new classes with the old CONF_* functions, 8507 the function CONF_set_default_method is provided. 8508 [Richard Levitte] 8509 8510 *) Add '-tls1' option to 'openssl ciphers', which was already 8511 mentioned in the documentation but had not been implemented. 8512 (This option is not yet really useful because even the additional 8513 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.) 8514 [Bodo Moeller] 8515 8516 *) Initial DSO code added into libcrypto for letting OpenSSL (and 8517 OpenSSL-based applications) load shared libraries and bind to 8518 them in a portable way. 8519 [Geoff Thorpe, with contributions from Richard Levitte] 8520 8521 Changes between 0.9.5 and 0.9.5a [1 Apr 2000] 8522 8523 *) Make sure _lrotl and _lrotr are only used with MSVC. 8524 8525 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status 8526 (the default implementation of RAND_status). 8527 8528 *) Rename openssl x509 option '-crlext', which was added in 0.9.5, 8529 to '-clrext' (= clear extensions), as intended and documented. 8530 [Bodo Moeller; inconsistency pointed out by Michael Attili 8531 <attili@amaxo.com>] 8532 8533 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length 8534 was larger than the MD block size. 8535 [Steve Henson, pointed out by Yost William <YostW@tce.com>] 8536 8537 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument 8538 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set() 8539 using the passed key: if the passed key was a private key the result 8540 of X509_print(), for example, would be to print out all the private key 8541 components. 8542 [Steve Henson] 8543 8544 *) des_quad_cksum() byte order bug fix. 8545 [Ulf M��ller, using the problem description in krb4-0.9.7, where 8546 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>] 8547 8548 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly 8549 discouraged. 8550 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>] 8551 8552 *) For easily testing in shell scripts whether some command 8553 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX' 8554 returns with exit code 0 iff no command of the given name is available. 8555 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases, 8556 the output goes to stdout and nothing is printed to stderr. 8557 Additional arguments are always ignored. 8558 8559 Since for each cipher there is a command of the same name, 8560 the 'no-cipher' compilation switches can be tested this way. 8561 8562 ('openssl no-XXX' is not able to detect pseudo-commands such 8563 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.) 8564 [Bodo Moeller] 8565 8566 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration. 8567 [Bodo Moeller] 8568 8569 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE 8570 is set; it will be thrown away anyway because each handshake creates 8571 its own key. 8572 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition 8573 to parameters -- in previous versions (since OpenSSL 0.9.3) the 8574 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining 8575 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro. 8576 [Bodo Moeller] 8577 8578 *) New s_client option -ign_eof: EOF at stdin is ignored, and 8579 'Q' and 'R' lose their special meanings (quit/renegotiate). 8580 This is part of what -quiet does; unlike -quiet, -ign_eof 8581 does not suppress any output. 8582 [Richard Levitte] 8583 8584 *) Add compatibility options to the purpose and trust code. The 8585 purpose X509_PURPOSE_ANY is "any purpose" which automatically 8586 accepts a certificate or CA, this was the previous behaviour, 8587 with all the associated security issues. 8588 8589 X509_TRUST_COMPAT is the old trust behaviour: only and 8590 automatically trust self signed roots in certificate store. A 8591 new trust setting X509_TRUST_DEFAULT is used to specify that 8592 a purpose has no associated trust setting and it should instead 8593 use the value in the default purpose. 8594 [Steve Henson] 8595 8596 *) Fix the PKCS#8 DSA private key code so it decodes keys again 8597 and fix a memory leak. 8598 [Steve Henson] 8599 8600 *) In util/mkerr.pl (which implements 'make errors'), preserve 8601 reason strings from the previous version of the .c file, as 8602 the default to have only downcase letters (and digits) in 8603 automatically generated reasons codes is not always appropriate. 8604 [Bodo Moeller] 8605 8606 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table 8607 using strerror. Previously, ERR_reason_error_string() returned 8608 library names as reason strings for SYSerr; but SYSerr is a special 8609 case where small numbers are errno values, not library numbers. 8610 [Bodo Moeller] 8611 8612 *) Add '-dsaparam' option to 'openssl dhparam' application. This 8613 converts DSA parameters into DH parameters. (When creating parameters, 8614 DSA_generate_parameters is used.) 8615 [Bodo Moeller] 8616 8617 *) Include 'length' (recommended exponent length) in C code generated 8618 by 'openssl dhparam -C'. 8619 [Bodo Moeller] 8620 8621 *) The second argument to set_label in perlasm was already being used 8622 so couldn't be used as a "file scope" flag. Moved to third argument 8623 which was free. 8624 [Steve Henson] 8625 8626 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes 8627 instead of RAND_bytes for encryption IVs and salts. 8628 [Bodo Moeller] 8629 8630 *) Include RAND_status() into RAND_METHOD instead of implementing 8631 it only for md_rand.c Otherwise replacing the PRNG by calling 8632 RAND_set_rand_method would be impossible. 8633 [Bodo Moeller] 8634 8635 *) Don't let DSA_generate_key() enter an infinite loop if the random 8636 number generation fails. 8637 [Bodo Moeller] 8638 8639 *) New 'rand' application for creating pseudo-random output. 8640 [Bodo Moeller] 8641 8642 *) Added configuration support for Linux/IA64 8643 [Rolf Haberrecker <rolf@suse.de>] 8644 8645 *) Assembler module support for Mingw32. 8646 [Ulf M��ller] 8647 8648 *) Shared library support for HPUX (in shlib/). 8649 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous] 8650 8651 *) Shared library support for Solaris gcc. 8652 [Lutz Behnke <behnke@trustcenter.de>] 8653 8654 Changes between 0.9.4 and 0.9.5 [28 Feb 2000] 8655 8656 *) PKCS7_encrypt() was adding text MIME headers twice because they 8657 were added manually and by SMIME_crlf_copy(). 8658 [Steve Henson] 8659 8660 *) In bntest.c don't call BN_rand with zero bits argument. 8661 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>] 8662 8663 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n] 8664 case was implemented. This caused BN_div_recp() to fail occasionally. 8665 [Ulf M��ller] 8666 8667 *) Add an optional second argument to the set_label() in the perl 8668 assembly language builder. If this argument exists and is set 8669 to 1 it signals that the assembler should use a symbol whose 8670 scope is the entire file, not just the current function. This 8671 is needed with MASM which uses the format label:: for this scope. 8672 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>] 8673 8674 *) Change the ASN1 types so they are typedefs by default. Before 8675 almost all types were #define'd to ASN1_STRING which was causing 8676 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING) 8677 for example. 8678 [Steve Henson] 8679 8680 *) Change names of new functions to the new get1/get0 naming 8681 convention: After 'get1', the caller owns a reference count 8682 and has to call ..._free; 'get0' returns a pointer to some 8683 data structure without incrementing reference counters. 8684 (Some of the existing 'get' functions increment a reference 8685 counter, some don't.) 8686 Similarly, 'set1' and 'add1' functions increase reference 8687 counters or duplicate objects. 8688 [Steve Henson] 8689 8690 *) Allow for the possibility of temp RSA key generation failure: 8691 the code used to assume it always worked and crashed on failure. 8692 [Steve Henson] 8693 8694 *) Fix potential buffer overrun problem in BIO_printf(). 8695 [Ulf M��ller, using public domain code by Patrick Powell; problem 8696 pointed out by David Sacerdote <das33@cornell.edu>] 8697 8698 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions 8699 RAND_egd() and RAND_status(). In the command line application, 8700 the EGD socket can be specified like a seed file using RANDFILE 8701 or -rand. 8702 [Ulf M��ller] 8703 8704 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures. 8705 Some CAs (e.g. Verisign) distribute certificates in this form. 8706 [Steve Henson] 8707 8708 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher 8709 list to exclude them. This means that no special compilation option 8710 is needed to use anonymous DH: it just needs to be included in the 8711 cipher list. 8712 [Steve Henson] 8713 8714 *) Change the EVP_MD_CTX_type macro so its meaning consistent with 8715 EVP_MD_type. The old functionality is available in a new macro called 8716 EVP_MD_md(). Change code that uses it and update docs. 8717 [Steve Henson] 8718 8719 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions 8720 where the 'void *' argument is replaced by a function pointer argument. 8721 Previously 'void *' was abused to point to functions, which works on 8722 many platforms, but is not correct. As these functions are usually 8723 called by macros defined in OpenSSL header files, most source code 8724 should work without changes. 8725 [Richard Levitte] 8726 8727 *) <openssl/opensslconf.h> (which is created by Configure) now contains 8728 sections with information on -D... compiler switches used for 8729 compiling the library so that applications can see them. To enable 8730 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES 8731 must be defined. E.g., 8732 #define OPENSSL_ALGORITHM_DEFINES 8733 #include <openssl/opensslconf.h> 8734 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc. 8735 [Richard Levitte, Ulf and Bodo M��ller] 8736 8737 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS 8738 record layer. 8739 [Bodo Moeller] 8740 8741 *) Change the 'other' type in certificate aux info to a STACK_OF 8742 X509_ALGOR. Although not an AlgorithmIdentifier as such it has 8743 the required ASN1 format: arbitrary types determined by an OID. 8744 [Steve Henson] 8745 8746 *) Add some PEM_write_X509_REQ_NEW() functions and a command line 8747 argument to 'req'. This is not because the function is newer or 8748 better than others it just uses the work 'NEW' in the certificate 8749 request header lines. Some software needs this. 8750 [Steve Henson] 8751 8752 *) Reorganise password command line arguments: now passwords can be 8753 obtained from various sources. Delete the PEM_cb function and make 8754 it the default behaviour: i.e. if the callback is NULL and the 8755 usrdata argument is not NULL interpret it as a null terminated pass 8756 phrase. If usrdata and the callback are NULL then the pass phrase 8757 is prompted for as usual. 8758 [Steve Henson] 8759 8760 *) Add support for the Compaq Atalla crypto accelerator. If it is installed, 8761 the support is automatically enabled. The resulting binaries will 8762 autodetect the card and use it if present. 8763 [Ben Laurie and Compaq Inc.] 8764 8765 *) Work around for Netscape hang bug. This sends certificate request 8766 and server done in one record. Since this is perfectly legal in the 8767 SSL/TLS protocol it isn't a "bug" option and is on by default. See 8768 the bugs/SSLv3 entry for more info. 8769 [Steve Henson] 8770 8771 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround. 8772 [Andy Polyakov] 8773 8774 *) Add -rand argument to smime and pkcs12 applications and read/write 8775 of seed file. 8776 [Steve Henson] 8777 8778 *) New 'passwd' tool for crypt(3) and apr1 password hashes. 8779 [Bodo Moeller] 8780 8781 *) Add command line password options to the remaining applications. 8782 [Steve Henson] 8783 8784 *) Bug fix for BN_div_recp() for numerators with an even number of 8785 bits. 8786 [Ulf M��ller] 8787 8788 *) More tests in bntest.c, and changed test_bn output. 8789 [Ulf M��ller] 8790 8791 *) ./config recognizes MacOS X now. 8792 [Andy Polyakov] 8793 8794 *) Bug fix for BN_div() when the first words of num and divsor are 8795 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0). 8796 [Ulf M��ller] 8797 8798 *) Add support for various broken PKCS#8 formats, and command line 8799 options to produce them. 8800 [Steve Henson] 8801 8802 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to 8803 get temporary BIGNUMs from a BN_CTX. 8804 [Ulf M��ller] 8805 8806 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont() 8807 for p == 0. 8808 [Ulf M��ller] 8809 8810 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and 8811 include a #define from the old name to the new. The original intent 8812 was that statically linked binaries could for example just call 8813 SSLeay_add_all_ciphers() to just add ciphers to the table and not 8814 link with digests. This never worked becayse SSLeay_add_all_digests() 8815 and SSLeay_add_all_ciphers() were in the same source file so calling 8816 one would link with the other. They are now in separate source files. 8817 [Steve Henson] 8818 8819 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'. 8820 [Steve Henson] 8821 8822 *) Use a less unusual form of the Miller-Rabin primality test (it used 8823 a binary algorithm for exponentiation integrated into the Miller-Rabin 8824 loop, our standard modexp algorithms are faster). 8825 [Bodo Moeller] 8826 8827 *) Support for the EBCDIC character set completed. 8828 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>] 8829 8830 *) Source code cleanups: use const where appropriate, eliminate casts, 8831 use void * instead of char * in lhash. 8832 [Ulf M��ller] 8833 8834 *) Bugfix: ssl3_send_server_key_exchange was not restartable 8835 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of 8836 this the server could overwrite ephemeral keys that the client 8837 has already seen). 8838 [Bodo Moeller] 8839 8840 *) Turn DSA_is_prime into a macro that calls BN_is_prime, 8841 using 50 iterations of the Rabin-Miller test. 8842 8843 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50 8844 iterations of the Rabin-Miller test as required by the appendix 8845 to FIPS PUB 186[-1]) instead of DSA_is_prime. 8846 As BN_is_prime_fasttest includes trial division, DSA parameter 8847 generation becomes much faster. 8848 8849 This implies a change for the callback functions in DSA_is_prime 8850 and DSA_generate_parameters: The callback function is called once 8851 for each positive witness in the Rabin-Miller test, not just 8852 occasionally in the inner loop; and the parameters to the 8853 callback function now provide an iteration count for the outer 8854 loop rather than for the current invocation of the inner loop. 8855 DSA_generate_parameters additionally can call the callback 8856 function with an 'iteration count' of -1, meaning that a 8857 candidate has passed the trial division test (when q is generated 8858 from an application-provided seed, trial division is skipped). 8859 [Bodo Moeller] 8860 8861 *) New function BN_is_prime_fasttest that optionally does trial 8862 division before starting the Rabin-Miller test and has 8863 an additional BN_CTX * argument (whereas BN_is_prime always 8864 has to allocate at least one BN_CTX). 8865 'callback(1, -1, cb_arg)' is called when a number has passed the 8866 trial division stage. 8867 [Bodo Moeller] 8868 8869 *) Fix for bug in CRL encoding. The validity dates weren't being handled 8870 as ASN1_TIME. 8871 [Steve Henson] 8872 8873 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file. 8874 [Steve Henson] 8875 8876 *) New function BN_pseudo_rand(). 8877 [Ulf M��ller] 8878 8879 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable) 8880 bignum version of BN_from_montgomery() with the working code from 8881 SSLeay 0.9.0 (the word based version is faster anyway), and clean up 8882 the comments. 8883 [Ulf M��ller] 8884 8885 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that 8886 made it impossible to use the same SSL_SESSION data structure in 8887 SSL2 clients in multiple threads. 8888 [Bodo Moeller] 8889 8890 *) The return value of RAND_load_file() no longer counts bytes obtained 8891 by stat(). RAND_load_file(..., -1) is new and uses the complete file 8892 to seed the PRNG (previously an explicit byte count was required). 8893 [Ulf M��ller, Bodo M��ller] 8894 8895 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes 8896 used (char *) instead of (void *) and had casts all over the place. 8897 [Steve Henson] 8898 8899 *) Make BN_generate_prime() return NULL on error if ret!=NULL. 8900 [Ulf M��ller] 8901 8902 *) Retain source code compatibility for BN_prime_checks macro: 8903 BN_is_prime(..., BN_prime_checks, ...) now uses 8904 BN_prime_checks_for_size to determine the appropriate number of 8905 Rabin-Miller iterations. 8906 [Ulf M��ller] 8907 8908 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to 8909 DH_CHECK_P_NOT_SAFE_PRIME. 8910 (Check if this is true? OpenPGP calls them "strong".) 8911 [Ulf M��ller] 8912 8913 *) Merge the functionality of "dh" and "gendh" programs into a new program 8914 "dhparam". The old programs are retained for now but will handle DH keys 8915 (instead of parameters) in future. 8916 [Steve Henson] 8917 8918 *) Make the ciphers, s_server and s_client programs check the return values 8919 when a new cipher list is set. 8920 [Steve Henson] 8921 8922 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit 8923 ciphers. Before when the 56bit ciphers were enabled the sorting was 8924 wrong. 8925 8926 The syntax for the cipher sorting has been extended to support sorting by 8927 cipher-strength (using the strength_bits hard coded in the tables). 8928 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod). 8929 8930 Fix a bug in the cipher-command parser: when supplying a cipher command 8931 string with an "undefined" symbol (neither command nor alphanumeric 8932 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now 8933 an error is flagged. 8934 8935 Due to the strength-sorting extension, the code of the 8936 ssl_create_cipher_list() function was completely rearranged. I hope that 8937 the readability was also increased :-) 8938 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 8939 8940 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1 8941 for the first serial number and places 2 in the serial number file. This 8942 avoids problems when the root CA is created with serial number zero and 8943 the first user certificate has the same issuer name and serial number 8944 as the root CA. 8945 [Steve Henson] 8946 8947 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses 8948 the new code. Add documentation for this stuff. 8949 [Steve Henson] 8950 8951 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from 8952 X509_*() to X509at_*() on the grounds that they don't handle X509 8953 structures and behave in an analagous way to the X509v3 functions: 8954 they shouldn't be called directly but wrapper functions should be used 8955 instead. 8956 8957 So we also now have some wrapper functions that call the X509at functions 8958 when passed certificate requests. (TO DO: similar things can be done with 8959 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other 8960 things. Some of these need some d2i or i2d and print functionality 8961 because they handle more complex structures.) 8962 [Steve Henson] 8963 8964 *) Add missing #ifndefs that caused missing symbols when building libssl 8965 as a shared library without RSA. Use #ifndef NO_SSL2 instead of 8966 NO_RSA in ssl/s2*.c. 8967 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf M��ller] 8968 8969 *) Precautions against using the PRNG uninitialized: RAND_bytes() now 8970 has a return value which indicates the quality of the random data 8971 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's 8972 error queue. New function RAND_pseudo_bytes() generates output that is 8973 guaranteed to be unique but not unpredictable. RAND_add is like 8974 RAND_seed, but takes an extra argument for an entropy estimate 8975 (RAND_seed always assumes full entropy). 8976 [Ulf M��ller] 8977 8978 *) Do more iterations of Rabin-Miller probable prime test (specifically, 8979 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes 8980 instead of only 2 for all lengths; see BN_prime_checks_for_size definition 8981 in crypto/bn/bn_prime.c for the complete table). This guarantees a 8982 false-positive rate of at most 2^-80 for random input. 8983 [Bodo Moeller] 8984 8985 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs. 8986 [Bodo Moeller] 8987 8988 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain 8989 in the 0.9.5 release), this returns the chain 8990 from an X509_CTX structure with a dup of the stack and all 8991 the X509 reference counts upped: so the stack will exist 8992 after X509_CTX_cleanup() has been called. Modify pkcs12.c 8993 to use this. 8994 8995 Also make SSL_SESSION_print() print out the verify return 8996 code. 8997 [Steve Henson] 8998 8999 *) Add manpage for the pkcs12 command. Also change the default 9000 behaviour so MAC iteration counts are used unless the new 9001 -nomaciter option is used. This improves file security and 9002 only older versions of MSIE (4.0 for example) need it. 9003 [Steve Henson] 9004 9005 *) Honor the no-xxx Configure options when creating .DEF files. 9006 [Ulf M��ller] 9007 9008 *) Add PKCS#10 attributes to field table: challengePassword, 9009 unstructuredName and unstructuredAddress. These are taken from 9010 draft PKCS#9 v2.0 but are compatible with v1.2 provided no 9011 international characters are used. 9012 9013 More changes to X509_ATTRIBUTE code: allow the setting of types 9014 based on strings. Remove the 'loc' parameter when adding 9015 attributes because these will be a SET OF encoding which is sorted 9016 in ASN1 order. 9017 [Steve Henson] 9018 9019 *) Initial changes to the 'req' utility to allow request generation 9020 automation. This will allow an application to just generate a template 9021 file containing all the field values and have req construct the 9022 request. 9023 9024 Initial support for X509_ATTRIBUTE handling. Stacks of these are 9025 used all over the place including certificate requests and PKCS#7 9026 structures. They are currently handled manually where necessary with 9027 some primitive wrappers for PKCS#7. The new functions behave in a 9028 manner analogous to the X509 extension functions: they allow 9029 attributes to be looked up by NID and added. 9030 9031 Later something similar to the X509V3 code would be desirable to 9032 automatically handle the encoding, decoding and printing of the 9033 more complex types. The string types like challengePassword can 9034 be handled by the string table functions. 9035 9036 Also modified the multi byte string table handling. Now there is 9037 a 'global mask' which masks out certain types. The table itself 9038 can use the flag STABLE_NO_MASK to ignore the mask setting: this 9039 is useful when for example there is only one permissible type 9040 (as in countryName) and using the mask might result in no valid 9041 types at all. 9042 [Steve Henson] 9043 9044 *) Clean up 'Finished' handling, and add functions SSL_get_finished and 9045 SSL_get_peer_finished to allow applications to obtain the latest 9046 Finished messages sent to the peer or expected from the peer, 9047 respectively. (SSL_get_peer_finished is usually the Finished message 9048 actually received from the peer, otherwise the protocol will be aborted.) 9049 9050 As the Finished message are message digests of the complete handshake 9051 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can 9052 be used for external authentication procedures when the authentication 9053 provided by SSL/TLS is not desired or is not enough. 9054 [Bodo Moeller] 9055 9056 *) Enhanced support for Alpha Linux is added. Now ./config checks if 9057 the host supports BWX extension and if Compaq C is present on the 9058 $PATH. Just exploiting of the BWX extension results in 20-30% 9059 performance kick for some algorithms, e.g. DES and RC4 to mention 9060 a couple. Compaq C in turn generates ~20% faster code for MD5 and 9061 SHA1. 9062 [Andy Polyakov] 9063 9064 *) Add support for MS "fast SGC". This is arguably a violation of the 9065 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with 9066 weak crypto and after checking the certificate is SGC a second one 9067 with strong crypto. MS SGC stops the first handshake after receiving 9068 the server certificate message and sends a second client hello. Since 9069 a server will typically do all the time consuming operations before 9070 expecting any further messages from the client (server key exchange 9071 is the most expensive) there is little difference between the two. 9072 9073 To get OpenSSL to support MS SGC we have to permit a second client 9074 hello message after we have sent server done. In addition we have to 9075 reset the MAC if we do get this second client hello. 9076 [Steve Henson] 9077 9078 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide 9079 if a DER encoded private key is RSA or DSA traditional format. Changed 9080 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional" 9081 format DER encoded private key. Newer code should use PKCS#8 format which 9082 has the key type encoded in the ASN1 structure. Added DER private key 9083 support to pkcs8 application. 9084 [Steve Henson] 9085 9086 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous 9087 ciphersuites has been selected (as required by the SSL 3/TLS 1 9088 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT 9089 is set, we interpret this as a request to violate the specification 9090 (the worst that can happen is a handshake failure, and 'correct' 9091 behaviour would result in a handshake failure anyway). 9092 [Bodo Moeller] 9093 9094 *) In SSL_CTX_add_session, take into account that there might be multiple 9095 SSL_SESSION structures with the same session ID (e.g. when two threads 9096 concurrently obtain them from an external cache). 9097 The internal cache can handle only one SSL_SESSION with a given ID, 9098 so if there's a conflict, we now throw out the old one to achieve 9099 consistency. 9100 [Bodo Moeller] 9101 9102 *) Add OIDs for idea and blowfish in CBC mode. This will allow both 9103 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to 9104 some routines that use cipher OIDs: some ciphers do not have OIDs 9105 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for 9106 example. 9107 [Steve Henson] 9108 9109 *) Simplify the trust setting structure and code. Now we just have 9110 two sequences of OIDs for trusted and rejected settings. These will 9111 typically have values the same as the extended key usage extension 9112 and any application specific purposes. 9113 9114 The trust checking code now has a default behaviour: it will just 9115 check for an object with the same NID as the passed id. Functions can 9116 be provided to override either the default behaviour or the behaviour 9117 for a given id. SSL client, server and email already have functions 9118 in place for compatibility: they check the NID and also return "trusted" 9119 if the certificate is self signed. 9120 [Steve Henson] 9121 9122 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the 9123 traditional format into an EVP_PKEY structure. 9124 [Steve Henson] 9125 9126 *) Add a password callback function PEM_cb() which either prompts for 9127 a password if usr_data is NULL or otherwise assumes it is a null 9128 terminated password. Allow passwords to be passed on command line 9129 environment or config files in a few more utilities. 9130 [Steve Henson] 9131 9132 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private 9133 keys. Add some short names for PKCS#8 PBE algorithms and allow them 9134 to be specified on the command line for the pkcs8 and pkcs12 utilities. 9135 Update documentation. 9136 [Steve Henson] 9137 9138 *) Support for ASN1 "NULL" type. This could be handled before by using 9139 ASN1_TYPE but there wasn't any function that would try to read a NULL 9140 and produce an error if it couldn't. For compatibility we also have 9141 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and 9142 don't allocate anything because they don't need to. 9143 [Steve Henson] 9144 9145 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS 9146 for details. 9147 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>] 9148 9149 *) Rebuild of the memory allocation routines used by OpenSSL code and 9150 possibly others as well. The purpose is to make an interface that 9151 provide hooks so anyone can build a separate set of allocation and 9152 deallocation routines to be used by OpenSSL, for example memory 9153 pool implementations, or something else, which was previously hard 9154 since Malloc(), Realloc() and Free() were defined as macros having 9155 the values malloc, realloc and free, respectively (except for Win32 9156 compilations). The same is provided for memory debugging code. 9157 OpenSSL already comes with functionality to find memory leaks, but 9158 this gives people a chance to debug other memory problems. 9159 9160 With these changes, a new set of functions and macros have appeared: 9161 9162 CRYPTO_set_mem_debug_functions() [F] 9163 CRYPTO_get_mem_debug_functions() [F] 9164 CRYPTO_dbg_set_options() [F] 9165 CRYPTO_dbg_get_options() [F] 9166 CRYPTO_malloc_debug_init() [M] 9167 9168 The memory debug functions are NULL by default, unless the library 9169 is compiled with CRYPTO_MDEBUG or friends is defined. If someone 9170 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which 9171 gives the standard debugging functions that come with OpenSSL) or 9172 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions 9173 provided by the library user) must be used. When the standard 9174 debugging functions are used, CRYPTO_dbg_set_options can be used to 9175 request additional information: 9176 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting 9177 the CRYPTO_MDEBUG_xxx macro when compiling the library. 9178 9179 Also, things like CRYPTO_set_mem_functions will always give the 9180 expected result (the new set of functions is used for allocation 9181 and deallocation) at all times, regardless of platform and compiler 9182 options. 9183 9184 To finish it up, some functions that were never use in any other 9185 way than through macros have a new API and new semantic: 9186 9187 CRYPTO_dbg_malloc() 9188 CRYPTO_dbg_realloc() 9189 CRYPTO_dbg_free() 9190 9191 All macros of value have retained their old syntax. 9192 [Richard Levitte and Bodo Moeller] 9193 9194 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the 9195 ordering of SMIMECapabilities wasn't in "strength order" and there 9196 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature 9197 algorithm. 9198 [Steve Henson] 9199 9200 *) Some ASN1 types with illegal zero length encoding (INTEGER, 9201 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines. 9202 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson] 9203 9204 *) Merge in my S/MIME library for OpenSSL. This provides a simple 9205 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough 9206 functionality to handle multipart/signed properly) and a utility 9207 called 'smime' to call all this stuff. This is based on code I 9208 originally wrote for Celo who have kindly allowed it to be 9209 included in OpenSSL. 9210 [Steve Henson] 9211 9212 *) Add variants des_set_key_checked and des_set_key_unchecked of 9213 des_set_key (aka des_key_sched). Global variable des_check_key 9214 decides which of these is called by des_set_key; this way 9215 des_check_key behaves as it always did, but applications and 9216 the library itself, which was buggy for des_check_key == 1, 9217 have a cleaner way to pick the version they need. 9218 [Bodo Moeller] 9219 9220 *) New function PKCS12_newpass() which changes the password of a 9221 PKCS12 structure. 9222 [Steve Henson] 9223 9224 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and 9225 dynamic mix. In both cases the ids can be used as an index into the 9226 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add() 9227 functions so they accept a list of the field values and the 9228 application doesn't need to directly manipulate the X509_TRUST 9229 structure. 9230 [Steve Henson] 9231 9232 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't 9233 need initialising. 9234 [Steve Henson] 9235 9236 *) Modify the way the V3 extension code looks up extensions. This now 9237 works in a similar way to the object code: we have some "standard" 9238 extensions in a static table which is searched with OBJ_bsearch() 9239 and the application can add dynamic ones if needed. The file 9240 crypto/x509v3/ext_dat.h now has the info: this file needs to be 9241 updated whenever a new extension is added to the core code and kept 9242 in ext_nid order. There is a simple program 'tabtest.c' which checks 9243 this. New extensions are not added too often so this file can readily 9244 be maintained manually. 9245 9246 There are two big advantages in doing things this way. The extensions 9247 can be looked up immediately and no longer need to be "added" using 9248 X509V3_add_standard_extensions(): this function now does nothing. 9249 [Side note: I get *lots* of email saying the extension code doesn't 9250 work because people forget to call this function] 9251 Also no dynamic allocation is done unless new extensions are added: 9252 so if we don't add custom extensions there is no need to call 9253 X509V3_EXT_cleanup(). 9254 [Steve Henson] 9255 9256 *) Modify enc utility's salting as follows: make salting the default. Add a 9257 magic header, so unsalted files fail gracefully instead of just decrypting 9258 to garbage. This is because not salting is a big security hole, so people 9259 should be discouraged from doing it. 9260 [Ben Laurie] 9261 9262 *) Fixes and enhancements to the 'x509' utility. It allowed a message 9263 digest to be passed on the command line but it only used this 9264 parameter when signing a certificate. Modified so all relevant 9265 operations are affected by the digest parameter including the 9266 -fingerprint and -x509toreq options. Also -x509toreq choked if a 9267 DSA key was used because it didn't fix the digest. 9268 [Steve Henson] 9269 9270 *) Initial certificate chain verify code. Currently tests the untrusted 9271 certificates for consistency with the verify purpose (which is set 9272 when the X509_STORE_CTX structure is set up) and checks the pathlength. 9273 9274 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour: 9275 this is because it will reject chains with invalid extensions whereas 9276 every previous version of OpenSSL and SSLeay made no checks at all. 9277 9278 Trust code: checks the root CA for the relevant trust settings. Trust 9279 settings have an initial value consistent with the verify purpose: e.g. 9280 if the verify purpose is for SSL client use it expects the CA to be 9281 trusted for SSL client use. However the default value can be changed to 9282 permit custom trust settings: one example of this would be to only trust 9283 certificates from a specific "secure" set of CAs. 9284 9285 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions 9286 which should be used for version portability: especially since the 9287 verify structure is likely to change more often now. 9288 9289 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions 9290 to set them. If not set then assume SSL clients will verify SSL servers 9291 and vice versa. 9292 9293 Two new options to the verify program: -untrusted allows a set of 9294 untrusted certificates to be passed in and -purpose which sets the 9295 intended purpose of the certificate. If a purpose is set then the 9296 new chain verify code is used to check extension consistency. 9297 [Steve Henson] 9298 9299 *) Support for the authority information access extension. 9300 [Steve Henson] 9301 9302 *) Modify RSA and DSA PEM read routines to transparently handle 9303 PKCS#8 format private keys. New *_PUBKEY_* functions that handle 9304 public keys in a format compatible with certificate 9305 SubjectPublicKeyInfo structures. Unfortunately there were already 9306 functions called *_PublicKey_* which used various odd formats so 9307 these are retained for compatibility: however the DSA variants were 9308 never in a public release so they have been deleted. Changed dsa/rsa 9309 utilities to handle the new format: note no releases ever handled public 9310 keys so we should be OK. 9311 9312 The primary motivation for this change is to avoid the same fiasco 9313 that dogs private keys: there are several incompatible private key 9314 formats some of which are standard and some OpenSSL specific and 9315 require various evil hacks to allow partial transparent handling and 9316 even then it doesn't work with DER formats. Given the option anything 9317 other than PKCS#8 should be dumped: but the other formats have to 9318 stay in the name of compatibility. 9319 9320 With public keys and the benefit of hindsight one standard format 9321 is used which works with EVP_PKEY, RSA or DSA structures: though 9322 it clearly returns an error if you try to read the wrong kind of key. 9323 9324 Added a -pubkey option to the 'x509' utility to output the public key. 9325 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*() 9326 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add 9327 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*()) 9328 that do the same as the EVP_PKEY_assign_*() except they up the 9329 reference count of the added key (they don't "swallow" the 9330 supplied key). 9331 [Steve Henson] 9332 9333 *) Fixes to crypto/x509/by_file.c the code to read in certificates and 9334 CRLs would fail if the file contained no certificates or no CRLs: 9335 added a new function to read in both types and return the number 9336 read: this means that if none are read it will be an error. The 9337 DER versions of the certificate and CRL reader would always fail 9338 because it isn't possible to mix certificates and CRLs in DER format 9339 without choking one or the other routine. Changed this to just read 9340 a certificate: this is the best we can do. Also modified the code 9341 in apps/verify.c to take notice of return codes: it was previously 9342 attempting to read in certificates from NULL pointers and ignoring 9343 any errors: this is one reason why the cert and CRL reader seemed 9344 to work. It doesn't check return codes from the default certificate 9345 routines: these may well fail if the certificates aren't installed. 9346 [Steve Henson] 9347 9348 *) Code to support otherName option in GeneralName. 9349 [Steve Henson] 9350 9351 *) First update to verify code. Change the verify utility 9352 so it warns if it is passed a self signed certificate: 9353 for consistency with the normal behaviour. X509_verify 9354 has been modified to it will now verify a self signed 9355 certificate if *exactly* the same certificate appears 9356 in the store: it was previously impossible to trust a 9357 single self signed certificate. This means that: 9358 openssl verify ss.pem 9359 now gives a warning about a self signed certificate but 9360 openssl verify -CAfile ss.pem ss.pem 9361 is OK. 9362 [Steve Henson] 9363 9364 *) For servers, store verify_result in SSL_SESSION data structure 9365 (and add it to external session representation). 9366 This is needed when client certificate verifications fails, 9367 but an application-provided verification callback (set by 9368 SSL_CTX_set_cert_verify_callback) allows accepting the session 9369 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK 9370 but returns 1): When the session is reused, we have to set 9371 ssl->verify_result to the appropriate error code to avoid 9372 security holes. 9373 [Bodo Moeller, problem pointed out by Lutz Jaenicke] 9374 9375 *) Fix a bug in the new PKCS#7 code: it didn't consider the 9376 case in PKCS7_dataInit() where the signed PKCS7 structure 9377 didn't contain any existing data because it was being created. 9378 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson] 9379 9380 *) Add a salt to the key derivation routines in enc.c. This 9381 forms the first 8 bytes of the encrypted file. Also add a 9382 -S option to allow a salt to be input on the command line. 9383 [Steve Henson] 9384 9385 *) New function X509_cmp(). Oddly enough there wasn't a function 9386 to compare two certificates. We do this by working out the SHA1 9387 hash and comparing that. X509_cmp() will be needed by the trust 9388 code. 9389 [Steve Henson] 9390 9391 *) SSL_get1_session() is like SSL_get_session(), but increments 9392 the reference count in the SSL_SESSION returned. 9393 [Geoff Thorpe <geoff@eu.c2.net>] 9394 9395 *) Fix for 'req': it was adding a null to request attributes. 9396 Also change the X509_LOOKUP and X509_INFO code to handle 9397 certificate auxiliary information. 9398 [Steve Henson] 9399 9400 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document 9401 the 'enc' command. 9402 [Steve Henson] 9403 9404 *) Add the possibility to add extra information to the memory leak 9405 detecting output, to form tracebacks, showing from where each 9406 allocation was originated: CRYPTO_push_info("constant string") adds 9407 the string plus current file name and line number to a per-thread 9408 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info() 9409 is like calling CYRPTO_pop_info() until the stack is empty. 9410 Also updated memory leak detection code to be multi-thread-safe. 9411 [Richard Levitte] 9412 9413 *) Add options -text and -noout to pkcs7 utility and delete the 9414 encryption options which never did anything. Update docs. 9415 [Steve Henson] 9416 9417 *) Add options to some of the utilities to allow the pass phrase 9418 to be included on either the command line (not recommended on 9419 OSes like Unix) or read from the environment. Update the 9420 manpages and fix a few bugs. 9421 [Steve Henson] 9422 9423 *) Add a few manpages for some of the openssl commands. 9424 [Steve Henson] 9425 9426 *) Fix the -revoke option in ca. It was freeing up memory twice, 9427 leaking and not finding already revoked certificates. 9428 [Steve Henson] 9429 9430 *) Extensive changes to support certificate auxiliary information. 9431 This involves the use of X509_CERT_AUX structure and X509_AUX 9432 functions. An X509_AUX function such as PEM_read_X509_AUX() 9433 can still read in a certificate file in the usual way but it 9434 will also read in any additional "auxiliary information". By 9435 doing things this way a fair degree of compatibility can be 9436 retained: existing certificates can have this information added 9437 using the new 'x509' options. 9438 9439 Current auxiliary information includes an "alias" and some trust 9440 settings. The trust settings will ultimately be used in enhanced 9441 certificate chain verification routines: currently a certificate 9442 can only be trusted if it is self signed and then it is trusted 9443 for all purposes. 9444 [Steve Henson] 9445 9446 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD). 9447 The problem was that one of the replacement routines had not been working 9448 since SSLeay releases. For now the offending routine has been replaced 9449 with non-optimised assembler. Even so, this now gives around 95% 9450 performance improvement for 1024 bit RSA signs. 9451 [Mark Cox] 9452 9453 *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2 9454 handling. Most clients have the effective key size in bits equal to 9455 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key. 9456 A few however don't do this and instead use the size of the decrypted key 9457 to determine the RC2 key length and the AlgorithmIdentifier to determine 9458 the effective key length. In this case the effective key length can still 9459 be 40 bits but the key length can be 168 bits for example. This is fixed 9460 by manually forcing an RC2 key into the EVP_PKEY structure because the 9461 EVP code can't currently handle unusual RC2 key sizes: it always assumes 9462 the key length and effective key length are equal. 9463 [Steve Henson] 9464 9465 *) Add a bunch of functions that should simplify the creation of 9466 X509_NAME structures. Now you should be able to do: 9467 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0); 9468 and have it automatically work out the correct field type and fill in 9469 the structures. The more adventurous can try: 9470 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0); 9471 and it will (hopefully) work out the correct multibyte encoding. 9472 [Steve Henson] 9473 9474 *) Change the 'req' utility to use the new field handling and multibyte 9475 copy routines. Before the DN field creation was handled in an ad hoc 9476 way in req, ca, and x509 which was rather broken and didn't support 9477 BMPStrings or UTF8Strings. Since some software doesn't implement 9478 BMPStrings or UTF8Strings yet, they can be enabled using the config file 9479 using the dirstring_type option. See the new comment in the default 9480 openssl.cnf for more info. 9481 [Steve Henson] 9482 9483 *) Make crypto/rand/md_rand.c more robust: 9484 - Assure unique random numbers after fork(). 9485 - Make sure that concurrent threads access the global counter and 9486 md serializably so that we never lose entropy in them 9487 or use exactly the same state in multiple threads. 9488 Access to the large state is not always serializable because 9489 the additional locking could be a performance killer, and 9490 md should be large enough anyway. 9491 [Bodo Moeller] 9492 9493 *) New file apps/app_rand.c with commonly needed functionality 9494 for handling the random seed file. 9495 9496 Use the random seed file in some applications that previously did not: 9497 ca, 9498 dsaparam -genkey (which also ignored its '-rand' option), 9499 s_client, 9500 s_server, 9501 x509 (when signing). 9502 Except on systems with /dev/urandom, it is crucial to have a random 9503 seed file at least for key creation, DSA signing, and for DH exchanges; 9504 for RSA signatures we could do without one. 9505 9506 gendh and gendsa (unlike genrsa) used to read only the first byte 9507 of each file listed in the '-rand' option. The function as previously 9508 found in genrsa is now in app_rand.c and is used by all programs 9509 that support '-rand'. 9510 [Bodo Moeller] 9511 9512 *) In RAND_write_file, use mode 0600 for creating files; 9513 don't just chmod when it may be too late. 9514 [Bodo Moeller] 9515 9516 *) Report an error from X509_STORE_load_locations 9517 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed. 9518 [Bill Perry] 9519 9520 *) New function ASN1_mbstring_copy() this copies a string in either 9521 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format 9522 into an ASN1_STRING type. A mask of permissible types is passed 9523 and it chooses the "minimal" type to use or an error if not type 9524 is suitable. 9525 [Steve Henson] 9526 9527 *) Add function equivalents to the various macros in asn1.h. The old 9528 macros are retained with an M_ prefix. Code inside the library can 9529 use the M_ macros. External code (including the openssl utility) 9530 should *NOT* in order to be "shared library friendly". 9531 [Steve Henson] 9532 9533 *) Add various functions that can check a certificate's extensions 9534 to see if it usable for various purposes such as SSL client, 9535 server or S/MIME and CAs of these types. This is currently 9536 VERY EXPERIMENTAL but will ultimately be used for certificate chain 9537 verification. Also added a -purpose flag to x509 utility to 9538 print out all the purposes. 9539 [Steve Henson] 9540 9541 *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated 9542 functions. 9543 [Steve Henson] 9544 9545 *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search 9546 for, obtain and decode and extension and obtain its critical flag. 9547 This allows all the necessary extension code to be handled in a 9548 single function call. 9549 [Steve Henson] 9550 9551 *) RC4 tune-up featuring 30-40% performance improvement on most RISC 9552 platforms. See crypto/rc4/rc4_enc.c for further details. 9553 [Andy Polyakov] 9554 9555 *) New -noout option to asn1parse. This causes no output to be produced 9556 its main use is when combined with -strparse and -out to extract data 9557 from a file (which may not be in ASN.1 format). 9558 [Steve Henson] 9559 9560 *) Fix for pkcs12 program. It was hashing an invalid certificate pointer 9561 when producing the local key id. 9562 [Richard Levitte <levitte@stacken.kth.se>] 9563 9564 *) New option -dhparam in s_server. This allows a DH parameter file to be 9565 stated explicitly. If it is not stated then it tries the first server 9566 certificate file. The previous behaviour hard coded the filename 9567 "server.pem". 9568 [Steve Henson] 9569 9570 *) Add -pubin and -pubout options to the rsa and dsa commands. These allow 9571 a public key to be input or output. For example: 9572 openssl rsa -in key.pem -pubout -out pubkey.pem 9573 Also added necessary DSA public key functions to handle this. 9574 [Steve Henson] 9575 9576 *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained 9577 in the message. This was handled by allowing 9578 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it. 9579 [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>] 9580 9581 *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null 9582 to the end of the strings whereas this didn't. This would cause problems 9583 if strings read with d2i_ASN1_bytes() were later modified. 9584 [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>] 9585 9586 *) Fix for base64 decode bug. When a base64 bio reads only one line of 9587 data and it contains EOF it will end up returning an error. This is 9588 caused by input 46 bytes long. The cause is due to the way base64 9589 BIOs find the start of base64 encoded data. They do this by trying a 9590 trial decode on each line until they find one that works. When they 9591 do a flag is set and it starts again knowing it can pass all the 9592 data directly through the decoder. Unfortunately it doesn't reset 9593 the context it uses. This means that if EOF is reached an attempt 9594 is made to pass two EOFs through the context and this causes the 9595 resulting error. This can also cause other problems as well. As is 9596 usual with these problems it takes *ages* to find and the fix is 9597 trivial: move one line. 9598 [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ] 9599 9600 *) Ugly workaround to get s_client and s_server working under Windows. The 9601 old code wouldn't work because it needed to select() on sockets and the 9602 tty (for keypresses and to see if data could be written). Win32 only 9603 supports select() on sockets so we select() with a 1s timeout on the 9604 sockets and then see if any characters are waiting to be read, if none 9605 are present then we retry, we also assume we can always write data to 9606 the tty. This isn't nice because the code then blocks until we've 9607 received a complete line of data and it is effectively polling the 9608 keyboard at 1s intervals: however it's quite a bit better than not 9609 working at all :-) A dedicated Windows application might handle this 9610 with an event loop for example. 9611 [Steve Henson] 9612 9613 *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign 9614 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions 9615 will be called when RSA_sign() and RSA_verify() are used. This is useful 9616 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available. 9617 For this to work properly RSA_public_decrypt() and RSA_private_encrypt() 9618 should *not* be used: RSA_sign() and RSA_verify() must be used instead. 9619 This necessitated the support of an extra signature type NID_md5_sha1 9620 for SSL signatures and modifications to the SSL library to use it instead 9621 of calling RSA_public_decrypt() and RSA_private_encrypt(). 9622 [Steve Henson] 9623 9624 *) Add new -verify -CAfile and -CApath options to the crl program, these 9625 will lookup a CRL issuers certificate and verify the signature in a 9626 similar way to the verify program. Tidy up the crl program so it 9627 no longer accesses structures directly. Make the ASN1 CRL parsing a bit 9628 less strict. It will now permit CRL extensions even if it is not 9629 a V2 CRL: this will allow it to tolerate some broken CRLs. 9630 [Steve Henson] 9631 9632 *) Initialize all non-automatic variables each time one of the openssl 9633 sub-programs is started (this is necessary as they may be started 9634 multiple times from the "OpenSSL>" prompt). 9635 [Lennart Bang, Bodo Moeller] 9636 9637 *) Preliminary compilation option RSA_NULL which disables RSA crypto without 9638 removing all other RSA functionality (this is what NO_RSA does). This 9639 is so (for example) those in the US can disable those operations covered 9640 by the RSA patent while allowing storage and parsing of RSA keys and RSA 9641 key generation. 9642 [Steve Henson] 9643 9644 *) Non-copying interface to BIO pairs. 9645 (still largely untested) 9646 [Bodo Moeller] 9647 9648 *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive 9649 ASCII string. This was handled independently in various places before. 9650 [Steve Henson] 9651 9652 *) New functions UTF8_getc() and UTF8_putc() that parse and generate 9653 UTF8 strings a character at a time. 9654 [Steve Henson] 9655 9656 *) Use client_version from client hello to select the protocol 9657 (s23_srvr.c) and for RSA client key exchange verification 9658 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications. 9659 [Bodo Moeller] 9660 9661 *) Add various utility functions to handle SPKACs, these were previously 9662 handled by poking round in the structure internals. Added new function 9663 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to 9664 print, verify and generate SPKACs. Based on an original idea from 9665 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified. 9666 [Steve Henson] 9667 9668 *) RIPEMD160 is operational on all platforms and is back in 'make test'. 9669 [Andy Polyakov] 9670 9671 *) Allow the config file extension section to be overwritten on the 9672 command line. Based on an original idea from Massimiliano Pala 9673 <madwolf@comune.modena.it>. The new option is called -extensions 9674 and can be applied to ca, req and x509. Also -reqexts to override 9675 the request extensions in req and -crlexts to override the crl extensions 9676 in ca. 9677 [Steve Henson] 9678 9679 *) Add new feature to the SPKAC handling in ca. Now you can include 9680 the same field multiple times by preceding it by "XXXX." for example: 9681 1.OU="Unit name 1" 9682 2.OU="Unit name 2" 9683 this is the same syntax as used in the req config file. 9684 [Steve Henson] 9685 9686 *) Allow certificate extensions to be added to certificate requests. These 9687 are specified in a 'req_extensions' option of the req section of the 9688 config file. They can be printed out with the -text option to req but 9689 are otherwise ignored at present. 9690 [Steve Henson] 9691 9692 *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first 9693 data read consists of only the final block it would not decrypted because 9694 EVP_CipherUpdate() would correctly report zero bytes had been decrypted. 9695 A misplaced 'break' also meant the decrypted final block might not be 9696 copied until the next read. 9697 [Steve Henson] 9698 9699 *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added 9700 a few extra parameters to the DH structure: these will be useful if 9701 for example we want the value of 'q' or implement X9.42 DH. 9702 [Steve Henson] 9703 9704 *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and 9705 provides hooks that allow the default DSA functions or functions on a 9706 "per key" basis to be replaced. This allows hardware acceleration and 9707 hardware key storage to be handled without major modification to the 9708 library. Also added low level modexp hooks and CRYPTO_EX structure and 9709 associated functions. 9710 [Steve Henson] 9711 9712 *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO 9713 as "read only": it can't be written to and the buffer it points to will 9714 not be freed. Reading from a read only BIO is much more efficient than 9715 a normal memory BIO. This was added because there are several times when 9716 an area of memory needs to be read from a BIO. The previous method was 9717 to create a memory BIO and write the data to it, this results in two 9718 copies of the data and an O(n^2) reading algorithm. There is a new 9719 function BIO_new_mem_buf() which creates a read only memory BIO from 9720 an area of memory. Also modified the PKCS#7 routines to use read only 9721 memory BIOs. 9722 [Steve Henson] 9723 9724 *) Bugfix: ssl23_get_client_hello did not work properly when called in 9725 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of 9726 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read, 9727 but a retry condition occured while trying to read the rest. 9728 [Bodo Moeller] 9729 9730 *) The PKCS7_ENC_CONTENT_new() function was setting the content type as 9731 NID_pkcs7_encrypted by default: this was wrong since this should almost 9732 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle 9733 the encrypted data type: this is a more sensible place to put it and it 9734 allows the PKCS#12 code to be tidied up that duplicated this 9735 functionality. 9736 [Steve Henson] 9737 9738 *) Changed obj_dat.pl script so it takes its input and output files on 9739 the command line. This should avoid shell escape redirection problems 9740 under Win32. 9741 [Steve Henson] 9742 9743 *) Initial support for certificate extension requests, these are included 9744 in things like Xenroll certificate requests. Included functions to allow 9745 extensions to be obtained and added. 9746 [Steve Henson] 9747 9748 *) -crlf option to s_client and s_server for sending newlines as 9749 CRLF (as required by many protocols). 9750 [Bodo Moeller] 9751 9752 Changes between 0.9.3a and 0.9.4 [09 Aug 1999] 9753 9754 *) Install libRSAglue.a when OpenSSL is built with RSAref. 9755 [Ralf S. Engelschall] 9756 9757 *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency. 9758 [Andrija Antonijevic <TheAntony2@bigfoot.com>] 9759 9760 *) Fix -startdate and -enddate (which was missing) arguments to 'ca' 9761 program. 9762 [Steve Henson] 9763 9764 *) New function DSA_dup_DH, which duplicates DSA parameters/keys as 9765 DH parameters/keys (q is lost during that conversion, but the resulting 9766 DH parameters contain its length). 9767 9768 For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is 9769 much faster than DH_generate_parameters (which creates parameters 9770 where p = 2*q + 1), and also the smaller q makes DH computations 9771 much more efficient (160-bit exponentiation instead of 1024-bit 9772 exponentiation); so this provides a convenient way to support DHE 9773 ciphersuites in SSL/TLS servers (see ssl/ssltest.c). It is of 9774 utter importance to use 9775 SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 9776 or 9777 SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 9778 when such DH parameters are used, because otherwise small subgroup 9779 attacks may become possible! 9780 [Bodo Moeller] 9781 9782 *) Avoid memory leak in i2d_DHparams. 9783 [Bodo Moeller] 9784 9785 *) Allow the -k option to be used more than once in the enc program: 9786 this allows the same encrypted message to be read by multiple recipients. 9787 [Steve Henson] 9788 9789 *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts 9790 an ASN1_OBJECT to a text string. If the "no_name" parameter is set then 9791 it will always use the numerical form of the OID, even if it has a short 9792 or long name. 9793 [Steve Henson] 9794 9795 *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp 9796 method only got called if p,q,dmp1,dmq1,iqmp components were present, 9797 otherwise bn_mod_exp was called. In the case of hardware keys for example 9798 no private key components need be present and it might store extra data 9799 in the RSA structure, which cannot be accessed from bn_mod_exp. 9800 By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for 9801 private key operations. 9802 [Steve Henson] 9803 9804 *) Added support for SPARC Linux. 9805 [Andy Polyakov] 9806 9807 *) pem_password_cb function type incompatibly changed from 9808 typedef int pem_password_cb(char *buf, int size, int rwflag); 9809 to 9810 ....(char *buf, int size, int rwflag, void *userdata); 9811 so that applications can pass data to their callbacks: 9812 The PEM[_ASN1]_{read,write}... functions and macros now take an 9813 additional void * argument, which is just handed through whenever 9814 the password callback is called. 9815 [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller] 9816 9817 New function SSL_CTX_set_default_passwd_cb_userdata. 9818 9819 Compatibility note: As many C implementations push function arguments 9820 onto the stack in reverse order, the new library version is likely to 9821 interoperate with programs that have been compiled with the old 9822 pem_password_cb definition (PEM_whatever takes some data that 9823 happens to be on the stack as its last argument, and the callback 9824 just ignores this garbage); but there is no guarantee whatsoever that 9825 this will work. 9826 9827 *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=... 9828 (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused 9829 problems not only on Windows, but also on some Unix platforms. 9830 To avoid problematic command lines, these definitions are now in an 9831 auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl 9832 for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds). 9833 [Bodo Moeller] 9834 9835 *) MIPS III/IV assembler module is reimplemented. 9836 [Andy Polyakov] 9837 9838 *) More DES library cleanups: remove references to srand/rand and 9839 delete an unused file. 9840 [Ulf M��ller] 9841 9842 *) Add support for the the free Netwide assembler (NASM) under Win32, 9843 since not many people have MASM (ml) and it can be hard to obtain. 9844 This is currently experimental but it seems to work OK and pass all 9845 the tests. Check out INSTALL.W32 for info. 9846 [Steve Henson] 9847 9848 *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections 9849 without temporary keys kept an extra copy of the server key, 9850 and connections with temporary keys did not free everything in case 9851 of an error. 9852 [Bodo Moeller] 9853 9854 *) New function RSA_check_key and new openssl rsa option -check 9855 for verifying the consistency of RSA keys. 9856 [Ulf Moeller, Bodo Moeller] 9857 9858 *) Various changes to make Win32 compile work: 9859 1. Casts to avoid "loss of data" warnings in p5_crpt2.c 9860 2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned 9861 comparison" warnings. 9862 3. Add sk_<TYPE>_sort to DEF file generator and do make update. 9863 [Steve Henson] 9864 9865 *) Add a debugging option to PKCS#5 v2 key generation function: when 9866 you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and 9867 derived keys are printed to stderr. 9868 [Steve Henson] 9869 9870 *) Copy the flags in ASN1_STRING_dup(). 9871 [Roman E. Pavlov <pre@mo.msk.ru>] 9872 9873 *) The x509 application mishandled signing requests containing DSA 9874 keys when the signing key was also DSA and the parameters didn't match. 9875 9876 It was supposed to omit the parameters when they matched the signing key: 9877 the verifying software was then supposed to automatically use the CA's 9878 parameters if they were absent from the end user certificate. 9879 9880 Omitting parameters is no longer recommended. The test was also 9881 the wrong way round! This was probably due to unusual behaviour in 9882 EVP_cmp_parameters() which returns 1 if the parameters match. 9883 This meant that parameters were omitted when they *didn't* match and 9884 the certificate was useless. Certificates signed with 'ca' didn't have 9885 this bug. 9886 [Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>] 9887 9888 *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems. 9889 The interface is as follows: 9890 Applications can use 9891 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(), 9892 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop(); 9893 "off" is now the default. 9894 The library internally uses 9895 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(), 9896 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on() 9897 to disable memory-checking temporarily. 9898 9899 Some inconsistent states that previously were possible (and were 9900 even the default) are now avoided. 9901 9902 -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time 9903 with each memory chunk allocated; this is occasionally more helpful 9904 than just having a counter. 9905 9906 -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID. 9907 9908 -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future 9909 extensions. 9910 [Bodo Moeller] 9911 9912 *) Introduce "mode" for SSL structures (with defaults in SSL_CTX), 9913 which largely parallels "options", but is for changing API behaviour, 9914 whereas "options" are about protocol behaviour. 9915 Initial "mode" flags are: 9916 9917 SSL_MODE_ENABLE_PARTIAL_WRITE Allow SSL_write to report success when 9918 a single record has been written. 9919 SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER Don't insist that SSL_write 9920 retries use the same buffer location. 9921 (But all of the contents must be 9922 copied!) 9923 [Bodo Moeller] 9924 9925 *) Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options 9926 worked. 9927 9928 *) Fix problems with no-hmac etc. 9929 [Ulf M��ller, pointed out by Brian Wellington <bwelling@tislabs.com>] 9930 9931 *) New functions RSA_get_default_method(), RSA_set_method() and 9932 RSA_get_method(). These allows replacement of RSA_METHODs without having 9933 to mess around with the internals of an RSA structure. 9934 [Steve Henson] 9935 9936 *) Fix memory leaks in DSA_do_sign and DSA_is_prime. 9937 Also really enable memory leak checks in openssl.c and in some 9938 test programs. 9939 [Chad C. Mulligan, Bodo Moeller] 9940 9941 *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess 9942 up the length of negative integers. This has now been simplified to just 9943 store the length when it is first determined and use it later, rather 9944 than trying to keep track of where data is copied and updating it to 9945 point to the end. 9946 [Steve Henson, reported by Brien Wheeler 9947 <bwheeler@authentica-security.com>] 9948 9949 *) Add a new function PKCS7_signatureVerify. This allows the verification 9950 of a PKCS#7 signature but with the signing certificate passed to the 9951 function itself. This contrasts with PKCS7_dataVerify which assumes the 9952 certificate is present in the PKCS#7 structure. This isn't always the 9953 case: certificates can be omitted from a PKCS#7 structure and be 9954 distributed by "out of band" means (such as a certificate database). 9955 [Steve Henson] 9956 9957 *) Complete the PEM_* macros with DECLARE_PEM versions to replace the 9958 function prototypes in pem.h, also change util/mkdef.pl to add the 9959 necessary function names. 9960 [Steve Henson] 9961 9962 *) mk1mf.pl (used by Windows builds) did not properly read the 9963 options set by Configure in the top level Makefile, and Configure 9964 was not even able to write more than one option correctly. 9965 Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended. 9966 [Bodo Moeller] 9967 9968 *) New functions CONF_load_bio() and CONF_load_fp() to allow a config 9969 file to be loaded from a BIO or FILE pointer. The BIO version will 9970 for example allow memory BIOs to contain config info. 9971 [Steve Henson] 9972 9973 *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS. 9974 Whoever hopes to achieve shared-library compatibility across versions 9975 must use this, not the compile-time macro. 9976 (Exercise 0.9.4: Which is the minimum library version required by 9977 such programs?) 9978 Note: All this applies only to multi-threaded programs, others don't 9979 need locks. 9980 [Bodo Moeller] 9981 9982 *) Add missing case to s3_clnt.c state machine -- one of the new SSL tests 9983 through a BIO pair triggered the default case, i.e. 9984 SSLerr(...,SSL_R_UNKNOWN_STATE). 9985 [Bodo Moeller] 9986 9987 *) New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications 9988 can use the SSL library even if none of the specific BIOs is 9989 appropriate. 9990 [Bodo Moeller] 9991 9992 *) Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value 9993 for the encoded length. 9994 [Jeon KyoungHo <khjeon@sds.samsung.co.kr>] 9995 9996 *) Add initial documentation of the X509V3 functions. 9997 [Steve Henson] 9998 9999 *) Add a new pair of functions PEM_write_PKCS8PrivateKey() and 10000 PEM_write_bio_PKCS8PrivateKey() that are equivalent to 10001 PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more 10002 secure PKCS#8 private key format with a high iteration count. 10003 [Steve Henson] 10004 10005 *) Fix determination of Perl interpreter: A perl or perl5 10006 _directory_ in $PATH was also accepted as the interpreter. 10007 [Ralf S. Engelschall] 10008 10009 *) Fix demos/sign/sign.c: well there wasn't anything strictly speaking 10010 wrong with it but it was very old and did things like calling 10011 PEM_ASN1_read() directly and used MD5 for the hash not to mention some 10012 unusual formatting. 10013 [Steve Henson] 10014 10015 *) Fix demos/selfsign.c: it used obsolete and deleted functions, changed 10016 to use the new extension code. 10017 [Steve Henson] 10018 10019 *) Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c 10020 with macros. This should make it easier to change their form, add extra 10021 arguments etc. Fix a few PEM prototypes which didn't have cipher as a 10022 constant. 10023 [Steve Henson] 10024 10025 *) Add to configuration table a new entry that can specify an alternative 10026 name for unistd.h (for pre-POSIX systems); we need this for NeXTstep, 10027 according to Mark Crispin <MRC@Panda.COM>. 10028 [Bodo Moeller] 10029 10030#if 0 10031 *) DES CBC did not update the IV. Weird. 10032 [Ben Laurie] 10033#else 10034 des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does. 10035 Changing the behaviour of the former might break existing programs -- 10036 where IV updating is needed, des_ncbc_encrypt can be used. 10037#endif 10038 10039 *) When bntest is run from "make test" it drives bc to check its 10040 calculations, as well as internally checking them. If an internal check 10041 fails, it needs to cause bc to give a non-zero result or make test carries 10042 on without noticing the failure. Fixed. 10043 [Ben Laurie] 10044 10045 *) DES library cleanups. 10046 [Ulf M��ller] 10047 10048 *) Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be 10049 used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit 10050 ciphers. NOTE: although the key derivation function has been verified 10051 against some published test vectors it has not been extensively tested 10052 yet. Added a -v2 "cipher" option to pkcs8 application to allow the use 10053 of v2.0. 10054 [Steve Henson] 10055 10056 *) Instead of "mkdir -p", which is not fully portable, use new 10057 Perl script "util/mkdir-p.pl". 10058 [Bodo Moeller] 10059 10060 *) Rewrite the way password based encryption (PBE) is handled. It used to 10061 assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter 10062 structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms 10063 but doesn't apply to PKCS#5 v2.0 where it can be something else. Now 10064 the 'parameter' field of the AlgorithmIdentifier is passed to the 10065 underlying key generation function so it must do its own ASN1 parsing. 10066 This has also changed the EVP_PBE_CipherInit() function which now has a 10067 'parameter' argument instead of literal salt and iteration count values 10068 and the function EVP_PBE_ALGOR_CipherInit() has been deleted. 10069 [Steve Henson] 10070 10071 *) Support for PKCS#5 v1.5 compatible password based encryption algorithms 10072 and PKCS#8 functionality. New 'pkcs8' application linked to openssl. 10073 Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE 10074 KEY" because this clashed with PKCS#8 unencrypted string. Since this 10075 value was just used as a "magic string" and not used directly its 10076 value doesn't matter. 10077 [Steve Henson] 10078 10079 *) Introduce some semblance of const correctness to BN. Shame C doesn't 10080 support mutable. 10081 [Ben Laurie] 10082 10083 *) "linux-sparc64" configuration (ultrapenguin). 10084 [Ray Miller <ray.miller@oucs.ox.ac.uk>] 10085 "linux-sparc" configuration. 10086 [Christian Forster <fo@hawo.stw.uni-erlangen.de>] 10087 10088 *) config now generates no-xxx options for missing ciphers. 10089 [Ulf M��ller] 10090 10091 *) Support the EBCDIC character set (work in progress). 10092 File ebcdic.c not yet included because it has a different license. 10093 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 10094 10095 *) Support BS2000/OSD-POSIX. 10096 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 10097 10098 *) Make callbacks for key generation use void * instead of char *. 10099 [Ben Laurie] 10100 10101 *) Make S/MIME samples compile (not yet tested). 10102 [Ben Laurie] 10103 10104 *) Additional typesafe stacks. 10105 [Ben Laurie] 10106 10107 *) New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x). 10108 [Bodo Moeller] 10109 10110 10111 Changes between 0.9.3 and 0.9.3a [29 May 1999] 10112 10113 *) New configuration variant "sco5-gcc". 10114 10115 *) Updated some demos. 10116 [Sean O Riordain, Wade Scholine] 10117 10118 *) Add missing BIO_free at exit of pkcs12 application. 10119 [Wu Zhigang] 10120 10121 *) Fix memory leak in conf.c. 10122 [Steve Henson] 10123 10124 *) Updates for Win32 to assembler version of MD5. 10125 [Steve Henson] 10126 10127 *) Set #! path to perl in apps/der_chop to where we found it 10128 instead of using a fixed path. 10129 [Bodo Moeller] 10130 10131 *) SHA library changes for irix64-mips4-cc. 10132 [Andy Polyakov] 10133 10134 *) Improvements for VMS support. 10135 [Richard Levitte] 10136 10137 10138 Changes between 0.9.2b and 0.9.3 [24 May 1999] 10139 10140 *) Bignum library bug fix. IRIX 6 passes "make test" now! 10141 This also avoids the problems with SC4.2 and unpatched SC5. 10142 [Andy Polyakov <appro@fy.chalmers.se>] 10143 10144 *) New functions sk_num, sk_value and sk_set to replace the previous macros. 10145 These are required because of the typesafe stack would otherwise break 10146 existing code. If old code used a structure member which used to be STACK 10147 and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with 10148 sk_num or sk_value it would produce an error because the num, data members 10149 are not present in STACK_OF. Now it just produces a warning. sk_set 10150 replaces the old method of assigning a value to sk_value 10151 (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code 10152 that does this will no longer work (and should use sk_set instead) but 10153 this could be regarded as a "questionable" behaviour anyway. 10154 [Steve Henson] 10155 10156 *) Fix most of the other PKCS#7 bugs. The "experimental" code can now 10157 correctly handle encrypted S/MIME data. 10158 [Steve Henson] 10159 10160 *) Change type of various DES function arguments from des_cblock 10161 (which means, in function argument declarations, pointer to char) 10162 to des_cblock * (meaning pointer to array with 8 char elements), 10163 which allows the compiler to do more typechecking; it was like 10164 that back in SSLeay, but with lots of ugly casts. 10165 10166 Introduce new type const_des_cblock. 10167 [Bodo Moeller] 10168 10169 *) Reorganise the PKCS#7 library and get rid of some of the more obvious 10170 problems: find RecipientInfo structure that matches recipient certificate 10171 and initialise the ASN1 structures properly based on passed cipher. 10172 [Steve Henson] 10173 10174 *) Belatedly make the BN tests actually check the results. 10175 [Ben Laurie] 10176 10177 *) Fix the encoding and decoding of negative ASN1 INTEGERS and conversion 10178 to and from BNs: it was completely broken. New compilation option 10179 NEG_PUBKEY_BUG to allow for some broken certificates that encode public 10180 key elements as negative integers. 10181 [Steve Henson] 10182 10183 *) Reorganize and speed up MD5. 10184 [Andy Polyakov <appro@fy.chalmers.se>] 10185 10186 *) VMS support. 10187 [Richard Levitte <richard@levitte.org>] 10188 10189 *) New option -out to asn1parse to allow the parsed structure to be 10190 output to a file. This is most useful when combined with the -strparse 10191 option to examine the output of things like OCTET STRINGS. 10192 [Steve Henson] 10193 10194 *) Make SSL library a little more fool-proof by not requiring any longer 10195 that SSL_set_{accept,connect}_state be called before 10196 SSL_{accept,connect} may be used (SSL_set_..._state is omitted 10197 in many applications because usually everything *appeared* to work as 10198 intended anyway -- now it really works as intended). 10199 [Bodo Moeller] 10200 10201 *) Move openssl.cnf out of lib/. 10202 [Ulf M��ller] 10203 10204 *) Fix various things to let OpenSSL even pass ``egcc -pipe -O2 -Wall 10205 -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes 10206 -Wmissing-declarations -Wnested-externs -Winline'' with EGCS 1.1.2+ 10207 [Ralf S. Engelschall] 10208 10209 *) Various fixes to the EVP and PKCS#7 code. It may now be able to 10210 handle PKCS#7 enveloped data properly. 10211 [Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve] 10212 10213 *) Create a duplicate of the SSL_CTX's CERT in SSL_new instead of 10214 copying pointers. The cert_st handling is changed by this in 10215 various ways (and thus what used to be known as ctx->default_cert 10216 is now called ctx->cert, since we don't resort to s->ctx->[default_]cert 10217 any longer when s->cert does not give us what we need). 10218 ssl_cert_instantiate becomes obsolete by this change. 10219 As soon as we've got the new code right (possibly it already is?), 10220 we have solved a couple of bugs of the earlier code where s->cert 10221 was used as if it could not have been shared with other SSL structures. 10222 10223 Note that using the SSL API in certain dirty ways now will result 10224 in different behaviour than observed with earlier library versions: 10225 Changing settings for an SSL_CTX *ctx after having done s = SSL_new(ctx) 10226 does not influence s as it used to. 10227 10228 In order to clean up things more thoroughly, inside SSL_SESSION 10229 we don't use CERT any longer, but a new structure SESS_CERT 10230 that holds per-session data (if available); currently, this is 10231 the peer's certificate chain and, for clients, the server's certificate 10232 and temporary key. CERT holds only those values that can have 10233 meaningful defaults in an SSL_CTX. 10234 [Bodo Moeller] 10235 10236 *) New function X509V3_EXT_i2d() to create an X509_EXTENSION structure 10237 from the internal representation. Various PKCS#7 fixes: remove some 10238 evil casts and set the enc_dig_alg field properly based on the signing 10239 key type. 10240 [Steve Henson] 10241 10242 *) Allow PKCS#12 password to be set from the command line or the 10243 environment. Let 'ca' get its config file name from the environment 10244 variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req' 10245 and 'x509'). 10246 [Steve Henson] 10247 10248 *) Allow certificate policies extension to use an IA5STRING for the 10249 organization field. This is contrary to the PKIX definition but 10250 VeriSign uses it and IE5 only recognises this form. Document 'x509' 10251 extension option. 10252 [Steve Henson] 10253 10254 *) Add PEDANTIC compiler flag to allow compilation with gcc -pedantic, 10255 without disallowing inline assembler and the like for non-pedantic builds. 10256 [Ben Laurie] 10257 10258 *) Support Borland C++ builder. 10259 [Janez Jere <jj@void.si>, modified by Ulf M��ller] 10260 10261 *) Support Mingw32. 10262 [Ulf M��ller] 10263 10264 *) SHA-1 cleanups and performance enhancements. 10265 [Andy Polyakov <appro@fy.chalmers.se>] 10266 10267 *) Sparc v8plus assembler for the bignum library. 10268 [Andy Polyakov <appro@fy.chalmers.se>] 10269 10270 *) Accept any -xxx and +xxx compiler options in Configure. 10271 [Ulf M��ller] 10272 10273 *) Update HPUX configuration. 10274 [Anonymous] 10275 10276 *) Add missing sk_<type>_unshift() function to safestack.h 10277 [Ralf S. Engelschall] 10278 10279 *) New function SSL_CTX_use_certificate_chain_file that sets the 10280 "extra_cert"s in addition to the certificate. (This makes sense 10281 only for "PEM" format files, as chains as a whole are not 10282 DER-encoded.) 10283 [Bodo Moeller] 10284 10285 *) Support verify_depth from the SSL API. 10286 x509_vfy.c had what can be considered an off-by-one-error: 10287 Its depth (which was not part of the external interface) 10288 was actually counting the number of certificates in a chain; 10289 now it really counts the depth. 10290 [Bodo Moeller] 10291 10292 *) Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used 10293 instead of X509err, which often resulted in confusing error 10294 messages since the error codes are not globally unique 10295 (e.g. an alleged error in ssl3_accept when a certificate 10296 didn't match the private key). 10297 10298 *) New function SSL_CTX_set_session_id_context that allows to set a default 10299 value (so that you don't need SSL_set_session_id_context for each 10300 connection using the SSL_CTX). 10301 [Bodo Moeller] 10302 10303 *) OAEP decoding bug fix. 10304 [Ulf M��ller] 10305 10306 *) Support INSTALL_PREFIX for package builders, as proposed by 10307 David Harris. 10308 [Bodo Moeller] 10309 10310 *) New Configure options "threads" and "no-threads". For systems 10311 where the proper compiler options are known (currently Solaris 10312 and Linux), "threads" is the default. 10313 [Bodo Moeller] 10314 10315 *) New script util/mklink.pl as a faster substitute for util/mklink.sh. 10316 [Bodo Moeller] 10317 10318 *) Install various scripts to $(OPENSSLDIR)/misc, not to 10319 $(INSTALLTOP)/bin -- they shouldn't clutter directories 10320 such as /usr/local/bin. 10321 [Bodo Moeller] 10322 10323 *) "make linux-shared" to build shared libraries. 10324 [Niels Poppe <niels@netbox.org>] 10325 10326 *) New Configure option no-<cipher> (rsa, idea, rc5, ...). 10327 [Ulf M��ller] 10328 10329 *) Add the PKCS#12 API documentation to openssl.txt. Preliminary support for 10330 extension adding in x509 utility. 10331 [Steve Henson] 10332 10333 *) Remove NOPROTO sections and error code comments. 10334 [Ulf M��ller] 10335 10336 *) Partial rewrite of the DEF file generator to now parse the ANSI 10337 prototypes. 10338 [Steve Henson] 10339 10340 *) New Configure options --prefix=DIR and --openssldir=DIR. 10341 [Ulf M��ller] 10342 10343 *) Complete rewrite of the error code script(s). It is all now handled 10344 by one script at the top level which handles error code gathering, 10345 header rewriting and C source file generation. It should be much better 10346 than the old method: it now uses a modified version of Ulf's parser to 10347 read the ANSI prototypes in all header files (thus the old K&R definitions 10348 aren't needed for error creation any more) and do a better job of 10349 translating function codes into names. The old 'ASN1 error code imbedded 10350 in a comment' is no longer necessary and it doesn't use .err files which 10351 have now been deleted. Also the error code call doesn't have to appear all 10352 on one line (which resulted in some large lines...). 10353 [Steve Henson] 10354 10355 *) Change #include filenames from <foo.h> to <openssl/foo.h>. 10356 [Bodo Moeller] 10357 10358 *) Change behaviour of ssl2_read when facing length-0 packets: Don't return 10359 0 (which usually indicates a closed connection), but continue reading. 10360 [Bodo Moeller] 10361 10362 *) Fix some race conditions. 10363 [Bodo Moeller] 10364 10365 *) Add support for CRL distribution points extension. Add Certificate 10366 Policies and CRL distribution points documentation. 10367 [Steve Henson] 10368 10369 *) Move the autogenerated header file parts to crypto/opensslconf.h. 10370 [Ulf M��ller] 10371 10372 *) Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of 10373 8 of keying material. Merlin has also confirmed interop with this fix 10374 between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0. 10375 [Merlin Hughes <merlin@baltimore.ie>] 10376 10377 *) Fix lots of warnings. 10378 [Richard Levitte <levitte@stacken.kth.se>] 10379 10380 *) In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if 10381 the directory spec didn't end with a LIST_SEPARATOR_CHAR. 10382 [Richard Levitte <levitte@stacken.kth.se>] 10383 10384 *) Fix problems with sizeof(long) == 8. 10385 [Andy Polyakov <appro@fy.chalmers.se>] 10386 10387 *) Change functions to ANSI C. 10388 [Ulf M��ller] 10389 10390 *) Fix typos in error codes. 10391 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf M��ller] 10392 10393 *) Remove defunct assembler files from Configure. 10394 [Ulf M��ller] 10395 10396 *) SPARC v8 assembler BIGNUM implementation. 10397 [Andy Polyakov <appro@fy.chalmers.se>] 10398 10399 *) Support for Certificate Policies extension: both print and set. 10400 Various additions to support the r2i method this uses. 10401 [Steve Henson] 10402 10403 *) A lot of constification, and fix a bug in X509_NAME_oneline() that could 10404 return a const string when you are expecting an allocated buffer. 10405 [Ben Laurie] 10406 10407 *) Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE 10408 types DirectoryString and DisplayText. 10409 [Steve Henson] 10410 10411 *) Add code to allow r2i extensions to access the configuration database, 10412 add an LHASH database driver and add several ctx helper functions. 10413 [Steve Henson] 10414 10415 *) Fix an evil bug in bn_expand2() which caused various BN functions to 10416 fail when they extended the size of a BIGNUM. 10417 [Steve Henson] 10418 10419 *) Various utility functions to handle SXNet extension. Modify mkdef.pl to 10420 support typesafe stack. 10421 [Steve Henson] 10422 10423 *) Fix typo in SSL_[gs]et_options(). 10424 [Nils Frostberg <nils@medcom.se>] 10425 10426 *) Delete various functions and files that belonged to the (now obsolete) 10427 old X509V3 handling code. 10428 [Steve Henson] 10429 10430 *) New Configure option "rsaref". 10431 [Ulf M��ller] 10432 10433 *) Don't auto-generate pem.h. 10434 [Bodo Moeller] 10435 10436 *) Introduce type-safe ASN.1 SETs. 10437 [Ben Laurie] 10438 10439 *) Convert various additional casted stacks to type-safe STACK_OF() variants. 10440 [Ben Laurie, Ralf S. Engelschall, Steve Henson] 10441 10442 *) Introduce type-safe STACKs. This will almost certainly break lots of code 10443 that links with OpenSSL (well at least cause lots of warnings), but fear 10444 not: the conversion is trivial, and it eliminates loads of evil casts. A 10445 few STACKed things have been converted already. Feel free to convert more. 10446 In the fullness of time, I'll do away with the STACK type altogether. 10447 [Ben Laurie] 10448 10449 *) Add `openssl ca -revoke <certfile>' facility which revokes a certificate 10450 specified in <certfile> by updating the entry in the index.txt file. 10451 This way one no longer has to edit the index.txt file manually for 10452 revoking a certificate. The -revoke option does the gory details now. 10453 [Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall] 10454 10455 *) Fix `openssl crl -noout -text' combination where `-noout' killed the 10456 `-text' option at all and this way the `-noout -text' combination was 10457 inconsistent in `openssl crl' with the friends in `openssl x509|rsa|dsa'. 10458 [Ralf S. Engelschall] 10459 10460 *) Make sure a corresponding plain text error message exists for the 10461 X509_V_ERR_CERT_REVOKED/23 error number which can occur when a 10462 verify callback function determined that a certificate was revoked. 10463 [Ralf S. Engelschall] 10464 10465 *) Bugfix: In test/testenc, don't test "openssl <cipher>" for 10466 ciphers that were excluded, e.g. by -DNO_IDEA. Also, test 10467 all available cipers including rc5, which was forgotten until now. 10468 In order to let the testing shell script know which algorithms 10469 are available, a new (up to now undocumented) command 10470 "openssl list-cipher-commands" is used. 10471 [Bodo Moeller] 10472 10473 *) Bugfix: s_client occasionally would sleep in select() when 10474 it should have checked SSL_pending() first. 10475 [Bodo Moeller] 10476 10477 *) New functions DSA_do_sign and DSA_do_verify to provide access to 10478 the raw DSA values prior to ASN.1 encoding. 10479 [Ulf M��ller] 10480 10481 *) Tweaks to Configure 10482 [Niels Poppe <niels@netbox.org>] 10483 10484 *) Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support, 10485 yet... 10486 [Steve Henson] 10487 10488 *) New variables $(RANLIB) and $(PERL) in the Makefiles. 10489 [Ulf M��ller] 10490 10491 *) New config option to avoid instructions that are illegal on the 80386. 10492 The default code is faster, but requires at least a 486. 10493 [Ulf M��ller] 10494 10495 *) Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and 10496 SSL2_SERVER_VERSION (not used at all) macros, which are now the 10497 same as SSL2_VERSION anyway. 10498 [Bodo Moeller] 10499 10500 *) New "-showcerts" option for s_client. 10501 [Bodo Moeller] 10502 10503 *) Still more PKCS#12 integration. Add pkcs12 application to openssl 10504 application. Various cleanups and fixes. 10505 [Steve Henson] 10506 10507 *) More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and 10508 modify error routines to work internally. Add error codes and PBE init 10509 to library startup routines. 10510 [Steve Henson] 10511 10512 *) Further PKCS#12 integration. Added password based encryption, PKCS#8 and 10513 packing functions to asn1 and evp. Changed function names and error 10514 codes along the way. 10515 [Steve Henson] 10516 10517 *) PKCS12 integration: and so it begins... First of several patches to 10518 slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12 10519 objects to objects.h 10520 [Steve Henson] 10521 10522 *) Add a new 'indent' option to some X509V3 extension code. Initial ASN1 10523 and display support for Thawte strong extranet extension. 10524 [Steve Henson] 10525 10526 *) Add LinuxPPC support. 10527 [Jeff Dubrule <igor@pobox.org>] 10528 10529 *) Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to 10530 bn_div_words in alpha.s. 10531 [Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie] 10532 10533 *) Make sure the RSA OAEP test is skipped under -DRSAref because 10534 OAEP isn't supported when OpenSSL is built with RSAref. 10535 [Ulf Moeller <ulf@fitug.de>] 10536 10537 *) Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h 10538 so they no longer are missing under -DNOPROTO. 10539 [Soren S. Jorvang <soren@t.dk>] 10540 10541 10542 Changes between 0.9.1c and 0.9.2b [22 Mar 1999] 10543 10544 *) Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still 10545 doesn't work when the session is reused. Coming soon! 10546 [Ben Laurie] 10547 10548 *) Fix a security hole, that allows sessions to be reused in the wrong 10549 context thus bypassing client cert protection! All software that uses 10550 client certs and session caches in multiple contexts NEEDS PATCHING to 10551 allow session reuse! A fuller solution is in the works. 10552 [Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)] 10553 10554 *) Some more source tree cleanups (removed obsolete files 10555 crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed 10556 permission on "config" script to be executable) and a fix for the INSTALL 10557 document. 10558 [Ulf Moeller <ulf@fitug.de>] 10559 10560 *) Remove some legacy and erroneous uses of malloc, free instead of 10561 Malloc, Free. 10562 [Lennart Bang <lob@netstream.se>, with minor changes by Steve] 10563 10564 *) Make rsa_oaep_test return non-zero on error. 10565 [Ulf Moeller <ulf@fitug.de>] 10566 10567 *) Add support for native Solaris shared libraries. Configure 10568 solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice 10569 if someone would make that last step automatic. 10570 [Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>] 10571 10572 *) ctx_size was not built with the right compiler during "make links". Fixed. 10573 [Ben Laurie] 10574 10575 *) Change the meaning of 'ALL' in the cipher list. It now means "everything 10576 except NULL ciphers". This means the default cipher list will no longer 10577 enable NULL ciphers. They need to be specifically enabled e.g. with 10578 the string "DEFAULT:eNULL". 10579 [Steve Henson] 10580 10581 *) Fix to RSA private encryption routines: if p < q then it would 10582 occasionally produce an invalid result. This will only happen with 10583 externally generated keys because OpenSSL (and SSLeay) ensure p > q. 10584 [Steve Henson] 10585 10586 *) Be less restrictive and allow also `perl util/perlpath.pl 10587 /path/to/bin/perl' in addition to `perl util/perlpath.pl /path/to/bin', 10588 because this way one can also use an interpreter named `perl5' (which is 10589 usually the name of Perl 5.xxx on platforms where an Perl 4.x is still 10590 installed as `perl'). 10591 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 10592 10593 *) Let util/clean-depend.pl work also with older Perl 5.00x versions. 10594 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 10595 10596 *) Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add 10597 advapi32.lib to Win32 build and change the pem test comparision 10598 to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the 10599 suggestion). Fix misplaced ASNI prototypes and declarations in evp.h 10600 and crypto/des/ede_cbcm_enc.c. 10601 [Steve Henson] 10602 10603 *) DES quad checksum was broken on big-endian architectures. Fixed. 10604 [Ben Laurie] 10605 10606 *) Comment out two functions in bio.h that aren't implemented. Fix up the 10607 Win32 test batch file so it (might) work again. The Win32 test batch file 10608 is horrible: I feel ill.... 10609 [Steve Henson] 10610 10611 *) Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected 10612 in e_os.h. Audit of header files to check ANSI and non ANSI 10613 sections: 10 functions were absent from non ANSI section and not exported 10614 from Windows DLLs. Fixed up libeay.num for new functions. 10615 [Steve Henson] 10616 10617 *) Make `openssl version' output lines consistent. 10618 [Ralf S. Engelschall] 10619 10620 *) Fix Win32 symbol export lists for BIO functions: Added 10621 BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data 10622 to ms/libeay{16,32}.def. 10623 [Ralf S. Engelschall] 10624 10625 *) Second round of fixing the OpenSSL perl/ stuff. It now at least compiled 10626 fine under Unix and passes some trivial tests I've now added. But the 10627 whole stuff is horribly incomplete, so a README.1ST with a disclaimer was 10628 added to make sure no one expects that this stuff really works in the 10629 OpenSSL 0.9.2 release. Additionally I've started to clean the XS sources 10630 up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and 10631 openssl_bio.xs. 10632 [Ralf S. Engelschall] 10633 10634 *) Fix the generation of two part addresses in perl. 10635 [Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie] 10636 10637 *) Add config entry for Linux on MIPS. 10638 [John Tobey <jtobey@channel1.com>] 10639 10640 *) Make links whenever Configure is run, unless we are on Windoze. 10641 [Ben Laurie] 10642 10643 *) Permit extensions to be added to CRLs using crl_section in openssl.cnf. 10644 Currently only issuerAltName and AuthorityKeyIdentifier make any sense 10645 in CRLs. 10646 [Steve Henson] 10647 10648 *) Add a useful kludge to allow package maintainers to specify compiler and 10649 other platforms details on the command line without having to patch the 10650 Configure script everytime: One now can use ``perl Configure 10651 <id>:<details>'', i.e. platform ids are allowed to have details appended 10652 to them (seperated by colons). This is treated as there would be a static 10653 pre-configured entry in Configure's %table under key <id> with value 10654 <details> and ``perl Configure <id>'' is called. So, when you want to 10655 perform a quick test-compile under FreeBSD 3.1 with pgcc and without 10656 assembler stuff you can use ``perl Configure "FreeBSD-elf:pgcc:-O6:::"'' 10657 now, which overrides the FreeBSD-elf entry on-the-fly. 10658 [Ralf S. Engelschall] 10659 10660 *) Disable new TLS1 ciphersuites by default: they aren't official yet. 10661 [Ben Laurie] 10662 10663 *) Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified 10664 on the `perl Configure ...' command line. This way one can compile 10665 OpenSSL libraries with Position Independent Code (PIC) which is needed 10666 for linking it into DSOs. 10667 [Ralf S. Engelschall] 10668 10669 *) Remarkably, export ciphers were totally broken and no-one had noticed! 10670 Fixed. 10671 [Ben Laurie] 10672 10673 *) Cleaned up the LICENSE document: The official contact for any license 10674 questions now is the OpenSSL core team under openssl-core@openssl.org. 10675 And add a paragraph about the dual-license situation to make sure people 10676 recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply 10677 to the OpenSSL toolkit. 10678 [Ralf S. Engelschall] 10679 10680 *) General source tree makefile cleanups: Made `making xxx in yyy...' 10681 display consistent in the source tree and replaced `/bin/rm' by `rm'. 10682 Additonally cleaned up the `make links' target: Remove unnecessary 10683 semicolons, subsequent redundant removes, inline point.sh into mklink.sh 10684 to speed processing and no longer clutter the display with confusing 10685 stuff. Instead only the actually done links are displayed. 10686 [Ralf S. Engelschall] 10687 10688 *) Permit null encryption ciphersuites, used for authentication only. It used 10689 to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this. 10690 It is now necessary to set SSL_FORBID_ENULL to prevent the use of null 10691 encryption. 10692 [Ben Laurie] 10693 10694 *) Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder 10695 signed attributes when verifying signatures (this would break them), 10696 the detached data encoding was wrong and public keys obtained using 10697 X509_get_pubkey() weren't freed. 10698 [Steve Henson] 10699 10700 *) Add text documentation for the BUFFER functions. Also added a work around 10701 to a Win95 console bug. This was triggered by the password read stuff: the 10702 last character typed gets carried over to the next fread(). If you were 10703 generating a new cert request using 'req' for example then the last 10704 character of the passphrase would be CR which would then enter the first 10705 field as blank. 10706 [Steve Henson] 10707 10708 *) Added the new `Includes OpenSSL Cryptography Software' button as 10709 doc/openssl_button.{gif,html} which is similar in style to the old SSLeay 10710 button and can be used by applications based on OpenSSL to show the 10711 relationship to the OpenSSL project. 10712 [Ralf S. Engelschall] 10713 10714 *) Remove confusing variables in function signatures in files 10715 ssl/ssl_lib.c and ssl/ssl.h. 10716 [Lennart Bong <lob@kulthea.stacken.kth.se>] 10717 10718 *) Don't install bss_file.c under PREFIX/include/ 10719 [Lennart Bong <lob@kulthea.stacken.kth.se>] 10720 10721 *) Get the Win32 compile working again. Modify mkdef.pl so it can handle 10722 functions that return function pointers and has support for NT specific 10723 stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various 10724 #ifdef WIN32 and WINNTs sprinkled about the place and some changes from 10725 unsigned to signed types: this was killing the Win32 compile. 10726 [Steve Henson] 10727 10728 *) Add new certificate file to stack functions, 10729 SSL_add_dir_cert_subjects_to_stack() and 10730 SSL_add_file_cert_subjects_to_stack(). These largely supplant 10731 SSL_load_client_CA_file(), and can be used to add multiple certs easily 10732 to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()). 10733 This means that Apache-SSL and similar packages don't have to mess around 10734 to add as many CAs as they want to the preferred list. 10735 [Ben Laurie] 10736 10737 *) Experiment with doxygen documentation. Currently only partially applied to 10738 ssl/ssl_lib.c. 10739 See http://www.stack.nl/~dimitri/doxygen/index.html, and run doxygen with 10740 openssl.doxy as the configuration file. 10741 [Ben Laurie] 10742 10743 *) Get rid of remaining C++-style comments which strict C compilers hate. 10744 [Ralf S. Engelschall, pointed out by Carlos Amengual] 10745 10746 *) Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not 10747 compiled in by default: it has problems with large keys. 10748 [Steve Henson] 10749 10750 *) Add a bunch of SSL_xxx() functions for configuring the temporary RSA and 10751 DH private keys and/or callback functions which directly correspond to 10752 their SSL_CTX_xxx() counterparts but work on a per-connection basis. This 10753 is needed for applications which have to configure certificates on a 10754 per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis 10755 (e.g. s_server). 10756 For the RSA certificate situation is makes no difference, but 10757 for the DSA certificate situation this fixes the "no shared cipher" 10758 problem where the OpenSSL cipher selection procedure failed because the 10759 temporary keys were not overtaken from the context and the API provided 10760 no way to reconfigure them. 10761 The new functions now let applications reconfigure the stuff and they 10762 are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh, 10763 SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback. Additionally a new 10764 non-public-API function ssl_cert_instantiate() is used as a helper 10765 function and also to reduce code redundancy inside ssl_rsa.c. 10766 [Ralf S. Engelschall] 10767 10768 *) Move s_server -dcert and -dkey options out of the undocumented feature 10769 area because they are useful for the DSA situation and should be 10770 recognized by the users. 10771 [Ralf S. Engelschall] 10772 10773 *) Fix the cipher decision scheme for export ciphers: the export bits are 10774 *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within 10775 SSL_EXP_MASK. So, the original variable has to be used instead of the 10776 already masked variable. 10777 [Richard Levitte <levitte@stacken.kth.se>] 10778 10779 *) Fix 'port' variable from `int' to `unsigned int' in crypto/bio/b_sock.c 10780 [Richard Levitte <levitte@stacken.kth.se>] 10781 10782 *) Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal() 10783 from `int' to `unsigned int' because it's a length and initialized by 10784 EVP_DigestFinal() which expects an `unsigned int *'. 10785 [Richard Levitte <levitte@stacken.kth.se>] 10786 10787 *) Don't hard-code path to Perl interpreter on shebang line of Configure 10788 script. Instead use the usual Shell->Perl transition trick. 10789 [Ralf S. Engelschall] 10790 10791 *) Make `openssl x509 -noout -modulus' functional also for DSA certificates 10792 (in addition to RSA certificates) to match the behaviour of `openssl dsa 10793 -noout -modulus' as it's already the case for `openssl rsa -noout 10794 -modulus'. For RSA the -modulus is the real "modulus" while for DSA 10795 currently the public key is printed (a decision which was already done by 10796 `openssl dsa -modulus' in the past) which serves a similar purpose. 10797 Additionally the NO_RSA no longer completely removes the whole -modulus 10798 option; it now only avoids using the RSA stuff. Same applies to NO_DSA 10799 now, too. 10800 [Ralf S. Engelschall] 10801 10802 *) Add Arne Ansper's reliable BIO - this is an encrypted, block-digested 10803 BIO. See the source (crypto/evp/bio_ok.c) for more info. 10804 [Arne Ansper <arne@ats.cyber.ee>] 10805 10806 *) Dump the old yucky req code that tried (and failed) to allow raw OIDs 10807 to be added. Now both 'req' and 'ca' can use new objects defined in the 10808 config file. 10809 [Steve Henson] 10810 10811 *) Add cool BIO that does syslog (or event log on NT). 10812 [Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie] 10813 10814 *) Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5, 10815 TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and 10816 TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher 10817 Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt. 10818 [Ben Laurie] 10819 10820 *) Add preliminary config info for new extension code. 10821 [Steve Henson] 10822 10823 *) Make RSA_NO_PADDING really use no padding. 10824 [Ulf Moeller <ulf@fitug.de>] 10825 10826 *) Generate errors when private/public key check is done. 10827 [Ben Laurie] 10828 10829 *) Overhaul for 'crl' utility. New function X509_CRL_print. Partial support 10830 for some CRL extensions and new objects added. 10831 [Steve Henson] 10832 10833 *) Really fix the ASN1 IMPLICIT bug this time... Partial support for private 10834 key usage extension and fuller support for authority key id. 10835 [Steve Henson] 10836 10837 *) Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved 10838 padding method for RSA, which is recommended for new applications in PKCS 10839 #1 v2.0 (RFC 2437, October 1998). 10840 OAEP (Optimal Asymmetric Encryption Padding) has better theoretical 10841 foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure 10842 against Bleichbacher's attack on RSA. 10843 [Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by 10844 Ben Laurie] 10845 10846 *) Updates to the new SSL compression code 10847 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 10848 10849 *) Fix so that the version number in the master secret, when passed 10850 via RSA, checks that if TLS was proposed, but we roll back to SSLv3 10851 (because the server will not accept higher), that the version number 10852 is 0x03,0x01, not 0x03,0x00 10853 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 10854 10855 *) Run extensive memory leak checks on SSL apps. Fixed *lots* of memory 10856 leaks in ssl/ relating to new X509_get_pubkey() behaviour. Also fixes 10857 in apps/ and an unrelated leak in crypto/dsa/dsa_vrf.c 10858 [Steve Henson] 10859 10860 *) Support for RAW extensions where an arbitrary extension can be 10861 created by including its DER encoding. See apps/openssl.cnf for 10862 an example. 10863 [Steve Henson] 10864 10865 *) Make sure latest Perl versions don't interpret some generated C array 10866 code as Perl array code in the crypto/err/err_genc.pl script. 10867 [Lars Weber <3weber@informatik.uni-hamburg.de>] 10868 10869 *) Modify ms/do_ms.bat to not generate assembly language makefiles since 10870 not many people have the assembler. Various Win32 compilation fixes and 10871 update to the INSTALL.W32 file with (hopefully) more accurate Win32 10872 build instructions. 10873 [Steve Henson] 10874 10875 *) Modify configure script 'Configure' to automatically create crypto/date.h 10876 file under Win32 and also build pem.h from pem.org. New script 10877 util/mkfiles.pl to create the MINFO file on environments that can't do a 10878 'make files': perl util/mkfiles.pl >MINFO should work. 10879 [Steve Henson] 10880 10881 *) Major rework of DES function declarations, in the pursuit of correctness 10882 and purity. As a result, many evil casts evaporated, and some weirdness, 10883 too. You may find this causes warnings in your code. Zapping your evil 10884 casts will probably fix them. Mostly. 10885 [Ben Laurie] 10886 10887 *) Fix for a typo in asn1.h. Bug fix to object creation script 10888 obj_dat.pl. It considered a zero in an object definition to mean 10889 "end of object": none of the objects in objects.h have any zeros 10890 so it wasn't spotted. 10891 [Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>] 10892 10893 *) Add support for Triple DES Cipher Block Chaining with Output Feedback 10894 Masking (CBCM). In the absence of test vectors, the best I have been able 10895 to do is check that the decrypt undoes the encrypt, so far. Send me test 10896 vectors if you have them. 10897 [Ben Laurie] 10898 10899 *) Correct calculation of key length for export ciphers (too much space was 10900 allocated for null ciphers). This has not been tested! 10901 [Ben Laurie] 10902 10903 *) Modifications to the mkdef.pl for Win32 DEF file creation. The usage 10904 message is now correct (it understands "crypto" and "ssl" on its 10905 command line). There is also now an "update" option. This will update 10906 the util/ssleay.num and util/libeay.num files with any new functions. 10907 If you do a: 10908 perl util/mkdef.pl crypto ssl update 10909 it will update them. 10910 [Steve Henson] 10911 10912 *) Overhauled the Perl interface (perl/*): 10913 - ported BN stuff to OpenSSL's different BN library 10914 - made the perl/ source tree CVS-aware 10915 - renamed the package from SSLeay to OpenSSL (the files still contain 10916 their history because I've copied them in the repository) 10917 - removed obsolete files (the test scripts will be replaced 10918 by better Test::Harness variants in the future) 10919 [Ralf S. Engelschall] 10920 10921 *) First cut for a very conservative source tree cleanup: 10922 1. merge various obsolete readme texts into doc/ssleay.txt 10923 where we collect the old documents and readme texts. 10924 2. remove the first part of files where I'm already sure that we no 10925 longer need them because of three reasons: either they are just temporary 10926 files which were left by Eric or they are preserved original files where 10927 I've verified that the diff is also available in the CVS via "cvs diff 10928 -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for 10929 the crypto/md/ stuff). 10930 [Ralf S. Engelschall] 10931 10932 *) More extension code. Incomplete support for subject and issuer alt 10933 name, issuer and authority key id. Change the i2v function parameters 10934 and add an extra 'crl' parameter in the X509V3_CTX structure: guess 10935 what that's for :-) Fix to ASN1 macro which messed up 10936 IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED. 10937 [Steve Henson] 10938 10939 *) Preliminary support for ENUMERATED type. This is largely copied from the 10940 INTEGER code. 10941 [Steve Henson] 10942 10943 *) Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy. 10944 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 10945 10946 *) Make sure `make rehash' target really finds the `openssl' program. 10947 [Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 10948 10949 *) Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd 10950 like to hear about it if this slows down other processors. 10951 [Ben Laurie] 10952 10953 *) Add CygWin32 platform information to Configure script. 10954 [Alan Batie <batie@aahz.jf.intel.com>] 10955 10956 *) Fixed ms/32all.bat script: `no_asm' -> `no-asm' 10957 [Rainer W. Gerling <gerling@mpg-gv.mpg.de>] 10958 10959 *) New program nseq to manipulate netscape certificate sequences 10960 [Steve Henson] 10961 10962 *) Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a 10963 few typos. 10964 [Steve Henson] 10965 10966 *) Fixes to BN code. Previously the default was to define BN_RECURSION 10967 but the BN code had some problems that would cause failures when 10968 doing certificate verification and some other functions. 10969 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 10970 10971 *) Add ASN1 and PEM code to support netscape certificate sequences. 10972 [Steve Henson] 10973 10974 *) Add ASN1 and PEM code to support netscape certificate sequences. 10975 [Steve Henson] 10976 10977 *) Add several PKIX and private extended key usage OIDs. 10978 [Steve Henson] 10979 10980 *) Modify the 'ca' program to handle the new extension code. Modify 10981 openssl.cnf for new extension format, add comments. 10982 [Steve Henson] 10983 10984 *) More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req' 10985 and add a sample to openssl.cnf so req -x509 now adds appropriate 10986 CA extensions. 10987 [Steve Henson] 10988 10989 *) Continued X509 V3 changes. Add to other makefiles, integrate with the 10990 error code, add initial support to X509_print() and x509 application. 10991 [Steve Henson] 10992 10993 *) Takes a deep breath and start addding X509 V3 extension support code. Add 10994 files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this 10995 stuff is currently isolated and isn't even compiled yet. 10996 [Steve Henson] 10997 10998 *) Continuing patches for GeneralizedTime. Fix up certificate and CRL 10999 ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print. 11000 Removed the versions check from X509 routines when loading extensions: 11001 this allows certain broken certificates that don't set the version 11002 properly to be processed. 11003 [Steve Henson] 11004 11005 *) Deal with irritating shit to do with dependencies, in YAAHW (Yet Another 11006 Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which 11007 can still be regenerated with "make depend". 11008 [Ben Laurie] 11009 11010 *) Spelling mistake in C version of CAST-128. 11011 [Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>] 11012 11013 *) Changes to the error generation code. The perl script err-code.pl 11014 now reads in the old error codes and retains the old numbers, only 11015 adding new ones if necessary. It also only changes the .err files if new 11016 codes are added. The makefiles have been modified to only insert errors 11017 when needed (to avoid needlessly modifying header files). This is done 11018 by only inserting errors if the .err file is newer than the auto generated 11019 C file. To rebuild all the error codes from scratch (the old behaviour) 11020 either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl 11021 or delete all the .err files. 11022 [Steve Henson] 11023 11024 *) CAST-128 was incorrectly implemented for short keys. The C version has 11025 been fixed, but is untested. The assembler versions are also fixed, but 11026 new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing 11027 to regenerate it if needed. 11028 [Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun 11029 Hagino <itojun@kame.net>] 11030 11031 *) File was opened incorrectly in randfile.c. 11032 [Ulf M��ller <ulf@fitug.de>] 11033 11034 *) Beginning of support for GeneralizedTime. d2i, i2d, check and print 11035 functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or 11036 GeneralizedTime. ASN1_TIME is the proper type used in certificates et 11037 al: it's just almost always a UTCTime. Note this patch adds new error 11038 codes so do a "make errors" if there are problems. 11039 [Steve Henson] 11040 11041 *) Correct Linux 1 recognition in config. 11042 [Ulf M��ller <ulf@fitug.de>] 11043 11044 *) Remove pointless MD5 hash when using DSA keys in ca. 11045 [Anonymous <nobody@replay.com>] 11046 11047 *) Generate an error if given an empty string as a cert directory. Also 11048 generate an error if handed NULL (previously returned 0 to indicate an 11049 error, but didn't set one). 11050 [Ben Laurie, reported by Anonymous <nobody@replay.com>] 11051 11052 *) Add prototypes to SSL methods. Make SSL_write's buffer const, at last. 11053 [Ben Laurie] 11054 11055 *) Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct 11056 parameters. This was causing a warning which killed off the Win32 compile. 11057 [Steve Henson] 11058 11059 *) Remove C++ style comments from crypto/bn/bn_local.h. 11060 [Neil Costigan <neil.costigan@celocom.com>] 11061 11062 *) The function OBJ_txt2nid was broken. It was supposed to return a nid 11063 based on a text string, looking up short and long names and finally 11064 "dot" format. The "dot" format stuff didn't work. Added new function 11065 OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote 11066 OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the 11067 OID is not part of the table. 11068 [Steve Henson] 11069 11070 *) Add prototypes to X509 lookup/verify methods, fixing a bug in 11071 X509_LOOKUP_by_alias(). 11072 [Ben Laurie] 11073 11074 *) Sort openssl functions by name. 11075 [Ben Laurie] 11076 11077 *) Get the gendsa program working (hopefully) and add it to app list. Remove 11078 encryption from sample DSA keys (in case anyone is interested the password 11079 was "1234"). 11080 [Steve Henson] 11081 11082 *) Make _all_ *_free functions accept a NULL pointer. 11083 [Frans Heymans <fheymans@isaserver.be>] 11084 11085 *) If a DH key is generated in s3_srvr.c, don't blow it by trying to use 11086 NULL pointers. 11087 [Anonymous <nobody@replay.com>] 11088 11089 *) s_server should send the CAfile as acceptable CAs, not its own cert. 11090 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 11091 11092 *) Don't blow it for numeric -newkey arguments to apps/req. 11093 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 11094 11095 *) Temp key "for export" tests were wrong in s3_srvr.c. 11096 [Anonymous <nobody@replay.com>] 11097 11098 *) Add prototype for temp key callback functions 11099 SSL_CTX_set_tmp_{rsa,dh}_callback(). 11100 [Ben Laurie] 11101 11102 *) Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and 11103 DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey(). 11104 [Steve Henson] 11105 11106 *) X509_name_add_entry() freed the wrong thing after an error. 11107 [Arne Ansper <arne@ats.cyber.ee>] 11108 11109 *) rsa_eay.c would attempt to free a NULL context. 11110 [Arne Ansper <arne@ats.cyber.ee>] 11111 11112 *) BIO_s_socket() had a broken should_retry() on Windoze. 11113 [Arne Ansper <arne@ats.cyber.ee>] 11114 11115 *) BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH. 11116 [Arne Ansper <arne@ats.cyber.ee>] 11117 11118 *) Make sure the already existing X509_STORE->depth variable is initialized 11119 in X509_STORE_new(), but document the fact that this variable is still 11120 unused in the certificate verification process. 11121 [Ralf S. Engelschall] 11122 11123 *) Fix the various library and apps files to free up pkeys obtained from 11124 X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions. 11125 [Steve Henson] 11126 11127 *) Fix reference counting in X509_PUBKEY_get(). This makes 11128 demos/maurice/example2.c work, amongst others, probably. 11129 [Steve Henson and Ben Laurie] 11130 11131 *) First cut of a cleanup for apps/. First the `ssleay' program is now named 11132 `openssl' and second, the shortcut symlinks for the `openssl <command>' 11133 are no longer created. This way we have a single and consistent command 11134 line interface `openssl <command>', similar to `cvs <command>'. 11135 [Ralf S. Engelschall, Paul Sutton and Ben Laurie] 11136 11137 *) ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey 11138 BIT STRING wrapper always have zero unused bits. 11139 [Steve Henson] 11140 11141 *) Add CA.pl, perl version of CA.sh, add extended key usage OID. 11142 [Steve Henson] 11143 11144 *) Make the top-level INSTALL documentation easier to understand. 11145 [Paul Sutton] 11146 11147 *) Makefiles updated to exit if an error occurs in a sub-directory 11148 make (including if user presses ^C) [Paul Sutton] 11149 11150 *) Make Montgomery context stuff explicit in RSA data structure. 11151 [Ben Laurie] 11152 11153 *) Fix build order of pem and err to allow for generated pem.h. 11154 [Ben Laurie] 11155 11156 *) Fix renumbering bug in X509_NAME_delete_entry(). 11157 [Ben Laurie] 11158 11159 *) Enhanced the err-ins.pl script so it makes the error library number 11160 global and can add a library name. This is needed for external ASN1 and 11161 other error libraries. 11162 [Steve Henson] 11163 11164 *) Fixed sk_insert which never worked properly. 11165 [Steve Henson] 11166 11167 *) Fix ASN1 macros so they can handle indefinite length construted 11168 EXPLICIT tags. Some non standard certificates use these: they can now 11169 be read in. 11170 [Steve Henson] 11171 11172 *) Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc) 11173 into a single doc/ssleay.txt bundle. This way the information is still 11174 preserved but no longer messes up this directory. Now it's new room for 11175 the new set of documenation files. 11176 [Ralf S. Engelschall] 11177 11178 *) SETs were incorrectly DER encoded. This was a major pain, because they 11179 shared code with SEQUENCEs, which aren't coded the same. This means that 11180 almost everything to do with SETs or SEQUENCEs has either changed name or 11181 number of arguments. 11182 [Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>] 11183 11184 *) Fix test data to work with the above. 11185 [Ben Laurie] 11186 11187 *) Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but 11188 was already fixed by Eric for 0.9.1 it seems. 11189 [Ben Laurie - pointed out by Ulf M��ller <ulf@fitug.de>] 11190 11191 *) Autodetect FreeBSD3. 11192 [Ben Laurie] 11193 11194 *) Fix various bugs in Configure. This affects the following platforms: 11195 nextstep 11196 ncr-scde 11197 unixware-2.0 11198 unixware-2.0-pentium 11199 sco5-cc. 11200 [Ben Laurie] 11201 11202 *) Eliminate generated files from CVS. Reorder tests to regenerate files 11203 before they are needed. 11204 [Ben Laurie] 11205 11206 *) Generate Makefile.ssl from Makefile.org (to keep CVS happy). 11207 [Ben Laurie] 11208 11209 11210 Changes between 0.9.1b and 0.9.1c [23-Dec-1998] 11211 11212 *) Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and 11213 changed SSLeay to OpenSSL in version strings. 11214 [Ralf S. Engelschall] 11215 11216 *) Some fixups to the top-level documents. 11217 [Paul Sutton] 11218 11219 *) Fixed the nasty bug where rsaref.h was not found under compile-time 11220 because the symlink to include/ was missing. 11221 [Ralf S. Engelschall] 11222 11223 *) Incorporated the popular no-RSA/DSA-only patches 11224 which allow to compile a RSA-free SSLeay. 11225 [Andrew Cooke / Interrader Ldt., Ralf S. Engelschall] 11226 11227 *) Fixed nasty rehash problem under `make -f Makefile.ssl links' 11228 when "ssleay" is still not found. 11229 [Ralf S. Engelschall] 11230 11231 *) Added more platforms to Configure: Cray T3E, HPUX 11, 11232 [Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>] 11233 11234 *) Updated the README file. 11235 [Ralf S. Engelschall] 11236 11237 *) Added various .cvsignore files in the CVS repository subdirs 11238 to make a "cvs update" really silent. 11239 [Ralf S. Engelschall] 11240 11241 *) Recompiled the error-definition header files and added 11242 missing symbols to the Win32 linker tables. 11243 [Ralf S. Engelschall] 11244 11245 *) Cleaned up the top-level documents; 11246 o new files: CHANGES and LICENSE 11247 o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay 11248 o merged COPYRIGHT into LICENSE 11249 o removed obsolete TODO file 11250 o renamed MICROSOFT to INSTALL.W32 11251 [Ralf S. Engelschall] 11252 11253 *) Removed dummy files from the 0.9.1b source tree: 11254 crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi 11255 crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f 11256 crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f 11257 crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f 11258 util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f 11259 [Ralf S. Engelschall] 11260 11261 *) Added various platform portability fixes. 11262 [Mark J. Cox] 11263 11264 *) The Genesis of the OpenSSL rpject: 11265 We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A. 11266 Young and Tim J. Hudson created while they were working for C2Net until 11267 summer 1998. 11268 [The OpenSSL Project] 11269 11270 11271 Changes between 0.9.0b and 0.9.1b [not released] 11272 11273 *) Updated a few CA certificates under certs/ 11274 [Eric A. Young] 11275 11276 *) Changed some BIGNUM api stuff. 11277 [Eric A. Young] 11278 11279 *) Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD, 11280 DGUX x86, Linux Alpha, etc. 11281 [Eric A. Young] 11282 11283 *) New COMP library [crypto/comp/] for SSL Record Layer Compression: 11284 RLE (dummy implemented) and ZLIB (really implemented when ZLIB is 11285 available). 11286 [Eric A. Young] 11287 11288 *) Add -strparse option to asn1pars program which parses nested 11289 binary structures 11290 [Dr Stephen Henson <shenson@bigfoot.com>] 11291 11292 *) Added "oid_file" to ssleay.cnf for "ca" and "req" programs. 11293 [Eric A. Young] 11294 11295 *) DSA fix for "ca" program. 11296 [Eric A. Young] 11297 11298 *) Added "-genkey" option to "dsaparam" program. 11299 [Eric A. Young] 11300 11301 *) Added RIPE MD160 (rmd160) message digest. 11302 [Eric A. Young] 11303 11304 *) Added -a (all) option to "ssleay version" command. 11305 [Eric A. Young] 11306 11307 *) Added PLATFORM define which is the id given to Configure. 11308 [Eric A. Young] 11309 11310 *) Added MemCheck_XXXX functions to crypto/mem.c for memory checking. 11311 [Eric A. Young] 11312 11313 *) Extended the ASN.1 parser routines. 11314 [Eric A. Young] 11315 11316 *) Extended BIO routines to support REUSEADDR, seek, tell, etc. 11317 [Eric A. Young] 11318 11319 *) Added a BN_CTX to the BN library. 11320 [Eric A. Young] 11321 11322 *) Fixed the weak key values in DES library 11323 [Eric A. Young] 11324 11325 *) Changed API in EVP library for cipher aliases. 11326 [Eric A. Young] 11327 11328 *) Added support for RC2/64bit cipher. 11329 [Eric A. Young] 11330 11331 *) Converted the lhash library to the crypto/mem.c functions. 11332 [Eric A. Young] 11333 11334 *) Added more recognized ASN.1 object ids. 11335 [Eric A. Young] 11336 11337 *) Added more RSA padding checks for SSL/TLS. 11338 [Eric A. Young] 11339 11340 *) Added BIO proxy/filter functionality. 11341 [Eric A. Young] 11342 11343 *) Added extra_certs to SSL_CTX which can be used 11344 send extra CA certificates to the client in the CA cert chain sending 11345 process. It can be configured with SSL_CTX_add_extra_chain_cert(). 11346 [Eric A. Young] 11347 11348 *) Now Fortezza is denied in the authentication phase because 11349 this is key exchange mechanism is not supported by SSLeay at all. 11350 [Eric A. Young] 11351 11352 *) Additional PKCS1 checks. 11353 [Eric A. Young] 11354 11355 *) Support the string "TLSv1" for all TLS v1 ciphers. 11356 [Eric A. Young] 11357 11358 *) Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the 11359 ex_data index of the SSL context in the X509_STORE_CTX ex_data. 11360 [Eric A. Young] 11361 11362 *) Fixed a few memory leaks. 11363 [Eric A. Young] 11364 11365 *) Fixed various code and comment typos. 11366 [Eric A. Young] 11367 11368 *) A minor bug in ssl/s3_clnt.c where there would always be 4 0 11369 bytes sent in the client random. 11370 [Edward Bishop <ebishop@spyglass.com>] 11371 11372